@seismicisolation

@Seismicisolation
@Seismicisolation
Front cover photo:
Rising above the grassy plain in the center of Chichén Itzá is the giant Pyramid of Kukulcán
constructed 1100-1300A.D. Early Spanish visitors to the city referred to the pyramid as EI Castillo
(“he Castle). Chichén Itzá, the ancient city whose name means “in the mouth at the Itzáe’s Well,”
was, in its time of grandeur (between 800 and 1200 A.D.), the center of political, religious and
military power in Yucatán.
@Seismicisolation
O J.Andrezj Wrotniak@Seismicisolation
Innovations in Design with
Emphasis on Seismic, Wind, and
Environmental Loading; Quality
Control and Innovations in
MaterialdHot-Weather
Concreting
E ditor
V. M. Malhotra
@
international
SP-209
@Seismicisolation
@Seismicisolation
DISCUSSION of individual papers in this symposium may be submitted in
accordance with general requirements of the AC1 Publication Policy to AC1
headquarters at the address given below. Closing date for submission of
discussion is June 2003. All discussion approved by the Technical Activities
Committee along with closing remarks by the authors will be published in the
September/October 2003 issue of either AC1 Structural Journal or AC1 Materials
Journal depending on the subject emphasis of the individual paper.
The Institute is not responsible for the statements or opinions expressed in its
publications. Institute publications are not able to, nor intended to, supplant
individual training, responsibility, or judgment of the user, or the supplier, of the
information presented.
The papers in this volume have been reviewed under Institute publication
procedures by individuals expert in the subject areas of the papers.
Copyright O 2002
AMERICAN CONCRETE INSTITUTE
P.O. Box 9094
Farmington Hills, Michigan 48333-9094
All rights reserved, including rights of reproduction and use in any form or by any
means, including the making of copies by any photo process, or by any electronic
or mechanical device, printed or written or oral, or recording for sound or visual
reproduction or for use in any knowledge or retrieval system or device, unless
permission in writing is obtained from the copyright proprietors.
Printed in the United States of America
Editorial production: Bonnie L. Schmick
Library of Congress catalog card number: 2002113749

ISBN: 0-87031-089-5
@Seismicisolation
@Seismicisolation
In December 1991, the American Concrete Institute sponsored an intemational conference
on Evaluation and Rehabilitation of Concrete Structures and Innovations in Design. This
was in keeping with the mission of the Institute, and the focus on emerging concrete
technology in the Pacific Rim region, and to provide technical support for a growing number
in AC1 international chapters and members. The conference was held in Hong Kong,
December 2-6, 1991. More than 80 papers were in the conference proceedings published by
AC1 as SP-128.
As a follow-up to the very successful Hong Kong conference, the AC1 Board of Direction
approved a second international conference to be held in 1994. The theme of the conference,
held in Singapore, was High Performance Concrete. The proceedings of this conference
were published as AC1 SP-149.
The third international conference in this series was held in 1997 in Kuala Lumpur. The
theme of the conference was High-Performance Concrete, Design and Materials; and
Recent Advances in Concrete Technology. The proceedings of the conference were
published as AC1 SP-172.
In September 2000, the fourth international conference was held in Seoul, Korea. This
conference was titled Repair, Rehabilitation, and Maintenance of Concrete Structures, and
Innovations in Design and Construction. Conference proceedings were published as AC1
SP-193.
The fifth AC1 international conference was held in Cancun, Mexico in December 2002. The
theme of the conference was Innovation in Design with Emphasis on Seismic, Wind and
Environmental Loading; Quality Control and Innovation in MaterialsíHot Weather
Concreting. Sixty-nine papers were reviewed in accordance with AC1 review procedures,
and 48 accepted for publication as AC1 SP-209. Twenty-one were selected for presentation
and for publication as supplementary papers.
To all those authors whose submissions could not be included in the conference program, the
Institute and the International Conference Executive Steering Committee extends their
sincere appreciation for their interest and hard work. Special thanks are offered to members
of the AC1 Review Panel, who met in December 2001 to review the submissions. Without
their prompt review and constructive comments, it would not have been possible to publish
AC1 SP-209 for distribution at the conference. The cooperation of the authors in accepting
reviewers’ suggestions and revising their manuscripts accordingly, is greatly appreciated.
The help and assistance of William R. Tolley, AC1 Executive Vice President, and his staff in
performing the administrative duties associated with this conference are gratefully
acknowledged. Particular thanks are extended to Phyllis Erebor, SpeakerManuscript Liaison
Officer, for her dedication to the conference work.
Editor
V. M. Malhotra, Chairman
AC1 2002 Fifth International Conference on
Innovation in Design with Emphasis on Seismic,
Wind and EnvironmentalLoading; Quality Control
and Innovation in MateriaMHot Weather Concreting.
@Seismicisolation
@Seismicisolation
EFFECT OF CURING IN A HIGH TEMPERATUREENVIRONMENT ON
COMPRESSIVE STRENGTH OF CONCRETE INCORPORATING DIFFERENT
COMPLEMENTARY CEMENTITIOUS MATERIALS
by R. Rivera-Villarrealand J. M. Rivera-Torres ............................. . l
MARINE EXPOSURE OF HIGH-STRENGTH LIGHTWEIGHT CONCRETE

byG.C.Hoff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
PERFORMANCE OF HIGH-VOLUME FLY ASH CONCRETE IN HOT WEATHER

by P. K. Mehta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
EFFECT OF FIBER REINFORCEMENT ON RESIDUAL PROPERTIESOF HIGH-

STRENGTH CONCRETE UNDER ELEVATED TEMPERATURE
by T. Horiguchi, T. Takano, N. Saeki, and T. D. Lin ......................... .53
THE 1999 TEHUACÁN (Mw = 7.0) AND OAXACA (Mw = 7.5) MEXICAN
EARTHQUAKES: LESSONS LEARNED
by S. M. Alcocer, C. Reyes, D. Bitrán, O. López-Bátiz, R. Durán-Hernández, and
L.Flores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
SHEAR STUD PERFORMANCE OF LIGHTWEIGHT POLYSTYRENE

CONCRETE COMPOSITE SECTIONS
by Y. Lin, J. M. Ingham, and J. W. Butterworth . . . . . . . . . . . . . . . . . . . . . . . . . . . . .89
PARCEL 6 DEVELOPMENT, CENTRAL ARTERYITUNNEL PROJECT, BOSTON,

MASSACHUSE'ITS
by B. S. Moffat and E. P. White . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .lo9
BEHAVIOR OF FLARED-COLUMN BENTS UNDER SEISMIC LOADING

by H. M. Nada, D. H. Sanders, M. Saiidi, and S. El-Azazy . . . . . . . . . . . . . . . . . ..133
MECHANICAL, THERMAL, AND ACOUSTICAL PROPERTIES OF CONCRETE

CONTAININGVEGETABLE PARTICLES
by L. Arnaud and V. Cerezo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .151
SUSTAINABLEAND DURABLE REINFORCED CONCRETE CONSTRUCTION

by V. Corinaldesi, G. Moriconi, and F. Tittarelli ........................... .169
EXPERIMENTAL RESULTS AND VALIDATION OF THE DESIGN TECHNIQUE

USED FOR FRP STRENGTHENING OF RC BRIDGES
byD.I.Kachlakev . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
INNOVATIVE SEISMIC RETROFIT SCHEME FOR 215 FREMONT STREET

BUILDING, SAN FRANCISCO, CALIFORNIA
by N. R. Amin, D. Figueira, and V. Wan ................................. ,207
@Seismicisolation
@Seismicisolation
vi Contents
SEISMIC STRENGTHENING OFA NON-DUCTILE CONCRETE FRAME
BUILDING-ADAPAZARI, TURKEY
by C. L. Thompson, B. T. Knight, and L. R. Redlinger. ..................... ,231
QUANTITATIVEINFRARED THERMOGRAPHY FOR QUALITY CONTROL OF

CONCRETE STRUCTURES STRENGTHENED WITH FXP COMPOSITES
by M. A. Stames, N. J. Carino, and E. A. Kausel .......................... ,247
DESIGN OF PILE CAPS WITH STRUT-AND-TIEMODELS

by J. Dimig, E. Rogalla, and J. Duntemann ............................... ,269
STUDY OF THE INFLUENCE OF CEMENT TYPE, CEMENT CONTENT AND

CONCRETE COVERTHICKNESS ON THE RESISTANCE AND DURABILITY
OF CONCRETES SUBJECTED TO CHLORIDE ION AGING
by A. Joukoski, K. F. Portella, C. M. Garcia, A. Sales, O. Baron, and
J.F.dePaula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
SELF-CONSOLIDATING CONCRETE PILES CONFINED IN FRP TUBES

by H. El-Chabib, M. Nehdi, and M. H. El Naggar ......................... .297
ANALYSIS OF SHRINKAGE REDUCING ADMIXTURES ON DRYING

SHRINKAGE OF CONCRETE
by C. Videla and C. Aguilar ........................................... .319
SUBSTITUTION OF FLY ASH, SLAG, AND ADMIXTURES IN FDOT

CONCRETE MIXTURE PROPORTIONS
byY.JinandN.Yazdani .............................................. 347
BIAXIAL TENSILE BEHAVIOR OF THE REACTIVE POWDER CONCRETE

by O. Plé, E. Astudillo de la Véga, G. Bernier, and O. Bayard . . . . . . . . . . . . . . . ..369
BEHAVIOR OF CONCRETE LIQUID CONTAINING STRUCTURES SUBJECTED

TO SEISMIC LOADING
by M. R. Kianoush, W. K. Tso, and M. Hamidi ........................... .389
THREE MISUNDERSTANDINGS IN UNIAXIAL TENSION TEST

OF CONCRETE
by H. Akita, H. Koide, M. Tomon, and S. M. Han ......................... .405
EVALUATION OF RESIDUAL STRENGTH OF REINFORCED CONCRETE

MEMBERS SUBJECTED TO REVERSED CYCLIC TORSION
by N. Kawaguchi, H.Kuga, and T. Yoda ................................. .415
EXTREME CONCRETE
by K. W. Awad, H. Mazen, and R. Fakih ................................. .435
CREEP OF CONCRETE WITH SUBSTITUTION OF NORMAL AGGREGATE BY

RECYCLED CONCRETE AGGREGATE
by J. M. V. Gómez-Soberón ........................................... .461
@Seismicisolation
@Seismicisolation
Cancun Conference Proceedings vi¡
SHRINKAGE OF CONCRETE WITH REPLACEMENT OF AGGREGATEWITH
RECYCLED CONCRETE AGGREGATE
by J. M. V. Gómez-Soberón ........................................... .475
COMPARATIVESTUDY OF THE MECHANICAL PROPERTIES AND

PERFORMANCE, UNDER ACCELERATED AGING, OF FOUR REPAIR
SYSTEMS FOR CONCRETE HYDRAULIC STRUCTURES
by A. C. M. Kormann, K. F. Portella, P. N. Pereira, R. P. Santos,
and A. Joukoski .................................................... .497
STRUT AND TIE MODELS FOR STRENGTH AND DEFORMATION OF

RC MEMBERS
by S. G. Hong and S. G. Lee .......................................... .513
SEISMIC PERFORMANCE OF A RC BUILDING WITH COLUMNS

REHABILITATED WITH STEEL ANGLES AND STRAPS
by S. M. Alcocer and R. Durán-Hernández ............................. .S31
SHAKE TABLE STUDIES OF EFFECTS OF FOUNDATION KEXIBILITY ON

SEISMIC DEMAND IN SUBSTANDARDBRIDGE PIERS
by M. Saiidi, B. Gopalakrishnan, and R. Siddharthan .................... SS3
EXPERIMENTAL STUDY ON THE BEHAVIOR OF REINFORCED CONCRETE

BUILDINGSWITH PRECAST FLOOR SYSTEM
by O. López-Bátiz and H. Silva-Olivera ............................... .S7 1
INFLUENCE OF THE VARIATION OF CEMENT SOURCE ON HOT WEATHER

CONCRETING
by G. A. Arenes-Oliva, M. Bergin, and A. M. Zayed ..................... .593
EVALUATION OF IN-PLACE FLEXURAL STRENGTH CONCRETE USING

COMPRESSIVE AND SPLITTING TENSILE STRENGTHS FOR CAST AND
CORED SPECIMEN
by W. W. Taylor, S. D. Palmer, and R. W. Kukes ........................ .609
MONITORING PRESTRESS LOSSES IN HIGH PERFORMANCE CONCRETE

BEAMS WITH A BUILT-IN FIBER OPTIC MONITORING SYSTEM
by R. L. Idriss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .62 1
PERFORMANCE-BASED EVALUATION OF RECTANGULAR RC

STRUCTURES USING A FLEXIBLE DIAPHRAGM MODEL
by J. M. Barron and M. B. D. Hueste .................................. .641
BEHAVIOR OF RCS BEAM-COLUMN-SLAB SUBASSEMBLIES UNDER

EARTHQUAKE-TYPE LOADING
by X. Liang, G. J. Parra-Montesinos, and J. K. Wight .................... .661
@Seismicisolation
@Seismicisolation
viii Contents
PERFORMANCE OF SACRIFICIAL EXTERIOR SHEAR KEYS UNDER
SIMULATEDSEISMIC LOADING
by P. F. Silva, S. Megally, and F. Seible . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..681
THERMAL EFFECTS ON CHLORIDE DIFFUSION IN CONCRETE

by B. D. Miller, M. A. Miltenberger, and E. K. Attiogbe . . . . . . . . . . . . . . . . . . ,701
SPECTRAL RESPONSE OF MORTAR SURFACES DURING CURING

by L. Kindervater, R. Luna, and A. Belarbi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,717
THE PERFORMANCE OF FIBER REINFORCEMENT IN CONCRETES

EXPOSED AT HIGH TEMPERATURES
by G. Giaccio and R. Zerbino ........................................ ,733
REINFORCED CONCRETE BEAM-COLUMN DESIGN: AN ARTIFICIAL

NEURAL NETWORK APPROACH
byM.E.Haque ..................................................... 757
ASSESSMENT OF STRUCTURAL SYSTEMS EVALUATION METHODS BASED

ON LOCAL SEISMIC DEMANDS
by H. S. Lew and S. K. Kunnath ...................................... ,771
HYBRID FIBERS AND THE DEVELOPMENT OF CRACKING IN CONCRETE

by J. S. Lawler, D. Zampini, and S. P. Shah. ............................ .791
FERRO-SILICATE SLAG FROM ISF ZINC PRODUCTION AS A SAND

REPLACEMENT: A REVIEW
by R. Hooper, C. McGrath, C. Momson, and K. Lardner .................... 311
EFFECT OF REINFORCING BAR CONTAMINATIONON STEEL-CONCRETE

BOND DURING CONCRETE CONSTRUCTION
by L. H. Taber, A. Belarbi, and D. N. Richardson ......................... ..839
SEISMIC BEHAVIOR OF ECCENTRIC R/C BEAM-COLUMN-SLAB

CONNECTIONS UNDER SEQUENTIAL LOADING IN TWO
PRINCIPAL DIRECTIONS
by B. Burak and J. K.Wight ......................................... ..863
STRENGTHENING RECTANGULAR CONCRETE COLUMNS USING FRP:

A NEW TECHNIQUE
by T. Rizk, I. Mahfouz, and S. Sarkani ................................. ..881
THERMAL CYCLIC TESTING ON A CONCRETE SANDWICH PANEL SYSTEM

by R. K. Devalapura, J. M. Seng, M. E McBride, D. W. Winiarski, and
B. D.Johnson ..................................................... .905
@Seismicisolation
@Seismicisolation
SP 209-1
Effect of Curing in a High

Temperature Environment on
Compressive Strength of Concrete
Incorporating Different
CompIe mentary Cementitio us
Materials
by R. Rivera-Villarreal and J. M. Rivera-Torres
Svnopsis: This paper provides results of different types of curing in hot weather
environment on the compressive strength of concrete made with portland cement
and complementary cementitious materials (CCM) such as natural pozzolans, fly
ash (FA), granulated blast furnace slag (GBFS), and silica fume (SF). In all con-
crete mixtures, a superplasticizing admixture (SPA) was used.
Nine series of concrete mixture were made. In seven of them (1,2, 3,4,5,
6 and 9) the normal portland cement (NPC) content was 200 kg/m3 and in two of
them (7 and 8) the same amount of cement was used but it was a portland-natural
pozzolan cement (PNPC). The CCM varies from 9.9 to 60.6% of the total cemen-
titious material. The W/C in all series was 0.70 when using NPC or PNPC. The
W/C+CM varied from 0.28 to 0.63. In all series the same amount of 1260 kg/m3 of
coarse aggregate was used.
Five different ways of curing were used. One was the initial and final
ASTM curing at 23°C up till the age of testing as reference, and four different
ways of curing in hot weather environment at 37°C for the first 24h were used.
These final curings were: A) ASTM; F) three day and G) seven-day water spray for
15 minutes every 2h; and (E) covered by two layers of membrane.
Adequate compressive strength development (CSD) can be obtained using
CCM but very good curing is necessary. Generally, by casting specimens at 37°C
and put them under ASTM curing next day at 2322°C (A), the strength at 28 days
was lowered by about 8% and at six months by about 8% lower than these casting
at 23°C.
Membrane curing was less effective at later ages mainly when fly ash was
used. There exist an optimum amount of fly ash to obtain maximum compressive
strength at later ages.
Kevwords: compressive strength; final curing; fly ash;

high temperature; membrane; portland cement;
pozzolans; silica fume; slag; superplasticizer
~
@Seismicisolation
@Seismicisolation
1
2 Rivera-Villarreal and Rivera-Torres
R. Rivera-Villarreal, Professor Emeritus in Civil Engineering at the
Universidad Autónoma de Nuevo León (UANL), is AC1 and RILEM Fellow
and AC1 Honorary Member, member of AC1 Committees 120, 236, E 701, E
802, E 803. Researcher Level II National Program of Researchers. President of
the Education Activities Committee and Advisor of the FIC-UANL AC1
Student Chapter of the North East México Chapter. Chairman of the RILEM
158 AHC. Chief of Concrete Technology Department at the Civil Engineering
institute, Titular Professor of Concrete Technology and Laboratory at the
School of Civil Engineering (FIC), UANL, Monteney.
J. Rivera-Torres: Civil Engineer, Research Assistant, and Concrete
Technology Laboratory Professor Assistant at the School of Civil Engineering
at UANL, Monterrey. Accredited as examiner for AC1 Certification of Field
Technicians on Reinforced Concrete Structures.
INTRODUCTION
in September 2000, the Strategic Development Council of the U. S.

concrete industry launched an interesting project for the hture of the concrete
industry; a Unified Vision for concrete technology and its social impact 30
years from now; it was called Vision 2030. Predicting that by the year 2030 the
industry will make processing improvements throughout the life cycle of
concrete, and that as a logic consequence of this, concrete will become the most
efficient and cost-effective material for construction.
This Vision expects that aggregates quarries, cement manufacturing,

ready-mixed and concrete product plants will be viewed as assets to their
neighboring communities. And one of the main strides is to use a variety of by-
products from other industries as well as recycled concrete as a constituent
materials in new concrete production [11.
Environmental issues will have a leading role in the sustainable

development of the cement and concrete industry in the 21" century. increased
emission of greenhouse gases to the atmosphere is no longer environmentally
and socially acceptable for overall sustainable development. The primary
greenhouse gas emissions are carbon dioxide (COZ) emissions. As the
manufacture of portland cement contributes significantly to the CO2 emissions,
it is imperative either to develop new environmentally fiiendly cement
manufacturing technologies, which should be economically acceptable; or to
find substitute materials to replace a major part of the portland cement used in
the concrete industry. The use of large volumes of fly ash and other CCM in
the construction industry contribute in reducing these emissions [2]. These
materials are cheaper than portland cement; therefore, the benefits of using
these materials are environmental as well as economical.
@Seismicisolation
@Seismicisolation
Cancun Conference Proceedings 3
Water permeability of concrete cover is related to durability and is
usually associated with good concrete mixture design and workmanship.
Microstructure of NPC is remarkably influenced when it is blended with a very
fine ground CCM. The CCM densifi the microstructure of hydrated cements,
decreasing porosity that influences the overall properties of the resulting
material. Good concrete practice should include adequate compaction and
suitable curing. Early and long moist curing has been identified as an
important requirement for durable concrete [3]. However, such curing
requirement is often ignored in the actual construction process mainly when
producing high strength concrete (HSC). Various references to the importance
of concrete curing in hot weather environments can be found in [4,5,6, & 71.
The appropriate use of a suitable CCM and a HRWRA is expected to

produce workable, strong and durable concrete. The pozzolanic CCM will
react with calcium hydroxide that is the weakest compound formed and is
liberated after the hydration of cement, which ends up in C-S-H hardened
compounds and in a refined micro pore structure in concrete. The combined
effect of pozzolanic reaction and the very fine particle size of CCM was
expected to improve the strength as well as the water permeability of concrete.
Capillary voids and micro cracks, which are generally connected or

interlinked, have a great influence on concrete permeability. The amount and
size as well as connectivity of capillary voids depend on the W/C and its
hydration [81.
The permeability of cement paste has also been related to concrete

durability. A great deal of work has been done measuring and predicting
permeability of cement paste. One aspect between permeability and pore
structure of hardened cement paste was presented by Nyame, et al [9].
Many researches are going on using ternary cementitious materials to

get sustainability of concrete materials with fly ash and slag as partial cement
replacement materials, looking for good strength development in high strength
and non-permeable concrete. One of these studies made with ash of Timber
Industrial (TI) included concrete subjected to different curing conditions [1O],
they obtained good results for concrete grade 80 MPa in the pre-cast concrete
industries. Autoclaved curing enhanced the potential of developing early
strength.
Several techniques have been applied to study the pore structure of

cement paste, such as the measurements of surface area and porosity by using
mercury intrusion porosimeûy (MIP), Brunaver-Emmett-Teller (BET) analysik,
and surface electron microscopy (SEM).
@Seismicisolation
@Seismicisolation
This study is an attempt to explore the possibility of producing durable
and sustainable concrete in hot weather with minimum moist curing, and
minimum amount of portland cement.
Materials
Two types of cement were used according the Mexican Standard NMX
C414-1999; CPO 30R similar to ASTM C 150 type I portland cement and CPP
30R similar to ASTM C 595-98 blended hydraulic cement type P, but with a
higher 28-day-strength-minimum of 30 MPa. The CCM used were: natural
pozzolan, FA, GBFS, and condensed SF. The chemical analysis of all
cementitious materials used is shown in Table 1 and physical properties in
Table 2. The FA used meets ASTM class F requirements, except for the
fineness, which is somewhat coarser, passing 45 pn 58% instead of 66% by
ASTM.
The aggregates used were crushed limestone from Monterrey, N. L.,

Mexico area. The fine aggregate had a dry relative density of 2.68 and SSD
absorption of 1.5%. The coarse aggregate had a dry relative density of 2.66 and
SSD absorption of 0.7%and a 20.0 mm nominal maximum size.
Since in this study the concrete consistency is a factor, it is of great

importance that the values measured through the consistency test should not be
affected by aggregate grading variations. To avoid this problem and to
maintain a uniform grading for each mixture, the fine and coarse aggregate
were separated into different size fractions, which were then recombined in the
required amount for each batch to provide the selected specific grading. The
mean fine and coarse aggregate grading is shown in Table 3.
A sulfonated formaldehyde naphthalene concentrate superplasticizer

admixture (SPA) with 1.12 density, was used in all nine series.
Mixture proportions
Nine concrete series (1 to 9) were made using different amounts of

CCM as shown in Table 4. In seven of them (1 to 6, and 9) 200 kg/m3 of NPC
was used; and in Series 7 and 8 the same amount of cement was used, but it
was PNPC. In all nine series the same amount of water was used to have W/C
= 0.70. In Table 4 the amount and percentages of each CCM used are shown.
In reference Series 1 only NPC was used; in Series 7 only NPPC (22 kg/m3)
was used. In all other seven mixtures using different amounts of CCM, the
same amount of SF was used. This amount corresponds to 9.9% of NPC 3 r
NPPC.
One of the reasons why all the mixtures have the same amount of water
(140 kg/m3) and 200 kg/m3 of NPC or NPPC with a W/C= 0.70, was to have
the same amount of calcium hydroxide produced during cement hydration. The
@Seismicisolation
@Seismicisolation
W/c+CC varies from 0.70 to 0.28 and the yield of cementitious materials from
200 to 507 kg/m3.
In all series, the amount of coarse aggregate was 1260 kg/m3. To

maintain the same amount of cement and water when using CCM, it was
necessary to adjust the absolute volume of CCM deducting it from the
corresponding amount of fine aggregate which in the reference series (without
CCM or SPA) was 891 kg/m3.
SPA was used in all nine series, the total amount added was adjusted to
give a final slump of 220 f 20 mm. The mixture proportions are shown in
Table 5.
Mixing and Casting
An Elrich concrete mixing machine 45L capacity was used, the drum
rotating to 30 rpm and the paddles in countercurrent 450 rpm. The mixing
procedure for reference concrete series 1 was as follows: To ensure the correct
final proportions in the test batch, just prior to mixing the first batch, the mixer
was buttered by mixing a batch proportioned to simulate closely the test batch.
The mortar adhering to the mixer after discharging is intended to compensate
for loss of mortar from the test batch. The dry aggregates and a part of the
mixing water were put into the drum and the machine was started for a half of
minute to saturate the aggregate; then, without stopping the machine the
cementitious materials (200 kg/m3 of portland cement) were added, including
the rest of the water needed to have 0.70 W/C. When all the materials were
into the drum,the mixing continued for an additional minute. The machine was
stopped for one minute, after that resting period, the mixing continued for one
more minute. The machine was stopped and immediately the slump test was
performed to measure the initial slump. After that, SPA was added to attain a
slump of 22W20 mm, mixing for one more minutes.
The mixing procedure for all other series was similar except that when
the remaining water was added, the fourth part of the SPA admixture was
included to facilitate the mixing. After the resting period of one minute, the
remaining SPA was added without stopping the machine and the mixing
continued for one more minute.
Initial slump for îhe reference mixture was 40-50 mm, and the final
slump was 220 f 20 mm after adding SPA. Except for the reference mixture
made at 23i2"C, all mixtures were made at 37°C & 1. Specimens were casted
in molds of 100 mm in diameter and 200 mm in height, three for each
compressive strength test at the specified ages of 3, 14, 28 days, 3, 6 months
and one year. However, only age 6 months results will be provided in this
paper. Additional specimens were made to measure porosity and permeability
up to 6 months, but the results are not included in this paper.
@Seismicisolation
@Seismicisolation
Specimens were stored under five different curing conditions up to the
age of testing. Daily information on ambient temperature, humidity, and
evaporation was recorded; and the average high and minimum monthly values
were determined as shown on Fig. 1.
Fresh Concrete Properties
All concrete series were casted at two different nominal concrete

temperatures, 23°C and 37"C, as shown in Table 6 , along with the determined
air content, laboratory temperature, concrete temperature, and slump.
The dosage of SPA was adjusted in ail the mixtures to obtain a final
slump of 220 f 20 111111, as shown in Tables 5 and 7.
Fresh Concrete
In Table 7 we can see that when the concrete temperature increases, so

does the dosage of SPA increases as percentage of the Cementitious Material
used. A very high increment is registered when no CCM were used, such
mixtures are reference series 1 and 7. A higher increment is observed for
reference Series 7 in which PNCP was used, this cement has natural pozzolans
intergrinded. Portland Cement is the coarser particle size of cementitious
materials, but to compensate the absolute volume when CCM is added, as
already mentioned in Mixture Proportions, CCM added is finer than the finer
aggregate that is substituting. As it can be seen in Table 5 where the % of SPA
is about the same to have the same final slump. There is a possibility that the
efficiency of a naphthalene based SPA is increased considerably when very fine
cementitious materials are used instead of very fine aggregates. Albeit, more
research is needed to confirm this possibility.
When the laboratory temperature was maintained at 23" C f 2, the

concrete temperature rose up from 30" to 33.5"C; an average increment of 7.4"
C; when the laboratory temperature was maintained at 37.5"C f 0.5, the
concrete temperature rose up from 38" to 40.5"C; an average increment of 2°C.
Curing
Five different ways of curing were used, it is important to define that we

call initial curing those made the first 24 h and final curing after 24 h.
For the reference concrete (Series i), ASTM C31M-96 curing (R) was
used for initial and final curing.
For the rest of the concrete the initial curing was under a temperature of
37°C f 1 and humidity of 60% rt 10, which were the same temperature and
@Seismicisolation
@Seismicisolation
humidity used when casting the concrete specimens, simulating a hot weather
environment.
TWEOF INITIAL
FINAL CURING
CURING CURING
R Dry 23°C Std ASTM curing room 23°C
A Dry 37°C Std ASTM curing 100% RH 23°C
F Dry 37°C 3 day water spray every 2h Ambient under shade
G Dry 37°C 7 day water spray every 2h Ambient under shade
E Drv 37°C Membrane Ambient under shade
Curing A Standard ASTM (23 and 100% RH after 24 h)

Curing F 3 days spraying specimens with water 15 minutes every 2 h in
ambient temperature under shade and no more water in same
ambient till the time of test.
Curing G 7 days spraying specimens with water 15 min every 2 h in ambient
temperature under shade till the time of test.
CuringE 2 layers of a membrane compound applied to specimens (200
mLlm2 per layer) and then ambient temperature under shade till the
time of testing. Carefulness was in applying the first layer of
membrane to the surface when concrete be saturated but surface
dry. To be sure not to lose any amount of water, prior to applying
the membrane the specimens were soaked in water and surface
dried.
Environmental Conditions
The environmental conditions of daily minimum, mean and highest

temperatures, the daily minimum, mean and highest relative humidity (RH) and
the daily minimum, mean and highest evaporation were recorded on site of
curing. In Fig. I average monthly values are shown.
During the six months of environmental curing, May through November

the highest temperature varies from 28.5"C to 37.5OC. The minimum relative
humidity varies from 32.8% to 53.8% and the maximum evaporation varies
from 0.24 L/m2/h to 0.46 Llm2/h.
Compressive Strength
Compressive strength was determined according to ASTM C-39-96

with the following modifications for those specimens under curing F, G and E.
Prior to testing of specimens they were soaked in water for two hours trying to
have saturated all the specimens at time of test. in the case where membrane
was used, it was removed by scraping it prior to putting the specimens in water.
@Seismicisolation
@Seismicisolation
In Table 7 compressive strength results for all series are shown, as well
as type of curing.
RESULTS AND DISCUSSION

In all series, the minimum compressive strength was obtained 'when
using the type F curing which was an intermittent spray with water 15 min.
every two hours curing for three days, and then lefî under shadow up to the age
of testing (Fig. 3 thru 11).
In all series, Fig. 3 to 11, the curing type G that was an intermittent
spray with water 15 min. curing every two hours for seven days, the
compressive strength increased very little compared to type F curing for higher
W/CtCM; for lower W/C+CM the difference was about 10%.
In all series, Fig. 3 to 11, the compressive strength with Type E

membrane curing was similar to type G curing in that little difference in
strength was observed. The membrane curing type E prevented any loss
of humidity after applying the first membrane layer.
In series 4, 5, and 6, fly ash was used and the compressive strength data
are shown in Fig. 6, 7, and 8. There was a high difference in the way of curing
in those having few water as curing F, G and E, the compressive strength is
lower than those having more water available as type curing R and A at 28 days
and strength increase of about 15% is observed and for later ages as 90 or 180
days, the strength increase is higher about 35%. This means that fly ash
concrete needs good water curing for a long time to increase strength
development.
The compressive strength development of the nine series of concrete

mixtures for each five types of curing is shown in Figures 12, 13, 14, 15 and
16.
In all series, the curing type R, in which the standard ASTM curing
were used all the time attained maximum compressive strength at ages 28 days
and higher, except Series 6 in which was used the maximum CCM of 307
kg/m3 as shown in Fig. 8. In this case, casting was at 37°C and being for 24h at
this temperature following with curing soaked in water curing at 23OC (Type A
curing).
Series 3 having less than 50% of cementitious materials than Series 5

has higher early strength and about the same late strength (Fig. 17 to 21). The
difference is that Series 3 has 27.7% GBFS and series 5 has 53.0% fly ash.
Both series have same amount of portland cement SF and water. The
effectiveness of GBFS was better than fly ash used.
@Seismicisolation
@Seismicisolation
Series 4 and 5 (see Fig. 6 and 7) have the same amount of portland
cement, SF, water and very little amount of SPA. The main difference is that
the amount of fly ash in Series 4 had 200 kg/m2 and Series 5 had 250 kg/cm3, a
difference of 50 kg/cm3 of more fly ash, but the compressive strength of Series
4 has about the same strength. This means that there exists an optimum amount
of fly ash for maximum strength.
In series 2 to 6 the same amount of PC (200 kg/m3), the same amount of

water (140 kg/m3), the same amount of SF (22 kg/m3) and same slump (220
f 20 mm) were maintained. The superplasticizer dosage varied with ambient
temperature from 1.22 to 1.35% of solids of total cementitious materials for
23°C and 1.30 to for 37OC. Series 2 and 6 have the same amount of PC
and SF and same initial hot curing (37 OC) but different W/C+CM, Series 2,
0.63 abd Series 6, 0.28, the last one has quaternary cementitious materials in
which we added furthermore FA and GBFS. For the same curing A, Series 2
(Fig. 4) the compressive strength at 6 months was 44.5 MPa and in Series 6
(Fig. 8) the compressive strength was 94.8 MPa, an increase of 213%.
The reference Series 1 (Fig. 3) with portland cement CPC 30R similar to
ASTM type 1 and Series 7 (Fig. 9) with PNPC without any CCM but with SPA
for reference R curing. The Series 7 with PNPC give more strength of 11.1% at
28 days and 22.1% at six months.
The Series 2 (Fig. 4) with portland cement and Series 8 (Fig. 10) with
same amount of PNPC and same amount of SF and SPA have no difference for
reference curing R in compressive strength at 28 days, but at six months Series
8 is higher 5.0%.
The Series 3 and Series 9 have same amount of portland cement and
same amount of SF and using SPA. The only difference was that series 3 has
85 kg/m3 of GBFS, and series 9 has 85 kg/m3 of NP at 28 days series 9 with
reference curing R had a 7.0% higher compressive strength and at six months
10.0%.
CONCLUSIONS
In hot weather environments usually the specimens are made and left at
the site for the first 24 h. After that, they are cured under >95% humidity and
23 f 2°C temperature, which is similar to curing type A. In this type of curing
against type R curing, which is ASTM curing (casted at 23OC), under the
conditions established in this study and for any amount of cementitious
materials, the compressive strength at 28 days was lower for a mean value of
10.0% at later ages about the same 12.0%, except for Series 3 and 6, where
GBFS was used, in Series 3 it was of only about 5.0% lower at 28 days and
13.0% at later ages, but in Series 6 , besides using the same amount of GBFS,
@Seismicisolation
@Seismicisolation
but a great amount of total cementitious material as about 60.0% by volume,
which lowers W/C+CM to 0.28 at 28 days, practically is the same for both
types of curing. For later ages as 180 days, the strength level was higher about
10.0% for curing A.
Two layers of good applied membrane were not as good as continuous

water curing mainly when fly ash was used as in series 4,5, and 6 when a 20%
less compressive strength at 28 days and about 45.0% at six months of concrete
casting at 37°C were obtained.
Intermittent water at 15.0 min for every 2h for 7 days is not enough to
gain strength development as continued water curing. It will be similar to two
layers of membrane.
Some GBFS, as used in this study, was more effective in developing

compressive strength. Series 3 had an additional 85 kg/m3 of GBFS and Series
5 had an additional 250 kg/m3 of fly ash and other materials the same, even the
higher amount of fly ash the compressive strength is around the same for any
age and type of curing, with less amount of water as curing F, G, and E. For
curing with water the Series 5 wiîh fly ash have 12.0% higher at 28 days and
about 27.0% at six months.
044-V
ACKNOWLEDGEMENT
The authors would like to thank Cemex for supplying the portland
cement and natural pozzolans used and Acon, S. A. de C.V. for supplying the
admixtures.
The authors also want to thank students Robert Montalvo, Sergio
Narváez, David Reyes and Iliana Garza for word processing and preparation of
illustrations.
REFERENCES
Vision 2030: A Vision for the U.S. Concrete Industry. Concrete

International, Farmington Hills, Mi, Vol. 23 No. 3 March 2001 pp. 25-34.
Malhotra V. M. “Making Concrete “Greener” with fly ash” Concrete
International Vol. 21, No. 5 May 1999, Farmington Hills, Mi pp. 61-66.
Mansur, M.A., Wee, T.H. & Chin, M.S. “Some Engineering Properties.of
Locally Produced High Strength Concrete.” Proc. 19th Conference on OÜr
World in Concrete & Structures. Singapore 1994, pp. 97- 106.
Mehta, P.K., Monteiro, P., “Concrete, Structure Properties and Materials”,
Prentice Hall 2”dEdition, Englewood Cliffs, NJ, 1993.
@Seismicisolation
@Seismicisolation
5. Powers, T.C., “A Discussion of Cement Hydration in Relation to the
Curing of Concrete”, Chapter 3, Proc. Highway Research Board. 27,
Washington, D.C., 1947,pp. 178-188.
6. AC1 305R-91, “Hot Weather Concreting”, AC1 Manual of Concrete
Practice, Part 2, pp. 1-20.
7. Rivera, R., “Effect of High Temperature Curing on the Compressive
Strength of Concrete Incorporating Large Volumes of Fly Ash.” V. M.
Malhotra, editor. Seventh CanmetIACI International Conference.
8. Shah, S.P. & Wang, K.“Microstructure, Microcracking, Permeability and
Mix Design Criteria of Concrete.” Proc. Fifth Intemational Conference on
Structural Failure Durability and Retrofitting (ICSF-97) 1997, pp 260-272.
9. Nyame, B.K & Illstone, J.M. “Relationships Between Permeability and
Pore structure of Hardened Cement Paste.” Magazine of Concrete
Research. Vol. 33, No. 116, Sept. 1981, pp 139-146.
10. Lee, Y.L., et al. “Strength Development and Water Permeability of High
Strength TIA Concrete.” Modem Concrete Materials - International
Congress Creating with Concrete. 1999, Dundee, Scotland, pp. 175-18 1.
List of Tables
1. Chemical analysis of all cementitious materials

2. Physical analysis of all cementitious materiais
3. Aggregate grading
4. Amount of complementary cementitious materials in each concrete series
5. Mixture proportions for all 9 series
6. Fresh concrete properties at nominal temperatures of 23°C and 37°C
7. Amount of SPA for all series at 23°C and 37°C in % of cementitious
materials
8. Compressive strength for all series and type of curing
List of Figures
1. Average monthly daily values of minimum, mean and highest evaporation

on site of curing from May to October 200 1.
2. Average monthly daily values of minimum, mean and highest temperatures
and relative humidity on site of curing from May to October 200 1.
3. CSD for series 1 with five type of curing. Reference concrete with
portland cement without any CCM.
4. CSD for series 2 with the five type of curing using portland cement and SF
and SPA.
5. CSD for series 3 with the five type of curing using portland cement GBFS
and SF and SPA.
6. CSD for series 4 with the five type of curing.
@Seismicisolation
@Seismicisolation
12. Compressive strength development (CSD) for the nine series using ASTM
standard curing "R" concrete made at 23°C.
13. CSD for the nine series for curing type A concrete made at 35°C
maintaining this temperature for 24h.
14. CSD for the nine series for curing type F 24h at 35"C, next three days
water spray every 24h and after under shade till time of testing.
15. CSD for the nine series for curing type G 24h at 35"C, next seven days
under humid burlap shade to time of testing.
16. CSD for the nine series for curing type E 24h at 37"C, after two membrane
layers and then under shade to time of testing.
17. Series 1 to 6 showing the content of CCM except PNPC, for curing R.
18. Series 1 to 6 showing the content of CCM except PNPC, for curing A.
19. Series 1 to 6 showing the content of CCM except PNPC, for curing F.
20. Series 1 to 6 showing the content of CCM except PNPC, for curing G.
21. Series 1 to 6 showing the content of CCM except PNPC, for curing E.
Table 1. Chemical analysis of cementitious materials
@Seismicisolation
@Seismicisolation
Table 2. Physical properties of cementitious materials
I
I ~ ~~~
Type Natural
Properties
L UPP 30R Pozzolan
Fineness
Passing 45pm 89.60 94.58 94.3
3.14 2.95 2.36
Vicat test
initiai set 94 95
Final set 132 202
Gilimore test
Initial set 105 106
143 179
of 50 mm cubes MPa
3d 24.3 23.7
7d 27.9 31.0
28 d 32.1 40.0
I
Autoclave Exmnsion ?4 .O525 -0.018
1
Strength activity index
with CPC 30R %
75.3
78dd 80.0
Table 3. Aggregate grading
Coarse, ‘YO passing

Sieve size
Maximum size 19 mm (3/4”)
1 ‘A’’ ---
%”
%,,
3.4
0.7
@Seismicisolation
@Seismicisolation
Table 4. Amount of complementary cementitious materials in each
concrete series
FA GBFS I NP I SF I TOTAL
CCM
31%
I
200 0.7
222 0.63
307 0.46
o(0
eries had a W/C= 0.70
* Total of cementitious materials.
All mixtures were made using 200 kg/m3 of Portland cement except
series 7 and 8, which were made using pozzolanic cement with aprox. 18%
naturai pozzolan.
Table 5. Mixture proportions for all 9 series
3.47 1.57
3.99 1.30
5.68 1.35
200 0.70 1260 2.68 1.34

222 0.7 0.63 1260 861 3.48 1.57
307 0.7 0.46 1260 784 4.48 1.46
* % of total cementitious material.
@Seismicisolation
@Seismicisolation
Table 6. Fresh concrete properties at casting nominal temperature of 23OC
and 37OC
~~~~ ~~
23OC 37OC
Series Air Lab Concrete Air Lab Concrete Slump
Content temp. temp. content Temp. temp
Table 7. Amount of SPA for all series at 23OC and 37OC in YOof ail
cementitious materiais
@Seismicisolation
@Seismicisolation
Table 8. Compressive strength for all series and type of curing
3CM :CM Lab

cí pressi Stren W 1
Series ig/m' % 'emp 180
3 7 28 90
OC -
1 O O 25.0 21.9 24.2 30.6 33.0 35.7
2 22 9.9 25.0 25.2 40.4 45.4 48.3 52.1
3 107 34.9 24.5 45.5 51.1 56.2 60.3 64.2
4 222 52.6 25.0 32.0 54.3 66.7 77.6 83.1
5 272 57.6 24.5 35.7 50.8 64.0 72.8 80.4
6 307 60.6 25.0 45.8 59.7 74.0 81.7 86.5
7 O O 24.5 19.9 24.0 34.0 39.8 43.6
8 22 9.9 24.5 25.0 36.9 45.9 50.0 54.8
9 -
107 34.9
- 23.5
- 30.6 44.9 60.1 -
65.7 69.8
-
1 O O 37.5 18.9 22.6 26.3 29.4 32.3
2 22 9.9 37 30.4 34.7 40.6 41.2 44.5
3 107 34.9 37 40.0 50.0 53.7 58.1 56.7
4 222 52.6 37 34.0 50.6 60.8 73.4 76.3
5 272 57.6 36.5 33.1 47.5 60.2 65.6 71.8
6 307 60.6 37 48.7 61.0 75.6 89.3 94.8
7 O O 37 19.4 23.2 32.6 38.2 40.0
8 22 9.9 36 24.6 35.7 45.0 48.2 50.5
9 107
- 34.9
- -
36.5 30.0 41.1 52.7 -
60.8 64.1
-
1 O O 37.5 18.0 19.1 21.3 28.3 29.5
2 22 9.9 37 27.6 36.3 35.0 39.3 40.3
3 107 34.9 37 39.8 43.1 45.6 47.5 46.5
4 222 52.6 37 34.3 49.7 52.2 52.1 54.3
5 F 272 57.6 36.5 33.1 43.3 52.8 52.1 52.4
6 307 60.6 37 47.9 51.7 64.3 61.1 57.0
7 O O 37 17.7 19.7 20.0 21.9 25.1
8 22 9.9 36 21.6 29.3 33.7 33.6 33.9
9 - 107 34.9 -
36.5 25.9 35.1 41.5 -
39.3 -39.4
1 O O 37.5 20.2 21.1 23.5 28.3 28.3
2 22 9.9 37 27.9 34.8 37.7 37.5 39.6
3 107 34.9 37 38.6 45.0 49.7 47.8 50.2
4 222 52.6 37 34.2 47.1 55.5 56.8 56.4
5 G 272 57.6 36.5 34.2 42.8 54.4 55.1 56.3
6 307 60.6 37 46.9 57.3 69.6 67.1 56.8
7 O O 37 17.8 22.4 25.0 28.0 29.6
8 22 9.9 36 21.3 29.4 37.6 37.1 37.5
9 - 107 34.9 -36.5 25.5 37.0 42.3 48.7 -48.2
@Seismicisolation
@Seismicisolation
Table 8. Compressive strength for all series and type of curing (cont.)
- -
Lab CE $ressi ;Streu !AY! L
Series remp
OC 3 7 28 90 180
- - -
1 37.5 17.7 20.1 21.8 28.3 28.1
2 37 21.7 38.4 39.5 38.8 41.8
3 37 44.7 48.5 51.6 51.8 52.4
4 37 41.5 50.9 54.7 57.8 56.9
5 36.5 39.9 44.4 49.8 53.5 53.8
6 37 56.4 64.7 69.6 71.5 67.9
7 37 18.4 27.0 28.6 27.4 27.1
8 36 29.7 34.9 40.3 44.9 43.6
-
9 36.5 31.7 40.0 51.6 51.7 -
53.3
a ............... ........................................
4 90
+L
~
>
Z 80 ....................................................................................................................
Y RH.moan
@Seismicisolation
@Seismicisolation
0.5 .
0.35
LT
U
Mag Jun Jid Agu Sept Oct NW

MONTHS
FIG.2. MûNTHLYAvERRiiES OF DAILY OBSEWATIONS OF NAPORATION AT
UNNERSIDRDRUTONOMA DE NUNO LEON CAMPUS YEAR 2001
O 20 40 60 80 100 120 140 160 180 200

AGE [DAYS1
FIG. 3. CONIPRESSNE STRENGTH DEVELOPMENT FOR SERIES 1WITH THE
FIVE TYPES OF EURING. REFERENCE CONCRETEWITH PORTLAND CEMENT.
@Seismicisolation
@Seismicisolation
O 20 40 60 80 100 120 140 160 180 200

AGE [DAYS)
FIG. 4. STRENGTH DEMLQPMENT FOR SERIES 2 WITH THE FNE W E S
OF CURING USING PORTLAND CEMENT AND SF.
@Seismicisolation
@Seismicisolation
O 20 40 60 80 100 120 140 160 180 200

&GE I M Y S l
FIG. 6. STRENGTH DMLOPMENT FOR SERIES 4 W T H THE FNE TYPES
OF CURING, USING TERNAWCEMENTITIOUS MATERIALS PORTLAND
CEMENT 47.4%, FLY ASH 47.4% AND SF 5.2%.
O 20 40 60 80 100 120 140 160 180 200

AGE [DAYS]
FIG. 7. STRENGTH DMLOPMENT FOR SERIES 5 WITH THE F M TYPES
OF CURING, USING TERNARY CEMENTITIOUS MATERIALS: PORTLAND
CEMENT 57.7%,FLY ASH 53.0% AND SF 4.7%.
@Seismicisolation
@Seismicisolation
O 20 40 60 80 100 120 140 160 180 200

AGE [DAYS1
FIG. 8. STRENGTH DEVELOPMENT FQR SERIES 6 WITH THE F I M TYPES
OF CURING, USING QUATERNAWCEMENTITIOUSMATERIALS
PORTLAND CEMENT 39.5%,FLY ASH 39.446, GBFS 16.8%,AND SF 4.3%.
O 20 40 60 80 100 120 140 160 180 10

AGE íOñYS1
FIG. 9. STRENGTH DMLOPMENl FOR SERIES 7 W T H THE FIVE MPES
OF CURING, W N C E CONCRETE WITH PNPC PRODUCED IN THE
CEMENT FACTOWAND SPA, NO CCM wns USED.
@Seismicisolation
@Seismicisolation
O 20 40 60 80 100 120 140 160 180 200

AGE IDAYSI
FIG. I O . STRENGTH DEVELOPMENTFOR SERIES 8 WITH THE FNE TYPES
OF CURING. SRME AS SENES 7 (FIG. 9) BUT INCLUDING SF 9.9%.
"-O 20 40 60 80 100 120 140 160 180 ZOO

AGE IDAYSI
FIG. 11. STRENGTH DMLOPMENT FOR SERIES 9 WITH THE FRIE TYPES
OF CURING USING TERNARY CEMENTITIOUS MATERIALS PORTLAND
CEMENT 65.1%,NATURIIL POZZOLIIN 27.796, AND SF 7.2%.
@Seismicisolation
@Seismicisolation
O 20 40 60 80 100 120 140 160 180 200

AGE (DAYS)
FIG.12. CONIPRESSNE STRENGTH DMLOPMENT FOR THE NINE SERIES
USING ASTM STANDARD CURING "R". CONCRETE MADE AT 23OC.
1
.................... L .........:......... ..........
.......... ....... ......... TYPE-A- ~URING~
............................. ........................
.......... I
j
........ ...........,..........i.
......... ....
a
____,__. :
............................
w '
.. - k
f 50.-
z
'i 40. ......................... ,,-i.- -
."".'.."......i..-...-i.-
-,;-..-..;-.-..-
......"a ....... .. ..-..o..---......
Li-.
8Y .......
30. ........i'. _-. ......
E
F! 201 .................... ....... ...... ......
U
i
.............................................................. :
+-1 4-2 &3
-4 -5 4 6
i --0--7~8 +3
O 20 40 60 80 100 120 140 160 180 200

AGE (DAYS1
FIG. 13. COMPRESSNE STRENGTH DMLOPYENT FOR THE NINE SERIES FOR
CURING PIPE "Ji". CONCRETE MADE AT 37°C AND MANTAINED THIS
TEMPERATURE FOR 24h.
@Seismicisolation
@Seismicisolation
24 Rivera-Villa rreal and Rivera-Torres
O 20 40 60 80 100 120 140 160 180 : i0

AGE (WAS1
FIG. 14. COMPRESSNE STRENGTHDMLOPMENT FOR THE NINE SERIES
FOR CURING TYPE "F". 24 h BT 37"ÇNEXT 3 DRIS WATER SRAY M R Y 2 h
RND RFTERUNDERSHRM TO TIME OF TESTING.
O 20 40 60 80 100 120 140 160 180 200

AGE (DAY SI
FIG. 15. COMPRESSNE STRENGTH FOR THE NINE SERIES FOR CURING TYPE
"G". 24 h AT 37T,NEXT i DAYS WñTER SPRBY M R Y 2 h AND IFTER UNDER
SHADE TO TIME OF TESTING.
@Seismicisolation
@Seismicisolation
O 20 40 60 80 103 120 140 160 180 200

AGE IDAYSI
FIG. 16. CONIPRESSNE STRENGTH DEMLOPMENT FOR THE NINE SERIES
FOR CURING TYPE E. 24 h AT 37%, AFTER TWO MEMERIINE LRIERS WES
APLICATED AND THEN UNDER SHADE TO TIME OF TESTING.
I80
AGE DAYS
FIG.17. SERIES 1 TO 6 SHOWING THE CONTENT OF CCM EXCEPT

PNPC ,FOR CURING "R"
@Seismicisolation
@Seismicisolation
80
ñGE DAYS
FIG.18. SERIES 1 TO 6 SHOWING THE CONTENT OF CCM EXCEPT

PNPC, FOR CURING "A"
80
AGE DAYS
Fig.19. SERIES 1 TO 6 SHMMNG THE CONTENT OF CCM EKCEPT

PNPC, FDR CURING "F"
@Seismicisolation
@Seismicisolation
DAYS
Fig. 20. SERIES 1TO 6 JHOVUING THE CONTENT OF CCM EXCEPT

PNPC, FOR CURING " G
80
AGE DAYS
FIG.21. SERIES 1 TO 6 SHDIMNG THE CONTENT DF CCBA EXCEPT

PNPC. FOR CURING "R"
@Seismicisolation
@Seismicisolation
SP 209-2
Marine Exposure of High-Strength

lightweight Concrete
by G. C. Hoff
Svnopsis:An extensive laboratory program was initiated in the early 1980’s to de-
velop high-strength lightweight concrete for use in offshore oil and gas structures
in severe marine environments. From the results of that development program, four
mixtures were chosen to be evaluated under field conditions. Large prisms (305 by
305 by 914mm) of each mixture were placed in the tidal zone at the U.S. Army
Corps of Engineers Severe Weather Exposure Station on the border between Cana-
da and the United States. The mixtures used an expanded slate aggregate from the
USA and a pelletized clay aggregate from Japan. All mixtures contained silica
fume and had total binder content from 494 to 556 kgku m with water-binder ra-
tios of 0.28 by mass. The concrete density varied from 1800 to 1990 kgícu m, with
90-day strength from 60 to 73 MPa depending on the mixture. The prisms under-
went annual visual and non-destructive evaluations. After 10-years exposure in the
tidal zone, the prisms were removed to the laboratory where they were examined
for strength, robustness and chloride ion penetration. This paper reports the results
of the test program. In general, the overall performance looks very good.
Keywords: freezing and thawing; high-strength;

lightweight aggregates; lightweight concrete; marine
exposure
@Seismicisolation
@Seismicisolation
29
30 Hoff
George C. Hoff is President, Hoff Consulting LLC, Clinton, Mississippi, USA.
Dr. Hoff is a past-President of the AC1 and an Honorary Member of the institute.
He has authored numerous papers on high-performancehigh strength concrete
and lightweight aggregate concrete.
INTRODUCTION
Between 1982 and 1986, eleven companies participated in a Joint Industry Project
to study the use of high-strength lightweight concrete for Arctic applications. The
results from that work are summarized in an earlier paper (1). Included in the
evaluation program was the freezing and thawing behavior of high-strength
lightweight concrete under a variety of laboratory test conditions. This
information is specifically summarized in another paper (2). To complement the
laboratory evaluations of freezing and thawing resistance, large specimens of
selected mixtures from the initial program were produced and installed at the U.
S. Army Corps of Engineers Severe Weather Exposure Station at Treat Island in
Cobscook Bay near Eastport, Maine, USA. The island is on the border of the
United States and Canada, and is at the entrance to the Bay of Fundy. At the
exposure site, the specimens were on a wharf that becomes submerged in
seawater with each tide. At the Northern latitude of Treat Island, the specimens
also undergo approximately 100 cycles of freezing and thawing during the winter
season when they alternate in and out of a submerged condition with the tide. The
specimens were subjected to the marine exposure for 10 years. During that time
period they underwent annual inspections. In 1995, after 10-years exposure, the
specimens were retrieved from the exposure site and evaluated in the laboratory.
The concrete specimens were fabricated during a period from March 9 to March
26, 1985 at Construction Technology Laboratories, Skokie, Illinois and were
shipped by truck to Eastport, Maine on May 30, 1985. Upon delivery, the truck
driver, finding no one to receive them, just dropped them iìom the truck into a
ditch on the side of the road and departed. They were subsequently retrieved and,
in July 1985, were transported by boat to the exposure site and installed on the
wharf. The initial installation of the specimens at Treat Island, their annual
evaluations, and final retrieval with laboratory evaluations were conducted by
researchers at the University of New Brunswick, Fredericton, New Brunswick,
Canada.
This paper describes the effect of the severe marine exposure on the high-strength
lightweight concrete that was evaluated, and relates the field exposure
observations to the laboratory studies on the same types of concrete.
@Seismicisolation
@Seismicisolation
MATERIALS
Cement
Type II portland cement from a single source was used for all the concrete
mixtures. The cement met the requirements of ASTM C 150, "Standard
Specification for Portland Cement,' for a Type IA1 cement. It had a specific
density of 3.15 and a low C3A content of 5.5%. Chemical and physical analysis
of the cement can be found in (1).
Sumlementarv Cementing Materials
Silica fume from a single supplier was used in all mixtures. The silica fume had a
silica content of 92-93% and a specific density of 2.02. Chemical and physical
analysis of the silica fume can be found in (1).
In the original test program (I), a screening of candidate lightweight aggregates

(LWA) was conducted to select the most appropriate aggregates for study. Two
aggregates were selected and were thoroughly studied (1). These same two
aggregates were used in the test specimens of this program.
The first LWA was a crushed aggregate, 19 mm nominal size manufactured

commercially from expanded slate in the USA, The oven-dry specific density
was 1.35 while the saturated surface-dry (SSD) specific density was 1.46. The
24-hour absorption was 7.60%, and the dry-rodded unit weight was 775 kg/cu m.
The second LWA was a pelletized clay aggregate produced in Japan. It also came
from a commercial production and had a nominal size of 16 mm. The oven-dry
specific density was 1.33 while the saturated surface-dry (SSD) specific density
was 1.45.The 24-hour absorption was 8.65%, and the dry rodded unit weight was
794 kg/cu m. For one of the mixtures evaluated (LWC-2 saturated)), the
aggregate manufacturer provided pressure saturated aggregates that had an as-
received moisture content of 21% and a specific density of 1.55 at that moisture
content.
The fine aggregate used in ail mixtures was a well-graded natural silica sand
(Elgin Sand) with a specific density of 2.69.
@Seismicisolation
@Seismicisolation
32 Hoff
Chemical Admixtures
Each mixture contained both a high-range water reducing admixture (HRWR4)

and an air-entraining admixture (AEA). The HRWRA was a sulphonated
napthalene formaldehyde condensate and met the requirements of ASTM C 494,
"Chemical Admixtures for Concrete," for a Type F high-range water reducing
admixture. It had a specific density of 1.20 and a solids content of 42%.
The AEA was a neutralized vinsol resin meeting the requirements of ASTM C
260, "Air-Entraining Admixtures for Concrete," and ASTM C 233, "Standard
Test Method for Air-Entraining Admixtures for Concrete." It had a specific
density of 1.05 and had a solids content of 6%. It was compatible with the
HRWRA used in this study.
CONCRETE MIXTURE PROPORTIONS
Four mixtures from the initial program were evaluated and are described in Table
1. They are designated HSLWC, LWCl, LWC2 and LWC2 (saturated). LWCl
used a crushed lightweight aggregate while HSLWC and LWC2 used a pelletized
lightweight aggregate. The fourth mixture, LWC2 (saturated) was similar to
LWC2 but used pressure-saturated pelletized lightweight aggregate. In actual
construction scenarios for these s p e s of concretes, there is a high probability that
the concrete will be pumped, and the pumping of lightweight concrete can be a
problem if the aggregates are not properly conditioned prior to batching. The
manufacturer of the pelletized aggregate of this study recommended that the
moisture content of the aggregate should be between 20% and 25% by oven
drying if the concrete was to be pumped. At those moisture levels, the amount of
mixture water forced into the lightweight aggregate by the pump pressures would
be minimized. The moisture content of the factory pressure-saturated aggregate
used was 21%. The mixture proportions of LWC2 (saturated) were adjusted from
LWC2 to account for the high moisture content of the aggregate.
TEST SPECIMENS
Two concrete prisms were made from each of the four basic mixtures of Table 1.
Each prism had nominal dimensions of 305 mm by 305- by 914 mm. A
separate concrete batch was used for each prism. The identification of the
mixtures and prisms is as follows:
@Seismicisolation
@Seismicisolation
Concrete Prism Concrete Prism
Mixture Number Mixture Number
HSLWC 11M LWC2 31M
12M 32M
LWCI 21M LWCZ 41M
22M (saturated) 42M
A summary of the fresh concrete properties for each batch is shown in Table 2.
After casting, the prisms were continuously moist cured for 7-days at 23OC and
then cured in laboratory air until shipped. Four 150 mm by 300 mm cylinders
were also made from each batch and were continuously moist cured at 23OC until
tested. Two cylinders each were tested for compressive strength at 28- and 90-
days age. The results are shown in Table 3.
FIELD SPECIMEN EVALUATIONS
Upon initial installation on the Treat Island exposure wharf, each specimen was
photographed, and had its longitudinal pulse velocity and resonant frequency
determined. The initiai measurements taken in 1985 are shown in Table 5 and
occurred when the specimens were between 4 and 5 months old.
Ultrasonic Pulse Velocity
The pulse velocity was determined in accordance with ASTM C 597, "Standard
Test Method for Puise Velocity Through Concrete." This method is used to
assess the uniformity and relative quality of concrete, to indicate the presence of
voids and cracks, and to indicate changes in the properties of the concrete. Fifty-
millimeter diameter transducers were used with a coupling medium made of
kaolinite and glycerol. The measuring meter was initially calibrated to a
reference bar and checked periodically throughout the testing. When possible,
five puise transit times were determined along the longitude of each specimen. A
measurement was taken at the center of each specimen and near each of the four
comers. All other things remaining fairly constant, continuing decreases in the
puise velocity (increased pulse transit time) indicate intemal deterioration of the
concrete, principally due to development of cracks and microcracks.
@Seismicisolation
@Seismicisolation
34 Hoff
Longitudinal Resonant Freauencv
The longitudinal resonant frequency was determined in accordance with ASTM C

215, "Standard Test Method for Fundamental Transverse, Longitudinal, and
Torsional Frequencies of Concrete Specimens." This test method is primarily
used to detect significant changes in the dynamic modulus of elasticity of concrete
that is undergoing exposure to weather or other types of deteriorating influences.
The dynamic modulus of elasticity is a function of the mass of the specimen, the
dimensions of the specimen, a correction factor that depends on the ratio of the
radius of gyration to the length of the specimen and Poisson's Ratio, and the
square of the resonant frequency. All other things remaining fairly constant, a
progressive decrease in the resonant frequency indicates continuing deterioration
of the concrete.
Annual Inspections
At approximately 1-year intervals, beginning in 1986 and continuing through

1995, the specimens were visually examined, photographed and had their pulse
velocity and resonant frequency measured. During that time, the specimens were
submerged and exposed to air with each tidal fluctuation and, during the winter
season, underwent an average of 109 cycles of freezing and thawing per year.
The actual observed number of freezing (internal concrete temperatures (-2.2OC)
or below) and thawing (internai concrete temperatures (+2.2OC) or above) cycles
was as follows:
Winter No. of Winter No. of

Season Cycles Season Cycles
1985-86 144 1990-91 89
1986-87 109 1991-92 75
1987-88 110 1992-93 123
1988-89 106 1993-94 97
1989-90 135 1994-95 102
Saecimen 21M
Sometime during the 1990-1991 exposure period, one end of specimen 21M
broke off from the specimen. The crack that precipitated the break was evident
during the 1990 annual inspection. The broken piece represented approximately
22% of the volume of the specimen. It was retrieved in December 1991 and taken
to the University of New Brunswick where it was evaluated for density, tensile
@Seismicisolation
@Seismicisolation
and compressive strength, alkaii-aggregate reactivity, hardened air-void content,
and air void spacing. The results of the evaluation are described next.
The broken piece measured approximately 305 by 305 by 203 mm and weighed
38.6 kg. The break was a clean fracture with no other pieces broken off. When
brought to the laboratory, it was weighed in both air and water and the densiîy
was calculated to be 1844 kglcu m. Subsequent storage in a moist curing room
did not increase the density.
Four 95 mm cores were drilled completely through the block. Two each were
drilled from both the end of the block and the side of the block to check for cracks
or other flaws in the concrete. No defects were visible. The cores were kept in a
moist room until tested.
Tensile strength tests were performed on three of the cores using an indirect
tensile strength test developed by the Building Research Station of Great Britain
(3,4). This test basically involves injecting nitrogen gas into a sealed off section
of core to build up an internal pore pressure. This pressure is increased until the
core fractures. The tensile strength is read directly as the pressure needed to break
the core. The results are shown in Table 6. Opposite ends of cores 1 and 2 were
tested to get an indication of the tensile strength throughout the specimen. These
two cores were taken from the side of the block and were parallel to each other.
Compressive strength tests were done according to ASTM C 39, "Standard Test
Method for Compressive Strength of Cylindrical Concrete Specimens," and
ASTM C 42, Obtaining and Testing Drilled Cores and Sawed Beams of
Concrete," on the remaining pieces of cores 1, 2 and 3 after the tensile tests and
on core 4 which was too short to do both a tension and compression test. The
results are shown in both Table 4 and 6.
The pieces from the broken cores from the tensile strength test were used to check
for the alkali-aggregate reaction (AAR) using the uranyl acetate fluorescent test.
The fracture surfaces were coated with uranyl acetate and let stand for 3 to 5
minutes. The excess uranyl acetate was then rinsed off and the core put under an
ultraviolet lamp. Any AAR on the surface then shows up as a bright green glow.
Only traces of U R were noted in fine aggregate (natural sand).
Optical and microchemical studies of the sawn and polished surfaces of concrete
were canied out. Determinations of the air-void content and air-void spacing
were done in accordance with ASTM C 457, "Standard Test Method for
Microscopical Determination of Parameters of the Air-Void System in Hardened
Concrete." The principal results were as follows:
Air Content, % 4.7

Specific Surface, mm-1 12.01
Spacing factor, mm 0.500
@Seismicisolation
@Seismicisolation
36 Hoff
Post-exposure Evaluations
After the field measurements were made in 1995, the specimens were taken from
the wharf and moved to the laboratory at the University of New Brunswick.
Discussion of specimen breakage is presented in the following section.
Specimens 11M and 3 1M experienced an additional break during retrieval that
had not fully manifested itself on the wharf. All other breaks had occurred during
the exposure period.
For strength testing, each of the specimens had three 95 mm cores drilled from it.
The cores were drilled from the side of the specimen with one core being
obtained, where possible, from each third of the specimen. One core from
specimen 22M was inadvertently drilled across a crack and could not be tested.
The cores were cut into 200 mm lengths, and the ends were ground for testing.
As with the earlier evaluation of a piece of Specimen 21M, compressive strength
tests of the cores were done according to ASTM C 39 and ASTM C 42. The
results of the compressive strength testing are shown in Table 4.
An additional 95 mm core was drilled from near the center of each specimen to
evaluate the extent of chloride ion penetration into the concrete over the ten-year
exposure period. The specimens were prepared following Canadian Standard
A23.2-4B, "Sampling and Determination of Water-Soluble Chloride Ion Content
in Hardened Grout or Concrete," and the Mercuric Nitrate Test Method was used
to determine the amount of soluble chloride. The chloride content was
determined in five layers as shown in Table 7. The chloride content shown is the
average from both specimens for that mixture. Any chloride content values below
100 mgkg are shown as 100 mgkg because this is the limit of precision of the
test method.
SPECIMEN BREAKAGE
Five of the eight specimens had lateral breaks in them at the time they were
removed from the wharf. All of the specimens had at least one of the top four
comers broken. The lateral breaks occurred from 200 to 330 mm from the ends of
the specimens. The orientation of the end of the specimen (towards the beach or
towards the sea) made no difference. In general, the location of the breaks
corresponded to the supporting timber that held the specimen off of the wharf
surface. The first observance of a lateral crack in the specimen and the time when
the specimen broke are as follows:
@Seismicisolation
@Seismicisolation
Specimen Observed Specimen
Number First Crack Broke
11M 1991 1993
12M 1992 1994
21M 1990 1991
22M 1991 1993
31M 1993 1995
The researchers who carried out the annual inspections believe that the cracks
were the result of rough handling of the specimens during their initial
transportation to the exposure station, that was followed by a period of rough
storms that displaced the specimens and caused a partial collapse of the
supporting structure. No plausible explanation based on materials behavior could
be developed. Once cracked, subsequent freezing and thawing cycles caused the
material to separate. The failure surfaces were fairly straight and uniform. Once
the specimens broke into smaller pieces, the resonant frequency measurements
could not be related to the original specimen size and configuration, and were
discontinued.
DISCUSSION OF FIELD EXPOSURE RESULTS
Chloride Ion Observations
The chloride ion contents observed in the first 11 mm indicate that there was little
difference for the three mixtures (HSLWC, LWC2 and LWC2 (saturated)) made
with the pelletized clay lightweight aggregate. The concrete made with the
crushed slate lightweight aggregate contained approximately 30% more chlorides.
At this close proximity to the surface of the concrete, it is probable that the open
cells on the surface of the crushed aggregate had taken in some chlorides from sea
water in addition to some that were present in the original concrete mixture.
For the second 11 mm, there was some ranking of the chloride levels with the
HSLWC having the highest level, followed by the crushed slate aggregate mixture
(LWC2), the pelletized clay mixture (LWC3) and the saturated pelletized clay
aggregate mixture (LWC2 (saturated)). At subsequent depths (30 to 105 mm), the
HSLWC had somewhat higher amounts of chloride than the other mixtures.
For the duration of this exposure period, the levels of chloride observed would not
appear to be a problem for reinforcing bars that had the usual amounts of concrete
cover (50 to 75 mm) typically specified for marine structures. This conclusion is
supported by the observation in the initial study (1) that LWC1, LWC2 and
HSLWC experienced density increases of less than 16 kg/cu m when submerged
in simulated seawater for periods up to 1-year. This suggests that very little
@Seismicisolation
@Seismicisolation
38 Hoff
moisture is entering the concrete at near-surface water depths experienced at Treat
Island.
Pulse Velociîv Measurements
Table 8 shows the ratio of the yearly pulse velocity measurement to the initial
pulse velocity measurement (See Table 5). For the first eight years, five pulse
velocity measurements were made (when possible) each year on each specimen.
As noted earlier, a measurement was made near each of the four comers and at the
center of the specimen. In years 9 and 1O, only one measurement was made at the
center of the surviving specimens because many of the comers had broken or had
become irregular so that good contact between the transducer and the concrete
could not be obtained. With the exception of Specimen 21M, which had an
observed transverse crack that caused a reduction in the pulse velocity at the Year
5 inspection, the other specimens appeared to begin reductions in pulse velocity at
Years 7 and 8. Because comer damage and some transverse cracking was
occurring at that time, the measurements made at the comers (particularly the top
two comers because the transverse cracking and comer cracking appeared to
initiate at the top surface) tended to be low values that reduced the overall average
value. Consideration of only the pulse velocities taken at the center of the
specimen indicated virtually no change in pulse velocity until the specimens
either broke or exceeded 9 years of exposure. The Year 10 pulse .velocities of
32M, 41M and 42M were reduced approximately 10%. This suggests that the
uncracked lightweight aggregate concrete was relatively unchanged in the marine
environment up to 1000 cycles (approx.) of freezing and thawing.
Resonant Frequency
Table 9 shows the ratio of the yearly resonant frequency measurement to the
initial resonant frequency measurement (See Table 5). As noted above, when all
things remain fairly constant for a test specimen, continuing decreases in the
resonant frequency indicate continuing deterioration of the concrete. Because the
resonant frequency is related to the length of the specimen, when the specimen
begins to appreciably shorten, the measured values of the shorter specimen can no
longer be directly compared to the longer specimen. This is what happened as the
some of the specimens began to have transverse cracks. The end points of the
results shown in Table 9 reflect when the resonant frequency measurements were
stopped for each specimen.
The results for LWC1 indicate that Specimens 21M and 22M began to experience
some deterioration after a few years of exposure. This gradual deterioration
continued at a fairly uniform rate until the specimens broke. After 3-years
@Seismicisolation
@Seismicisolation
exposure at Treat Island, the specimens had undergone 363 freezing and thawing
cycles and were beginning to show a slight decrease in resonant frequency. For
the laboratory test (1) with ASTM C 666, Procedure A, there was no change in
the dynamic modulus at 300 cycles. After 6 or 7 years exposure at Treat Island,
LWCl had been subjected to 693 or 768 freezing and thawing cycles, respectively
and had experienced approximately a 10-12% decrease in resonant frequency.
The "rapid freeze thaw test" used for LWCl in the initial program (1,2) was
identical to the ASTM C 666, Procedure A except that it used simulated seawater.
For the same number of cycles as Treat Island, the results in the laboratory also
indicate a 10% reduction in the dynamic modulus.
Specimens 11M and 12M began to experience some deterioration after 5- to 6-

years of exposure. This gradual deterioration also continued at a fairly uniform
rate until the specimens broke. These results are similar to the laboratory results
(1,2) with the field specimens performing slightly, but not significantly, better.
Specimens 31M, 32M, 41M and 42M all indicated exceptional performance after
1090 cycles of freezing and thawing.
Strength Considerations
From Tables 3 and 4, the following comparison of the compressive strength data
can be made:
28-b~ 10-Year Percent
Cylinder Average of Cylinder
Strength, Core Strength, Strength,
Mixture MPa -
MPa -
YO
HSLWC 70.0 59.8 85
LWCl 59.3 47.3 80
LWC2 59.8 38.1 64
LWC2 (sat.) 58.3 53.3 91
For LWCl, the 8-year (Specimen 21M) and 10-year core strengths were lumped
together as there was no significant difference in the average values.
The experience of this author (5) with cores from normal density high-strength
concrete, taken at later ages (5-years) and compared to continuously moist cured
cylinders, was that the core strength typically was only 70 to 80% of the cylinder
strength. For normal density high-strength concrete, it is generally believed that
due to the high binder content, the presence of silica fume and low water-binder
ratio, a self-desiccation can occur that leads to internal microcracking, and
associated strength regression with time. For lightweight concrete, which initially
contains more moisture in the aggregate, less effect due to self-desiccation should
be expected because the aggregates can supply additional water for the binder
@Seismicisolation
@Seismicisolation
40 Hoff
hydration, particularly at the interfacial zone of paste and aggregate. This is
evident to some degree for LWCl (80%) and particularly for HSLWC (85%) and
LWC2 (saturated) (91%). The core strengths of LWCL are lower (64%), thus
suggesting that the concrete may have begun to deteriorate due to the severe
exposure.
The tensile strength (Table 6) measured from the cores of the broken piece of
21M (LWC1) averaged 4.8 MPa. Corresponding compressive strengths from
cores of the same piece averaged 47.8 MPa. In the original work (i), for LWCl
the 28-day ratio of splitting tensile strength of cylinders (ASTM C 496, Splitting
Tensile Strength of Cylindrical Concrete Specimens) to compressive strength of
cylinders, both of which had been moist cured for 28-days, was only 6.6%. There
is not enough data here to make any specific conclusions but the observed tensile
strength appeared to be adequate so that there were no concerns about loss of
tensile strength.
Surface Conditions
For higher numbers of cycles in the laboratory "spectral" testing (1,2), LWCl and
LWC2 showed extensive cracking, spalling and pitting, and HSLWC was
experiencing extensive scaling. For the Treat Island exposure, HSLWC did not
exhibit any pitting or scaling. LWCl, LWC2 and LWCL (saturated) did have a
small amount of pitting but it occurred only on the top surface of the specimens.
That surface was the only unformed surface of the specimen, having been hand-
finished with a trowel. All of the specimens did experience a small loss of the
material at the comers. In a few instances, some of the comers were broken off.
In general, the surfaces were considered satisfactory.
CONCLUSIONS
I After exposing several mixtures of high-strength lightweight aggregate concrete

in the tidal zone to a severe marine environment for 10-years, the following
conclusions were reached:
1. Chloride contents in the concrete cover were quite low as would normally be
expected from properly designed and built marine structures. Concrete made
with a crushed slate aggregate had higher chloride contents near the surface,
than comparable concrete made with a pelletized clay aggregate. This may be
due to an open cell structure on the surface of the crushed aggregate that
would adsorb some chloride when compared to the sealed surface of the
pelletized aggregate. The higher strength concrete of this program (68.9 h4Pa)
@Seismicisolation
@Seismicisolation
generally had somewhat higher chloride content than the lower strength
concrete (57.9 to 60.0 m a ) .
2. Based on ultrasonic pulse velocity and resonant frequency measurements
made annually over a 10-year period, the high-strength lightweight concrete
prisms have performed quite satisfactorily on exposure to numerous wetting-
drying and fieezing-thawing cycles (an average of 109 freezing and thawing
cycles per year).
3. With the exception of two prisms (31M and 32M) made with mixture LWC2,
the core strengths of the concrete after 10-years were within expected values
(80 to 90 percent) of the 28-day cylinder strength. Companion prisms (41M
and 42M) to LWC2, made with the same lightweight aggregate but in a
saturated condition, gave the highest core strength ratio (91 percent) thus
suggesting that there may have been self-desiccation internally in the concrete
mixture LWC2 made with unsaturated aggregate.
4. In general, the surface condition of the concrete was good after 10-years of
severe marine exposure. A number of the unreinforced prisms experienced
some through-thickness breakage and broken corners. This was attributed to
rough handling prior to their installation and being tossed about on the wharf
during severe storms.
5. While the concrete appears to give satisfactory performance up to 10-years
age, these limited results are not sufficient to draw conclusions about long-
term durability (50- to 100-years service life) in a severe marine environment.
REFERENCES
1. Hoff, G.C., "High Strength Lightweight Aggregate Concrete for Arctic

Applications, Parts 1, 2 and 3," Structural Lightweight Agrrregate Concrete
Performance, Editors: T. A. Holm and A. M. Vaysburd, Published as AC1
SP-136, American Concrete Institute, Farmington Hills, Michigan, 1992, pp
1 - 245.
2. Hoff, G.C., "Freezing and Thawing Behavior of High-Strength Lightweight

Concrete," Proceedings, International Conference on Concrete Under Severe
Conditions," Sapporo, Japan, August 1-4, 1995, Vol. 1, pp 1368-1377.
3. Lydon, F.D. and Balendran, R.V., "Some properties of higher strength

lightweight concrete under short-term tensile stress," International Journal of
Lightweight Concrete, Vol. 2, No. 3, 1980, pp125-139.
@Seismicisolation
@Seismicisolation
42 Hoff
4. Boyd, A.J., "The Indirect Tensile Strength Test," Technical Memo,
University of New Brunswick, Department of Civil Engineering, Fredericton,
New Brunswick, Canada, March 9,1992.
5. Hoff, G.C. and Elimov, R., "Core Strengths of High-Strength Concrete,"

Proceedings, International Conference on Hi&-Performance Concrete:
Desien and Materials and Recent Advances in Concrete Technologv, Kuala
Lumpur, Malaysia, December 2-5, 1997, pp 705-731. (Published as SP-172,
American Concrete Institute, Farmington Hills, Michigan.)
Table 1 -- Concrete Miture Proportions

I kdcu
._... m
~~. I
Mixture HSLWC LWCl LWCZ LWCZ (Sat.)
Specimen 11M I 12M 21M I 22M 31M I 32M 41M I 42M
I I I I
Admixture liter liter liter liter liter liter liter liter

Mix Water* 148 149 133 136 132 132 141 139
WaterKement 0.315 0.314 0.307 0.307 0.303 0.303 0.304 0.304
Waterícementitious 0.279 0.278 0.279 0.279 0.276 0.276 0.277 0.277 ,
*Water added in additionto that contained in the chemical admixtures.
Concrete Specimen Slump, Air Content, Unit Weight,

Mixture Number mm percent kg/cu m
HSLWC 11M 165 5.8 1983

12M 200 5.0 1977
LWCl 21M 105 6.7 1799
32M 1780
LWC2 41M 155 5.O 1805
130 1804
@Seismicisolation
@Seismicisolation
Table 4 -. Iompressi :Strength lata fr< 'es

lomp. 3mp.
Concrete Specimen core Str., Concrete jpecimen Core str.,
Mixture Number Number MPa Mixture Number Number MPa
HSLWC 11M 1 73.8 LWC2 31M 1 48.3

2 46.0 2 32.1
4 76.3 4 37.6
Average 65.3 Average 39.3
HSLWC 12M 1 46.9 LWC2 32M 1 37.6
2 57.6 2 29.3
4 58.0 4 43.6
Average 54.2 Average 36.8
LWC1* 21M* 1 (95)* 55.2* LWC2 41M 1 52.5
3 (95)* 52.0* (sat.) 2 53.8
4 (95)* 49.8* 4 48.8
~
Average* 52.3* Average 51.7
LWCI* 21M* 1 (93)* 41.3* LWC2 42M 1 58.4
2 (93)* 46.1* (sat.) 3 44.4
3 (93)* 46.0* 4 61.7
4 (93)* 57.9* Average 54.9
Average* 47.8*
LWC1 22M 2 39.2
4 44.2
Average 41.7
*Cores01 Uned and i ;ted from Specimen LM in both 1 3 (93) a n d 995 (95).
@Seismicisolation
@Seismicisolation
44 Hoff
Table 5 -- Initial Pulse Velocity and Resonant
Pulse Resonant
Mixture Number
HSLWC 11M 4353 1142
22M 4106
LWC2 31M 4122 1090
32M 4055 1067
4223 1107
4167 1094
Table 6 -- Tensile and Compressive Strengths of Cores
Core Compressive Strength, Tensile Strength,

Number MPa MPa
1 41.2 5.1
2 46.1 5.0
3 46.0 4.4
4 57.9 ----
HSLWC, LWC-1, LWC-2, LWC-2 (sat.),

Depthof avg., avg., avg., avg.5
Layer,mm mgkg mgkg mgkg mgkg
0-1 1 3500 4640 3330 3535

15-26 1875 1550 1155 600
30-41 725 1O0 225 265
45-55 530 115 125 1O0
95-105 280 1O0 105 1O0
@Seismicisolation
@Seismicisolation
6 1.00 1.00 0.98 1.02 1.01 1.00 1.00
7 1.00 1.00 0.89 1.00 1.01 1.00 1.00
8 0.92 0.88 0.80 1.00 0.97 0.99
9 1.02 0.96 0.89
10 0.93 0.88 0.89
Table 9 - Ratio of Yearly Resonant Frequency Measurement to
@Seismicisolation
@Seismicisolation
SP 209-3
Performance of High-Volume Fly Ash

Concrete in Hot Weather
by P. K. Mehta
SvnoDsis: Thermal cracking is a serious problem with modem portland-cement

concrete structures, especially cast-in-place massive elements made with relatively
high cement content and cured under hot-weather conditions. From theoretical
considerations and field experience it is concluded that blended portland-cement
concrete mixtures containing 50 percent or more ASTM Class F fly ash by mass of
the total cementitious material perform much better under these conditions.
Keywords: fly ash; heat of hydration; hot weather;

portland cement; thermal cracking
@Seismicisolation
@Seismicisolation
47
48 Mehta
P. Kumar Mehta is Professor Emeritus in the Civil and Environmental
Engineering Department at the University of California at Berkeley. He is the
author or coauthor of numerous papers on cementitious materiais and properties of
concrete including three books. A Fellow of the American Ceramic Society and
the American Concrete Institute, he has received several awards including the
ACI’s Wason Medal for materials research, and AC1 Construction Prentice
Award.
INTRODUCTION
The term, “maturity,” implies the development of strength in freshly-

mixed concrete. To achieve a desired level of maturity, the cement hydration
reactions have to be facilitated through proper curing of concrete after
placement. The two objects of curing are to prevent the loss of moisture and to
control the concrete temperature for a sufficient length of time until it has
reached an acceptable level of maturity. Effects of both cold weather and hot
weather conditions on maturity of cement-based products are well documented
in the published literature. Modem portland cements appear to be better suited
for concrete construction in mild or colder climates. From theoretical
considerations and field experience discussed in this paper it seems that, under
warm-weather conditions, blended portland cements containing very high
volumes of fly ash provide a better and more cost-effective approach for
building durable concrete structures.
THEORETICAL CONSIDERATIONS
Since the 1930s, it is well known that the rates of hydration and therefore
strength development in cement-based products have been greatly accelerated
by altering the compound composition and fineness of cement. For instance,
ASTM Type I portland cements of today typically contain 50% of the highly
reactive compound, tricalcium silicate, that is almost twice as much as it was 70
years ago. Similarly, modem cements are ground to a fineness that is close to
400 m2kg Blaine. This is also considerably higher than the typical fineness of
older cements, namely 250-300 m2kg. As a result, according to Mehta and
Burrows (i), the 7-day compressive strength of ASTM C109 mortar cubes has
doubled from about 2500 psi (17 Mpa) to 5000 psi (34 Mpa).
The hydration reactions of portland cement are exothermic. The

increased reactivity of cement, which is responsible for high-early strength, is
associated with the liberation of high heat of hydration. With modem portland
cements, approximately 50 percent of the potential heat is liberated within the
first 3 days; in fact, it is estimated that 70 to 80 c d g heat is generated within
@Seismicisolation
@Seismicisolation
7 days (2). in cold-weather concreting, the temperature rise associated with the
heat of hydration can be a help. For example, the cement hydration rate at 20°C
is twice the rate at 1O"C, and at 35°C it becomes four times the rate at 10°C (3).
However, under hot-weather conditions, the adiabatic temperature rise in
concrete mixtures, particularly with those containing high cement content, can
become a liability. in massive structural members, the heat dissipation is slow
and a large temperature difference between the interior and the surface of
concrete can often result in thermal cracking which would lead to loss of
strength and impermeability. With 350-450 kg/m3 (600-760 lb/yd3) cement
content, it is not unusual to have an adiabatic temperature rise of the order of 50-
8OOC. Experience shows that, unless the fresh concrete is precooled, thermal
cracks are virtually unavoidable in such concrete mixtures. Published literature
contains many reports of thermal cracking with high-early strength concrete
mixtures including those used for the construction of recently built bridges in
Colorado and Texas (1).
Mock-up test blocks of 1.2 by 1.2 by 2.4 m size made with trial-batch
concrete mixtures to be used for the construction of a cathedral in the Los
Angeles area showed 45 to 70°C temperature rise and severe thermal cracking
(4). To avoid cracks due to thermal gradients the designers specified that the
maximum temperature difference between the interior and the surface of
concrete should be less than 25OC. This was achieved by a combination of
methods that included formulation of concrete mixtures with chilled water and
ice, 15% cement replacement with a Class F fly ash, and leaving the forms in
place for at least 2 Vi days after the concrete placement.
EXPERIENCE WITH HIGH-VOLUME FLY ASH CONCRETE
The high-volume fly ash (HVFA) concrete developed by Malhotra (3,

offers a convenient and cost-effective approach to the solution of the thermal
cracking problem in massive structural concrete members made with modem
cements. Typically, the HVFA concrete mixtures are characterized by 50
percent or more cement replacement with fly ash by mass. A 50 percent
substitution of cement with a pozzolan is expected to bring down the 7-day heat
of hydration from about 70-80 to 45-50 caYg (2). Bilodeau and Malhotra (6)
reported that, in a large HVFA concrete block of 3.05 by 3.05 m by 3.05 m size
the temperature rise was limited to 35"C, compared to 65°C reached in a block
of the same size made of concrete containing ASTM Type I portland cement
only.
Mehta and Langley (7) reported how, under hot-weather conditions,

thermal cracking in a recently installed raft foundation was compleiely
eliminated by the use of a HVFA concrete mixture containing 57% fly ash by
mass of the total cementitious materials. This foundation for a stone temple in
Kauai is composed of two, parallel, unreinforced, monolith slabs, each 36 by 17
by 0.61 m (117 by 57 by 2 feet). It was built during the summer when the
@Seismicisolation
@Seismicisolation
50 Mehta
ambient temperature was 25 to 30"C, and the relative humidity was 80 to 100%.
The specifications for the concrete mixture required a 100-125 mm slump, a
minimum of 20 MPa compressive strength at 90-day, and a maximum
temperature rise of 15°C in the interior of concrete. Laboratory trial mixtures
and a field test showed that, with local materials, the specifications could be met
by the use of a superplasticized, air-entrained, concrete containing 106 kg/m3
ASTh4 Type I portland cement and 142 kg/m3 ASTM Class F fly ash. The
temperature history of concrete in the slab-on-ground showed that the concrete
temperature rose from 27 to 40°C within the first 24 hours after placement, and
thereafter dropped gradually to the ambient temperature of about 30°C during
the next 6 days. A similar temperature rise of 13°C was recorded for the upper
slab which took two weeks to cool down to the ambient temperature. Careful
examination of the exposed surface of slabs at regular intervals until the time of
construction of the superstructure, almost two years after the installation of the
foundation, did not show evidence of any cracking. Also, unlike portland-
cement concrete mixtures which do not show substantial strength gain between
90 day and 1-year, the test cores of the HVFA concrete showed a doubling of
the compressive strength from 20 to 40 MPa during the same period.
The satisfactory behavior of the HVFA system under warm and humid
curing conditions is not limited to the low-early strength concrete mixtures of
the type used for the Kauai temple foundation. Recently, Manmohan and Mehta
(8) reported the use of high-early strength, HVFA concrete mixtures for the
construction of a post-tensioned, reinforced concrete belt foundation and heavily
reinforced shear walls for the seismic strengthening of Barker Hall - a six-storey
building at the University of California campus at Berkeley. A superplasticized,
non-air-entrained, concrete mixture containing 200 kg/m3 ASTM Type I
portland cement and 200 kg/m3 ASTM Class F fly ash gave 20 MPa
compressive strength at 7-day in standard-cured cylinders. However, the 7-day
strength of concrete cores obtained from a 1 m3, site-cured, test cube made from
the same concrete mixture was found to be 27 MPa, that was more than
adequate to meet the construction schedule. It should be noted that concrete
cores from site-cured mock up test blocks, not the standard-cured cylinders, are
a better representative of the properties of concrete in the structure. With HVFA
concrete, a modest temperature rise of the order of 15 to 30°C is beneficial for
acceleration of this cement hydration and the pozzolanic reactions; at the same
time the concrete temperature is not too high to cause thermal cracking. On the
contrary, cores obtained from structures built with high-early strength concrete
mixtures containing portland-cement only generally show much lower strength
at early age than the standard-cured cylinders because of excessive temperature
rise and subsequent thermal cracking.
Due to the massive size of the Barker Hall foundation, 3.3 m deep and
1.8 m wide, an adjustment was made in the concrete mixture proportioning in
order to reduce thermal stresses from the heat of cement hydration. The
portland cement content of this concrete was reduced from 200 to 160 kg/m3 but
@Seismicisolation
@Seismicisolation
the amount of fly ash was kept unchanged at 200 kg/m3; thus the fly ash content
in the concrete mixture was 55 percent by mass of the total cementitious
material. With this adjustment, the temperature rise of concrete in the wall
interior was limited to 24°C at approximately 80 hours after placement.
CONCLUSION
Modem portland cements, characterized by high tricalcium silicate

content and high fineness, hydrate rapidly, and therefore develop high-early
strength. They also develop high heat of hydration which, in the case of massive
structural elements, usually results in thermal cracking unless costly measures
are taken to address the issue. Concrete placement under hot-weather conditions
further aggravates the problem. It is, therefore, generally accepted now that
modem portland cements are not suited for the fabrication of large cast-in-place
concrete elements under hot-weather conditions. From theoretical
considerations and field experience, it is clear that blended portland cement
mixtures containing 50 percent or more ASTM Class F fly ash by mass of the
cementitious material provide a better and more cost-effective approach for
building durable concrete structures in warm climates. In the case of the HVFA
system, portland cement and fly ash complement each other in the sense that;
first, the adiabatic temperature rise is much lower due to a drastic reduction in
the portland cement content of the concrete mixture, and secondly the heat of
hydration is beneficially used up for acceleration of the pozzolanic reaction
instead of causing excessive temperature rise and subsequent thermal cracking.
It is also concluded that an accurate assessment of the early strength and
impermeability of concrete in massive, cast-in-place structural members can
only be made from cores obtained from site-cured test blocks, not from
laboratory-cured concrete cylinders.
REFERENCES
(1) Mehta, P.K., and Burrows, R. W., “Building Durable Structures in the
21” Century,” Concete Intemational, Vol. 23, No. 3, March 2001, pp.
57-63.
(2) Mehta, P. K., and Monteiro, P. J. M., Concrete: Microstructure,

Properties, and Materials, McGraw-Hill, College Custom Series, 1993,
550 pages.
(3) Gotfredsen, H. H. and Idom, G. M., “Curing Technology at the Faroe

Bridge, Denmark,“ ACI, SP-95, 1985, pp 17-31.
(4) Selna, D., and Monteiro, P. J. M., “Cathedral of Our Lady 01 tne
Angels,” Concrete International, Vol. 23, No. 11, Nov. 2001, pp. 27-33,
@Seismicisolation
@Seismicisolation
52 Mehta
(5) Malhotra, V. M., ?Superplasticized Fly Ash Concrete for Structural
Applications,? Concrete International, Vol. 8, No. 12, Dec. 1986, pp 28-
31.
(6) Bilodeau, A. and Malhotra, V. M., ?High-Volume Fly Ash Systems:

Concrete Solution for Sustainable Development,? ACI Mat. Jour., Vol.
97, NO.1, 2000, pp. 41-48.
(7) Mehta, P. K., and Langley, W. S.? ?Monolith Concrete Foundation:

Built to Last a 1000 Years,? Concrete International, Vol. 22, No. 7, July
2000, pp. 27-32.
(8) Manmohan, D., and Mehta, P. K.,?Heavily Reinforced Shear Walls and
Mass Foundations Built with Green Concrete,? to be published in
Concrete Intemational, July 2002.
@Seismicisolation
@Seismicisolation
SP 209-4
Effect of Fiber Reinforcement on

Residual Properties of
High-Strength Concrete Under
Elevated Temerature
by T. Horiguchi, T. Takano, N. Saeki, and T. D. Lin
Svnopsis: High-strength concrete shows particular characteristic behavior at ele-

vated temperatures, such as explosive spalling, that is rarely observed in normal-
strength concrete. This behavior has been attributed to the very dense concrete ma-
trix usually associated with high-strength concrete.
This paper presents the effect of fiber reinforcement on mitigation of ex-
plosive spalling and residual properties of high-strength concrete under elevated
temperatures (600°C). The experimental work has been carried out on the influ-
ence of three parameters associated with fire resistance. These three parameters are
an addition of polypropylene fibers andior polymer beads, the moisture content of
the concrete and the water to cement ratio.
The experimental results showed that the internal temperature elevation
was mitigated in the test specimens, that contained the fibers or beads, that were
melted by heating, by low cement-water ratio and by high moisture content. The
accompanying strains due to heating at these conditions were reduced in the test
specimens.
Keywords: density; fiber reinforced concrete; fire

resista nc e; high-stre ngt h concrete; polypropylene
fibers; explosive spalling; thermal diff usivity
@Seismicisolation
@Seismicisolation
53
54 Horiguchi et al.
AC1 member Dr. T. Horiguchi is an associate professor at Department of
Civil Engineering, Hokkaido University, Sapporo, Japan. He received his Dr.
Eng. from Hokkaido University in 1975. His research interests are in the area
of concrete durability, lunar concrete, fiber reinforced concrete and Controlled
Low Strength Materials.
T. Takano is a doctoral student in the Department of Civil Engineering,

Hokkaido University, Sapporo, Japan. He has a Master degree in High
Performance Concrete Engineering from Hokkaido University, Japan.
AC1 member Dr. N. Saeki is Professor at Department of Civil Engineering,

Hokkaido University, Sapporo, Japan. He received his Dr. Eng. in 1975 from
Hokkaido University. His research interests include durability of concrete,
development of recycling systems and rehabilitation and strengthening of
existing concrete structures.
AC1 member Dr. T. D. Lin is a president of LinTek Corporation, IL. He

initiated lunar concrete research and served as the chairman of AC1
Committee 125, Lunar Concrete. He was the recipient in 1994 of the AC1
Delmar Bloem Award and the AC1 award for significant technology
contribution.
INTRODUCTION
A large number of studies on the properties of concrete at high

temperature have been carried out since the 1940s (1-3). The test specimens
were usually made of either low or normal-strength concrete, since these
studies were conducted before the advent of high-strength concrete for use in
construction. Recent results of fire tests have shown that there are well-
defined differences between the properties of high-strength concrete and
normal strength concrete. However what is more important about high-
strength concrete is the occurrence of explosive spalling when high-strength
concrete is subjected to rapid heating, as in the case of a fire (4, 5). High-
strength concrete is generally made with a low water to cement ratio and
typically includes silica fume. Thus High-strength concrete has lower
permeability and water content compared with normal strength concrete. It is
pointed out that the higher susceptibility of high-strength concrete to explosive
spalling when subjected to high temperature is due to the lower permeability
of high-strength concrete, which limits the ability of moisture to escape from
the pores. As heating increases, the pore pressure also increase. This increase
in vapor pressure continues until the internal stresses become so large as to
cause explosive spalling of the heated concrete. Experimental studies (6-9)
have shown that explosive spalling of high-strength concrete is affected by the
following factors: a) rate of temperature rise, b) mineralogical composition of
the aggregates, c) thermal induced mechanical stresses, d) reinforcement
@Seismicisolation
@Seismicisolation
arrangement, e) moisture content, and f ) density of the concrete matrix.
Recently the addition of polypropylene fibers into high-strength
concrete was reported to be very effective against the explosive spalling (10-
13). As heating increases, the fibers in the cement matrix start to melt at about
170°C and increase the total pore area. This melting effect mitigates the
explosive spalling as it provides pore space in which moisture vapor can
accumulate at lower vapor pressures. However, it is difficult to predict with
certainty when high-strength concrete will fail by explosive spalling, because
the explosive spalling doesn’t always occur. The object of this study is to
investigate the explosive spalling mechanism and the residual properties of the
concrete at elevated temperatures. In order to evaluate the melting effect,
polypropylene fiber and polymer beads were used in this study. These two
materials have about the same melting point ( 150 to 170°C) but different
geometry ( fiber and beads ). Compressive strength of 60 to 100 MPa (8700 to
14500 psi) was designed for test specimens with varying water to cement ratio
(W/C). Effect of the designed strength, fiber or polymer beads addition and
the moisture content of the specimen on the residual properties and the
explosive spalling mechanism were evaluated in this study.
EXPERIMENTAL PROCEDURES
Mixture DroDortion
The mixture proportion of the test specimens is shown in Table 1. Unit
water content (W) and sand aggregate ratio (da) were kept constant (W=170
kg/m3 and s/a=50%). The water to cement ratio (W/C) varies two levels
(W/C=0.25,and 0.45). Either polypropylene fiber (PP) or polymer beads (PB)
were added with 0.25% by volume. The melting point of PP fiber is
approximately 170°C and that of polymer beads is about 150°C. This indicates
that mixed PP fiber or polymer beads may melt to form voids to allow the
internal water evaporation. Two different levels of moisture conditions were
selected in this study. The specimens of wet condition (OPC45-wet, PB25-wet,
PP25-wet and PP45-wet) were cured by covering with a wet sheet for one day
after demoulding. The average moisture contents of OPC45-wet, PB25-wet,
PP25-wet and PP45-wet were 5.1%, 6.0%, 6.6%, and 5.1%, respectively. The
specimens for dry condition (PP45-dry) were cured by underwater curing at 20
*2”C for 27 days and then dried in a electric oven (60°C) for one day. The
average moisture content of PP45-dry was 3.4%.
Heating test Drocedure

An electric furnace, as shown in Fig.1, was used for heating the test
specimens. The heating procedure is shown in Fig.2. The heating rate was kept
10Wmin for an hour, followed by holding a test specimen at a constant
temperature of 600°C for two hours and cooling it afterward. The cooling step
was slowed down to prevent test specimens cracking.
@Seismicisolation
@Seismicisolation
56 Horiguchi et al.
Measurements of tenmerature distributions and deformations
Three cylindrical test specimens of 4 100 X 200 mm ( Q 4 X 8 inches)
were prepared for each mixture. In each specimen, five thermocouples of K
type were embedded at every 10 mm (0.4 inch) from the surface to the center
of specimen in order to measure the dynamic distribution of internal
temperatures. Figure 3 shows the detail of thermocouples and the specimen.
Thermal diffusivity K was calculated by the finite differential method (14)
from the following equation;
where 6x: the measuring distance,

tit: unit step time ( 1/60 hr.),
6T,: increment of temperature at point 2 from t to t+6t,
Ti, Tz, T3: measured temperature at point 1,2, and 3,
The finite difference equation was used to calculate the thermal

diffusivity of the measuring point at the center of three measuring points. For
this purpose, when temperature rose at a constant rate, the denominator of
equation (1) approaches O, and the thermal diffusivity has a tendency to
becoming large. When the gradient of the three measuring points is extremely
severe, a similar trend occurs, and then it is necessary to increase number of
measurement points.
The thermal deformation of the test specimens was measured during
heating by the compensated steel rod method (15, 16) at 100°C steps from
room temperature up to 600°C. The specimen and the displacement
measurement apparatus are shown in Figs.3 and 4. As shown in Fig.4, the
specimen has two fixed rings, which are 100 mm (4 inch) distance from each
other and made of stainless steel. Each ring is attached to two rods made of
stainless steel, projecting into the outside. The thermal expansion of the steel
rod itself was measured to cancel the thermal expansion of the rod, and then
calculate the concrete deformations. This method compensates for mechanical
errors, which originate from the different thermal coeficients between the rod
and the specimen.
EXPERIMENTAL RESULT AND DISCUSSION
Residual comDressive strength and exdosive sDalling

Compressive test results of before and after the heating stage are
summarized in Table 2. The residual compressive strength rates, Z were
calculated with the following equation;
Z=SaíSb X lOO(%) (2)
@Seismicisolation
@Seismicisolation
where Sa: average compressive strength after heating @Pa)
Sb: average compressive strength before heating @Pa)
From the table, the effect of the PP fiber reinforcement on the residual
strength is clearly shown by comparing the results of plain concrete mixture
(OPC45-wet) and mixture of fiber reinforced concrete (PP45-wet). It is found
that the PP fiber addition improves the residual strength from 39.5% to 49.6%.
From the test results of PB25-wet and PP45-wet, it can be estimated that the
polymer beads addition improves the residual strength but less than that of the
PP fiber. In the same manner, the effect of the moisture content on the residual
strength can be shown by comparing the results of PP45-dry and PP45-wet. As
for the high-strength fiber reinforced concrete, the residual strength shows
higher value with the higher moisture content.
The observations of surface cracking and explosive spalling were also
shown in Table 2. Surface cracks were observed in all the specimens. Only
two of three OPC45-wet specimens had the explosive spalling. This indicates
that the addition of polypropylene fiber as well as the polymer beads can
mitigate the explosive spalling. This implies that the geometry of the melting
material (fibers or beads) is not of great importance for the mitigation of
explosive spalling, and that the effective action against the explosive spalling
is to melt and form voids that allow vapor pressure reductions.
TemDerature distribution and thermal diffusivitv

Figure 5 shows the relationship of the surface temperatures and the
internal temperatures of the specimens. The specimen with the dry condition
(PP45-dry) showed the biggest difference temperature between surface and
internal temperature among the other specimens at wet conditions. At the
same surface temperature of OPC45-wet and PP45-wet specimen, the internal
temperatures show different values. As an example, at the surface temperature
of 300°C, the internal temperature of OPC45-wet and PP45-wet show a
difference of about 200"Cand iOO"C, respectively. This significant difference
in thermal conductivity is attributed to the addition of PP fibers, since all of
the PP fiber mixtures (PP25-wet, PP45-dry and PP45-wet) showed identical
behavior. From Fig.5, it is also seen that the internal temperatures of PP fiber
mixtures increase gradually up to 170-190°C, above which they increase
abruptly. This observation suggests that the PP fiber melting system is
operating since the melting temperature of PP fiber is coinciding with the
turning-point temperature. In the other hand, the effect of polymer beads is not
clearly observed from this figure, since the thermal conductivity behavior of
PB25-wet shows the same as that of plain concrete (OPC45-wet). The effeq
of the moisture content on the thermal property is also shown in this figure. At
the same surface temperature, the specimen with high moisture content (PP45-
wet) showed higher internal temperature, which may increase the potential for
explosive spalling.
Figure 6 shows the summary of the thermal diffusivity calculations.
From this figure, it is shown that the specimen with low moisture content
@Seismicisolation
@Seismicisolation
58 Horiguchi et al.
(PP45-dry) showed the lowest thermal diffusivity. This implies that the lower
moisture content showed the lower thermal diffisivity. This result meets good
agreement with the result of Fig.5, which indicate that the specimen with the
dry condition (PP45-dry) had the biggest difference temperature between
surface and internal temperature. In the same manner, it is shown that the
specimens with higher water to cement ratio showed the lower thermal
diffusivity. Ail thermal diffisivities of the specimens with PP or PB in wet
condition decreased significantly with increasing heating temperature up to
350 C, and stayed constant. It is well known that the thermal diffisivity is
affected by the moisture content and by the density. From the test results this
is confirmed in the case of high-strength fiber reinforced concrete. Moreover,
it is found that the thermal diffusivity of polypropylene fiber reinforced
concrete decreased significantly with increasing temperature after the melting
point of the fiber.
Longitudinal strain develor>ment

Figure 7 shows the longitudinal strain development due to elevated
temperature. It is shown that the longitudinal strain in the test specimens
increased when the furnace temperature had risen. The highest strain was
obtained with the dry mix condition (PP45-dry). Among the other mixes of
wet-condition, plain concrete (OPC45-wet) showed the highest ultimate strain
and the others showed similar ultimate strain. From these results, it can be said
that the lower moisture content showed the higher strain, and that the addition
of polypropylene fiber or polymer beads decreased the thermal strain. The
effect of water to cement ratio on the longitudinal strain didn’t show a
significant difference. However it seems that the specimen with the lower
water to cement ratio showed the lower longitudinal strain.
CONCLUSIONS
From the experimental test results, the following conclusions were

made:
1) Polypropylene short fibers added in the mixture improve the residual
strength of high-strength concrete after a high temperature test. The
effect of polymer beads addition on improving the residual strength is
the same as the addition of polypropylene fibers, but is less in
magnitude than that of the PP fiber. As for the high-strength PP fiber
reinforced concrete, the residual strength shows higher strength
retention with the higher moisture content.
2) Based on the observation of both surface cracking and explosive

spalling after the heating test, surface cracks were observed in all the
specimens. However only two of three OPC45-wet specimens failed
by explosive spalling. This indicates that the addition of
polypropylene fiber as well as the polymer beads mitigates the
@Seismicisolation
@Seismicisolation
explosive spalling.
From the test results, it is confirmed that the thermal diffusivity of the
high-strength fiber reinforced concrete decreases as the moisture
content decreases, and as the water to cement ratio increases.
Moreover, it is found that the thermal diffusivity of polypropylene
fiber reinforced concrete decreased significantly with increasing the
temperature after the melting point of the fiber.
Addition of the polypropylene fiber or polymer beads in high-strength

concrete decreases the ultimate thermal strain.
The highest thermal ultimate strain was obtained with specimen of

low moisture content. The lowest ultimate strain was observed with
specimen containing polymer beads. From these results, it can be said
that the lower moisture content showed the higher strain.
REFERENCES
1. Menzel, C.A., “Tests of the Fire Resistance and Thermal Properties Solid
Concrete Slabs and Their Significance,” Proceedings, American Society
for Testing And Materials (ASTM), Vo1.43, pp. 1099-1153, 1943
2. Malhotra, H.L., “The Effect of Temperature on The Compressive Strength
of Concrete,” Magazine of Concrete Research, Vo1.8, N0.22, pp.85-94,
1956
3. Saemann, J.G., and Washa, G.W., “Variation of Mortar and Concrete
Properties and Temperature,” AC1 Journal, Proceedings Vol.54, No. 5,
pp.385-397, 1957
4. Takano, T, Shimura, K., Horiguchi, T. and Saeki, N., “Deformation and
Deterioration Behavior of High Strength Concrete under Elevated
Temperature,” Transactions of the Japan Concrete Institute, V01.22,
pp. 105-112,2000
5. Takano, T., Horiguchi, T. and Saeki, N., “ Residual Properties of High
Strength Fiber Reinforced Concrete Exposed to High Temperature,” 4th
Joint Symposium on the Structural Materials between Korea & Japan,
pp.73-78,2000
6. Ali, F.A., Connolly, R. and Sullivan, P.J.E., “ Spailing of High Strength
Concrete at Elevated Temperatures,” Journal of Applied Fire Science,
V01.6, No.1, pp.3-14, 1996-97
7. Noumowe, A.N., Clastres, P., Debicki, G. and Costaz, J.L., “ Thermal
Stresses and Water Vapour Pressure of High Performance Concrete at
High Temperature, 4’h Int. Symposium on Utilization of High-Strength/
High-Performance Concrete, pp.561-570, 1996
8. Harmathy, T.Z.,” Effect of Moisture on the Fire Endurance of Building
Element,” ASTM STP-385, 1965
@Seismicisolation
@Seismicisolation
60 Horiguchi et al.
9. Phan, L.T., “Fire Performance of High-Strength Concrete: A Report of the
State-of-the-Art,” NISTIR 5934, NIST 1996
10. Jensen, B.C. and Aarup, B., “ Fire Resistance of Fibre Reinforced Silica
Fume Based Concrete,” 4” Int. Symposium on Utilization of High-
StrengthíHigh-PerformanceConcrete, pp.55 1-560, 1996
11. Hoff, G.C. ‘‘ Fire Resistance of High-Strength Concrete for Offshore
Concrete Platforms”, Concrete in Marine Environment, AC1 SP-163,
pp.53-88, 1996
12. Bilodeau, A., Malhotra, V.M. and Hoff, G.C.,” Hydrocarbon Fire
Resistance of High-Strength Normal-Weight and Lightweight Concrete
Incorporation Polypropylene Fibers, 4* CANMET/ACI/JCI Int. Conf. On
Recent Advances in Concrete Technology, pp. 505-530, 1998
13. Takano, T., Shimura, K., Horiguchi, T. and Saeki, N., “ Fire Resistance of
High-strength Mortars with Several Types of Fibers and Polymers,” 3rd
Joint Symposium on the Structural Materials between Korea & Japan,
pp.65-72,2000
14. Khouny, G.A., Sullivan, P.J.E. and Grainger, B.N., “ Radial Temperature
Distributions within Solid Concrete Cylinders under Transient Thermal
States, “ Magazine of Concrete Research, Vo1.36, No.128, pp. 146-156,
1984
15. Sullivan, P.J.E. and Khourry, G.A., “ Fire Performance of Various
Concretes Deduced from Strain Measurements during First Heating,” AC1
SP-92, pp. 175-196, 1986
16. Barragan, B., Maio, A.D., Giaccio, G., Traversa, L. and Zerbino, R.,
“Effect of High Temperature on Residual, Mechanical, and Transport
Properties of Concrete, 5” Int. Conf. on Durability of Concrete, AC1 SP-
192, pp.983-1000,2000
@Seismicisolation
@Seismicisolation
Table 1 : Mix proportion of test specimen
PP45-Wet 378 170 50 0.45 PP = 0.25 5.1

pP45-D~ 378 170 50 0.45 PP = 0.25 3.4
PP25-Wet 680 170 50 0.25 PP = 0.25 6.6
PB25-Wet 680 170 50 0.25 PB = 0.25 6.0
Table 2 : Summary of test results
1 I 1
@Seismicisolation
@Seismicisolation
62 Horiguchi et al.
Fig. 1 Electric furnace
-
o
L 600
700
I I
O 60 120 180
Heating time (minutes)
Fig.2 Illustrative heating curve
@Seismicisolation
@Seismicisolation
Fig. 3 Thermocouples in specimen
Fig. 4 Deformation measurement
@Seismicisolation
@Seismicisolation
Horiguchi et al.
600 I -
300
200
1 O0
O
O 1O0 200 300 400 500 600
Surface Temperatures@)
Fig. 5 Relationship between internal and surface temperature
-f 1816
N
14
4 12
10
.-).
a-.
.> v)
8
2 6
r:
U
3 4
i 2
e n
" __________
~~
150 250 350 450 550

Furnace temperatures r C )
Fig. 6 Thermal diffusivity and furnace temperatures
2. 5
-s? 2
1
.-c
-
21.5
-ulm
c
5 1
3
.-
c
F
so. 5
O
O 1 O0 200 300 400 500 600
Furnace temperature ("Cl
@Seismicisolation
@Seismicisolation
Fig. 7 Longitudinal strain and furnace temperature
SP 209-5
The 1999 Tehuacán (Mw = 7.0) and

Oaxaca (Mw = 7.5) Mexican
Earthquakes: lessons learned
by S. M. Alcocer, C. Reyes, D. Bitrán,
O. López-Bátiz, R. Durán-Hernández,
and L. Flores
Svnopsis: On June 15,1999, a magnitude Mw = 7.0 earthquake occurred with epi-

center located southwest from the city of Tehuacán, Puebla, and on September 30,
1999, a Mw = 7.5 magnitude earthquake occurred with epicenter located northeast
from the city of Puerto Escondido, Oaxaca, Mexico. The earthquakes affected a
wide geographical area, including seven states. Few casualties were recorded. In
the first earthquake, three-quarters of the total losses occurred in the State of
Puebla, 15 percent in the State of Oaxaca and only the remaining ten percent in the
States of Morelos, Mexico, Tlaxcala, Veracruz and Guerrero; the second earth-
quake only affected the State of Oaxaca. Historical buildings, dwellings, hospitals,
schools and roads were mostly impacted. Some damages were also experienced in
bridges and lifelines. The economic activities were not severely impaired. This pa-
per summarizes the seismological characteristics of the event, its economic impact
and the emergency measures adopted by the authorities. Observations made during
several visits to the affected regions of Puebla and Oaxaca on the behavior of soils
and foundations, lifelines, schools, dwellings, historical monuments and commer-
cial buildings are described. Recommendations for the development of vulnerabil-
ity reduction programs are presented.
Kevwords: columns; ea rthq ua ke engineering; historic a I

bu iidings; hos p ita Is; hous¡n g; ma son ry; reinforced
concrete; repair; schools; strengthening
@Seismicisolation
@Seismicisolation
65
66 Alcocer et al.
Sergio M. Alcocer, FACI, is a research professor at the Inst. of Engng. of the
Natl. Univ. of Mexico (UNAM) and Director of Research at the Natl. Center for
Disaster Prevention (CENAPRED). He is a member of AC1 Committees 318,
Building Code Requirements, and 374, Performance Based Seismic Design of
Concrete Buildings, as well as a member of Joint ACI-ASCE Committee 352,
Joints and Connections in Monolithic Concrete Structures. His research interests
include earthquake-resistant design of reinforced concrete and masonry
structures.
Carlos Reyes is Head of the Structural Engineering and Geotechnical Area

CENAPRED. He holds M.Sc. and Ph.D. degrees from UNAM. His research
interests include the seismic risk assessment, seismic behavior and rehabilitation
of reinforced concrete structures.
Daniel Bitrán is Head of the Area of Socioeconomic Studies at CENAPRED.

He holds B.Sc. and M.Sc. degrees in Economics from the University of Chile.
His professional and research interests include the social and economic
consequences of disasters.
Oscar López-Bátiz is a Research Engineer in the Structural Engineering and

Geotechnical Area at CENAPRED. He holds B.Sc. and M.Sc. degrees from
UNAM, and a Ph.D. degree from the University of Tokyo. His research interests
include the seismic behavior of reinforced and precast concrete structures.
Roberto Duran-Hernández is a Research Engineer at CENAPRED. He holds

B.Sc. and M.Sc. degrees from UNAM, where he is a Ph.D. candidate at present.
His research interests include the seismic behavior and rehabilitation of
reinforced concrete structures.
Leonardo Flores is a Research Engineer in the Structural Engineering and

Geotechnical Area at CENAPRED. He holds B.Sc. and M.Sc. degrees from
UNAM. He is member of the committee on Masonry Standards of the Mexico
City Building Code. His research interests include the seismic behavior and
rehabilitation of reinforced concrete and masonry structures.
INTRODUCTION
The June 15 and September 30, 1999 earthquakes (Mw 7.0 and 7.5, respectively)
mainly affected the Mexican states of Puebla and Oaxaca, respectively. Just over
300 million dollars were granted for reconstruction of damaged structures and
infrastructure. As has occurred in past events, most damaged sectors were
@Seismicisolation
@Seismicisolation
housing, historical buildings and roads. These sectors used 75% of the total funds
spent in reconstruction (220 million dollars). The remainder was assigned for the
rehabilitation of power and water supply lines. With regards to reinforced
concrete structures, losses were 23% of the total, and were mostly concentrated
in schools and hospitals.
in order to better understand and identifj the social and physical vulnerabilities
of the different regions affected by the earthquakes, a reconnaissance team of
researchers of the National Center for Disaster Prevention, CENAF'RED, visited
the most damaged villages and cities. This paper is aimed at summarizing the
observations of the type and causes of distress; emphasis is given to reinforced
concrete and masonry structures. Detailed reports can be found elsewhere (1 &
2).
SEISMIC HAZARD IN MEXICO
Mexico is a country exposed to the effects of natural phenomena of different

kinds that could cause disasters. This is attributed to its geographical position,
volcanic and seismic activities, as well as to the characteristics of its weather,
topography and hydrology. Regarding the earthquake phenomena, most seismic
events are the result of the interaction of tectonic plates; this is the case of the
subduction zone of the Cocos and Rivera plates under the North American plate,
and the Pacific Plate that displaces laterally with respect to the North American
plate along the Baja California Peninsula (Fig. la). Normal fault earthquakes
within the Cocos plate are less frequent. For seismic design purposes, Mexico is
divided into four large regions (Fig. lb). Region A has the lowest hazard,
whereas region D corresponds to the highest hazard. Design peak ground
accelerations for soft soils in regions A to D are 0.05, 0.10, 0.64 and 0.86 g,
respectively (3). Design maximum spectral accelerations for soft soils in regions
A to D are 0.20, 0.36, 0.64 and 0.86 g, respectively. Maximum design spectra
accelerations and size of the spectra plateau vary with the soil characteristics.
Evidently, such zoning does not consider site effects, similar to those observed in
alluvial valleys and in the valley of Mexico. For those locations, micro-zoning
maps are available. It is not uncommon to find zones of highest hazard due to
site effects, located hundreds of kilometers from the coast, that are comparable to
those along the Pacific coast. Mexico City is a prime example of this situation.
Seismological Characteristics of the June 15 and September 30, 1999

Earthuuakes
The main seismological characteristics of the earthquakes are summarized in

Table 1. Normal faulting within the Cocos plate produced both earthquakes.
@Seismicisolation
@Seismicisolation
68 Alcocer et al.
According to reference 4, up to 1999, 10 earthquakes M6.5 or larger had
occurred in the epicentral regions of the June 15 and September 30 events. Based
on this, the average return period of M6.5 events is of the order of 20 to 30 years.
A detailed analysis of the recorded seismographs indicates that the number of
aftershocks was small (which actually is typical of normal fault earthquakes) and
that some directivity towards the Northwest of the epicenter existed. For the
September 30 event, the number of aftershocks was considerably larger; it is
believed that this anomaly should be credited to the lower depth of the focus.
Based on the damage recorded, isoseismal maps were produced for both
earthquakes (1, 2 & 4). The shape of the curves for the June 19 event also
indicates some directivity along a NW-SE axis. This is consistent with the
analysis of seismic records.
Although their magnitude was a little different, the highest Modified Mercalli
Intensity index for both events was VIII. However, for same MMI‘s, that is for
same level of damage, the peak ground acceleration for the September 30 event
was about twice that generated by the June 15 motion (Fig. 3). This has been
credited to constructions of higher strength and superior quality in the State of
Oaxaca, where the epicenter ofthe second earthquake was located.
EARTHQUAKE DAMAGE
Most severe damage was concentrated on housing, hospitals, schools and

historical buildings. Damage intensity in the first three was moderate, although
some collapses were recorded. Due to the characteristics and intensity of the
damage, it is important to point out the performance of hospitals and schools for
their impact on disaster management activities (1). The term ‘%istorical
buildings” is used in this paper to refer to churches, palaces, museums, and
others that are of significance for the region and the country, particularly because
of their aesthetics, architecture and historical heritage.
Loss estimations of the direct and indirect economic losses of the June 15
earthquake indicated that total losses amounted to approximately 150 million
dollars. Almost all losses occurred in the State of Puebla, with minor losses in
the States of Morelos, Mexico, Tlaxcaia, Veracruz and Guerrero. This value is
almost equal to the losses estimated for the September 30 event (5). The relative
losses for the different sectors affected by the earthquakes are shown in Fig. 4.
Relative to the socioeconomic consequences of the earthquakes, a very

significant aspect that must be highlighted is that most negative effects (injuries,
material losses, business interruption) were concentrated in the poorest
population in the region. This supports the lesson, several times affirmed, that
@Seismicisolation
@Seismicisolation
poverty is a factor that increases the risk to natural phenomena. Few casualties
were recorded in both earthquakes: 50 deaths and over 1O0 persons injured.
Damage to Historical Buildings
The most noteworthy damage, both for its quantity and intensiîy, concentrated on
historical buildings. It is important to point out that the most severe damage
occurred in the State of Puebla during the Tehuacán earthquake. in past
earthquakes, the large inventory of seismically vulnerable structures had
registered comparable damage, like in 1980 during the Huajuapan de León
earthquake. Thus, damage in historical buildings was not a unique feature of the
1999 event. Additionally, a simple review of the commemorative plaques
installed in several churches, gives a fair idea of the magnitude and type of
damage in previous earthquakes, even those that occur~edduring Colonial times.
in general, damage in historical buildings can be explained by the intrinsic

weaknesses of the masonry materials and structural layout, as well as by
weathering and lack of proper maintenance, among other factors. An schematic
view of typical damage patterns in churches in presented in Fig. 5.
Looking at damage statistics, it is apparent that most vulnerable elements, in

order of frequency, were the towers, vaults, cupolas, arches and walls.
Ornamental objects were also vulnerable. Horizontal cracking at the ends of the
columns of bell towers, as well as inclined cracking in the base of the towers and
buttresses, although at a smaller extent, were significant. The direction of
cracking was from the upper portion of the structure down and away from the
church longitudinal axis (Fig. 6). Horizontal cracks at the base of the pediment,
and vertical and inclined cracks through window and door arch keystones in the
façade were also observed.
Less common were the vertical cracks along the joint of the towers and the
frontispiece, as well as the cracking between windows of the drum and along the
meridians of the cupolas. Longitudinal cracks along the vault and arch keystones
were recorded in most churches visited and for which access to the interior was
allowed.
in the State of Oaxaca, the September 30 earthquake inflicted, comparatively,. a

lower amount of damage to historical buildings. in general, this category Öf
buildings in the lower two-thirds of the state, such as churches and convents, are
shorter and have more robust walls and buttresses than those in the State of
Puebla. Furthermore, bell towers are also shorter.
@Seismicisolation
@Seismicisolation
70 Alcocer et al.
Damage to Housing
As it has been typical in past earthquakes in Mexico (e.g. the 1985 Michoacán,
1995 Chiapas and Manzanillo, and 1997 Arteaga earthquakes) housing was one
of the hardest hit sectors in both events. Damage types are also very similar to
those recorded in previous earthquakes. Plain masonry dwellings, including those
made of adobe walls, were mostly affected. Damage was also recorded in
buildings with inadequate lateral force resisting systems, such as those with soft
stories at the ground level, as well as in structures in which, due to an assumed
incorrect interpretation of buildings standards, the amount of walls was low and
the construction had some gross errors. In very few cases, the superior
performance of rehabilitated (strengthened) houses by new confinement elements
was observed. This was the case of plain brick and adobe masonry houses whose
walls had been rehabilitated by adding reinforced concrete tie-columns and bond
beams. It was the aim of this strengthening technique to tie the walls together, as
well as to increase the in-plane tensile strength of the masonry wall.
One possible explanation of the high risk found in the housing sector is that, at
present, a significant number of States are located in zones of high and very high
seismic hazard, i.e. zones C and D, respectively (Fig. Ib), which, in turn, have a
large number of houses built with substandard materials. in Fig. 7, the percentage
of houses built with durable materiais is shown per each State (6). According to
this data, durable materials for walls are stone/brick/blockmasonry and concrete;
non-durable materials are bahareque, adobe masonry, timber and cardboard
sheeting, typically used in very poor neighborhoods and communities within
Mexico.
Damaee to Schools
Although the performance of most school structures was adequate, there was a
significant number damaged due to construction deficiencies and/or structural
design (1 & 2). No new lessons were to be learned, causes of damage, as well as
the basic criteria for school rehabilitation are well known. Damage in schools
designed with old codes was relevant; such codes have less stringent
requirements than those in present regulations.
In the older part of towns, damaged buildings with inadequate structural systems.
for earthquake resistance and unacceptable detailing for ductile behavior were
observed; these structures have been used, after some modifications, as schools
over the past decades, and even centuries. The lack of proper maintenance of this
type of buildings is apparent; leaks, growth of vegetation inside the cracks, and
unstable parapets are distinctive. The classical failure of short columns due to the
presence of partial-height masonry walls was observed in the long direction of
school buildings. Typical damages in schools are shown in Fig. 8. Common
@Seismicisolation
@Seismicisolation
rehabilitation schemes included replacement of the masonry walls by reinforced
concrete walls in the short direction. in the long direction, due to operational
constraints, reinforced concrete jacketing of frame elements was the option most
often selected.
Damape to health facilities
in regards to health facilities, although few were damaged, the type and level of
damage was noteworthy, particularly in those that presumably were designed and
built in accordance with modem building codes (Fig. 9). in fact, after the very
severe damage in this type of facilities in the 1985 Mexico earthquakes, it had
been assumed that new hospitals had been designed and constructed to limit the
non-structural damage, i.e. partitions had been separated from ñ-ame members.
Also, that the seismic safety of large hospitals had been evaluated.
In the city of Puebla, two hospitals were the most damaged Hospital General
San Alejandro and Hospital de Traumatología y Ortopedia, both from the
Mexican institute of Social Welfare (IMSS). in both hospitals, partition walls
participated in resisting the lateral loads, so that under cyclic displacements,
nonstructural and structural damage was apparent. in the surgery building of
H.G. San Alejandro inclined cracking in the RC columns occurred due to the
interaction with partition walls under lateral demands. Although partition walls
were separated &om the frame members, the joint was filled with a flexible
wood-based material. This material, highly absorbent of water, impregnated from
the masonry mortar of the partitions, losing its flexibility. Therefore, for
structural purposes, the partitions were attached to the columns (7). A detailed
seismic evaluation study of the hospital was carried out; it included the
development of site-specific design spectra and measurement of the fundamental
vibration characteristics of the buildings. Only the damaged two-story surgery
building was rehabilitated in both directions through steel Chevron braces and
jacketing of columns made of steel angles and straps (7). Diagonals were made
of 324-mm diameter circular pipes with 14.3-mm and 12.7-mm thickness at the
ground and first floor, respectively. Braces were connected to the existing
concrete beams through 32-mm steel dowels embedded into new concrete
elements added on the beam sides. Only the 450-mm square columns at the
ground story were jacketed. Steel jackets consisted of 152 x 152 x 10 mm angles
and 1O0 x 1O mm straps spaced at 400 mm.
in the case of the Hospital de Traurnatologia y Ortopedia, hospital authorities

considered that the widespread distribution and type of damage affected the basic
functions and decided to evacuate the hospital completely. Even though the
hospital did not suffer significant structurai damage, fiom a disaster management
point of view, the facility failed because it could not provide the care for which it
was designed (Fig. 9). in the figure it is apparent the nonstructural damage in the
@Seismicisolation
@Seismicisolation
72 Alcocer et al.
façade of the building, as well as a surgical table that tilted over during the
quake.
In the State of Oaxaca, similar damages were observed. Damage attributed to an

inappropriate application of codes and standards, as well as to an inadequate
selection of the construction site were also observed. In Figs. 10 and 11, damage
in two facilities in the State of Oaxaca is shown. The clinic in Fig. 10 had not
been open to the public when the quake occurred; the hospital unit in Fig. 11 was
built on an inadequately compacted fill made on a steep slope.
It is important to recall that health care and education facilities, as well as all
other infrastructure needed for emergency management, must be designed, both
at the structure and content levels, to remain operational under moderate
(frequent to rare events) and under the design earthquake (very rare events)
expected in that region. This implies that the probability of evacuation of health
care facilities is very small.
Mexican building codes are intended to satisfy the abovementioned performance

objectives through the use of an importancefactor. The magnitude of such factor
varies from 1.33 to 1.5. However, it is not clear what is the effect of this factor
on the design, particularly for moderate seismic demands. in general, the
established practice for the design of hospitals in the country consists of the
verification of the strength and deformation capacities of the structure under the
design earthquake; however, the demands for frequent-to-rare events are seldom
explicitly considered. Even more, the structural members of these facilities are
detailed similarly to members in which large inelastic deformations and
concentrated damage is expected. The design and detailing philosophy
contradicts the performance objective established for emergency facilities.
Rehabilitation of Historical Buildings
As a consequence of the large number of historical buildings damaged in the

States of Puebla and Oaxaca, rehabilitation programs were immediately
implemented. Distinctly different techniques were used. In general, two
performance objectives were established:
a) In some cases, the performance objective was to restore the conditions

that the buildings had prior to the earthquake, but without improving its strength,
deformation capacity or its stifíñess. Thus, the monuments were just repaired.
Generally, such objective was satisfied by repairing the cracks and by
substituting crushed and spalled materials with similar materials.
b) The second performance objective of the rehabilitation implied

strengthening of the building. Besides restoring its original strength, additional
@Seismicisolation
@Seismicisolation
capacity (Le. load carrying, stiffness, deformation or a combination of them) was
provided. It is the aim of this performance objective that under similar or larger
events, less damage would be inflected and that this damage could be repaired at
reduced costs. In general, strengthening of historical buildings implies
nonreversible processes, i.e. techniques and schemes that once applied, are
difficult or impossible to remove. Therefore, these techniques are pretty much
permanent. One example of this would be the addition of exterior buttresses, well
anchored to the existing structure.
A significant aspect of the rehabilitation procedure (both repair and

strengthening) was the lack of complete and systematic information about
previous rehabilitation works. Such information would have been very valuable
to improve the decision making process, by providing evidence on the techniques
that had exhibited adequate performance, and on those that must be avoided.
Moreover, a databank of this kind would aid restorers to better comprehend the
building response, as well as to better define the rehabilitation schemes.
Although, in general, limited and simple modifications are the basic premise for
the restoration of a historical building, the fact that large amounts of damage and
costly repairs occur should be considered. The simple restoration of the original
condition (first performance objective) implies the acceptance that, in future
events of similar or larger intensity, at least comparable damage would be
recorded. For the region under study, this means that every 20 to 30 years
(assuming that this is the average return period of intense earthquakes), a similar
rehabilitation program, with comparable or, even, increased costs would have to
be implemented. The rehabilitation of the building, using nonreversible
techniques, offers the advantage of increasing the level of safety under larger
earthquakes.
In economical terms, a major investment for rehabilitation of buildings to satisfy

the second performance objective would allow the use of funds to reduce the
seismic vulnerability of housing, schools and health care facilities. Even though
the preservation of the architecturai, historical and aesthetic heritage of the
country should not be negotiated, the negative cycle of repairs and associated
expenses should be avoided. The socioeconomic condition of the country forces
the adoption of compromise solutions, that is, to maintain or slightly improve the
building performance without altering the appearance of the building.
RECOMMENDATIONS DERIVED FROM LESSONS LEARNED
Departing from the fact that the most efficient approach for risk reduction and
disaster prevention is to reduce the vulnerability of damageable systems through
Vulnerability Reduction Programs (VRP’s), the lessons observed allowed the
development of the following recommendations. A conservative estimation of
@Seismicisolation
@Seismicisolation
74 Alcocer et al.
the benefit-to-cost ratio (or effectiveness-to-cost ratio) of W ' s is of the order
of five, but in the health care system, values up to eight have been recorded (8).
Since the effects of disasters are negatively concentrated in the most vulnerable
strata of the population (poor people, children, women, elder and handicapped
persons), it is recommended the inclusion, in al VRP's, of means to improve the
level and quality of life of beneficiaries. Along this thinking, the participation of
the population in all stages of the program should be encouraged; their
contributions should be reflected by naming them as benefìciaries and not as
victims. in the design and implementation of such programs, persons with
distinctly different levels of education, expertise, interests, and even cultural
background, should be recognized. With regards to structural safety, it is
important to accept up front that it may not be possible to fulfill all design and
detailing requirements established in codes for new construction. Examples of
the latter are historical buildings and non-engineered construction.
General Recommendations for the Implementation of Vuinerabiiitv

Reduction Programs
It is recommended that VRP's be organized with education and training

subprograms for masons at the community level. The application of these
courses is aimed at improving the local construction practice by incorporating
techniques of the same kind, but of superior quality.
Robust and permanent technical support should exist at the community level.
This support, at the design and implementation phases of the program, should
come from the local academic and professional sectors.
Structural solutions must be compatible with the local practice. Furthermore, its
complexity must be consistent with the level of training of the work force in each
region. Solutions that involve clockwork detailing for simple structures should be
avoided; it is recommended the participation of Toca1 people in the design of the
w.
It is well known that after a major disaster, the sensitivity of the population
towards improved safety and more information about the phenomena, its
consequences and mitigation measures, increases. Then, it is recommended to
make use of such periods to implement specifically tailored programs. It is
advisable to carry out pilot programs in small towns; these programs would
provide feedback for the development of programs at a larger scale. Cutting-edge
technology must be used as much as possible; remote sensing, satellite imagery
and GIS are technologies readily available.
@Seismicisolation
@Seismicisolation
Recommendations for Housing
In this area, there is a large wealth of information and technical experience on

vulnerability assessments and structural rehabilitation. It is advisable to adapt
such information and expertise as soon as possible, as well as to disseminate it,
among emergency managers and among the population living in moderate and
very high seismic hazard areas. Information on how to tie together the masonry
walls, to increase the in-plane masonry tensile strength, and to improve the
durability of masonry is available (Fig. 12) (9).
Firstly, it is essential to develop regional or local programs, both in urban and

rural areas, to identi@ the housing types and to assess its vulnerability. With
regards to low-cost housing, it is necessary the development and distribution of
damage evaluation and rehabilitation guides. Similarly, for the most part,
information is accessible.
Recommendations for Schools and Health Care Faciüties
The development of an inventory of schools and hospitals located in the zones of

highest seismic hazard in the country, and the evaluation of vulnerability of
prototype schools and hospitals must be done immediately.
It is desirable to develop andíor adapt rehabilitation guidelines akin to the

Mexican design and construction practices, as well as to adopt a performance-
based design approach. In such criteria both the structural system and building
contents must be included, especially for the case of hospitals. It is convenient to
revise the concept and magnitude of the importance factor, in light of the
performance objectives established. Analytical and experimental studies on
hospital prototypes aimed at assessing the building performance (including
structural and nonstructural components) under earthquakes of distinct intensity
and recurrence should be performed at once.
Recommendations for Historical Buildiu~zs
Firstly, it must be recognized that general solutions for rehabilitation of historical

buildings are not easily applied. Furthermore, rehabilitation should be performed
on a case-by-case basis. However, such statements must not be interpreted as to
cancel or delay the need for developing general and simple evaluation anil
rehabilitation guidelines.
It is largely advisable to develop a national geo-referenced databank of the

structural characteristics of historical buildings. Although in recent times, large
efforts have been devoted to cataloging aesthetic and architectural features of
@Seismicisolation
@Seismicisolation
76 Alcocer et al.
these buildings, almost nothing has been done from the structural engineering
point of view. As a complement, it is relevant to develop a catalog of previous
rehabilitations, in which the level and amount of damage, as well as a rating of
their level of success or failure are included. This register would help to identifv
the adequacy of rehabilitation schemes.
Based on national and international experiences, it is recommended the

development of assessment and rehabilitation guidelines. Such guides would
help as a technical source for the development of specific VRF”s.
Finally, the development of programs aimed at rehabilitating historical buildings

at highest risk of with highest aesthetic, architectural and historic value should be
carried out. The active participation of professional restorers, historians and
engineers, among others, should be promoted. It is not sustainable for the country
to periodically cope with the large expenses involved in restorations. Thus,
strengthening efforts must be encouraged so that, over time, the cost-benefit
ratios would favor risk mitigation.
ACKNOWLEDGMENTS
The support of all Municipal, State and Federal authorities of the National
System of Civil Protection is gratefully acknowledged.
REFERENCES
Alcocer S.M., Aguilar G., Flores L.E., Bitrán D., Durán R., López-Bátiz
O., Pacheco M.A., Reyes C . , Uribe C.M., and Mendoza M.J., “El sismo
de Tehuacán del 15 de junio de 1999,” (in Spanish), Report IEGl02199,
Centro Nacional de Prevención de Desastres, ISBN 970 628 601-2,
December, 2001,159 pp.
López-Bátiz O., Reyes C., Durán R., Bitrán D. and Lermo J., “El sismo
de Oaxaca del 30 de septiembre de I999 @,,,=7.5)”, (in Spanish), Report
IEG/06/01, Centro Nacional de Prevención de Desastres, ISBN 970 628
609-8, December, 2001,71 pp.
Comisión Federal de Electricidad (CFE, 1994), ‘‘Manual de Obras
Civiles. Diseñopor Sismo, (in Spanish), Mexico, D.F.
”
Singh S.K., Ordaz M., Pacheco J.F., Quaas R., Alcántara L., Alcocer S.;-
Gutiérrez C., Meli R., and Ovando E., “A Preliminary Report on the
Tehuacán, México, Earthquake of June 15, 1999,” Seismological
Research Letters, Seismological Socìeo of America, Vol. 70, No. 5 ,
Sept.-Oct., 1999, pp. 489-504.
@Seismicisolation
@Seismicisolation
5. Bitrán, D. “Evaluación del impacto socioeconómico de los principales
desastres naturales ocurridos en la República Mexicana durante 1999’:
(in Spanish), Cuaderno de Investigación No. 50, Centro Nacional de
Prevención de Desastres, November, 2000,194 pp.
6. Instituto Nacional de Estadística, Geografia e Informática (INEGI), XII
General Census of Population and Housing, 2000. www.ineczi.gob.mx
7. Garcia Jarque Ingenieros. “Dictamen, conclusiones y recomendaciones
estructurales del Hospital General de Zona 35, San Alejandro, del IMSS
en Puebla, Pue., (in Spanish), August 1999,45 pp.
”
8. Programa Especial para la Prevención y Mitigación del Riesgo de

Desastres en México 2001-2006,124 pp., www.cenaured.unam.mx
9. Centro Nacional de Prevención de Desastres, “Métodos de reJUerzopara
la vivienda rural de autoconstrucción, ”(In Spanish),
www.cenapred.unam.mx
TABLE 1 - SEISMOLOGICAL CHARACTERISTICS (4)

Focus location
Event De h. km Magnitude Local time
LaiRude, O N LongRude, ”w
June 15.1999 18.20 97.47 60 to 80 7.0 15:41:06
September 30,1999 7.5 11:31:00
Harvard CMT 16.230 96.970 53.9
NEIC CMT 16.148 96.708 53.0
üniv ofMichigan 16,100 96.700 42.0
Kikuchi 16.100 96.700 55.0
UNAM 16.000 97.020 39.7
@Seismicisolation
@Seismicisolation
78 Alcocer et al.
Pacific Plate
a. Tectonic setting of Mexico
b. Seismic hazard zones in Mexico

Figure 1 -Tectonic plates and seismic zoning of Mexico according to the
Federal Commission of Electricity (3)
@Seismicisolation
@Seismicisolation
106' 104' 102' 100' 98' 96' 94' 92.

Longitude W
a. Isoseismal curves of the June 15 earthquake
b. Isoseismal curves of the September 30 earthquake
Figure 2 - Isoseismal maps for the 1999 Mexico earthquakes (1, 2 ¿A 3)
@Seismicisolation
@Seismicisolation
80 Alcocer et al.
P
70
v)
æ 60
-8 50
.-
*E 40
f
<n
30
J 20
-
3 10
+ o
Housing School Hospital Churches and Roads and
Buildings Buildings Public Bridges
Buildigs
Figure 4 - Losses from the Tehuacan and Oaxaca earthquakes
@Seismicisolation
@Seismicisolation
I façade
In ornaments
Figure 5 - Typical damage in historical buildings (churches) due to

theTehuach earthquake (1)
Tensile diagonal
cracking
I I
Unsupparted suie Tower support6d
by the facade
Figure 6 - Possible failure mechanisms responsible for the cracking

of bell towers, bases of towers and arches in the façade (1)
@Seismicisolation
@Seismicisolation
82 Alcocer et al.
...........
...........
........... 76%-100%
-
60 % 75.9 %
0 45%-55.9%
[7 30%-44.9%
0%-29.9%
Figure 7 - Percentage of houses, per state, made of walls built

with durable materials and with concrete or masonry roofs (6)
@Seismicisolation
@Seismicisolation
a. Inadequate structure for a school building
b. Column shear failure in a school building
Figure 8 -Typical damage in school buildings
@Seismicisolation
@Seismicisolation
-- __
Alcocer et al.
c
b. Surgical
- table that tiItd over
during the earthquake
Figure 9 - Damage in health care facilities due to the

Tehuacán earthquake (1)
@Seismicisolation
@Seismicisolation
Figure 10 - Masonry walls and RC columns damaged in a small

hospitalin Santa María Colotepec, Oaxaca (2)
@Seismicisolation
@Seismicisolation
86 Alcocer et al.
a. View ofa hospital unit built on a loose fill
b. Typical damage in partition masonry walls
Figure 11 - Inadequate selectjon of. the constryction site and

damage tononstructural partitions in a hospital unit in Santa
@Seismicisolation
Catarina Juquila, Oaxaca (2)
@Seismicisolation
Jacketing on both sides
Steel
Place
aroun
wire mesh to foundation

@ First 10-mm mortar cover, 1:%:4% (cement:lime:sand)
@ Wire mesh (welded-wire or chicken wire), 10 stapleslm'
@) Final lCmm mortar cover
Figure 12 - Rehabilitation of adobe houses with wall jacketing (9)
@Seismicisolation
@Seismicisolation
SP 209-6
Shear Stud Performance of

Lightweight Polystyrene Concrete
Composite Sections
by Y. Lin, J. M. Ingham, and J. W. Butterworth
SvnoDsis: An experimental investigation was conducted at the University of Auck-

land to quantify the performance of shear studs embedded in composite ribbed
slabs, formed by casting lightweight polystyrene concrete onto profiled steel deck-
ing. In total, 18 tests were conducted using a new type of push-off test rig, with test
units composed of either lightweight polystyrene concrete or normal weight con-
crete (to facilitate performance comparison). This paper reports and discusses the
main results of those 18 tests. Results indicated that in most cases designs utilizing
lightweight polystyrene concrete had considerable ductility and shear capacity.
Recommended design provisions based on the experimental data are presented.
Kewvords: composite section; ductility; polystyrene

lightweight concrete; shear capacity; shear studs; steel
beam
@Seismicisolation
@Seismicisolation
89
90 Lin et al.
Yang Lin is a Post-graduate student, University of Auckland, New Zealand. He
received his BS (civil) from Harbin Ship Engineering University in 1991 and his
MCom (economics) from Southwest JiaoTong University in 1996.
Jason M Ingham is the Cement & Concrete Association Fellow, and a senior
lecturer in the Department of Civil and Environmental Engineering, University
of Auckland, New Zealand. He received his BE (Hons) and ME (Dist.) from the
University of Auckland, and PhD from the University of California at San
Diego in 1995.
John W Buttenvorth is a senior lecturer in the Department of Civil and

Environmental Engineering, University of Auckland, New Zealand. He
received his BE from the University of Auckland in 1965 and PhD from the
University of Surrey in 1975.
INTRODUCTION
The commercial advantages of lightweight concrete, derived from aspects such

as'reduced.seismic lateral loads, áie well understood. This'has lead to a growing
interest in lightweight concrete in New Zealand. Since 1995 one of the main
concrete suppliers in New Zealand, Firth Industries Ltd (Firth), has pursued
alternative lightweight concrete solutions to satisfy increasing demand within
the New Zealand concrete industry. One such example of high strength
lightweight concrete was expanded shale lightweight concrete employed in the
construction of the Wellington stadium in 1998 [I]. However, organizing and
transporting the specified aggregates made the use of high strength lightweight
concrete uneconomical in smaller projects or those demanding lower strength.
B ST' polystyrene lightweight concrete, using easily transportable aggregates,
was subsequently chosen for further development in New Zealand.
BST concrete was first developed by Australian scientists in 1990 [2]. BST
aggregate can be manufactured either from expanded balls or from recycled
polystyrene granulate, with 3-4 mm diameter expanded polystyrene beads
commonly used for higher strength demand mixtures. The polystyrene is treated
via a coating process with a patented chemical compound, called BM-90/2.
BST concrete with a density from 300 kg/m3 to 1800 kg/m3 and compressive
strength from 0.5 MPa to 25 MPa has superior thermal and insulation properties
and in excess of 4 hours tire rating. BST concrete is also easy to mix, pour and
pump. In practice, BST concrete has been used as a floor or roof topping in
' Named after the organization sponsoring development, BST Holdings Pty Limited, Austraiia.
@Seismicisolation
@Seismicisolation
Australia, and in pre-cast internal or external walling systems in Malaysia.
However, prior to commencement of this study, BST concrete had not been used
as a structural concrete.
In parallel, the recent impact on New Zealand of global market forces has
resulted in growing use of composite steel beams with profiled steel decking and
in-situ concrete floors. Incorporating this increasing demand for composite
construction solutions, BST concrete was proposed as an in-situ concrete to cast
slabs over profiled steel decking and steel beams.
For a composite floor system, achievement of composite action is crucial, and

doubts initially existed about the performance of shear studs when embedded in
polystyrene concrete. This prompted a study at the University of Auckland to
consider the performance of BST concrete cast on profiled steel decking,
focusing on determining the shear capacity and ductile properties of shear studs.
EVALUATION OF SHEAR DEMAND AND CAPACITY
A typical composite floor system involves composite beam spans of 8-12 m, and
composite slab spans of between 2-3 m. A typical composite floor system, in
this case using BST polystyrene lightweight concrete, is shown in Fig. 1.
Demand
The shear demand on headed studs located at the concrete-steel interface derives
from two parts: composite action and diaphragm action. Diaphragm action is
typically less significant, and is generated by transferring lateral loads induced
by earthquake from the floor to the lateral load resisting frames. Usually, the
lateral loads are assumed to be evenly distributed to all studs connecting the
diaphragm to the relevant lateral load resisting elements.
The main component of shear demand is due to composite action, which

depends on the plastic flexural strength of a composite beam in conjunction with
the capacity design philosophy. The shear force generated at the composite
interface is determined by a plastic analysis of the composite beam. For
instance, for partial composite action and a plastic neutral axis in the steel
section, the shear force required of shear studs located between maximüm
bending moment and adjacent zero bending moment locations is:
@Seismicisolation
@Seismicisolation
92 Lin et al.
corresponding to the nominal moment capacity of the composite beam, being
M, =C,e’+C,e (2)
C, = 0.85f:a,bl (3)
A,f, - 0.85fc’alb,
c, = 2
(4)
where e and e’ are the lever arms as shown in Fig. 2(c), a1 is the depth of the
rectangular concrete stress block, bi is the effective width of concrete slab, A, is
the cross-sectional area of steel section, C, is the concrete compression force
resultant, C, is the steel compression force resultant, fy is the yielding strength of
steel, and f: the specified compressive strength of the concrete.
Capacity
Ollgaard, Slutter and Fisher [3] pioneered research to quantify the shear capacity
of shear studs. Using statistical analyses, they derived an equation (see (5)) to
determine the shear strength (qr) of a single shear stud in push specimens in
which the concrete slab did not fail prematurely through splitting, shear or
embedment failure:
where the units are in N and mm, Ash is the cross-sectional area of the stud
shank, f,’ is the specified compressive strength of the concrete, and E, is the
Young‘s modulus of elasticity of the concrete. This equation has been adopted
by many national standards as a basic formula to calculate the nominal shear
capacity of each shear stud. For instance, for a 19-mm diameter headed stud,
embedded in a constant thickness slab cast of 25 MPa normal weight concrete,
qr is 109 kN from textbooks or standards used in New Zealand, Canada, and
USA [4-61, and is evaluated as 95 kN in the British Standard, where
recommendations have been modified to account for recent test data [7].
For shear studs embedded in a ribbed concrete slab, a reduction factor b b has
been established to modify Eq. 5 [ 8 ] :
@Seismicisolation
@Seismicisolation
where N is the number of studs in a rib, 4t is the total stud height, hrib is the rib
height, and Wnb is the mean width of the rib, as shown in Fig. 2(a). However,
more recent research shows that the reduction factor suggested above may
underestimate the negative influences of the profile [9-131.
TEST SET-UP AND TEST UNIT DETAILS
Drawbacks of Eurocode 4 í1994) Method
The standard method of conducting push-off tests is covered in section 10 of

Eurocode 4 (1994) [14], where the test rig, detailing of the push specimen, and
the test procedure have been specified. An example of a push specimen
complying with Eurocode 4 is'shown in Fig. 3 .
There are some potential drawbacks when using the standard method. Firstly,
the eccentric transfer of force between the steel beam and the concrete slab may
induce unexpected forces in the shear connector, which may either increase or
reduce the stud strength. The load applied to the steel section is dispersed to the
base of the element at a distance of h3/2 from the interface, where h3 is the total
slab thickness, forming a couple Fh3l2. When the base of the concrete element
is fixed, a horizontal frictional force H is induced to resist this couple, as shown
in Fig. 3(a). Hence, the push-test induces a compressive force in the shear studs
[15]. When the base of the concrete element is free to slide, both of the two
levels of studs could be under tensile forces, as shown in Fig. 3(d), which does
not occur in an in-service composite beam.
Another inadequacy of the standard test is that the assumed symmetrical

distribution of load from the steel beam into the two nominally identical slabs is
unlikely to occur, especially in the inelastic range. Furthermore, the test Set-up
leads to a stud test in which concrete splitting is suppressed, making it
unsuitable for determining the shear capacity of a stud in a system where global
splitting may occur prior to stud failure.
The inaccuracy of the standard push-off test method has also been reported by
several other researchers [15-181, and alternative testing rigs were used in their
tests.
@Seismicisolation
@Seismicisolation
94 Lin et al.
Test Rip and Units
In this study, the test Set-up adopted corresponded to that previously used at the
Uiliversity of Auckland [19]. The shear force was applied by blocking one side
of the test unit and adding a push force on the other side. The push force and
the reaction force were positioned collinear with the steel-slab interface, such
that the position of the test unit on the test rig had no influence on test results.
The test rig with mounted unit is shown in Fig. 4.
The test units had a consistent slab dimension of 1005x840 mm, were built on a
1.3 m length of universal steel column designated 200UC60, and used profiled
steel decking with a rib height of 55 mm and average rib width of 156 mm.
Headed studs with a shank diameter of 19 mm and welded length of 95 mm or
122 mm were used. A typical shear stud test unit is shown in Fig. 5. Specific
details of the 18 test units are listed in Table i . As shown in Figs. 4 and 5, a
reinforced concrete corbel allowed the shear force to be collinearly transferred.
The reinforcement used in the concrete corbel was designed to be compatible
with the largest force generated in push-off tests. The reinforcing bars extended
into the slab but were spread away from the studs to avoid influencing
performance at the location of the stud, as shown in Fig. 6(a).
Test units could be divided into four main groups: normal weight concrete (units
1-6), BST concrete (units 7-18), two studs per rib (units 13-15), and parallel
laying of steel decking (units 16 and 17). The position of additional transverse
or longitudinal bars or mesh is shown in Fig. 5. A specially designed “bowl”-
like reinforcing assemblage was placed around each stud in unit 18, in an
attempt to enhance the tensile capacity of the pullout cones through the failure
surface, as shown in Fig. 6(b).
The tests protocol was the same as specified by Erocode 4 [14]. Interface slip
was measured by portal gauges in four representative locations, with interface
separation being measured.
BST Concrete Properties
For units 7 through 18, BST polystyrene lightweight concrete with a target
characteristic compressive strength of 20 MPa and a target mean dry density of
@Seismicisolation
@Seismicisolation
1800 kg/m3 was used to cast the slabs. The mixture proportions are shown in
Table 2.
General properties of BST concrete were distinctive from normal weight

concrete. Slump of fresh BST concrete in most batches was low. Lack of heavy
coarse aggregates made the fresh concrete cone stand well, resulting in little
slump. Fresh BST concrete was shoveled directly into the moulds and
compacted using a $26 mm vibrator, Compaction of the fresh concrete was a
problem, as indicated by voids that remained in the slab when the vibration was
not sufficiently intensive. Large quantities of voids, with a maximum diameter
of 5 mm, were observed around the edge of unit 7 where less compaction had
been applied. As the BST polystyrene beads tended to float while the concrete
was vibrated, the surface of the finished siab had a bubbly appearance. The
properties of BST concrete used in each shear stud test unit are shown in Table
3 . From these data it was established that the BST concrete used had a
characteristic compressive strength of 19.0 MPa at the 95 % confidence level
and a mean elastic modulus of 15.0 GPa.
Observations and Results
Delamination between concrete slab and profiled steel decking was the main
phenomenon in testing, as'showln in Fig. -7(a). Delamination first occufred.at the
rear portion of a concrete rib, behind the thrusting stud. Delamination then
occurring at the fiont portion of the concrete rib, normally indicated that the
peak shear force had been reached. Subsequent diagonal cracking at the rib
represented final failure of the shear stud. Transverse cracking at slab troughs
was also observed in most tests (Fig. 7(b)). Typical force-slip histories are
shown in Fig. 8 through to Fig. 10, with key results listed in Table 4 (see also
Fig. 11). For units 7, 8, 10, and 11, the characteristic shear capacity of 19-mm
studs embedded in 120-mm thick BST ribbed composite slabs was determined
to be 70.5 kN per stud, at the 95% confidence level.
It can be seen in Fig. 8 through to Fig. 10 that in most cases, the combination of
Eqs. 5 and 6 overestimated the shear capacity of studs embedded in a ribbed
slab. Studs with 122 mm welded length always had larger shear capacity and
more ductility than studs with a 95 mm welded length. Studs positioned at the
front portion of the concrete rib (unfavorable position) had less shear capacity
and ductility than studs positioned at the rear portion of the rib (favorable
position). Additional reinforcing improved the performance of shear studs by
enhancing concrete confinement around the studs (unit 10, 11, and 18).
@Seismicisolation
@Seismicisolation
96 Lin et al.
Comparison with Normal Weight Concrete
According to Eq. 5, as the elastic modulus of BST concrete was about 60% of
that of normal weight concrete, the strength of a shear stud embedded in BST
concrete then should be about 77% of that embedded in normal weight concrete.
However, results indicated that shear capacity and stability of shear studs using
20 MPa BST concrete exceeded that using 20 MPa normal weight concrete
(NC) for a given geometric scenario and f i , as shown in Fig. 12. Initial
stiffness and final ductility of the shear connection using 20 MPa BST concrete
was also larger than that using 20 MPa NC.
Through this comparison, it could be seen that in this range of tests there was no
obvious correlation between the elastic modulus of concrete and shear stud
strength. Also, density of concrete did not influence shear stud strength.
Failure Modes
Stud pullout with concrete cones was the dominant failure mode of the 18 tests,
indicated by studs separating from the concrete slab with concrete cones (Fig.
13(a)). The concrete cones were broken into pieces due to longitudinal and
transverse cracks. The stud base moved.fonvard relative to both the concrete
and the steel decking while the shear force was applied, leaving a gap between
the stud and the concrete near the stud base, as shown in Fig. 14(a).
Concrete shear-off failure occurred in units 13 and 14, which had two studs in
each rib, and in unit 4, which was composed of 20 MPa normal weight concrete
(Fig. 13(b)). This failure was similar to pullout failure, except that the concrete
cones around the studs were intact without any crack in any direction, and the
studs stayed integral with the steel decking with no significant slipping, rotation
or deformation (Fig. 14(b)). The shear face followed the upper surface of the
steel decking and studs. This type of failure was accompanied by low shear
capacity and rather brittle behavior of the shear stud connection.
Stud fracture failure occurred in units either with longer studs embedded in a
thicker slab or concrete being strong or reinforcing enhanced. Studs fractured
just above the weld collar after severe shear deformation occurred near the stud
base (Fig. 15(a)). This type of failure was accompanied by larger maximum
shear force per stud, compared with pullout failure mode.
Concrete splitting failure only occurred in units 15 and 16, in which the profiled
steel decking was laid with the profiles parallel to the steel beam. The studs
were directly welded to the beam flange at a longitudinal spacing of 150 mm
and embedded in the solid concrete haunch that had an average width of 281
@Seismicisolation
@Seismicisolation
mm. Local splitting and crushing of the concrete first occurred at the concrete
bearing zone near the stud base, and the stud base moved forward, causing a
longitudinal ripping crack behind it. The longitudinal splitting and ripping crack
then propagated forward to the previous stud and backward to the next stud until
these cracks linked together forming global splitting of the concrete (Fig. 15(b)),
which indicated failure of the unit.
CONCLUSIONS
1. BST concrete with a characteristic compressive strength of 19 MPa (as

achieved in this research) was suitable to be used as a concrete slab, with
regard to composite action with steel beams.
2. For a given steel decking profile, 20 MPa BST concrete performed at least as
well as 20 MPa normal weight concrete.
3 . Concrete failure, including pullout and shear-off failure, dominated tests of
120-mm thick composite ribbed slabs for given steel decking profile and f,'.
4. For concrete failure modes, Eqs. 5 and 6 should be reviewed to reduce the
risk of overestimating the shear capacity of shear studs, or alternative
equations should be established.
5. Based on the experimental data, 70.5 kN can be adopted as nominal shear
capacity of 19-mm studs embedded in 120-mm thick ribbed composite slabs
for given steel decking profile and using BST concrete.
6 . Two studs per rib was an unsuitable structural form when the concrete is too
weak relative to the combined strength of two adjacent studs.
NOTATION
cross-section area of steel section

cross-section area of stud shank
depth of the rectangular concrete stress block
effective width of concrete slab
compression force resultant in concrete
compression force resultant in steel
overall depth of steel beam
specified compression strength of concrete.
yielding strength of steel
height of profiled decking rib
factored moment resistance of composite section
number of studs in a rib
number of studs between maximum and zero moment
@Seismicisolation
@Seismicisolation
98 Lin et al.
Qcomp = nqr
qr nominal shear capacity of shear stud given by NZS 3404
strength reduction factor of shear studs embedded in ribbed slab
Wfib mean width of profiled decking rib
REFERENCES
1. McSaveney, L. and Leonard, G. (2000). ?The Wellington stadium - New

Zealand?s first use of high strength lightweight precast concrete?, Second
International Symposium on Structural Lightweight Aggregate Concrete, 18-22
June, Norway, pp. 385 - 395.
2. Edwards, C. (1993). BST Lightweight Concrete - Technical Summary, BST

Holdings PTY Ltd, Report 6* edition, Australia, 88 p.
3. Ollgaard, J. G., Slutter, R. G. and Fisher, J. W. (1971). ?Shear strength of

stud connections in lightweight and normal-density concrete?, Engineering
Journal, American Institute of Steel Construction, Vol. 8, pp. 55-64.
4. Clifton, C. (1998). ?Reminders on good detailing and construction practice?,

Steel Design and Construction Bulletin, HERA, No. 42, April, p. 8.
5. Chien, E. L. and Ritchie, J. K.(1984). Design and Construction of Composite

Floor System, Canadian Institute of Steel Construction, Canada, 323 p.
6. Deck Design Manual (2001). United Steel Deck, Inc., a subsidiary of Bouras
Industries Inc., Manual No. 303-15, 63 p.
7. BS 5950 (i 990). Structural use of steelwork in building, Part 3: Design in

composite construction, Section 3. I : Code of practice for design of simple and
continuous composite beams, British Standards Institution, London.
8. Grant, J., Fisher, J. and Slutter, R. (1977). ?Composite beams with formed
metal deck?, Engineering Journal of the American Institute of Steel
Construction, Vol. 14, 1? quarter, pp, 24-42.
9. Jayas, B. S. and Hosain, M. U. (1988). ?Behaviour of headed stud in

composite beams: push-out tests?, Canadian Journal of Civil Engineering, Vol.
15, No. 2, April, pp. 240-253.
10. Robinson, H. (1988). ?Multiple stud shear connections in deep ribbed metal
deck?, Canadian Journal of Civil Engineering, Vol. 15, pp. 553-569.
@Seismicisolation
@Seismicisolation
11. Lloyd, R. M. and Wright, H. D. (1990). “Shear connection between
composite slabs and steel beams”, Journal of Constructional Steel Research,
Elsevier Science Ltd., Vol. 15, No. 4, pp. 255-285.
12. Mottram, J. C. and Johnson, R. P. (1990). “Push tests on studs welded

through profiled steel sheeting”, The Structural Engineer, The Institution of
Structurai Engineers, Vol. 68, No. 10, May, pp. 187-193.
13. Johnson, R. P. and Yuan, H. (1998). “Existing rules and new tests for stud
shear connectors in troughs of profiled sheeting”, Proceedings of the Institution
of Civil Engineers, Structures and Buildings, Vol. 128, No. 3, August, pp. 244-
251.
14. Eurocode 4: ENV 1994-1-1:1994, (1994). Design of Composite Steel and

Concrete Structures Parts 1-1 General Rules and Rules for Buildings, BSI
Standards, London, pp. 143-148.
15. Oehlers, D. J, and Bradford, M. A. (1995). Composite Steel and Concrete

Structural Members, Elsevier Science Ltd, Oxford, U.K., 549 p.
16. Oehlers, D. J. and Johnson, R. P. (1987). “The strength of stud shear

connections in composite beams”, The Structural Engineer, The Institution of
. .
Structural Engineers, Vol. 65B, No. 2, June, pp. 44-48.’
17. Gattesco, N. and Guriani, E. (1996). “Experimental study on stud shear

connectors subjected to cyclic loading”, Journal of Constructional Steel
Research, Elsevier Science Ltd., Vol. 38, No. 1, pp. 1-21.
18. Esterling, W. S. and Porter, M. L. (1994). “Steel-deck-reinforced concrete

diaphragms. I”, Journal of Structural Engineering, ASCE, Vol. 120, No. 2, Feb.,
pp, 560-576.
19. Butterworth, J. (2000). “Push-off tests on shear studs used with a deep
ribbed deck”, Auckland UniServices Limited, University of Auckland, Report
No. 8258.00, 25 p.
20. Park, R. (1989). “Evaluation of ductility of structures and structural

assemblages from laboratoq testing”, Bulletin of the New Zealand National
Society for Earthquake Engineering, Vol. 22, No. 3, September, pp. 155-166.
@Seismicisolation
@Seismicisolation
100 Lin et al.
TABLE I-LIST OF SHEAR STUD TEST UNITS
unit150/B20/19FU12 150 BST 20 2 19 122 0.95 thicker decking

unit
120/820/2X19/13 120 BST 20 4 19 95 0.75
unit two studs per
120/B20/2x19/14 120 EST 20 4 19 95 0.75 rib
unit
150/E20/2X19/15 150 BST 20 4 19 122 0.75
unit 120/B20/19P/16 120 BST 20 4 19 95 0.75 parallel laying

of steel decking
unit 150/B20/19P/17 150 BST 20 4 19 122 0.75
unit "bowl"
,12O/B20/19R/6M/l8 120 BST 20 2 19 95 0.75 reinforcing
@Seismicisolation
@Seismicisolation
TABLE 2-MIXTURE PROPORTIONS OF BST POLYSTYRENE
LIGHTWEIGHT CONCRETE
BST polystyrene concrete volume: 30 litres
Material Quantity
3.5 mm BST 12 litres
Cement
PAP7 (coarse sand)
P3020 superplasticiser (Rheobulid) 120 ml
Water 5.5 to 6.1 litres
Note: PAP7 means percentages ail passing 7 mm diameter sieve (e.g. 35%
between 2.38 mm and 4.75 mm, and 40 % under 1.18 mm).
14 1800 18.4 13.9

21 1763 19.5 14.3
28 1749 19.4 2.66 15.2
unit 18 0.42 30 1.11 28 1779 1761 18.8 2.19 15.2
@Seismicisolation
@Seismicisolation
102 Lin et al.
TABLE 4-RESULTS OF SHEAR STUD TESTS
lunit 18 I 1
97 I 18.8 15.2 I 115 I 86 IP&F I 1.1 I 6.4 I 5.8
Note. P = concrete cones pullout; F = stud fracture; S = concrete shear-oc
GS = concrete global splitting.
Ay is the yield slip of the shear stud connection, as defined in Fig. 1 i(a).
4 is the ultimate slip of the shear stud connection, as defined in Fig.
1l(b).
@Seismicisolation
@Seismicisolation
h.1 h'-
a PN
._
sheaf mnne
h.
Fig. 3. An example of a push specimen complying with Euro Code 4 [14]
@Seismicisolation
@Seismicisolation
104 Lin et al.
(a) Plan new
u 1 310UCi58HO bolls lo slwng flow I II

(b) Elevaticm
Fig. 4. Test rig with mounted test unit
ar
sh
@Seismicisolation
@Seismicisolation
(a) Delamination and deformation of steel decking (unit 08) (b) Transverse cracks at slab troughs (unit 10)
Fig. 7. Delamination and deformation of the steel decking
160 - 125
F iza- - l :
5
- 075:
- - 0 5 -
P
~ O25 e
Unit o1 -Uni1 02 -Unil 03
o
o 5 10 15 20 25 30
Mean slip (mm)
.
Fi 8. Sh ar force-slip history - 30 Ml'a normal weight concrete (unit 01,
08and 035
1O0
80
D
i5 40
B
5 20
-UnIl08 -Uni109 ---
O
O 5 10 15 20 25 30 35
Mean slip (mm)
Fig. 9. Shear force-slip history - 20 MPa BST concrete (unit OS, 09, and 10)
@Seismicisolation
@Seismicisolation
106 Lin et al.
100, 1.25
1 :
E
0.75
0.5 f
m
0.25
i
O
O 5 10 15 20 25 30 35
Mean slip (mm)
Fi 10. Shear force-slip history - 20 MYa iW1 concrete (unit 13, 14, 16,
an%18)
Interface slip iilecpe- &

(a) Yield inledaœ slip (b)U ö W irterácastip
Fig. 11. Definition of yield interface slip and ultimate interface slip 120J
- 1O0
-NC (unil 04) -EST (unit 07) -NC (unil 06) -EST (unitoe)
O
O 5 10 15 20 O 5 10 15 20 25 30
Mean slip (mm) Mean sllp (mm)
Fig. 12.20 MPa normal weight concrete and BST concrete
@Seismicisolation
@Seismicisolation
(b) Shear-off concrete cones (unit 14)

Fig. 13. Pullout or shear-off concrete cones
@Seismicisolation
@Seismicisolation
108 Lin et al.
(a) Movement and deformation of stud (pullout failure) (b) Intact studs (shear-off failure)
Fig. 14. Studs after testing
(a) Stud fractured just above weld collar (unit 03) (b) Global splitting along studs (unit 17)
Fig. 15. Fracture failure and global splitting failure
@Seismicisolation
@Seismicisolation
SP 209-7
Parcel 6 Development, Central

Arte rypunneI Project,
Boston, Massachusetts
by B. S. Moffat and E. P. White
SvnoDsis: Completion of the Central Arterymunnel Project includes development

of more that 20 distinct parcels along the original, elevated artery corridor. This
paper presents challenges encountered during the design of Parcel 6, which is lo-
cated between New Chardon and Sudbury Streets in Boston, Massachusetts. The
Parcel 6 lid will cover five adjacent cast-in-place concrete boat section ramps. The
final use of the lid is not yet known; however, since construction of the ramps is
currently underway, the walls were re-designed to accommodate either a land-
scaped deck, or a five-story building. Design included both wind and seismic later-
al analyses of the two types of potential lid structures, with the reinforcing in the
ramp walls modified to accommodate both options in order to minimize impacts or
retrofits required in the future. Changes to the original design included modifying
wall heights and reinforcing, and inclusion of interim backwalls for temporary
earth support. The landscaped option included preliminary designs of four separate
deck structures, skewed portal beams, overhead impact attenuators, and a merge
gore area. The building option presumed asymmetrical column loadings and loca-
tions, with a comparative analysis of column base shears used to determine maxi-
mum loading on the existing ramp walls.
Keywords: boat section ramp; development parcel;

lateral analysis; lid structure; seismic load; wind load
@Seismicisolation
@Seismicisolation
1o9
110 Moffat and White
AC1 member Barbara S. Moffat is a Senior Structural Engineer and Project
Manager for Jacobs Civil Inc. in Bellevue, Washington. She is a licensed
structurd' engineer in the Commonwealth of Massachusetts and a licensed
engineer in Washington and Oregon States. Her design experience includes
bridges, tunnels and shafts, including inspections, rehabilitation and demolition.
Ellen P. White is a Project Engineer and Project Manager for Jacobs Civil Inc.
in Boston, Massachusetts. She is a licensed structural engineer in the
Commonwealth of Massachusetts. She is experienced in the design of bridges,
cut-and-cover tunnels, underground structures, and demolition of existing
structures.
INTRODUCTION
The final phases of the Central ArteryITunnel Project involve the

development of more than 20 distinct parcels along the comdor in which the
original, elevated Artery structure is currently located (Fig. 1). Parcel 6 extends
for a length of approximately 360 feet (109.7 m) between New Chardon and
Sudbury Streets in the heart of downtown Boston, Massachusetts. The potentid
future development and use of this Parcel presented unique design challenges.
The Central Artery/Tunnel Project is committed to meeting certain

provisions contained within the Massachusetts Environmental Protection Agency
Cerhficate (MEPA) obtained after completion of the Final Supplemental
Environmental Impact Report (1). One of these requires that Parcel 6, which
consists of five adjacent boat-section ramps over an approximate 61 meter (200
fi) width (Fig. 2), be covered to mitigate its impacts to the surface environment.
This requirement must be met even if federal funds are not available for this
improvement.
Parcel 6 is a transition point between three distinct districts of Boston:

Government Center, Bulfmch Triangle, and the North End community.
Covering this parcel will greatly enhance both pedestrian flow and handicap
accessibility between each of these diverse sections of the City. Capping these
open ramps will also greatly reduce noise and be more aesthetically pleasing.
Parcel 6 is unique from the other parcels along the comdor in that it ulil
cover a series of U-shaped boat section ramps, all of which are in close
proximity to the surface streets. While the foundation systems of other parcels
can be independent of the artery tunnel and ramp structures, Parcel 6 must be
supported by these ramp structures themselves without impact to the mainline
northbound and southbound tunnels below.
@Seismicisolation
@Seismicisolation
The Boston Redevelopment Authority, which has created the overall final
development plan for the corridor, has designated Parcel 6 as “Recreation Open
Space” (2) which would allow the development of either a landscaped deck (Fig.
3) or a building (Fig. 4) over the ramps. Potential building uses include a
community center or a small high school. The Massachusetts Turnpike
Authority, which owns and will operate the Central ArteryíTunnel, is committed
to constructing a deck over the ramps in order to satisfj the aforementioned
environmental commitments if a building developer cannot be identified.
The Section Design Consultant* for this portion of the project,

investigated both deck and building options. Preliminary deck design and lateral
analyses were performed for both of these alternates to determine impacts to the
ramps below. The ramps, originally designed in accordance with preliminary
loads provided by the Central ArteryíTunnel Management Consultant (3), were
then modified to accommodate either option.
LANDSCAPED DECK OPTION
Four deck layouts were presented in the final draft of the “Preliminary
Design Study, Parcel 6 Ramp Cover” (4). These options varied primarily with
respect to the extent of ramp coverage. During development of these
alternatives, the profile was kept as low as possible in order to minimize the
visual impact of the deck.
The option that provided the maximum deck area over the ramp
structures was used for analysis purposes (Figs. 5,6). Its diagonal configuration
and skewed portals provided lower walkway slopes and allowed pedestrian paths
to blend with the street pattern of the adjoining North End community. The low
profile also allowed the wall heights to be kept to a minimum. Portal locations
were set so as to achieve maximum possible coverage of the ramps without
introducing additional mechanical ventilation.
A minimum 4.42 m (14.5 ft) traffic envelope was maintained for the
ramps below. Steel roof girders were spaced at 1.52 m (5 fi) centers and were
designed as composite with the .23 m (9 in.) concrete roof slab. Four spans,
from 8.53 m (28 fi) to 21.0 m (69 fi), were chosen to optimize girder designs
with corresponding section depths varying fiom 0.61 to 1.14 m (24 to 45 in.)
deep. Design parameters included a 0.91 m (3 fi) soil surcharge to account fm
future landscape needs, a 28.7 kPa (600 lb/fi*) live load surcharge to
accommodate potential vehicular loads above, and a 2.4 kPa (50 lb/f?) ceiling
surcharge load. Due to the relatively low depth of landscape cover, wind load
was not investigated; however, seismic lateral loads were calculated for
@Seismicisolation
@Seismicisolation
comparison against the seismic loading determined fiom the building analysis, as
defined below.
Preliminary designs were also developed for the two skewed portal
beams, the largest of which spanned 29 m (95 It) and supported 14 deck girders.
The ramp walls supporting these beams experienced end reactions of up to 4,737
kN (1,065 kips). Confinement reinforcement was added at each bearing location
to force the vertical wall reinforcing to function as an internal column.
To reduce the possibility of structural damage to the roof system ftom

trucks ignoring the posted height limits, overheight portal attenuators were
introduced at each tunnel entrance. These concrete attenuators, composite with
the first full-width interior girder, are designed to stop a truck with a gross
vehicle weight (GVW) of up to 356 kN (80 kips) up to a speed of 48 kph (30
mph). This truck load equated to a 472 kN (106 kips) horizontal load across a
0.8 m (2.5 It) width. Simultaneous vehicle impacts were not considered.
BUILDING OPTION
The design of the building will be the responsibility of the building

developer should one be identified. The developer will also design any decking
required to access this centrally located structure. Zoning requirements h i t the
height of the building to 16.8 m (55 It). The building is limited to a gross area of
13,006 m2 (140,000 fi2). The Preliminary Design Study included conceivable
layouts and loadings for each floor of this building. One of the potential building
configurations is shown in Fig. 7. Each unique floor configuration indicated
tributary areas for application of a dead load of 6.5 kPa (135 lb/ft2), a floor live
load of 4.8 kPa (100 lb/ft2), and a roof live load of 1.7 kPa (35 lb/ft2). Transfer
beams were spaced at 7.62 m (25 It), resulting in a total of 85 tabulated column
point loads upon the ramp walls.
It was critically important to determine if the building column loads

adversely impacted the mainline northbound and southbound tunnels directly
below Parcel 6. The mainline structures were originally designed with the 1995
preliminary joint development loads (JDLs). A comparison between the JDLs
and the loads in the Preliminary Design Study confirmed that the revised
magnitudes and locations did not adversely impact the roof girders, soldier piles,
cast-in-place walls or invert slabs of the mainline structure. Loading at the
critical roof girder was 9% less than that used for the original design.
Due to the use of lean concrete as an intermediate fill between the ramps
and mainline tunnels, settlement of the ramps was not a concern.
@Seismicisolation
@Seismicisolation
Lateral Analvsis
After review of the JDLs, a lateral load analysis of the building was
performed to determine impacts to the surface ramps. Although neither the
material types nor member stifhesses of the building were known, the following
assumptions were made in order to perform this analysis:
0 The future building shall be designed as a moment-resisting space

fiame.
No moment shall be transferred to the ramp walls; the building will
be required to be pinned at its base.
0 Column loads are assumed to act at the centerline of each ramp wall.
0 Each floor shall act as a rigid diaphragm.
0 Shear in the longitudinal direction will be neglected due to continuity
and length of ramp walls below.
The Massachusetts State Building Code (MSBC)( 5 ) was used for both
wind and seismic analyses of the building.
Wind Load Desim Parameters
Boston is located within the Massachusetts Zone 3 Wind Load Map. For
this preliminary investigation, an Exposure ?A? level was used for Parcel 6.
Exposure ?A? is appropriate for use in the centers of large cities where the
terrain for at least 0.8 km (0.5 mi) upwind of the structure is heavily built up,
With at least 50% of the buildings being in excess of four stories, and when
Exposure ? B (urban development and rolling terrain) prevails beyond this
boundary. The design wind pressure for Zone 3, Exposure ?A?, is 0.67 kPa (14
lb/fl?). This loading was applied to the projected elevation of each floor and roof
ignoring setbacks (Fig. 8). In accordance with the MSBC, a wind load
distribution factor of 1.0 was applied in the transverse direction in combination
with a 0.7 factor applied longitudinally.
Seismic Design Parameters
Seismic Hazard Exposure Group II, which is applicable for buildings

having a substantial public hazard due to occupancy or use, was chosen for this
site. Due to structural irregularities in both building plan and elevation a Seismic
Performance Category (SPC) of ?D? was used. This SPC designation requires
100% lateral loading and 30% longitudinal loading be applied simultaneously to
@Seismicisolation
@Seismicisolation
the structure. The potential structural irregularities also required that the
building be designated as Type 3, which allowed either the Equivalent Lateral
Force Method (static analysis) or the Modal Analysis Procedure (dynamic
analysis) to be used. The Equivalent Lateral Force Method was chosen due to
the high level of uncertainty with respect to the future building design. With the
assumption that this building will be a moment-resisting space fiame, a torsional
amplification factor was not applicable. Story drift determinations will be the
responsibilityof the building developer during final design.
Several assumptions were made to determine the shear force applied at

each floor. level. To calculate the fundamental period of the building, T, the
following variables were used:
CT, a coefficient reflecting the structural properties and deformational

characteristics of the resisting elements, was not yet decided;
therefore, the most conservative value of CT=0.02 was used.
The MSBC defines the coefficient of the upper limit on calculated

period, Ca, as equal to 1.6 for an acceleration of 0.12g.
Hn,building height, was assumed to be 16.5 m (54 ft).
Utilizing these variables, the period, T, was then calculated.

Ta (approximate fundamental period) = C T H ~ ’ ~
Ta = 0.02(54)3’4 = 0.4 seconds
And therefore,
T = TaCa= (0.4)(1.6) = 0.64 seconds
in conjunction with the period, T, calculated above, the following
coefficients were used to determine the static seismic coefficient factor, C,, used
in the Equivalent Lateral Force Method.
Per the MSBC, a peak-velocity acceleration of O. 12g was used.
0 S , the site coefficient, was conservatively taken as 1.5 for an Sj soil:

a soil profile containing 12.2 m (40 fi) to 30.5 m (100 fi) in thickness
of medium stiff clay with undrained shear strength of 47.9 kPa (1000
lb/fi2) or greater, with or without intervening layers of granular soils;
or up to 12.2 m (40 ft) of soft clay with undrained shear strength of
less than 47.9 Wa (1000 lb/ft2); or up to 12.2 m (40 ft) of very loose
to loose gravel, sand or silt; or up to 6 m (20 ft) of organic soil or
loose or soft fill.
@Seismicisolation
@Seismicisolation
A response modification factor, R, of 3.5 was chosen, which is the
most conservative value for an SPC ‘D” for a building with an
unknown stnictural fiaming system.
These factors were combined into the code-specified formula for

calculation of the seismic coefficient, C,, as follows:
C, = (1 .2AVS)/(RT2”)
C, = (1.2)(0.12)(1.5)/(3.5)(0.4)2’3 = 0.083
C, was limited to a maximum value of (2.5)(&)/R, or (2.5)(0.12)/3.5

which equals 0.086. Therefore, the value of C, used was 0.083 to determine the
resulting seismic base shear for application at each floor level:
V = 0.083 W, where W is the applicable dead load at each floor level.
Column Base Shear
Based upon the tabulated column locations and loads given in the
Preliminary Design Study, the centers of gravity (C.G.) and rigidity (C.R.) were
individually calculated for each of the six building levels. Eccentricities in both
the x- and y-directions, e, and %, respectively, were then determined. A
minimum eccentricity of 5% fiom the C.G. to the C.R. was used to account for
possible torsional effects. Lateral forces were distributed to the columns based
upon the following formula, which is typically used in the design of bolt groups:
where, P= P, (Transverse Loading) or P, (Longitudinal Loading)

N= Number of load bearing columns at the floor level under
consideration
R= (p: + P 2)”2
e = (ex2 + 2)’”
d= distance between column and C.G. (described below)
12= (Zx2 + i?) at each floor, where x and y are the distances
fiom each column to the C.G. of that floor
The building was divided into four zones (Fig. 9) based on the proximity
of each column to the overall C.G. of the building. The distance between that
C.G. and the farthest column within each zone was the distance, d, used in the
above calculation. Each column within that zone received that calculated shear.
These shears were tabulated for each floor level and column location, and then
combined to compute the overall base shears.
@Seismicisolation
@Seismicisolation
RESULTS OF ANALYSIS
The wind and seismic base shears determined flom the lateral analyses
d e h e d above were compared and the critical loading was applied transversely to
the top of the supporting ramp wails. The seismic design for the building
controlled, with a maximum unfactored critical column base shear of 378 kN (85
kips). These walls were then investigated to determine adequacy of original
reinforcing. Effects of overturning due to both wind and seismic forces on the
structure were also investigated. Neither one controlled or impacted the wall
designs.
The vertical column loads provided in the Preliminary Design Study were
used in conjunction with the controlling column base shears for re-analysis of
each section of boatwall. Three load cases were considered: maximum vertical
with maximum lateral loading; minimum vertical with maximum lateral loading;
and minimum vertical with minimum lateral loading. Load combinations were
used as specified in the Central Artery's Project Design Criteria for Cut-and-
Cover Structures (6):
GroupA= 1.3@+1.67L+1.3E+H)
Group C (Operating Design Earthquake, ODE) =
1.0 (D + 1.3E + H + EQ)
Group C (Maximum Design Earthquake, MDE) =
1.3@+1.3E+H+EQ)
(where D = Dead Load, L = Live Load, E = Earth Pressure, H =
Hydrostatic Pressure, and EQ = Earthquake Load.)
The unfactored vertical column loads and base shears were combined
with lateral static loads and dynamic soil, water and surcharge pressures in 2-
dimensional plane frame models. These flames were analyzed with both deck
and building loads. In order to accommodate either a deck or building structure,
modifications to ramp geometries and wall thickness and heights were required
although rebforcement in the invert slabs was not impacted, Wall designations
are shown in Fig. 10, with Table 1 summarizing the changes required in these
walls.
RAMP MODIFICATIONS
Planned development of the future parcel impacted Contracts currently

under construction. In addition to the modifications resulting from re-design of
the ramp structures, the following changes were implemented:
@Seismicisolation
@Seismicisolation
The original ramp wall heights were designed to retain the adjacent
surface street grades. To accommodate the depth required for the
potential lid structures and their connections, wall heights had to be
lowered. Backwalls up to 2.13 m (7 ft) in height were introduced in
order retain the interim soil grades (Fig. 11).
Original cast-in-place boatwall thicknesses were decreased by 5 1 mm

(2 in.) in several locations to accommodate direct-applied tile
finishes. For this project, tile finishes were not required in the open
ramp sections; modiijmg these to covered lids necessitated the
application of these tiles. In conjunction with the vertical and lateral
loading at the top of the ramp walls from the lid structure, this change
significantly impacted the reinforcing.
It was critical that the addition of new ramp walls and walkways not
impact the existing steel underpinning system of the elevated viaduct
structure directly above Parcel 6 . Therefore, notches and blockouts
were introduced into the ramp modifications to avoid potential
interference.
0 Walkways and planter boxes originally separated several of the

ramps. Modifications were made to the walls in these areas to allow
full-height ramp walls in the future should they be required.
One of the tunnel egress structures is located within the Parcel 6

limits. Its header beam required re-design due to additional vertical
and torsional loading determined from analysis of the lid structure.
0 A new merge gore area, where two independent ramps converge, was
designed, including cap beam and gore columns, due to decreased
visibility and sight distance fiom raised ramp walls.
Architectural modifications included alterations to the barrier types

and locations, new layouts of walkways and toe holes, and re-design
of the wall rustication joints.
Mechanical changes included provisions for standpipe niches and the

relocation of trench drains to the new portal locations.
0 In addition to accommodating future tunnel lighting, modifications

were made to electrical pull box and conduit layouts. Also, the
locations of sign and light pole foundations were altered.
@Seismicisolation
@Seismicisolation
Incorporation of these changes prior to constniction of the ramps
minimized the impacts of future parcel development both with respect to cost
and schedule.
COST EFFECTIVENESS
There are two separate, adjacent construction contracts that are

responsible for the mainline tunnels and ramps within the boundaries of Parcel 6.
Construction of the lid structure will be completed by a follow-on Contract if a
building developer is not identified.
Implementation of changes into contracts already under construction was

advantageous for several reasons. Designing for the worst-case scenario permits
the flexibility for either the deck or building option to be chosen in the future.
Performing a full preliminary analysis permitted the inclusion of only those
changes that were absolutely necessary. Early incorporation of these: changes
into the construction contracts mitigated impacts to the construction schedule and
minimized the potential for costly change orders. Additionally, future
modifications or retrofits are not anticipated since the structural capacities of the
ramps have been increased to adequately handle the planned development of
Parcel 6.
CONCLUSION
The planned, future development of Parcel 6 has offered unique

challenges both with respect to analysis and design primarily due to the fact that
the character of the fume lid structure has not yet been determined. Both
seismic and wind analyses were performed on a potential five-story structure to
ensure that the ramp walls below would be able to adequately support the
development of a building. It was also necessary to analyze and perform the
preliminary design of a deck structure should a building structure not be
constructed. It was important that changes necessary for the support of this lid
be implemented into contracts under construction to minimize future impacts. In
the opinion of the authors, although the lid type was not known, the resulting
designs offered an extremely cost-effective solution able to accommodate both
scenarios.
* Jacobs Civil Inc.
@Seismicisolation
@Seismicisolation
REFERENCES
1. US. Dept. of Transportation, “Central Artery (Interstate 93)/Third

Harbor Tunnel (Interstate 90) Project, Boston, Massachusetts: Final
Supplemental Environmental Impact Statement”, Boston: Mass. Dept.
of Public Works, January 1991
2. Boston Redevelopment Authority, “Boston 2000 Plan”, 1991.
3. Centrai Artery/Tunnel Management Consultant, “Joint Development

Loads (JDL’s)”, July 1995.
4. “F‘reliminary Design Study, Parcel 6 Ramp Cover”, July 1998.
5. Massachusetts State Building Code, 780 CMR - Sixth Edition, 1997.
6. Central Artery/Tunnel Management Consultant, ‘Troject Design

Criteria for Cut-and-Cover Structures”, 1995.
@Seismicisolation
@Seismicisolation
Table 1 -Wall Reinforcement Summary
WALL REINFORCING REINFORCING

2-#11 @ 150 mm (6 in.) West Face
No Change
#8@ 150 mm (6 in.) East Face
#11 @ 150 mm (6 in.) West Face
1B #7 @ 150 mm (6" in.) EF
#8 @ 150 mm (6 in.) East Face
2A No Wall #9 @ 150 mm (6 in.) EF
2B No Wall #7 @ 150 mm (6 in.) EF
3A(1) #7 @ 300 mm(12 in.) EF #11 @ 150mm(6in.)EF
3A(2) #7 @ 300 mm (12 in.) EF #9 @ 150 mm (6 in.) EF
#7 @ 150 mm (6 in.) West Fact
3B #9 @ 150 mm (6 in.) EF
3C #7 @ 150 mm (6 in.) EF
#9 @- 150 mm (6 in.) East Face
@ 150 mm (6 in') +#10@300 mm (I2 #I1 @ 150 mm (6 in.) West Face
3D in.) West Face #9 @ 150 mm (6 in.) East Face
4A #7 @ 150 mm (6 in.) EF #11@ 150mm(6 in.)+#8@450mm(18
in.) East Face
4B( 1) #7 @ 300 mm (12 in.) EF #11 @ 150 mm (6 in.) + #8@ 450 mm (18
in.) East Face
4B(2) #7 @ 300 mm (12 in.) EF #8 @ 150 mm (6 in.) EF
4C #7 @ 300 mm (12 in.) EF #8@300mm(12in.)EF
5A #8 @ 150 mm (6 in.) West Face #8 @3OOmm(12in.)EF

2-#11@ 150 mm (6 in.) East Face
#8 @ 150 mm (6 in.) West Face; 2-#11 @ #8 @ 150 mm (6 in.) West Face
5B 150 mm (6 in.) East Face #11@ 150mm(6in.)EastFace
5c #11 @ 150 (6 )'.i + # l o @ 300 mm #I1 @ 150 mm (6 in,) East Face
(12 in.) East Face
#7 = 22.225 mm diameter #10 = 32.258 mm diameter

#8 = 25.400 mm diameter #11 = 35.814 mm diameter
#9 = 28.65 1 mm diameter
@Seismicisolation
@Seismicisolation
Fig. 1 - Parcel 6 Location, Boston, Massachusetts.
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
Fig. 2- Uncovered Ramp Design.
Fig. 3 - Landscaped Deck Option.
@Seismicisolation
@Seismicisolation
Fig. 4 - Building Option.
@Seismicisolation
@Seismicisolation
Fig. 5 - Ramp Cover Deck.
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
Fig. 7 - Ramp Cover Building.
@Seismicisolation
@Seismicisolation
Fig. 8 - Building Section.
@Seismicisolation
@Seismicisolation
Fig. 9 - Zone Designations for Column Shears.

@Seismicisolation
@Seismicisolation
c
w
>
I-
A
4
3
__
(D
4-
w
LL >
I-
iI
I-
I $
i I
Fig. 10 - Ramp Wall Designations.

@Seismicisolation
@Seismicisolation
Fig. 11 - Interim Backwall.
@Seismicisolation
@Seismicisolation
SP 209-8
Behavior of Flared-Column Bents

Under Seismic loading
by H. M. Nada, D. H. Sanders, M. Saiidi,
and S. El-Azazy
Svnopsis: Past Earthquakes have shown that flared columns are susceptible to pre-
mature shear failures. In the 1999 Northridge Earthquake, Shear failures have been
caused by a plastic hinge formation occurred at the base of the flare and a subse-
quent increase in the level of column shear demand above design levels. This paper
presents an experimental and analytical study that is examining new details for
flared columns and joints. The primary feature is a gap at the top of the column and
the amount of transverse flare reinforcement. Two two-column bents were tested
on the shake table to examine dynamic effects as well as column and beam inter-
action. The test specimens had different amounts of confining reinforcement in the
flare. The detail proved an overall ductility that is above the minimum limit recom-
mended by the current specifications. Test also showed that gap started to close at
ductility level that is below the recommended limit.
Kevwords: bent; bridges; circular columns; flared;

seismic; shake table
@Seismicisolation
@Seismicisolation
133
134 Nada et al.
Hisham M. Nada, Ph.D. Candidate at the University of Nevada, Reno. He
received his Civil Engineering BS and MS degrees form El-Mansoura
University, Egypt.
David H. Sanders, Associate Professor at the University of Nevada, Reno. He

received his Civil Engineering BS degree from Iowa State University and his
MS and PhD degrees from the University of Texas at Austin. He is AC1
Fellow, chair of AC1 341 Earthquake Resistant Concrete Bridges, secretary for
ACIIASCE 445 Shear and Torsion, and a member of TAC.
AC1 Fellow, M. Saiid Saiidi, Professor of Civil Engineering at the University

of Nevada, Reno. He is the former chair and a current member of AC1
Committee 341, Earthquake-Resistant Concrete Bridges. He is a member of
AC1 342, Evaluation of Concrete Bridges and Bridge Elements, and 352,
Joints and Connections in Monolithic Concrete Structures.
Saad El-hazy, Ph.D., PE manages seismic research program in the Office of

Earthquake Engineering, Caltrans. He received his B.Sc. from Cairo
University, Egypt, and his M.Sc. and Ph.D. from Ohio State University,
Columbus, Ohio. His research focuses on seismic performance and retrofit of
highway structures and its incorporation into design and retrofit procedures.
INTRODUCTION
Design of flared columns to resist earthquake loading is a complicated

issue due to the changing cross section along the column length. Past
earthquakes have shown that flared columns could have poor behavior due to
the large increase of column flexural capacity at the top of the column. The
increased capacity can cause the plastic hinge region to shift away from the
column-beam interface into the column. This causes the column to behave like
a much shorter column and therefore have increased shear demand. The
shifting of the hinge and the larger shear demand was not included in many
previous designs and has led to brittle shear column failures. Brittle shear
behavior may lead to significant damage andor complete collapse of the
bridge.
It was believed that if the flares had low longitudinal and transverse
reinforcement ratios, they would fail during earthquakes. Therefore, the
@Seismicisolation
@Seismicisolation
column core would be the element remaining to resist the earthquake.
Experiments and past experience proved that even lightly reinforced flares
contribute to increasing the column’s flexural capacity (1). To solve this
problem, Caltrans has developed new details attempting to separate the flared
portion of the column from the overall flexural stiffness and capacity of the
column. This is achieved by separating the flare from the bent cap using a gap
between the top of the flare and the beam bottom surface, Fig. 1. Caltrans’
recommendations came as a result of single column slow-cyclic tests and not
on system bent tests (1). The purpose in conducting shake-table bent tests is to
study the effectiveness of the flare detail and to examine the behavior within
the joint region as well as the overall system response. Four bents will be
tested as part of the overall study. Two bents will be flexurally dominated
while two will have shorter height that has much higher shear demand. This
will be accomplished by using two different column heights. The columns are
pinned at the base with a two-way hinge. Within each bent of similar height,
the amount of flare transverse reinforcement will be varied. In one case the
amount of transverse reinforcement in the flare will be in accordance with
current Caltrans’ guidelines while in the second case, the amount of
reinforcement will be reduced in accordance to a proposed Caltran’s detail.
This document will focus on the tall bents tested (LFCD1 &LFCD2), which
are flexurally dominated. The experiments were conducted at the Large-Scale
Structures Laboratory at the University of Nevada, Reno.
MODEL DESIGN
A survey was made of existing bridges in California to develop a

prototype. Because the model will be used for new bridge detailing, only
bridges that were designed according to the latest code provisions were
considered in the survey. Based on these details and the 50-ton vertical load
capacity of a single shake table, a 1R-scale model was developed. The vertical
axial load acting on each column was calculated according to Eq. 1.
Where;
Ag = Gross column cross section and
fc’ = Concrete compressive strength (3 1.O MPa).
The 10% level was set as a typical upper range for the majority of new
@Seismicisolation
@Seismicisolation
136 Nada et al.
columns in regular bridges. This level resulted in a total axial load of
451.5 KN. In order to distribute this load along the length of the bent, lead
buckets in combination with small rams were used. The other key details for
the columns are given in Table 1 and shown in Fig. 2. All reinforcement is
Grade 60 to conform to current design. Some dimensions that cannot be
scaled due to practical issues, like concrete cover, are kept within the
minimum range that allows for practical use. The flexurally-dominated tall
columns had a clear height of 1,625.6 mm while the shorter columns had a
height of 939.8 mm. The bents described in this paper had the taller columns.
Flare reinforcement layout was calculated by scaling the prototype

dimensions (2,3). Current Caltrans recommendations state the following (4):
The gap width is calculated based on the estimated rotation at the top
of column so as to prevent gap closure. The minimum height of the
gap is 50 mm.
The longitudinal flare reinforcement is nominal. Spacing between
longitudinal flare reinforcement shall not exceed 450 mm and not less
than 150 mm.
The transverse flare reinforcement ratio in the upper 113 of the flare
height is ph=0.45%+0.05 while that ratio for the lower 213 of the flare
height is ph=0.075%k0.025.
Where;
ph = 2 A b / S D
S = Spiral pitch
D = Column diameter
Ab = Area of spirals
To investigate these provisions and their effectiveness, two flare details

were developed. The first one had a flare lateral reinforcement ratio that
complied with the above-motioned provisions. The second detail had constant
lateral flare reinforcement ratio along the total height of the flare. This was set
to the lowest value of ph=O.O75%.
To match the prototype, the center-to-center span of the bent should have
been 2,286 mm. Since it was decided to use lead to apply the axial load, there
@Seismicisolation
@Seismicisolation
was insufficient spacing between the columns to accommodate all of the lead.
For this reason, the span was increased to 2,718 mm. The beam’s depth was
also increased in order to have the same relative flexural stiffness between the
columns and the beam as in the prototype. The bent cap was designed
according to Caltrans’ recommendations, which states that the beam capacity
should be 1.2 times the capacity of the plastic hinge in the column. This is
done to assure that a plastic hinge would form in the columns rather than the
beam.
ANALYSIS
An analysis of the specimens was made in order to calculate the system

natural frequency, initial stiffness, and the maximum load capacity. The
analysis was also used to determine the earthquake record to be used to test the
specimens. The system is complicated due to the abrupt change in geometry at
the top of the column gap region and the base hinge region. A finite element
non-linear analysis was used in order to estimate the load-displacement curve
of the bent. Another advantage of using the finite element method that it takes
into account the variation that occurred in the axial load acting along each
column due to the horizontal load applied to the whole system. A simple
hysteretic model was also used to determine the earthquake response.
A simplified plane stress model was used to model the bent. The concrete
elements were 2-D plane stress elements having a thickness equal to the
structure’s thickness. The reinforcement was modeled as a 2-D beam element
in order to model the dowel action. Contact elements were used at the top of
the flare and the bottom of the beam within the gap region. Gap elements were
also used in the base hinge region. Since the footing was heavily reinforced, it
was modeled as a rigid linear elastic region. The footing was prevented from
moving in any direction by restraining the degrees of freedom of the outer
nodes, Fig. 3. ADINA, a non-linear finite element program, was used for
performing the analysis (5).
The lateral load was applied as two concentrated loads on each end of the
beam. Several options are provided for the concrete model. The Kupfer model
was selected. For the reinforcement elements a bilinear stress-strain curve was
used. A pushover analysis was performed by incrementing the lateral load
gradually. The program goes unstable at the beginning of the longitudinal
@Seismicisolation
@Seismicisolation
138 Nada et al.
reinforcement yielding due to the fact that a small number of elements become
crushed resulting in a change in the global stiffness matrix. In order to prevent
the instability, the crushed elements were removed manually and multiple runs
were performed in order to get the load-displacement curve. To verify this
technique an analysis was performed using plane stress elements and the
multiple runs technique for a previously tested circular column (6). The result
of the analysis showed good consistency between ,the experimental and
analytical results. Fig. (4) shows the deformed shape and the load-deflection
curve for the flared column bent structure.
The finite element analysis results showed that:
The failure occurred in the gap located at the top of the columns due to
yielding of the reinforcement and concrete crushing.
Extensive cracking occurred in the beam-column region.
Cracks spread along the flare height while no crushing occurred in this
region.
The analysis using ADINA was stopped before gap closure at the top of the
flare or at the base of the column. Currently, more sophisticated analysis is
being done using a program called DIANA in order to extend the analysis.
Analysis was also done using a program called RCSHAKE (7) that was
developed at the University of Nevada, Reno. This analysis was used to excite
the model with different earthquake records. The program uses an idealized
bilinear load-deflection curve, the mass of the specimen and a scaled
earthquake record under consideration. The earthquake time step is scaled with
a scale factor equal to the square root of the model scale factor. This enables
the same behavior from the model as from a prototype subjected to the same
earthquake. The Northridge Sylmar earthquake in 1994 record was found to be
the most suitable.
TESTING
The specimens were constructed at the University of Nevada, Reno. A total

of 110 strain gauges were attached to reinforcement in the columns, flares and
the beam-column connections. Material properties for the specimens are
summarized in Tables 2 and 3.
The footing was tied to the shake table using prestressed steel bars to
@Seismicisolation
@Seismicisolation
prevent any uplift during the test. The axial load was applied using steel
buckets filled with lead blocks. The maximum vertical load capacity the shake
table can sustain limited the weight of the specimen plus lead to 444.8 KN.
The specimen weight was about 133.4 KN including the footing. The total
axial load for both columns was 451.4 KN. In order to reduce the weight on
the table, 178 KN of axial load was modeled using self equilibrating rams in
combination with the mass rig, Fig. 5. The mass rig is a steel structure
supported by frictionless pins so that it allows the system to move laterally
without restraining it. The four rams were mounted over the two column and
the mass rig was connected to the specimen with a rigid link.
Beside the strain gauges, 34 transducers were used to measure the curvature
along the beam height, beam-column connection in-plane displacements, out-
of-plane displacements and horizontal movement of the bent and the column
base. Accelerometers were mounted on the table and on top of the specimen.
The loading procedure for the specimen consisted of a series of earthquake

runs. This loading sequence was chosen so that it would capture the important
points like cracking, yield and ultimate. The specimens were excited with
scaled versions of the Sylmar record (Northridge Earthquake, 1994). The
acceleration of the original record was multiplied by a factor. The following
numbers are the factors used for the series of earthquake records: 0.15, 0.5,
0.75, 1.0, 1.25, 1.5, 1.75, 2.0, 2.125, 1.75, and 1.75 times Sylmar. During the
test of LFCD1, it was observed that after 2.125 Sylmar the frequency of the
specimen dropped dramatically from an initial value of 3.2 Hz. tol.0 Hz.
Because of this large drop it was decided not to increase the motion but to
repeat lower motions to characterize the system. The 1.75 Sylmar record was
repeated twice for LFCD1 and once for LFCD2. Fig. 6 shows final crack
patterns for LFCD1 while Fig. 7 shows final crack patterns for LFCD2.
Observations during the test and maximum strain level reached are
summarized in Tables 4 and 5. Comparing the final results of the two
specimens showed that the initial stiffness, final achieved displacements and
the experimentally calculated yield displacements are identical.
The increase in relative rotation between the beam and the column
increased dramatically after yielding of the reinforcement in the gap region
causing the gap to be closed for both specimens, Figs. 8 and 9. The curvature
trend along the flare height is proportionally increasing with the increase in
@Seismicisolation
@Seismicisolation
140 Nada et al.
elevation for LFCD2 while this is not the case for LFCD1. LFCD1 had
increased curvature at the base of the flare caused by increased confinement in
the flare. The accumulative load-deflection curves of the two specimens for ail
runs are shown in Fig. 10.
Fig. 11 shows the envelope curve for the two specimens and the results
from the finite element analysis as well. The graph shows a kink in the load-
deflection envelope at approximately 70 mm for LFCD1 and LFCD2 followed
by a large increase in the overall stiffness. This behavior was due to the gap
closure that caused an increase in the load carrying capacity that was predicted
to be 160.1 KN. The difference of gap closure effect on the load-displacement
envelope of the two specimens was because of concrete crushing of LFCD2 at
the top portion of the flare. This behavior reduced the capability of the flare to
transfer force to the beam.
The experiments were stopped due to concerns about the overall stability
of the specimen. Since the reinforcement at the top of the columns was near
the strain-hardening region, bar fracture could cause the top of the column to
quickly lose moment capacity. It was also observed that while the last two runs
had smaller input accelerations, the displacement histories were the same as
2.125 Sylmar. Increased moment capacity did lead to shear cracking in the
middle portion of the flare and a shear crack in the beam as well as concrete
crushing in the middle of the flare (see Figs. 6c, 7c).
LFCD2 shows wider and longer shear crack plus many small shear and
flexural cracks spread over the beam span. Another important observation is
that the shear crack appearing in LFCD1 started after a distance of about d/2
from the end of the flare. In LFCD2, this distance is measured from the column
core. This shows that the extensive concrete spalling shifted the primary load
path to the core of the column.
CONCLUSIONS
Both of Caltrans details is above the target limit of 6 that is

recommended by Caltrans.
While the ductility level is adequate, the closure of the gap increased the
capacity at the top of the column that led to shear cracks in the columns.
This behavior is undesirable for it might lead to brittle shear failure the
columns.
@Seismicisolation
@Seismicisolation
o The gap closure affected the plastic hinge formation at the top of the
column, and hence, transferred higher loads to the bent cap that resulted
in shear cracks in the bent cap.
0 The lateral reinforcement in the top third of the flare should be
proportioned so that it insures the spalling of the concrete in this region
to prevent further load transfer through the flare and insures the flexural
plastic hinge behavior in the gap region.
o An analysis or design based on only the column core stiffness did not
accurately model the observed behavior. Analysis that is more
representative using Strut-and-Tie model of the system is under
development.
ACKNOWLEDGEMENT
The authors would like to express their appreciation to design

engineers that assisted greatly in this joint effort between Caltrans and
the University of Nevada, Reno. Caltrans is also acknowledged for their
financial support. The authors would like to thank Patrick Laplace and Paul
Lucas for their work in the Large-Scale Structures Laboratory at the
University of Nevada, Reno. The opinion expressed in this paper are those
of the authors and do not necessarily represent the opinions of the sponsor.
REFERENCES
Anthony V. Sanchez, Frieder Seible, and M.J. Nigel Priestley (1997).

Seismic Performance of Flared Bridge Columns, Report No. SSRP-
97/06, The University of California, San Diego.
Caltrans Bridge Design Specification, California Department of

Transportation, Division of Structures ,2000.
Caltrans Bridge Design Details, California Department of

Transportation, Division of Structure, 1994.
Caltrans Seismic Design Criteria, Version 1.1 (July 1999). Caltrans,

Sacramento, CA
ADINA Reference Manuals (1999). ADINA R&D, inc, WatertÖwn,

MA
@Seismicisolation
@Seismicisolation
142 Nada et al.
6. Ashraf El-Bahy, Sashi K. Kunnath, Andrew W. Taylor, and William C.
Stone (1996). Cumulative Seismic Damage of Reinforced Concrete
Bridge Piers, Report NCEER-96-OOXX, University of Centrai Florida,
Orlando.
7. Laplace, Patrick, David Sanders, M. "Saiid" Saiidi, and Bruce Douglas

(1999). Shake Table Testing of Flexure Dominated Reinforced
Concrete Bridge Columns. CCEER #99-13, University of Nevada,
Reno, Department of Civil Engineering, pp 152.
Table 1. Summary of model and prototype dimensions.
Description Prototype Model Factor
Column Diameter (mm) 1,524 304.8 0.2

Flare Width (mm) 3,175 635 0.2
Column Reinforcement 30#44mm 1 4 # 13mm 1
Column Reinforcement Ratio 2.39 2.4 1
l 9 mm @ 14.88 mm @ 31.751 N/A

101.6
0.0037 0.0039 1
Column Height (mm) 8,128 I 1,626 I 0.2
~~
I
LFCD1 LFCDL
I
Transverse Reinforcement at
4.88 Dia. @ 27.9 mm 3.76 Dia. @ 96.5 mm
top 1/3 of the flare
Transverse Ratio 0.439 ?o' 0.075 %
Transverse reinforcement at
3.76 Dia. @ 96.5 mm 3.76 Dia. @ 96.5 mm
remaining 213 of flare height
Transverse Ratio 0.075 % 0.075 %
6 Wires - 6 Wires -
Vertical Flare Reinforcement
3.76 mm Dia. 3.76 mm Dia.
@Seismicisolation
@Seismicisolation
Table 2. Reinforcement yield stress
Yield Point
Reinforcement Diameter
íMPa)
15.9 mm 482
12.7 mm 427
9.5 m 448
Dia. 4.88 mm 418
Dia. 3.73 mm 49 1
Table 3. Concrete compressive strength on testing day (MPa)
Member LFCD1 LFCD2
Footing 35.2 47.4

Columns 37.4 43.6
Girder 38.4 36.2
@Seismicisolation
@Seismicisolation
144 Nada et al.
Table 4. Summary of test observations
LFCD1 LFCDL
Factor Observations Observations
0.15 - Minor cracking in the beam- Minor cracking appears in beam-
0.5 column connection and the column connection and in flare
column. region.
0.75 & Cracks increased in the A significant increase of
1.o beam-column connection cracking in the bearn-column
and initial cracks in the flare connection occurred. Flare
region. cracks increased in length and
new cracks started to form
1.25 A significant increase of Minor crack development in
cracking in the beam-column bearn-column connection. Shear
connection occurred and flexural cracks appeared in
flare region. Concrete cover
spallinp in the top of the flare.
1.5 & Cracks formation in the Crack formation stopped in
1.75 beam-column connection beam-column connection while
drops while the number and flexural and shear cracks
extension of diagonal cracks increased in the flare and in the
in the flares increases bottom of the column.
2.0 & Flare cracks increased and Flare cracks increased and
2.125 widened. Concrete spalling widened. Shear cracks along the
was observed in the middle column height. Extensive
third of the column and at concrete spalling occurred in top
the top of the flare portion of the flare region.
Last two Extensive shear cracking Flare cracks increased
1.75 extends along the column extensively and widened.
height. Displacement was Concrete spalling in top portion
equivalent to 2.125 Sylmar of the flare region. Second 1.75
Svlmar is not included
After Many wide shear cracks Shear cracks were found in the
bucket were found in the beam and beam And few flexural cracks in
removal extensive flexural cracks in the middle of the span.
the middle of the span.
Yield Displacement
(experimentallv calculated)
12.7 mm 1 12.7mm I
Yield Force 93.4KN 93.4 KN
Maximum Achieved Displacement 184 mm 1 8 4 m .-
Maximum Force 282.5 KN 266.9 KN .
Ductilitv Ratio (includes base slip) 14.5 14.5
Disdacement at Gap Closure 69.9 mm 69.9 mm
Ductilitv at Gap Closure 5.5 5.5 ~
@Seismicisolation
@Seismicisolation
Table 5. Summary of maximum strain
IFactor I LFCD2 LFCD2

I I
Hoops
0.15 &
0.0037
0.75 &
0.0053
0.0013
1.5 &
0.0023
1.75
2.125
1.75 0.05 0.0019 0.0403 0.014 NIA 0.0085 0.039 0.0033
1.75 0.05 0.003 0.0403 0.014 NIA NIA NIA NIA
@Seismicisolation
@Seismicisolation
146 Nada et al.
Fig. 1. Layout of the proposed detail
Fig. 2. Specimen LFCD1 layout and reinforcement detailing
@Seismicisolation
@Seismicisolation
Fig. 3. Finite element model
(a) (b)
Fig. 4. (a) Deformed shape (b) Load-displacement curve
@Seismicisolation
@Seismicisolation
148 Nada et al.
Fig. 5. Test setup
--I-. . ... ..
(b) (cl
Fig. 6. LFCD1 crack patterns, (a) shear cracks in the beam, (b) beam column connection
cracks, (c) column cracks
@Seismicisolation
@Seismicisolation
(b) (cl
Fig. 7. LFCD2 crack patterns, (a) shear cracks in the beam, (b) beam column connection
cracks, (c) column cracks
L a t column
Fig. 8. Curvature along flare height for LFCDI
@Seismicisolation
@Seismicisolation
150 Nada et al.
LfeCoiumn
e lea 1 r 1
Right Column
Fig. 9. Curvature along flare height for LFCD2
(a) (b)
Fig. 10. Accumulative load-displacement curve, (a) LFCD1, (b) LFCD2
Fig. 11. Load-displacement envelope curve
@Seismicisolation
@Seismicisolation
SP 209-9
Mechanical, Thermal, and

Acoustical Properties of Concrete
Containing Ve geta bIe Partic les
by L. Arnaud and V. Cerezo
Svnopsis: This paper deals with the measurement of physical properties (mechan-
ical, thermal, acoustical) of various formulations of concrete containing vegetable
particles. Such material is made up with hemp shives mixed with lime binders.
Shives are very porous and so lightweight. Thus, this concrete presents a high
porosity related to the microscopic porosity of the shives and the macroscopic
porosity due to the arrangement of particles. Moreover, this material presents a
ductile behavior and can bears high strain without been destroyed. Depending on
the binder proportion, the mechanical properties of vegetable concrete cover a
wide range: maximum stress in between 0.4 and 1.2 MPa, Young modulus in be-
tween 20 and 90 MPa, strain at maximum stress in between 4 and 10%.The ther-
mal conductivity ranges from 0.06 to 0.1 1 W.m-'.K-', sound absorption between
0.5 and 1. The final aim of this study is to optimize the formulation of vegetable
concrete according to its use (wall, floor, roof. . .). A theoretical model made with
self-consistent method allows to calculate precisely the coefficient of conductivity
h as a function of the mixture proportion and the compactness level. A comparison
with experimental measurements shows a good accuracy of the results.
Keywords: binder; physi ca I properties; s hives;

vegetable particles
@Seismicisolation
@Seismicisolation
151
152 Arnaud and Cerezo
Laurent ARNALJD received his Diploma of Civil Engineer from the National School of Public
works in 1989. He is Doctor of Engineering in Mechanics (1993) from the Ecole Centrale in
Paris. His research themes deal with the study of the mechanical behavior of heterogeneous
materials, including experimental and modeling aspects based on homogenization methods. He is
working in the fields of setting phenomena and of the physical characterization of building
materials.
Veronique CEREZO received his Diploma of Civil Engineer h m the National School of Public
Works in 2000. She is preparing a thesis dealing with heterogeneous material containing
vegetable particles (experimental and modeling aspects based on homogenization methods) at
the National School of Public Work in Lyon.
National School of Public Work, Department of Civil Engineering and Building, Laboratory of
Geomaterials, U M CNRS 1652
Rue Maurice AUDM, 69518 VAULX-EN-VELM, FRANCE
Email : laurent.amaud@entpe.fr, cerezo@entpe.fr
1 INTRODUCTION
Considering the issues of waste recycling and consequences of pollution on human health,
people have been trying to develop more efficient and environment-friendly building materials.
Concrete containing vegetable particles mixed together with a binder enters in this category.
Indeed, vegetable particles respect environment (bio-degradable). Moreover, it appears that such
material offers a high quality of life by creating a pleasant atmosphere, reguiating hygrometry
and temperature. In this paper, the particular case of hemp particles (shives) and lime is
presented. This material is both used in France for building new houses using a bearer wood
structure and rehabilitating old houses. Nevertheless, very similar study could be applied with
very similar results to the case of other vegetable concretes or other binders.
From a scientificpoint of view, the special structure of vegetable concrete is of interest. On the
one hand, the shives are very porous and therefore deformable, which is not the case for most
other civil engineering building materials. In addition, this concrete shows a macroscopic
porosity due to the arrangement of shives. Thus,this both microscopic and macroscopic porosity
provides specific mechanical, thermal and acoustical properties.
Nowadays, a global approach of the characteristics is favored including these three aspects. The
aim is to optimize the formulations considering the uses, because mechanical, thermal and
acoustical properties are somehow confiicting. For example, a material with a low thermal
conductivity contains a large part of vegetable particles to increase the porosity. However, the
mechanical properties decrease with this increasing porosity. Thus, a study is proposed here to
determine the best formulation for a given application (insulating wall or roof, bearing walls,
acoustic screen...). The performances of hemp concrete in each field are first studied and the
performances level in thermal are modeled.
@Seismicisolation
@Seismicisolation
This paper deals with the results so far obtained (1). in the first section, the raw materials are
presented and their properties are described. Then, the mechanical, themial and acoustical
behaviors are successively assessed and explained.
2 THE TESTED MATERIALS
2.1 The Shives
Shives were obtained from the stem of the hemp plant, which is glazed and cut. They are from 1
to 3cm long and OScm wide &.i). A shive is characterized by a microscopic porosity due to
the network of capillaries. This specificity explains the feature of the particles. They are very
light as compared to other mineral aggregates. The density of a particle equals to 320 kglm’ and
the bulk density of a shives pile ranges from 110 and 150kg/m3 depending on the compaction
level. In addition, the microscopic porosity of vegetable particle leads to a significant
compressibility to shives which can be highly strained -2): due to inter-particle rubbing,
samples of shives in bulk (Table 1) have been tested. Shives in bulk present a ductile behavior
and a simple after compressive test can reach up to 30% of the initial height very easily (final
axial strain = 0.7dm). This point constitutes an original property of the material.
Moreover, due to capillary tubes, shives are highly sensitive to water: a given mass of shives
absorbs two times its dry density after one minute and half and two times its dry density after ten
minutes.
Finally, after drymg, the air contained in shives constitutes a good heat insulation (hh4.058
W/(m.K) for ~ ~ 1 kg/m3).
5 5 Therefore, shives are used alone in bulk to fill up walls or to put
them under the floor in view of insulating rooms.
2.2 The Binder
The binder tested is a mixture of aerated lime, quicklime and some additives. Quicklime has a
rapid set to develop mechanical properties at early age concrete, whereas aerated lime provides
long term properties. The additives contain an au entrainer admixture to have a thin porosity of
the binder. The whole components of the binder explain its mechanical properties and its brittle
behavior -3). After one month, the compressive strength Rc stabilized and reached a value
from 5 to 6 MPa and the failure strain is less than 2% m.4). It must be noticed that other
binders can be used with hemp particles to make vegetable concrete.
@Seismicisolation
@Seismicisolation
-
2.3 Manufacturing process ani Formu ations c Concrete ontaininq
veqetable particles
This concrete is made of hemp particles, water and binder. A wide range of mixture proportions
are tested: binder content ranges from 1O0 to 530kg for 11Okg of hemp. Very different kinds of
concrete are then obtained, thus expecting si&canîly differingproperties &5,6). In the case
of a low proportion of binder (used as insulahg material for roof), the material appears like a
pile of shives with small link of binder between them. In the case of a high propodon of binder
(floor), particles are entirely coated inside the binder.
Considering the absorption properties of hemp, the particles and two third of the water are first
mixed and the rest of the components are added after. This proceed is chosen to favor the set of
quicklime.
In view of obtaining homogeneous samples, different layers 50 mm in depth are successively
compressed with an electro-mechanic press. indeed, shives are characterized by a high
compressibility and a high Poisson’s effect (v = O.l), so it is difficult to compress in an
homogeneous way thick layers. A compressive stress of 0.05 MPa is applied. Three kinds of
samples are realized for measuring material properties: cylinders (32cm high and 16cm in
diameter) for mechanical tests, slabs (section of 27*27cmz and 5cm of depth) for thermal tests
and prism (section of 8.5*8.5 cm2 and lOcm, 20cm or 30cm long) for acoustical tests. A good
reproducibility of the physical characteristics is observed (Table 3). After, the samples are dried
in a climatic chamber with constant temperature (T=2OoC) and hygrometry (H~=50%).
3 MECHANICAL PROPERTIES
-
3.1 Behavior and Measured Parameters
The properties of the material are evaluated by means of compressive tests. The tests are carried
out using an electromechanical press with a displacement of iOmm per minute. Other strain rates
were studied but the influence on the results is not significant. Samples are tested after 21 days,
3,6,9, 12 and 24 months to evaluate the parameters during a long period due to the setting time
of lime.
The global behavior of concrete is proved to be ductile m . 7 ) . Three fields can be determined
on the mechanical tests curves -4. The first one corresponds to the quasi-elastic domain
(strain inferior to 0.2 mim) where the Young modulus E and the Poisson coefficient v of the
material are measured. The end of this domain corresponds to the yield point. The second one is
the domain between the yield point and the point M, where the stress is maximum. Thus, the
maximum compressive strength CTM and the strain EM are determined. The third domain
corresponds to strains after the maximum stress. The material continues to bear mechanical
efforts without any failure. Repetitive compressive tests give similar values for the parameters
-8).
@Seismicisolation
@Seismicisolation
Concrete containing hemp is a mixture between compressible particles and binder with a brittle
behavior. So, the mixture exhibits an intermediate behavior according to the proportion of
binder. With a low proportion of binder, the behavior recorded is close to that of the shives in
bulk with a ductile behavior. The binder does not play a significant role in the mechanical
strength of concrete. Very high strain can be applied (more than 0.5 d m ) without any failure.
For higher levels of binder content, while the setting is occurring (‘3 months) the behavior is
ductile, but after the setting (9 months) hemp concrete exhibits brittle behavior similar to the
binder one.
3.2 Experimental Results
- Evolution of the Parameters with Time -
The mechanical characteristics’ evolution is presented versus time and quantity of binder in the
mixture -9,lO). The legend indicates both the name of the formulation and the binder
content in mass. The higher the binder content is the higher the maximum stress. During six
months, the maximum compressive strength OM quickly increases. Lime binder is gradually
setting. Then, an inflection of the curve is recorded. After nine months, the properties got
stabilized and the level reached is related to the quantity of binder. The value for GM ranges
between 0.4 MF’a and 1.2 MPa after two years. Simiiar conclusions can be drawn considering E.
The frnal value ranges between 20 and 90MPa. These characteristics are low as compared to
those of other building materials 4).
On the contrary, as expecting the strain EM decreases both with time and binder content. Mixtures
with a significant amount of binder present an increasing rigidity of the samples due to the
continuous matrix of binder surrounding the particles. The final value of EM is around 0.05 mim.
Mixtures with a low proportion of binder presents a higher compressibility level due to porous
vegetable particles. Therefore, the strain will be more important. The binder acts only as a link
between the particles, but has a very small impact on the rigidity. The frnal value of EM is around
0.08 d m . These values are high when compared to conventional materiais used in civil
engineering (Table 4).
- Influence of the Binder Content -
This is the main parameter for the mechanical properties of hemp concrete 1- 1,12). and a
first modeling of the behavior. The maximum stress OM increases strongly with the binder
content (multipíication by three for 15% of binder added). Similar results are obtained for E
(multiplication by five for 15% of binder added). The strain EM. is divided by two for 15% of
binder added.
@Seismicisolation
@Seismicisolation
-
3.3 Conclusions
Hemp concrete demonstrates original mechanical properties as compared to other conventional

building materials. It is characterized by a ductile behavior: the applied strain is 10 to 1000 times
greater without any failure. This is an interesting property in order to prevent crack in case of
differential strains for example. Thus, the risks of a sudden collapse are really limited for houses
ma& with this product. Moreover, the mixtures studied show a large range of maximum stress
OH (between 0.4 and 1.2MF'a) and Young modulus E (40 to 9OMpa). Nevertheless, these values
are not Sumcient to use this concrete in load-beaxing structures.
4 THERMAL PROPERTIES
The aim is to evaluate the themial conductivity h of vegetable concrete. The tests were carried
out using an isothermal box where a sample is submitted to a constant temperature gradient. A
permanent heat flow circulates between the two faces of the sample. h is deduced considering
thermal balance and evaluating heat losses from the study of referenced materials. Then, a
theoretical approach based on self-consistent method gives the conductivity X as a function of the
density and the binder content. First, the raw materials are studied, then the hemp concrete.
4.1 ExperimentalResults
- Raw Materials -
First, the conductivity of the particles used alone is measured. A value of h=0.058 W.m-'.K-' is
obtained for p=155kg.m" which corresponds to the usual density o f partides used as insuiating
material for filling in the walls.
Then, the conductivity of the dried binder is measured to avoid the infiuence of water in the test.
A value of h=0.271 W.m-l.R' for a mean density p=1050 kg.m-' is obtained.
- Concrete containina vegetable particles

Conductivity of various formulations are evaluated during three measurement campaigns. Fo
measurements were made. Density of the dried hemp concrete ranges from 250 and 400 kg.m . Y
-
The values of h lie in between 0.070 and 0.100 W.m-l.K-' &.13), which is of the same order as
dry autoclaved aerated concrete (h CI 0.115 W.m".K' for p 515 kg.m-3) and wood concrete
(p -400 kg.m-' and h 0.15 W.m-'.K') 0. As expected the high porosity of hemp concrete
LJ
leads to good thermal insdating property. The mixtures for low density (210 kg.~n-~) show
thermal conductivity close to those of polystyrene or some mineral wools (rockwool ...)
%0,06 W.m-'.K".
@Seismicisolation
@Seismicisolation
It is important to notice that for materials with high permeability the influence of the convection
phenomenon is limited by the use of copper plate on each face of the samples. The experimental
measurements which are not on the theoretical curves -13) are made without the plates to
have an idea of the influence of the convection phenomenon. The value of the deviation between
theory and experimental measurements is around 15%.
4.2 Theoretical approach
The theoretical approach is based on the self-consistent method (3,4). From the properties of
each components of the mixture, the conductivity coefficient of the macroscopic medium is
determined as a function of the mixture proportion and the density. The determination of the
macroscopic law for heat transfer is based on the energy preserving principle at the local scale.
The thermal energy contained in the heterogeneous cell coated by the equivalent homogenous
medium is assumed to be the same as the energy contained in the equivalent homogenous
medium submitted to the same limit conditions. Two models are used a two-phases model for
shives used in bulk and a three-phases model for hemp concrete.
- Conducüvitv of chives in bulk -

For the conductivity of shives used in b u k without any binder, the model is composed by a cell
of air (L= 0.026 W.m-'.K') embedded in hemp shell (hp, php= 320 kg.m-'). The macroscopic
conductivity of the resulting medium is (5,6,7):
r 1
wheres = &/&p)3= 1 - ph I php
1
+
J*
3 AIAp-1
Finally,
This formula gives the conductivity &, of shives as a function of the density Ph of the shives in
bulk used for insulation. An experimental measurement of shives with the density of 155kg.m.'
gave hh4.058 W.m-'.K-'. Thus, xhp=0.102 W.m.'.R' is deduced considering the density of one
hemp particle php=320kg.m-'.
- Conductivitv of concrete containinq veqetable particles

in this case, the three phase modeling is used -14). The geometric model of the composite
consisted of spherical inclusions embedded in an equivalent homogenous medium, whose
properties should be determined. A three-phase model is used to take into account the whole air
contained on dried vegetable concrete. At the center, there is air with a conductivity ha=0.026
W.m-'.K' and a radius Ra. Then, there is hemp particle with hh=0.058 W.m.'.K-' and a radius
@Seismicisolation
@Seismicisolation
Rh. After, binder with &=0.271 W.m-'.K-' and a radius %. The conductivity of the homogenous
medium is expressed as function of density ratio for each component (5,6).
A coefficient k is defmed by:
k = mass of shives (kg) I mass of dry binder (kg)
This value is fixed for each mixture. Then, two coefficient E and 6 are determined as a function
of k,ph, Pb. They depend on the compaction level of each mixture for k fixed.
The experimental measurements and the theoretical model (continuous curve) give close results
-13). The conductivity of the model tends to the conductivity of air when density decreases
to O. 5 gives a comparison between experiments and theory for the three conventional
formulations roof, wall and floor after drying. The deviation is less than 10%. This model seems
adapted to concrete made with hemp particles for a wide range of formulation.
4.3 Conclusions
The thermal study highlighted some important facts relating to the properties of hemp concrete.
First, the value of thennal conductivity is equivalent to that of other well insulating materials
usually found in civil engineering. h ranges between 0.06 W.m-'.K-' and O.lIW.m-'.F' for p
between in 200 and 450 kg/m3. Then, the theoretical model obtained with the self-consistent
method fits well the experimental data for the various mixture proportions tested.
@Seismicisolation
@Seismicisolation
5 ACOUSTIC PROPERTIES
Acoustical properties for building materials are based on the capacity to absorb, to reflect or to
transmit noise (8). It appears that the property of absorption is fundamental for the global
acoustic comfort in houses. Very often, it is related to the porosity of material, while reflection or
transmission is related to mass and elastic properties. Concrete containing vegetable particles has
a small density (less than 4O0kg.ms3) and a great macroscopic porosity. So, the absorption
coefficient a is evaluated by means of an experimental device, the ?Kundt tube?. The impact of
various parameters on acoustical performance levels is determined.
5.1 Parameters
The coefficient of absorption a is defined as the ratio of the energy reflected by the material to
the energy provided by the source. It is considered that there is no transmission of energy
through the material. The whole energy is absorbed or reflected. The source transmitted a
standing wave with a frequency ranging from 100 and 2000 Hz,which corresponds to the range
of frequencies used to characterize building materials. Various sample formulations floor, wall
and roof and lengths 1O, 20 and 30cm are tested In a first part, experimental values recorded for
a are given. In a second part, the influence on a of the two parameters chosen is evaluated.
5.2 Experimental results
For each case, a is plotted F&.15,16,17 for each frequency in the range [IOOHz- 2kHz]. The
mean curves obtained for each kind of samples considering the tested formulations and lengths
are given. The legend gives both the name of the formulation and to the thickness of the sample
tested in cm.
First, the coefficient of absorption is higher than 0.5 for almost all the cases considered more
than a half of the energy is absorbed by hemp concrete. Moreover, there are at least two peaks of
absorption for the different formulations with high values (between 0.7 and 1). Thus, this
concrete can be used to absorb totally particular frequencies. This fact gives the possibility of
building acoustic screen focused on special frequencies.
These results are obtained by testing directly the surface made with vegetable concrete (high
open porosity of the matenal) without any coat. It must be noticed that the frequencies used
corresponds to wavelengths in between 16.5 cm and 3.3 m. This lengths are very high in
comparison with the size of the pores of hemp concrete.
@Seismicisolation
@Seismicisolation
-
5.3 Influence of the parameters
Considering the thickness of the samples, an evolution of the curves is noticed for the different
formulations. Indeed, the peaks of absorption globally move back to low frequencies when the
length increases from 10 to 20 crn and the maximum values of the peaks tend to decrease. in fact,
low frequencies correspond to high wavelengths. So, when the length of the sample is too short,
the mechanical compressibility has a more important effect than the absorption in the pores. The
waves cannot enter in the pores of the surface and are less absorbed. For high frequencies (short
wavelengths), the waves can more easily enter in the material and be absorbed by the capillary
tubes. Then, the properties of absorption seems to stabilize when the length of samples with a
high rate of binder increases to 30 cm.
Considering the quantity of binder, the effect on acoustical absorption seems to be linked to the
porosity of the material. indeed, whatever the length, the mixture roof, which is the most porous,
absorbs between 10 and 20% more than the other formulations. Concrete containing a great part
of particles presents a macroscopic structure close to the shives one. On the contrary, concrete
with a high rate of binder presents a less porous structure and a higher rigidity of the samples.
The continuous die surrounds the particle and the size of the pores of the binder is smaller. So,
the sound is less absorbed.
To conclude, F&18 shows the values for a of classical materiais. This values are given by
octave. The range of frequencies in between 1OOHz and 2 kHz can be divided into five areas
characterized by (fi+,-fi)/fi =constant and f+1=2*f,. It can be noticed that the coefficient of
absorption is lower to 0.5 whereas it is rarely the case for concrete made with vegetable particles.
-
5.4 Conclusions
Conventional building materials have a high density and a small porosity. Thus, they reflect
sounds instead of absorbing them. On the contrary, concrete containing vegetable particles
proved to be really efficient considering the absorption properties. Whatever formulation, a
higher than 0.5 are obtained, which means that 50% at least of the energy is absorbed. This fact
entails a good acoustical comfort. The last stage of the study addressed the performance of more-
various formulations.
@Seismicisolation
@Seismicisolation
6 CONCLUDING REMARKS
Through concrete containing vegetable particles, a new enViroment-f?iendly material is

studied, taking into account the mechanical, thennai and acoustical properties. The porosity of
the vegetable aggregates leads to interesting non usual properties in each field. Due to the shives
compressibility, this concrete with hemp particles has a ductile behavior and can bear very high
strain without any failure.
According to the binder content and the compaction level, various formulations with densities
ranging between 250 and 400 kgm-' are tested. Indeed, the different properties of vegetable
concrete goes on opposite ways according to the formulation.
A maximum stress O M lying between 0.4 and 1.2MPa is recorded, whereas the Young Modulus
lies between 20 and 90MPa and the strain EM ranges between 0.04 and 0.1 d m . The thermal
conductivity IC ranges from 0.06 to 0.10 W.m".K-'. A theoretical model developed through the
self-consistent method fits well the experimental measurements and enables to evaluate h for
other mixture proportion. Lastly, the acoustical properties recorded far better than those of
conventional building materials. The coefficient of absorption is superior to 0.5 for the whole
formulation.
Globally, this concrete containing hemp particles is interesting considering the fact that the good
properties in each field can be modified by changing the binder content or the compaction level
t o adapt it for specific uses.
ACKNOWLEDGEMENTS
We thank the LCDA company, manufacturing hemp, and the Strassevil company, a binder
manufacturer for their funding of the study, ADEME for their support and X. Olny for the
acoustical measurements.
REFERENCES
ARNAUD, L. CEREZO, V. Qualification physique des matériaux de construction à base de

chanvre. E.N.T.P.E.(Laboratoire de Géomatériaux), 2001.95 pages. Rapport de recherche, Contrat
CNRS no 0711462.
GARCIA-JALDON,C. Caracthisation morphologique et chimique du chanvre (cannabis sativa),

prkaitement à la vapeur et valorisation .Thèse doctorale :Universitt Joseph Fourier (Grenoble i),
1995.217 pages.
HASHIN, Z. Assessment of &e self-consistentscheme approximation: conductivity of particulate

composites. J.Comp.Mater, 1968, v01.2, p.284-300.
HERVE, E. et ZAOUI, A. Modelling the effective behavior of nonlinear matrix-inclusion
composites. European Journal of Mechanics. MSolids, 1990, ~01.6,p.505-5 15
@Seismicisolation
@Seismicisolation
[ 5 ] BOUTIN, C. Conductivité thermique du béton cellulaire autoclavé : modélisation par méthode
autocohérente. Matériaux et Constructions, vo1.29, p609-615.
[ 6 ] ARNAUD, L. MONNET, H.SALLET, F. Mechanical and themial properties of vegetal concrete.

In proc. Sustainable construction into the next Millennium: EI'IV¡IOMe.Inentdly ñiendly and
innovation cernent based materials, 2-5Nov. 2000, Elsevier Ed. N. P. Barbosa, R.N. S w a y et C.
Lynsdaie, pages 302-311.
[ 7 3 ARNAUD, L. MONNET, H. SALLET, F. Modélisation par homogénéisation autocohérente de ia
conductivité thermique des bétons et laines de chanvre. In proc. Congrés fiançais de thermique,
15-17 mai 2000, Eisevier Ed. A. Lallemand et J.F. Leone, pages 543-548.
[ 8 ] OLNY, X. , Thèse doctorale MEGA, 1999,281~
[ 9 ] ARNAUD, L. Properties of Hemp Concrete, 3" Internationnal Symposium Bioresource Hemp

2000. Hannovre (Allemagne). 2000.
@Seismicisolation
@Seismicisolation
-
Table 1: Density of shives in bulk
(kg.rn-3
r1
I
I
2
3
I
I
I
123.6
126.0
130.3
I
I
I A
Table 2: Some special mixture proportions of fresh concrete containing hemp particles
given in kg
NUMER t K t S H MtAN ü t N p
ROOF 15 456 kglrn" 210 kglrn'
WALL 15 678 kglrn" 330 kglm'
FLOOR 13 797 kglrn' 400 kglrn'
Table 4: Mechanical characteristics of some examples of materials used in civil engineering
=e 5: Theoretical and experimental conductivity of vegetable concrete
@Seismicisolation
@Seismicisolation
_. - - /
/
O 0.1 0.2 03 0,4 0.5 0.6 0.7

AXIAL STRAIN
F& 1 :Mechanical behavior of hemp alone
4, I
&F 2: Hemp particles

F&3 :Comparison between two kinds of
compressivestrength test for the binder
7 , I
sms
BMDER
O 2 MONRI4 6 8
&4 :Re of the binder F& 5: Extreme stmctures of concrete

containing vegetable particles
@Seismicisolation
@Seismicisolation
-
Fig. 6 Different kind of concrete with vegetable particles with a growing proportion of
binder
0.5
O .4
2
Y
0.3 :s:
Y
o
0.2 E
1
0.1
o
o 0.02 0.04 0.06 0.08 0.1 0.1 2
AXIAL STRAIN (mrm)
Fig. 7: Example of compressive curves and Poisson’s coefficientfor vegetable concrete

-
STRAIN
O 0.05 0.1 0.15 0.2

AXIAL STRAIN (mlm)
Fig.
-8: Ductile behavior of concrete containing hemp particles
@Seismicisolation
@Seismicisolation
I.4 042.87%
- - - oo - - - - - - - 8
1.2 : A3725'h
1.0 1
,..
... .. - 0 - L
FLOOR (34.8%)
E :
, A 035.67%
B A *WALL(33.7%)
A
-
u
+J3.60%
0 030.51%
0 -29.92%
.ROOF (24.6%)
' . " ' . """"-$
O 5 10 15 20 5 30
-
Fie. 9
I
1O0
80
e
Z 60 A FLOOR (34.8%
v>
3 .WALL (33.7%)
8
I
40 33.60%
8
$ 20
ROOF (24.6%]
O
. .
O 5 10 MONTH 15 20 25 30
-
Fig. 10 : Young modulus of different mixtures of concrete made with hemp particles
A 15 motnr
i 9 months
o 6 months
= 3 months
- ~~
20% 25% 30% 35% 40% 45%

% OF BINDER
-
Fig. 11: Maximum compressive strength as a function of the binder content (Yo in mass)
@Seismicisolation
@Seismicisolation
120 -
20% 25% 40% 4 %

30%96 OF BINDER
-
Fig. 12: Young Modulus as a function of the binder content (YOin mass)
0,120
-THEORY FLOOR
0,110
-THEORY WALL
u^ -THEORY ROOF
E 0,100
ROOF
E
E
P
r
2
0.090
0,080
.WALL
FLOOR
o
z
O
O 0.070
-I
2
5 0.060
+
L
0,050
0.040
175 225 275 325 375 425
DENSKY (kg.m-3)
Fig. 13: Thermal conductivity of vegetable concrete with different mixtures

-
@Seismicisolation
@Seismicisolation
Homogenous medium (kq)

Binder (Lb;RJ
Hemp particles (h,
Rd
Air (La)
Fig. 14: Three-phase model
1,o 1.o
0.9 0.9
0.8 oa
0.7 0.7
0.6 O8
f 0.5 5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
loo0 1MM .im HEñlZ 1wO IWO0
Fig. 15: Coefficient of absorption for roof
- -
Fig. 16: Coefficient of absorption for waii
rm rn lmi rom
im fo3 m lua m
Fig. 17: Coefíkient of absorption for floor
- -
Fig. 1 8 Coefficient of absorption of other
materiais
@Seismicisolation
@Seismicisolation
SP 209-10
Sustainable and Durable Reinforced

Concrete Construction
by V. Corinaldesi, G. Moriconi, and F. Tittarelli
Svnopsis: Recycled aggregates used in combination with high-volume fly ash is an

example of sustainable construction material because it represents a judicious use
of resources by recycling by-products, that results in a lower environmental impact
through reduced carbon dioxide emissions and reduced natural aggregate extrac-
tion from quarries. Furthermore, the related concrete mixtures yield satisfactory
mechanical performance. The goal of the experimental work reported here was to
investigate the effect of recycled aggregate and/or fly ash on carbonation and chlo-
ride penetration depth, as well as the effect on corrosion behavior of either bare or
galvanized steel in cracked reinforced concrete. The concrete mixture contained
equal amounts of fly ash and cement. The results show that the introduction of the
sustainability concept in concrete technology by using recycled aggregate and/or
fly ash did not cause any deleterious effects on durability of reinforced concrete
specimens in terms of both chloride and carbon dioxide penetration, and reinforce-
ment corrosion in cracked concrete.
Kevwords: cracked reinforced concrete; durability;

galvanized reinforcement; high volume fly ash;
recycled-aggregate concrete; sustainable construction
@Seismicisolation
@Seismicisolation
169
170 Corinaldesi et al.
Valeria Corinaldesi is a research civil engineer and has a Ph.D. in Materials
Engineering. She is developing a research project on building materials and their
recycling at the Department of Materials and Environment Engineering and
Physics at the University of Ancona, Italy.
Giacomo Moriconi is Professor of Materials Science and Technology at the

University of Ancona, Italy and is the Head of the Department of Materials and
Environment Engineering and Physics. He is the author or co-author of
numerous papers in the field of cement and concrete technology and building
materials performance and durability.
Francesca Tittarelli has a Ph.D. in Materials Engineering, and is Researcher in

Materials Science and Technology at the Department of Materials and
Environment Engineering and Physics at the University of Ancona, Italy.
INTRODUCTION
The manufacture of durable concrete for buildings can contribute to

sustainable construction (I) based on technology and materials that preserve
natural resources and improve the environmental quality. For example, some
building materials that are no longer able to fulfill their original task can be
processed in such a way that they can be used as aggregate for concrete (2).
Replacing natural aggregate with recycled aggregate in concrete protects the
environment by reducing the impact on quarries for virgin aggregate and by
reducing of the volume of rubble disposed to landfills. Similarly, the use of fly
ash in concrete enables to recycle an industrial waste product. In particular, due
to its pozzolanic activity, fly ash can be used as a partial replacement for
cement, thus reducing the energy consumption and carbon dioxide emissions
related to cement production (3).
Previous experiments (4) showed the feasibility of making structural

concrete mixtures with recycled aggregate and high-volume fly ash. In
particular, the results showed that the use of fly ash in recycled-aggregate
concrete refined the pore structure by reduction of macro-pores volume. This
had beneficial effects on mechanical properties such as compressive, tensile, and
bond strength. In this way, performance similar to an ordinary concrete mixture
prepared with natural aggregate was achieved except for somewhat lower
stifbess of the recycled-aggregate concrete mixture ( 5 ) .
In this study, the durability aspects are reported both in the absence
and presence of cracks which often appear in concrete structures as a
@Seismicisolation
@Seismicisolation
consequence of plastic settlement, shrinkage, creep, thermal stress, dynamic
loading, etc. (6-8). A review of codes of recommended practice for the design of
concrete structures, in particular British Standard 81 10 (9) and Eurocode 2: Part
1 (IO), reveals that crack width should not exceed 0.3 mm. In short, the aim of
this work was to determine whether the introduction of sustainability concepts
in concrete manufacturing, by using recycled aggregate and high-volume fly
ash, as a cement replacement material would have any adverse effect on the
durability of reinforced concrete both in terms of chloride and carbon dioxide
penetration, and corrosion of bare or galvanized reinforcement embedded in
cracked concrete.
MATERIALS
Portland Cement
A commercial portland-limestone blended cement type CEM IIIA-L

42.5 R according to the European Standards EN-I9711 was used. The Blaine
fineness of cement was 0.415 m21g and density was 3.05 kglm3. The chemical
composition of cement is reported in Table 1.
Natural Awreaates
Crushed limestone aggregate (15 mm maximum size) and natural sand

(6 mm maximum size) were used. The gradation of both crushed aggregate and
sand are shown in Fig.1 and the physical properties are shown in Table 2.
Recycled AagrePates
A coarse fraction (15 mm maximum size) and a fine fraction (6 mm

maximum size) of recycled aggregate were used. These fractions were directly
supplied by an industrial crushing plant in Villa Musone, Italy, where the rubble
from building demolition were suitably selected, crushed, cleaned and sieved.
The gradations of both coarse and fine recycled aggregate are shown in Fig. 1 and
the physical properties are shown in Table 2.
Superplasticizer
A 30% aqueous solution of an acrylic-based admixture was added,

when required, to achieve desired workability.
Mineral Admixture
A low calcium fly ash (ASTM C 618 Class F) produced by a thermal

power plant was used. The Blaine fineness of fly ash was 0.448 m’lg and its
@Seismicisolation
@Seismicisolation
specific gravity was 2.25 kg/m3. The chemical composition of fly ash is reported
in Table 1 .
MIXTURES PROPORTIONS
The mixtures proportions are reported in Table 3.
When proportioning concrete, the two kinds of aggregate - recycled or

natural - were used with the same gradation (15 mm maximum diameter),
obtained by combining the fine and the coarse sieve fractions, according to the
Bolomey particle size distribution curve (1 1) and assuming the physical property
values reported in Table 1 for each kind of aggregate. Moreover, the mixture
workability in relation to the particles shape, as well as the cement dosage, were
taken into account in order to optimize the gradation.
Since no detection of the actual moisture content of the aggregates was

carried out, the amount of mixing water was adjusted to obtain the same 150-180
mm slump.
Reference concrete, labeled as Nat-0.6, was manufactured by using

only natural aggregates with a water to cement ratio of 0.60 as commonly used
in practice. Further concrete (Nat+FA-0.6) containing only natural aggregate
was prepared with the same wic of 0.60, by adding to the mixture as much fly
ash (FA) as cement. However, in this way a water to cementitious material ratio
of 0.30 was obtained. In order to maintain the same workability level, a water-
reducing admixture was added; actually, an acrylic-based superplasticizer was
used at a dosage of 2.0% by weight of cement.
Then, a recycled-aggregate concrete (Rec-0.30) was prepared, by

completely substituting natural aggregate with recycled aggregate, with a wic of
0.30 when recycled aggregate was used to achieve the same strength grade as
the reference concrete. In order to decrease the cement content an acrylic-based
superplasticizer was added at a dosage of 1.O% by weight of cement.
Finally, a recycled-aggregate concrete (Rec+FA-0.6) was prepared with

a wlc of 0.60, by adding to the mixture as much fly ash as cement to replace the
fine aggregate fraction. However, in this way a water to cementitious material
ratio of 0.30 was obtained. In the presence of fly ash a dosage of 1.8% by
weight of cement of an acrylic-based superplasticizer was also necessary to
achieve the desired workability.
@Seismicisolation
@Seismicisolation
PREPARATION, CURING AND TESTING OF SPECIMENS
Compression Tests
Cubic specimens, 100 mm in size, were manufactured according to

Italian Standards UNI 6130-72 Part I; these specimens were cast in polystyrene
forms and wet cured at 20°C (UNI 6130-72 Part II).
Compression tests according to Italian Standards UNI 6132-72 were

carried out. The bearing faces were sufficiently planar and smooth as not to
require any capping or grinding. The specimens were loaded at a constant strain
rate until failure.
For each mixture and each curing time (3, 7, 28 and 56 days) three
specimens were used for mechanical tests, according to Italian Standards.
Carbonation Depth
The carbonation depth was measured by the phenolphthalein test

(RILEM CPC - 18) on cubic concrete specimens, 100 mm in size, exposed to
air at a temperature of 20°C after demoulding at 1 day.
Chloride Penetration
The chloride penetration into concrete was evaluated through the silver
nitrate and fluorescein test (12). Both solutions were sprayed on the two
fractured surfaces obtained by splitting concrete specimens, 100 mm in size,
exposed to a 10% sodium chloride aqueous solution after wet-curing for 1 week
and air-curing for 3 weeks at 20°C.
Corrosion Tests
Prismatic concrete specimens (280 mm x 70 mm x 70 mm) were

manufactured for electrochemical measurements. Each prismatic specimen was
reinforced with a steel plate (210 mm x 40 mm x 1 mm), bare or galvanized,
embedded with 3 cm concrete cover. Reinforcing plates instead of common bars
were used for two reasons. Firstly, plates can assure that the specimen cracked
without splitting; secondly, steel plates can assure more significant anodic area
at the crack apex. The zinc coating on galvanized steel plate was 100 pm thick,
obtained by molten zinc immersion, with an outer pure zinc layer about 20 pm
thick. The galvanized steel plates, just before embedment in the fresh concrete,
were submerged for 5 sec in a 15% sodium hydroxide solution to dissolve the
zinc carbonate layer formed during atmospheric storage. The electric contacts
between the reinforcing plates and the measurement equipment were arranged
according to previously reported methodology (13).
@Seismicisolation
@Seismicisolation
After 1 month of air curing at 20 OC, the specimens were cracked by
flexural stress so that a crack width of 1 mrn was produced in a pre-formed
notch area with the apex crack reaching the reinforcement. Then the specimens
were exposed to weekly wet-dry cycles (2 days of drying followed by 5 days of
wetting) in a 10% sodium chloride solution.
The corrosion risk of the reinforcement in the concrete specimens

exposed to the aggressive environment was evaluated by corrosion potential
measurements with a saturated calomel electrode (SCE) as reference, ?whilethe
rate of the corrosion process was monitored by polarization resistance
measurements through the galvanodynamic method using an external graphite
bar as counter-electrode and by calculating the average value between the anode
and the cathode.
The electrochemical values reported in the following graphs are averages

of the measurements carried out on three specimens of each type during the full
immersion period.
In order to validate and to complete the evaluation of the electrochemical

behavior, the concrete specimens were saw-cut after 7 wet-dry cycles in the
chloride solution to assess the corrosion by visual observation. All the steel
plates were removed after splitting the concrete specimens. The surface of the
corroded area on bare steel plates was .evaluated after pickling, while
metallographic analysis was carried out on the cross section of the galvanized
steel plates to evaluate the coating thickness decrease due to the corrosive
attack.
Cornmession Tests
Figure 2 shows the compressive strength as a function of time for the

four concrete specimens. It can be noticed that after 28 days of curing the
compressive strength was in the range of 27-31 MPa except for Nat+FA-0.6,
which showed higher values up to 38 MPa. It can be also observed that, at early
ages, the W/C had more effect than the type of aggregate on the strength
development. However, the use of a weaker aggregate (the recycled aggregate),
had more influence on the compressive strength as the cement matrix became
stronger. This was therefore the ?weak link? in the chain (14).
Effective pozzolanic activity is shown by fly ash in the presence of

both recycled and natural aggregate even at early ages, and more particularly at
longer ages. As a matter of fact, fly ash enabled a strength increase of about
59% for the natural-aggregate concrete with the same W/C.
@Seismicisolation
@Seismicisolation
Carbonation Depth
Figure 3 shows the carbonation depth (x, in mm) as a function of the

time of exposure to air (t, in days). A linear relationship between the
carbonation depth and the square root of time ( x = k - &) can be drawn; the
extrapolated values of x after 1 year were 8.5, 6.8, 5.7 and 2.8 mm respectively
for the concretes Nat-0.6, Rec+FA-0.6, Rec-0.3 and Nat+FA-0.6.
A strong positive contribution of fly ash addition is detectable (by

comparing the results obtained for Nat-0.6 and Nat+FA-0.6) mainly due to the
refinement of the pore system.
In general, it is confirmed that, for concretes containing high volumes

of fly ash, carbonation did not present any problems for corrosion of the
reinforcements, because of the very low permeability of concrete, even when a
porous aggregate, such as the recycled aggregate, was used (see in Fig. 3 the
carbonation depth in Rec+FA-0.6).
Chloride Penetration
Figure 4 shows the chloride penetration depth as a function of time of

exposure to the 10% sodium chloride aqueous solution afler water saturation of
concrete specimens. In Fig. 4 data are reported after about one month immersion
because, before this time, instability phenomena due to chloride binding with
the cement matrix (15) causes interference of chloride migration.
Collepardi et al. (1 2) found that chloride penetration depth (x) varies

with elapsed time (t) following an equation [i] obtainable from the solution of
Fick's second law under non-steady-state conditions for diffusion in a semi-
finite solid:
where D is the diffusion coefficient of chloride ions into concrete pores filled of
water, expressed in cm2 * s-' . lo-*. The values of D yobtained from [i] by
interpolating the results showed in Fig. 4 for the different concretes, are reported
in Table 4.
It is quite evident the strong beneficial effect due to the presence of fly
ash on the chloride penetration depth measured for Rec+FA-0.6 and Nat+FA-0.6
concretes; in fact, the chloride ion diffusion coefficients into them are one order
of magnitude less than those into the other concretes.
Collepardi et al. (12) also reported experimental results supporting the

hypothesis that the different nature of pore surfaces formed in the presence of
pozzolanic materials played an essential role in affecting the concrete behavior
during the penetration of chloride ions. Actually, when the concrete contains fly
@Seismicisolation
@Seismicisolation
ash, chloride binding by the cement matrix significantly increases, as compared
to reference concrete made without fly ash.
A positive influence of the lower waterlcement on the resistance

against chloride penetration can be also observed. As a matter of fact, the
diffusion coefficient of chloride ions into Nat-0.6 concrete is more than double
with respect to Rec-0.3, although recycled instead of natural aggregate was
used. Nevertheless, the aggregate porosity certainly had some effect on the
concrete permeability to chloride (compare Nat+FA-0.6 and Rec+FA-0.6 in Fig.
4). In fact, Zhang et al. (16) tested some lightweight aggregate concretes and
concluded that the permeability of concrete was more dependent on the cement
matrix than the lightweight aggregate porosity.
Corrosion Tests
Bare Steel -- Figure 5 and 6 show respectively the free corrosion potential
values and corrosion rates of bare steel plates embedded in cracked concrete as a
function of wet-dry cycles. Just after the exposure to the chloride environment, all the
steel plates assumed activation values lower than -500 mV/SCE, reflecting a general
high corrosion risk, regardless of the cement matrix type. At the same time, the related
polarization resistance values did not change significantly with the cement matrix,
indicating similar corrosion rates for the different concretes. Therefore, the
opportunity of manufacturing ecocompatible concrete, by replacing natural with
recycled aggregate andor by adding high fly ash volume, does not seem to
negatively affect the corrosion behavior of embedded steel reinforcements when
an adequate strength class is guaranteed. Moreover, the results obtained could
dispel any doubt on the uncertain corrosion behavior of steel in high volume fly
ash concrete, due to the concrete pore solution alkalinity reduction caused by the
fly ash pozzolanic activity, in the presence of cracks in concrete.
The electrochemical tests represents a partial feature of the

reinforcement corrosion behavior since instantaneous measurements can only be
carried out during the wet period and there is no chance to estimate the metal
surface area effectively corroding. Therefore, after 7 wet-dry cycles all the
specimens were saw-cut in order to visually evaluate the corroded area and to
assess the weight loss of the bare steel plates after pickling.
All the reinforcements were characterized by a strong corrosive attack

in the crack area justified by the high chloride concentration detected on the
reinforcement surface (2-5% by cement weight) widely overcoming the
concentration threshold (0.4% by cement weight) generally reported as the
critical value able to induce the corrosion process in bare steel. However, from
a morphological point of view (Fig. 7), in the presence of recycled aggregate
and, particularly, of fly ash the corrosive attack appears more diffuse and less
penetrating with respect to that detected on the steel plate embedded in the
reference concrete.
@Seismicisolation
@Seismicisolation
Galvanized Steel -- Figures 8-9 show respectively the free corrosion
potential and the related polarization resistance of galvanized steel plates
embedded in cracked concrete as a function of wet-dry cycles. As bare steel,
also in this case the electrochemical measurements showed no significant
decrease of the corrosion resistance of galvanized steel embedded in concrete
mixtures containing recycled aggregate and fly ash. Indeed, the corrosion risk
seems to be even lowered in recycled-aggregate concrete (see Rec-0.3 in Fig. 8).
The evaluation of saw-cut concrete specimens, as already described for

bare steel, revealed additional unexpected information to that obtained by the
electrochemical measurements. As a matter of fact, in the reference concrete
(Fig. loa), far from the crack apex, Fe-Zn alloy appeared on the plate surface,
meaning total consumption of the pure zinc layer due to the corrosive attack, as
later confirmed by metallographic analysis (Fig. lob). On the other hand, the
galvanized steel plates extracted from recycled-aggregate concrete showed a
lower corrosive attack. Zinc grains are still visible on the galvanized plate
surface after the aggressive exposure (Fig. l i a ) and the metallographic
observation revealed a continuously thick pure zinc layer still present on the
reinforcement (Fig. 11b). Similar results were obtained when high-volume fly
ash was used.
CONCLUSIONS
The introduction of the sustainability concept in concrete manufacturing

by using recycled aggregate andor high-volume fly ash showed no deleterious
effect on durability of reinforced concrete specimens.
In fact, the addition of fly ash was very effective in reducing the
carbonation and chloride ion penetration depths in concrete, even when recycled
instead of natural aggregate was used. In the presence of cracks in the concrete
cover, the use of sustainable concrete mixtures did not decrease the corrosion
resistance of steel reinforcement, as long as the concrete strength was adequate.
REFERENCES
1. Mehta, P.K. 1997, Bringing the Concrete Industry into a New Era of
Sustainable Development, Proceedings of the Mario Collepardi Symposium
on “Advances in Concrete Science and Technology”, P.K. Mehta Ed., Rome,
Italy, October 7-10, 1997, pp. 49-67.
2. O’Brien, K. 1998, Using Recycled Aggregate Concrete as Part of an Overall

Approach to Sustainable Construction, in “Use of Recycled Concrete
@Seismicisolation
@Seismicisolation
Aggregate”, R.K. Dhir, N.A. Henderson, M.C. Limbachiya Eds., Thomas
Telford Publishing, London, Great Britain, 1998, pp. 459-470.
3. Malhotra, V.M. and Bilodeau, A. 1999, High-Volume Fly Ash System: the
Concrete Solution for Sustainable Development, Proceedings of the
International Symposium on “Concrete Technology for Sustainable
Development in the Twenty-First Century”, P.K. Mehta Ed., Radha Press,
New Delhi, India, 1999, pp. 43-64.
4. Corinaldesi, V. and Moriconi, G . 2001, Role of Chemical and Mineral

Admixtures on Performance and Economics of Recycled Aggregate
Concrete, Proceedings of the Seventh CANMET/ACI International
Conference on “Fly Ash, Silica Fume, Slag and Natural Pozzolans in
Concrete”, V.M. Malhotra Ed., July 22-27,2001, Madras, India, pp 869-884.
5. V. Corinaldesi, F. Tittarelli, L. Coppola and G. Moriconi, Feasibility and

Performance of Recycled Aggregate in Concrete Containing Fly Ash for
Sustainable Buildings, Proceedings of the CANMET/ACI Intemational
Symposium on “Sustainable Development of Cement and Concrete”, San
Francisco, U.S.A., September 16-19,2001, Ed. by V.M.Malhotra.
6. Arya, C. and Smith, D.N. Crack Control and its Relevance to the Durabiliw
of Reinforced Concrete, Third CANMET/ACI International Conference on
“Durability of Concrete”, Nice, France, 1994, Supplementary papers, pp.
329-340.
7. Coppola, L., Fratesi, R., Monosi, S. and Troli, R. Corrosione dell Acciaio in
Strutture in Calcestruzzo Immerse in Acqua di Mare, Proceedings of the
Third National Congress AIMAT, Napoli, Italy, September 25-27, 1996, pp.
46 1-473.
8. Suzuky, K., Ohno, Y., Praparntanatorn, S. and Tamura. H. Corrosion of

Reinforcement in Concrete Construction, C.L. Page, K.W.J. Treadaway and
P.B. Bamforth Eds., Elsevier Applied Science, London, 1980, 167.
9. British Standard Institute: BS8110: Part 1: 1985 “Code of practice for the
structural use of concrete”, London, BSI, 1985.
10. British Standard Institute: “Eurocode 2: Part 1: 1992 Design of concrete

structures: General rules and rules for buildings”, London, BSI, 1992.
1 1. Collepardi, M., “Concrete Science and Technology” (in Italian), Third

Edition, Hoepli, Milano, Italy, 1991.
12. Collepardi, M., Marcialis, A. and Turriziani, R., ‘Penetration of Chloride

Ions into Cement Pastes and Concretes’, Journal of American Ceramic
@Seismicisolation
@Seismicisolation
Socieíy, V.53, 1972, pp.534-535.
13. Fratesi, R., Moriconi, G. and Coppola, L. 1996, The influence of steel
galvanization on rebars behaviour in concrete? Corrosion of Reinforcement
in Concrete Construction, C.L. Page, P.B. Bamforth and J.W. Figg Eds., SCI
Special Publication No. 183, 1996, pp. 630-641.
14. Corinaldesi, V., Isolani, L. and Moriconi, G., Use of rubblefiom building
demolition as aggregates for structural concretes, Proceedings of the 2nd
National Congress on Valorization and Recycling of Industrial Wastes,
L’Aquila, Italy, 1999, pp. 145-153.
15. Pauri, M., Monosi, S., Alverà, I. and Collepardi, M. ‘Assessment of Free and
Bound Chloride in Concrete’, Materials Engineering, V.l, No 2, 1990, pp.
497-501.
16. Zhang, M.H. and G j m , O.E. ‘Permeability of high-strength lightweight

concrete’, ACIMaterials Journal, V.88, No 5 , 1991, pp. 463-469.
Table 1 - Chemical analyses of cement and fly ash (LOI-free basis).
Oxide (%) Cement Fly Ash

Si02 29.67 59.94
A1203 3.74 22.87
Fe203 1 .80 4.67
Ti02 0.09 0.94
Ca0 59.25 3.08
Mg0 1.15 1.55
so3 3.25 0.35
K20 0.79 2.19
Na20 0.26 0.62
L.O.I. 11.62 3.34
@Seismicisolation
@Seismicisolation
Table 2 - Physical properties of the aggregate.
Specific Gravity Water Absorption Passing 75 pm

Aggregate Fractions
(kg/m3> (%I ("/.I
Natural Sand 2620 3 0.9
Crushed Aggregate 2680 2 0.2
Fine Recycled Fraction 2150 10 0.5
Coarse Recycled Fraction 2320 8 0.3
Table 3 - Concrete mixture proportions.
Mixture Nat-0.6 Nat+FA-0.6 Rec-0.3 Rec+FA-0.6

WIC 0.60 0.60 0.30 0.60
WICM 0.60 0.30 0.30 0.30
Mixture proportions, kg/m3
Water 230 165 165 230
Cement 380 275 550 380
Fly Ash 275 380
Natural Sand 314 298
Crushed Aggregate 1338 1269
Fine Recycled Fraction 372
Coarse Recycled Fraction - 1060 1057
Superplasticizer 5.5 5.5 6.8
Table 4 - Diffusion coefficientsof chloride ions into different concretes at 20°C.
Mixture Nat-0.6 NatiFA-0.6 Rec-0.3 ReciFA-0.6

D
0.556 0.017 0.230 0.043
(cm2* s-' .
@Seismicisolation
@Seismicisolation
1O0
90
80
.-
rA
70
2 60
.I 50
v)
& 40
30
20
10
O
0.01 0.1 1 10 1 O0
Sieve opening (mm)
Fig. 1 - Grain size distribution curves of the aggregate fractions.
45
5
O
O 10 20 30 40 50 60
Curing Time (days)
Fig. 2 - Compressive strength of different concretes as a function of curing time.
@Seismicisolation
@Seismicisolation
10
9
- 8
E
E
u 7
’ 6
Iy
LEI
c 5
.-
O
c,
2 4
O
- 2 3
B 2
1
O
O 5 10 15 20
Exposure time (daYsIn)
Fig. 3 - Carbonation depth as a function of the time of exposure to air.
30 -_
h
E 25
v
5
U
E
2o
--
--
ONat+FA-0.6
ONat-0.6
ARec-0.3
+Rec+FA-O.ó
t _ _ _ ~ _---
-
.o ..
111
--- .__
,d / ... I
.-
2* 15
10 -_
;
8 s ___-_
A
A -, .
A
_-__
v
Fig. 4 - Chloride penetration as a function of the time of exposure

to a 10% sodium chloride aqueous solution.
@Seismicisolation
@Seismicisolation
-400 ~
-500
W
.-c
I
(d
C
8
c
.-m -600
L
e
u
-700 ! I I I I I
I
1 2 3 4 5 6 7
Wet-Dry Cycles
Fig. 5 - Corrosion potential of bare steel plates embedded in cracked concrete

as a function of wet-dry cycles.
160
150
n
E 140
c
130
8C
8
m
120
4
.I
110
C
.o
.L
100
(d
-2
(d
90
I
2 80
70
1 2 3 4 5 6 7
Wet-Dry Cycles
Fig. 6 - Polarization resistance of bare steel plates embedded in cracked

concrete as a function of wet-dry cycles.
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
180
-
E
a
160
g 140
::
:
.I
120
p:
t 100
O
.I
c)
-
'5
80
2 60
40
1 2 3 4 5 6 7
Wet-Dry Cycles
Fig. 9 - Polarization resistance of galvanized steel plates embedded in cracked

concrete as a function of wet-dry cycles.
Fig. 1 O - Visual observation (a) and metallographic cross section (b)

of the galvanized steel plates embedded in reference concrete (Nat-0.6).
@Seismicisolation
@Seismicisolation
(4 (b)
Fig. 1 1 - Visual observation (a) and metallographic cross section (b)

of galvanized steel plates embedded in recycled-aggregate concrete (Rec-0.3).
@Seismicisolation
@Seismicisolation
SP 209-1 1
Experimental Results and Validation

of the Design Technique Used for
FRP Strengthening of RC Bridges
by D. I. Kachlakev
Svnopsis: The Horsetail Creek Bridge (HCB), constructed in 1912, is located

along the Historic Columbia River Highway in Oregon. The cross beams of this
historic structure were found to be 50 percent deficient in flexure and 94 percent
deficient in shear, mainly due to the traffic loads increase. Analysis of the altema-
tive designs indicated that glass FRP (GFRP) laminates would be most suitable for
shear strengthening, while carbon FRP (CFRP) laminates would be best for flexur-
al capacity enhancement. Concurrently, four full size beams, as similar as possible
to the actual bridge beams, were constructed to simulate the retrofit of the bridge.
One of the beams served as a control; one beam was strengthened for shear capac-
ity increase only; one beam was reinforced with CFRP for flexure; and one beam
was reinforced with CFRP for flexure and GFRP for shear. Results revealed that
addition of either GFRP or CFRP composites strengthening provided static capac-
ity increase of 45 percent compared to the control beam. The beam strengthened
with CFRP for flexure and GFRP for shear, which simulated the HCB cross beams
after the retrofit, exhibited nearly 100 percent of moment capacity increase. Post
cracking stiffness of all beams was increased, primarily due to the flexural CFRP
laminates. Results suggested that capacity of the experimental beam, retrofitted in
the same fashion as the bridge, should exceed the bridge design load of 720 kN-m
(after strengthening), sustaining up to 868 kN-m of applied moment. The addition
of GFRP for shear alone was sufficient to offset the lack of steel stirrups in the ac-
tual bridge, allowing for a conventionally reinforced concrete beam with signifi-
cant shear deficiency to fail by yielding of the tension steel. The ultimate deflec-
tions of the shear GFRP reinforced beam were nearly twice those of the control
shear-deficient beam. The experimental beam retrofitted with only CFRP for flex-
ure failed as a result of diagonal tension cracking at a load 45% greater than the
control beam. A design method for flexure and shear was proposed before the on-
set of this experimental study and used on the HCB. The design procedure for flex-
ure was refined and allowed for predicting the response of the beam at any applied
moment. The flexural design procedure includes provisions for non-crushing fail-
ure modes, and was shown to be slightly conservative using the design material
properties.
Keywords: bridges; composites; fiber reinforced

polymers; repair and retrofit; structures
@Seismicisolation
@Seismicisolation
187
188 Kachlakev
AC1 member Dr. Damian I. Kachlakev is a Professor in the Department of Civil
and Environmental Engineering at California Polytechnic State University, San
Luis Obispo. His research interests include FRP composite materials, FE
modeling and durability of FRP in concrete. Professor Kachlakev is a member
of AC1 Committee 440, ASCE, ASTM, IABSE, and ISPC.
INTRODUCTION
In recent years, fiber reinforced polymers (FRP) have been used to
increase the capacity of reinforced concrete structures [(1), (2)]. FRPs are
typically comprised of high strength fibers (e.g. aramid, carbon, glass)
impregnated with an epoxy or vinyl ester resin. Many experimental studies have
been conducted using FRP reinforcing on both beams and columns (3). The
number of research studies involving FRP-shear strengthening have grown
dramatically in the last few years, since many reinforced concrete structures
constructed in the first half of the 20' century were inadequately reinforced for
today's traffic demands [(4), (9,(6)]. Field applications of FRP for shear
strengthening of RC beams have also been recorded [(7), (13)]. Flexurai
behavior of FRP reinforced beams has long been researched and better
understood. Basic, yet essential studies involving flexural bending of FRP
reinforced RC beams have been reported [(S), (9), (lo), (li), (12)]. Load
capacity increases are usually approximately 1.5 to 2.5 times that of the
unstrengthened beam, depending upon FRP system type, thickness of the
composite reinforcement, fiber angle orientation, and geometry and condition of
the original beam. Many of these studies do not involve MI-scale specimens.
Field application of FRP is becoming common practice in the U.S., but a
complete understanding of the behavior of reinforced concrete (RC) beams
retrofitted with FRP is still lacking. This paper investigated the bending
behavior of full-size beams in more detail than previously known studies.
The full size beams were modeled after the cross beams of Horsetail
Creek Bridge, a historic structure located in the Columbia Gorge, near Portland,
Oregon (Figure 1). The bridge was built in 1912, and consists of three 6.10 m
spans, a total of 18.3 m. The two cross beams were found to be 95 percent
deficient in shear, relative to the current traffic loads. Since the beams were
constructed without any steel stirrups, the concrete provided the entire shear
resistance. Two of the longitudinal and ail cross beams were also 50 percept
deficient in flexural capacity. The bridge, noted to be a historic structure,
needed a retrofit scheme that maintained its original appearance. After careful
considerations, FRPs were selected to strengthen the beams (13). The retrofit
was completed in the fall of 1998.
@Seismicisolation
@Seismicisolation
EXPERIMENTAL PROGRAM
To verifi the strengthening design of Horsetail Creek Bridge, four full-
scale beams with similar geometry and rebar placement as the bridge
crossbeams were constructed [(14), (15), (16)]. Figure 2 shows the beam
dimensions and the location of the rebar. There were three main flexural steel
bars extending the full length and two bars that bent up to reinforce negative
moment regions of the beam. Smaller diameter bars were positioned near the
compression face of the beam. The beams were designed to match the strength
rather than the serviceability requirements of the bridge beams. For load rating
purposes, AASHTO specifies the concrete strength of a bridge constructed
before 1959 to be 17.2 MPa and the steel yield stress to be 228 MPa [(17)].
Concrete and steel are not readily available at these low strength levels. In an
effort to construct beams with similar ultimate strength as the bridge beams,
reinforcement bars with smaller cross-sectional areas were used to account for
the higher yield strength of today’s steel.
The four beams were cast and cured separately under similar conditions.
Type I ready-mix concrete with nominal 28-day strength of 20.7 MPa and 152
mm slump was used. The beams were cast in the same form to ensure the
dimensions were as similar as possible. Each beam was cured in a moist
condition until removed from the form 7 to 14 days after pouring. Ambient
conditions during casting and curing did not vary significantly from beam to
beam. After curing, three of the four full-size beams were strengthened with
FRP. A description of each beam is given in Table 1, and the FRP
configurations are shown in Figure 3 (a, b, c). Material properties are presented
in Table 2.
All beams were tested in third-point bending as shown in Figure 3-d. No

restraint was provided against rotation along any axis. Supports did not provide
any fixity aside from friction due to normal forces. Thus, the beams could be
analyzed as simply supported beams. All beams spanned 5.49 m with a shear-
span of 1.83 m. A 2670 kN, internal-frame, hydraulic press with a load cell was
used to load the beams. This machine was designed to compress test specimens
by transferring all forces into its own fiame. For beams that spanned beyond the
frame of the machine, the maximum applied force was limited to 712 kN.
Deflections were measured at three locations. Critical strains were

recorded at the mid-span section and at two sections in high shear regions.
Gauges were placed on the concrete surface, FRP surface and inside the beam
on the steel, and were appropriately protected before casting of the beam-s
(Figure 3, d). Fiber optic gauges were utilized only on the FRP reinforced
beams. Long gauge lengths (1000 to 1700 mm) collected strain data for the
experimental beams, exactly replicating the fiber optics installed on the bridge.
@Seismicisolation
@Seismicisolation
190 Kachlakev
EXPERIMENTAL RESULTS
The control, flexure (F-only) and shear (S-only)-retrofitted beams were
loaded to their respective capacities. Failure modes are presented in Table 3.
The flexure and shear-retrofitted (S&F) beam was loaded up to the capacity of
the testing equipment of 712 kN, which corresponded to a maximum applied
moment at the midspan of 651 kN-m. No signs of failure were observed. The
beam was loaded a second time to 712 kN with the load points closer to
midspan, corresponding to an applied moment of 868 kN-m. Again, no signs of
pending failure were observed.
A summary of the experimental load and deflection results is presented

in Table 4. The load value of 67 kN was selected for pre-cracking comparison
of strains and deflections. Deflections recorded at midspan of the S-only beam
went beyond the limit of the measuring equipment near ultimate loads and were
extrapolated. Figure 4 presents the experimental plots of load vs. deflection.
Figure 5 shows the load vs. strain plots at midspan of the S&F beam. Similar
plots were generated for three sections (Figure 3, d) of all four beams.
EXPERIMENTAL RESULTS INTERPRETATION AND DISCUSSION
Experimental Observations
Many benefits of using the FRP reinforcing were evident from the
experimental data. A comparison of the experimental load and deflection results
is presented in Table 5. All FRP- reinforced beams exhibited first cracking at
higher loads than the unstrengthened control beam. Pre-cracking stiffness is
mostly dependent on the elastic modulus of the concrete. The F-only and S-only
beams had a lower concrete elastic modulus (25.74 GPa and 26.22 GPa,
respectively) than the control beam (30.05 GPa), likely resulting in lower
stiffness before the onset of cracking.
The addition of FRP as a shear and flexure external reinforcement both

independently and as a combined system substantially increased the load
carrying capacity of the beams and improved ductility at ultimate moment
capacity. Although unproven by the experiment due to limitations of the testing
equipment, the S&F beam would also exhibit greater deflection at failure than
the unstrengthened section, primarily because the control beam failed in shear.
The addition of the shear GFRP composite laminates compensated for

the lack of stirrups in the control beam and converted the mode of failure from
diagonal tension to ductile failure. The F-only and S-only beams achieved
equivalent static load capacity, 45% greater than the control beam, albeit failing
in different modes. The S&F beam sustained over 49% higher load (or moment)
compared to the unstrengthened (control) beam in the third-point loading
configuration. During a second testing, the S&F beam sustained a moment 98%
higher than the unstrengthened beam without pending failure.
@Seismicisolation
@Seismicisolation
Since the maximum experimental shear force and maximum moment are
directly proportional to the total applied load, percent comparisons are valid for
shear capacities and moment capacities. It should be noted that the high
experimental gains resulting from using the FRP strengthening would not be as
significant if the beam were only deficient in flexure. However, the S&F beam
has a higher capacity than the S-only beam, which is evidence that the flexural
capacity is also increased by the addition of FRP to a beam Without shear
deficiency. This fact has been documented in [(9), (lo), (1 i), (12)].
Cracking Patterns
Most of the behavioral understanding of the cracking mechanism came
from the control and F-only beams. Diagonal tension cracking occurred in these
beams. Such cracks are the result of combined bending and shear forces
realigning the principal tension direction. S-only and S&F beam cracks were
mostly concealed under the FRP reinforcing. Loading from zero to 66.7 kN did
not produce any notable cracking in neither beam. First cracks appeared at about
80 kN near the midspan. The flexural cracks in the control beam were noted to
increase in length and quantity up to 156 kN at which time the first evidence of
shear cracks were visible. The critical shear cracks did not completely develop
until near the ultimate load of 476 kN (Figure 6, a). The critical shear crack in
the F-only beam developed at a load of approximately 267 kN. An accurate
assessment of the F-only beam cracking was not possible, since the CFRP
covered the section of the beäm where tension cracks initiated (Fig. 6, b).
The visible cracks in the F-only beam were fewer and did not appear to
propagate as high as for the control beam. The shear crack that developed in the
F-only beam was visibly wider than for the control beam just prior to failure.
This behavior can be attributed to the additional resistance provided by the
CFRP laminates, allowing for larger deflections after the formation of the
diagonal tension crack. The failing crack formed on the F-only beam occurred
somewhat more explosively than for the control beam, apparently due to the
dissipation of the internally stored energy.
Very little evidence of cracking of the concrete could be seen on the S-

only beam through the GFRP laminate that covered the sides of the beam
(Figure 6, c). Vertical cracks occurred just prior to ultimate load, propagated
through the GFRP reinforcement and were visible as a color change. Since the
composite is unidirectional, the vertical glass fibers had little effect on the
appearance and propagation of the flexural cracks. The only evidence of
cracking on the fully strengthened beam (S&F beam) was visible at the
midspan. These tension cracks were slightly audible and visible at
approximately 534 kN. As the load approached the machine limit, these cracks
only increased in height. Even during the second loading with an applied
moment of 868 kN-m, cracking was only visible at the midspan section (Figure
6, d). Figure 7 suggests that the presence of the GFRP in the S&F beam stopped
@Seismicisolation
@Seismicisolation
Kachlakev
the crack growth occurring at 156 kN. In the control beam, similar crack
development initiated at about the same load level (156 kN), which was
believed to develop into the shear failure.
Failure Mechanisms of the Experimental Beams

Ail failures occurred in the predicted modes and no premature or non-
traditional failure mode occurred as the result of FRP strengthening (Table 3).
The control beam was deficient in shear as suggested by the load rating
calculations of Horsetail Creek Bridge and failed by diagonal tension cracking.
The F-only beam also failed in shear as anticipated (since no shear

strengthening was added), but at a higher load than the control beam. The F-
only beam sustained 45 % higher load than the control beam, although the
failure of the F-only beam was related to the s.ame deficiency as that of the
control beam. Since shear reinforcing was absent, the addition of CFRP for
flexure should not be expected to add shear strength capacity to the beam. In the
classical sense, the CFRP added no additional design shear strength. However,
since the CFRP was wrapped up the sides of the beam, the layers of carbon
bridged the critical diagonal tension cracks on both ends of the beam, which
resulted in additional load-carrying capacity. The addition of the CFRP also
increased the flexural rigidity of the beam. Thus, for any given load, strains
were lower in comparison to the control beam, and helped to offset the shear
failure. The CFRP fibers were able to maintain integrity of the beam in the
presence of the shear crack. The failure of the beam was accompanied by
transverse rupture of the composite. The failure was believed to be a tension
failure of the matrix in the direction perpendicular to the carbon fibers, or shear
failure in the direction parallel to the carbon fibers. Since the composite was
intended to be horizontally unidirectional in strength, excessive straining of the
composite in the vertical direction would likely result in failure. These
observations suggest that it may be advantageous to apply a composite with bi-
directional strength in order to provide higher horizontal and vertical strength.
The most effective resistance to diagonal tension cracks would be an FRP
whose principal direction is oriented orthogonal to the crack (aligned with the
principal tension strains). The difficulty is then predictability of the beam
response, since a composite with unisotropic properties will be applied .to a
beam supposedly homogeneous and isotropic [(15), (16 ) ] .
The addition of GFRP along the sides of the S-only beam resulted in a
classical reinforced concrete beam failure. The GFRP was sufficient to mitigate
a diagonal tension crack and elevate the beam strength to cause a flexural failure
at the midspan section. The sequence of events leading to the flexural failure of
the S-only beam is evident in Figure 8. The main flexural steel yielded at 534
kN. Consequently, the flexural rigidity of the beam was reduced causing a rapid
increase in deflection (Figure 4). Ultimately, the concrete crushed at the top at
midspan. A considerable amount of “ductility” was present in the S-only beam,
as apparent from the large deflections that occurred after the steel yielded.
@Seismicisolation
@Seismicisolation
The fully-FRP reinforced (S&F) beam did not fail or show signs of
impending failure while the applied moment of 868 kN-m was held constant and
essential strain gauges were observed in real time. Deflections were greater than
these achieved under third point loading (651 kN-m applied moment), but did
not increase under a constant load. It is believed that the S&F beam would fail
from crushing of the concrete in compression as a secondary effect, after
yielding of the primary tension steel. Rupture of the FRP composite would
possibly occur after crushing of the concrete.
Concrete strains at the top-midspan location of the S&F beam were

approaching 0.0014 at the maximum applied load (Figure 5). Strains in the
CFRP reinforcing at midspan were approaching 0.003 and strains in the main
tension reinforcing steel were slightly above 0.002. This is clear evidence as to
the projected failure sequence of the beam, where the steel yields, extended
deflections result, the concrete crushes and then the FRP ruptures from
excessive strain. The result would be an “over-reinforced” beam in the classical
sense in which the concrete crushes before the FRP would rupture. The FRP has
taken the role of the steel reinforcing, once the steel has yielded, and the beam is
no longer “over-reinforced”.
The S&F beam clearly was not near failure at the maximum applied
moment as seen in Figures 9 and 10. Strains recorded in the top concrete did not
show impending collapse, nor was any mechanism evident that would lead to
collapse. Figure 9 suggests that the strain in the top midspan concrete was
approximately 1350 micro strain at 712 kN. If the observed trends suggested
that the beam was at approximately half its capacity at the maximum applied
load in this study. The failure of the beam will likely result in crushing of the
concrete, at about 3000 micro strain. However, as seen from Figure 10, the
bottom steel reinforcing was about to yield (2700 micro strain at 712 kN). This
would have caused the stiffness of the beam to be reduced, increasing the top
concrete and bottom FRP strains at a higher rate.
Fiber Optic Strain Gauges

Comparison between typical measurements obtained from fiber optic
(FO) vs. resistance strain gauge is presented in Figure 11. Much of the data
collected from the FO was not very useful for behavioral interpretation of the
experimental beams. This was the result of two specific shortcomings: the gauge
length used being too long and the resolution of strain too rough to extract
representative strain response from the beam testing. There was a specific
problem with gauges placed in high shear regions of the experimental beams.
As shown in the sketch in Figure 11, the fiber optic gauges in the shear region
collected data over a 700 mm gauge length. This length included a shear crack
near the beam bottom and near-zero strains near the top of the beam. Strains
near the bottom of the beam and near the support are very high due to diagonal
tension cracking. In contrast, strains near the top of the beam over the supports
@Seismicisolation
@Seismicisolation
194 Kachlakev
are very small, particularly after growth of the shear crack. Thus, providing a
gauge over this distance only produced average strains, which were not useful
for structural analysis of these concrete beams.
This is not to say the existing Horsetail Creek Bridge fiber optics are not
useful. In contrast, the opposite is quite true. On the bridge, there are two
separate gauges in the high shear region of the beams. These gauge lengths are
slightly shorter than those used in the experimental studies. Regardless of their
lengths, however, they can be used to decipher if the bridge beam is
approaching failure. This can be accomplished by establishing a strain limit
beyond which the fiber optic system (being monitored “real time”) can signal
impending failure.
DESIGN OF BEAMS EXTERNALLY REINFORCED WITH FRP

LAMINATES
Design Recommendations Based on Test Results

Flexural failure of concrete beams strengthened with FRP is likely to
occur due to yielding of the tension steel, followed by crushing of the concrete
and ultimately rupture of the FRP composite. This arises from the large
difference in ultimate strains for the constituent materials. CFRPs can sustain
strains around 0.02 whereas concrete crushes at 0.003 and steel yields at
approximately 0.002 (a factor of 7 to 10 difference). This behavior is anticipated
for any FRP reinforced beam that has adequate adhesiodanchoring and was
originally designed as an “under-reinforced” beam (where the steel yields before
the concrete crushes).
One very important point every designer should realize is that the added
flexural capacity of FRP to existing RC beams will often not be a dramatic
structural enhancement. Most concrete beams were designed to fail by yielding
of the tension steel before the concrete is able to crush. This results in a
“ductile” beam, which is able to undergo visible deflections before ultimately
losing load carrying capacity. Most retrofit projects where FRP is used to
enhance an existing deficient member will involve originally under-reinforced
concrete beams. The FRP strengthening is then producing additional capacity by
taking the beam to a level of over-reinforcement. In an FRP strengthened RC
beam, it may be good practice to account for the likelihood that adding FRP will
result in failure by crushing of the concrete. This deficiency can be offset by
either using a smaller phi factor or by decreasing the allowable stresses,
depending on the design philosophy adopted.
Recommended Design Technique

Table 6 presents a comparison of the S&F beam and the actual HCB
beam capacities as predicted by the design [(15), (i6)]. The adopted design
method predicts a moment capacity slightly higher than that achieved during the
experimental studies. The maximum applied experimental moment of 868 kN-
@Seismicisolation
@Seismicisolation
m, which did not cause failure of the beam, suggests that the proposed design
method does not under-predict the strength. Shear capacities after strengthening
are difficult to compare with experimental, since the GFRP shear strengthened
beam failed in flexure (as expected).
Once the validity of the design method was verified, it was used to
compared the responses of both the fully reinforced S&F beam and the F-only
beam. Since the design is flexure-based, these two beams should respond
similarly to applied load. However, the F-only failed in shear while the S&F
beam was predicted to fail in flexure. Results for three different strengths of
concrete are shown in Figure 12. The method describes the response with good
accuracy. Most reassuring is the fact that the material properties used to predict
the response are design and not experimental values and produce slightly
conservative results. In particular, only the dimensions are experimental values.
The design concrete strength of 20.7 MPa produces a slightly conservative
prediction.
The experiments and the design calculations of the S&F Beam verified
that the strengthened HCB beams have at least the required moment capacity
(Table 7).
CONCLUSIONS
Application of GFRP laminates for shear and CFRP laminates for
flexure resulted in a static load capacity increase by 49% over the original beam
capacity. Moment capacity was increased by 98%. The experimental beams
retrofitted with only the designed flexural carbon FRP still resulted in diagonal
tension failure albeit at a more substantial 689 kN total applied load, compared
to the 476 kN total load sustained by the unstrengthened beam. Since the CFRP
was intended to provide flexural reinforcing, it was horizontally unidirectional.
The CFRP was wrapped up the sides a sufficient amount to provide resistance
across the diagonal tension crack. In addition, the increased stiffness provided
by the CFRP decreased the deflection and thus offset cracking by reducing
strain in the beam.
The addition of GFRP for shear was sufficient to offset the lack of
stirrups and cause a conventional RC beam failure by steel yielding at the
midspan. This allowed ultimate deflections to be approximately 110% higher
than for the shear deficient control beam, which failed due to a significant
diagonal tension crack.
The flexural CFRP strengthening, in addition to the shear GFRP

reinforcing, increased the load that caused yielding of the primary tension steel
by more than 33%.
Post-cracking stiffness was increased by 17% and 2 1% for the addition of GFRP
in shear and CFRP in flexure, respectively. The addition of both systems
increased post-cracking stifhess by 30%.
@Seismicisolation
@Seismicisolation
196 Kachlakev
The original HCB beams were likely deficient in shear prior to retrofit,
as suggested by the failure of the control beam. The original HCB beams would
have failed at approximately 238 kN shearing force. Total unfactored dead load
and HS20 vehicle live load shear acting on the bridge cross beams was
estimated to be 287 kN. The total factored load was estimated to be 397 kN.
This bridge has likely never seen a live load equal to or greater than the design
HS20 vehicle.
The original HCB beams retrofitted with only the designed flexural
CFRP might still have resulted in diagonal tension failure, achieved at a higher
load. The experimental beam strengthened with only flexural CFRP failed at
689 kN total applied load.
The original HCB beams retrofitted with only the designed shear GFRP likely
would have failed in flexure at the midspan. The experimental beam
strengthened with only the shear GFRP failed at 689 kN total applied load.
Yielding of the main flexural steel initiated prior to crushing of the concrete.
The original HCB beams retrofitted with both the GFRP for shear and
CFRP for flexure should well exceed the static demand imposed by the new
design traffic loads. The fully reinforced experimental beam exceeded the total
factored load demand (required by the load rating of the HCB) of 720 kN-m
sustaining up to 868 kN-m applied moment, without failure.
REFERENCES
1. Cooper, J. M., (2000). “Advanced Composite Materials for the 21StCentury
Bridges: The FHWA Prospective”, “Innovative Systems for Seismic Repair And
Rehabilitation of Structures”, pp. 131-142, March 2000
2. Rizkalla, S . , Labossiere P. (1999). “Planning for a New Generation of

Infrastnicture: Structural Engineering with FRP in Canada” Concrete
International, 25-2 8.
3. Kachlakev, D. I., (1998). “Strengthening Bridges Using Composite

Materials”, US DOT FHWA and Oregon DOT.
4. Cheng, R.J.J., Hutchinson R., Rizkalla S. (1997). “Rehabilitation of Concrete

Bridges for Shear Deficiency Using CFRP Sheets“ 42ndInternational SAMPE
Symposium, 325-335.
5. Arduini, M., D’Ambrisi A., Di Tommaso A. (1999). “Shear Failure of

Concrete Beams Reinforced with FRP Plates” Infiastructure Repair Methods,
123-130.
6. Kachlakev, D., and Barnes, W. (1999). “Flexural and Shear Performance of

Concrete Beams Strengthened with Fiber Reinforced Polymer Laminates“, ACI,
SP-188, 959-972.
@Seismicisolation
@Seismicisolation
7. Crasto, A., and Kim, R. (1997). “Rehabilitation of Concrete Bridge Beams
with Fiber-Reinforced Composites” 42& International SAMPE Symposium, 77-
83.
8. Kachlakev, D., B. Green and W. Barnes (1999). “Behavior of Concrete

Specimens Reinforced with Composite Materials-Laboratory Study”, US DOT
FHWA and Oregon DOT.
9. GangaRao, H.V.S., Vijay P. (1998). “Bending Behavior of Concrete Beams

Wrapped with Carbon Fabric” Journal of Structural Engineering, ASCE, Vol.
124, No. 1, 3-10.
10. Rostasy, F.S., Hankers C., Ranisch, E.H. (1992). “Strengthening of WC-
and P/C-Structures with Bonded FRP Plates” Advanced Composite Materials in
Bridges and Structures, Canadian Society for Civil Engineering, 255-263.
11. Ritchie, P.A. (1991). “External Reinforcement of Concrete Beams Using

Fiber Reinforced Plastics” Structural Journal, ACI, Vol. 88, No. 4,490-496.
12. Saadatmanesh, H., Ehsani, M. (1991). “RC Beams Strengthened with GFRP
Plates I: Experimental Study” Journal of Structural Engineering, ASCE, Vol.
117,No. 11, NOV.,3417-3433.
13. Laylor, H.Martin and Kachlakev, D.I., (2000). “Fiber-Reinforced Polymer

Composites for Strengthening Bridges in Oregon”, TR News, Transportation
Research Board, National Research Council, Number 208, May-June, 2000
14. McCurry, D.D. and Kachlakev D.I., (2000). “Strengthening of Full-Scale

Reinforced Concrete Beams Using FRP Laminates and Monitoring with Fiber
Optic Strain Gages“, “Innovative Systems for Seismic Repair And Rehabilitation
of Structures, Technomic Publishing Co, Inc., pp. 131-142, March 2000
15. McCurry, D.D., (2000). “Strengthening Reinforced Concrete Beams Using

FRP Composite Fabrics: Full-scale Experimental Studies and Design Concept
Verification, Master of Science Thesis, Oregon State University, Corvallis,
Oregon, USA
16. Kachlakev, D. and D. McCuny, (2000). “Testing of Full-size Reinforced

Concrete Beams Strengthened with FRP Composites: Experimental Results and
Design Methods Verification,” US DOT FHWA and Oregon DOT.
17. AASHTO Subcommittee on Bridges and Structures (1996), Standkd

Specifications for Highway Bridges. Including up to 1999 Interim Revisions.
AASHTO, Washington D.C.
@Seismicisolation
@Seismicisolation
198 Kachlakev
Table 3 : Beams failure modes
Table4: Summ y of experimt itai load and deflection

Measurement Control
I_
F-only 1
S-only
Post-cracking Stifmess 20.1 kN/mm 26.3 kNlm
. i - - - .
Midspan Den. @Yield Not observed Not observed 23 mm 1 Not observed
Maximum Deflection 24.5 mm 25 mm

Not
Load @ Failure 476 kN
observed
Load @First Cracking 78.3 kN 96.5 kN 87.6 kN 96.1 kN
651 kN-m"-
Max. Applied Moment 435 kN-m
... - - (868 E-E)
Max.Shear Force 234 kN 345kN 1345kN 356 kN
P
1- No Failure observed
@Seismicisolation
@Seismicisolation
Table 5: Comparison of experimental load and deflection
I Percent Gain Over Control
Item
i
ControlBeam B~~
1 Data F-only S-only S&F
1
~~ ~~~
Midspan Defl. @67 kN 5.2% -6.5%

._ - - __- - - -- - _ _
Max. Recorded Deflection 24 % 44% 3.8%'
failure
Load @ failure I476m 45 Yo ' 45% 49%
cracking
Maximum Applied Shear
. -
1 238 kN 45 Yo 45% 49%
45% 49%
. No failure observed; 2- Extrapolated fi-om recorded trend.
Table 6: Design predictions for experimental and HCB beams
Required' Before After

Strengthening2 strengthening3
Failure Mode Ductile/ Flexure Diagonal Tension Ductile/ Flexure
Shear Capacity, 436 238 485
kN
Moment 720 462 777
Capacity, kN-m
@Seismicisolation
@Seismicisolation
200 Kachlakev
Figure 1 : Horsetail Creek Bridge
64
l
#19 Steel rebar
6096
#22 Main Tension Steel
305 A
2#19 & 1 #16
Steel rebar
1 #1G Steel
rebar
3 #22 Steel
rebar
I
i
SECTION A-A SECTION B-B
Figure 2: Experimental beams geometry and position of the steel

reinforcement
@Seismicisolation
@Seismicisolation
h&q--l
203
~~
\u&&od(FRF
(3 i a Y 4
I
2layen
\
iiayer
Figure 3(a): Flexural strengthening with carbon FRP unidirectional

laminates
I I
4
<
1676
6Wb
\
4 iaym, appiied
Unidirectional in "U shape
152 GFRF' (2 layers)
Figure 3(b): Shear strengthening with glass FRP vertically applied

laminates
4L I ,
Unidirectional
under GFRP (see F-only) GFRP (see S-only)
Figure 3(c):Flexural and shear strengthening- combination of (a) and (b)
2134 3048
-
,
II 1500
1067
'
Figure 3(d): Test Beams loading and strain gauges locations
@Seismicisolation
@Seismicisolation
202 Kachlakev
(4
O 5 10 15 20 25 30 35 40 45 50
180 801
160 712
- 140 623
2P 120 534
3 100
;Y 445
2 80 356
n
60 267
*
L o 178
20 89
O O
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
Mispan Downwani Deûection (in)
Figure 4: Load vs. deflection experimental results [After (1 5)]
. 801
160 - 712
- 140 - 623
.a% 120 - 534
2 100 - 445
P
2 80 - 356
9
’
3
d
.
60
40
20
- 267
- 178
- 89
.n
I -
-2000 -1500 -1000 -500 O 500 1000 1500 2000 2500 3000
Microstrain
Figure 5: S&F Beam load vs. strain at mid-span [After (1 S)]
@Seismicisolation
@Seismicisolation
a) Control beam at failure b) F-only beam at failure
c) S&F beam at ultimate load d) Shear-only beam at

failure
Figure 6: Experimental beams at ultimate loading [After (1 5)]
Microsuain
F i m e 7: S&F beam strain (1500 mm from beam end) [After (14)]
@Seismicisolation
@Seismicisolation
204 Kachlakev
801
712
623
534
445
356 5
267
178
89
O
-3000 -2000 -1000 O 1000 2000 3000
Microstrain
Figure 8: S- only beam load vs. strain at midspan [After (14)]
801
712
623
534
445
356
267
178
89
Microstrain
Figure 9: S&F beam compressive strain comparison [After (1 5)]
@Seismicisolation
@Seismicisolation
180 801
160 712
h
140 623
8a 120 534
3 loo 445
u22 80
80 356 zi
3
’
3 60
40
20
267
178
89
O O
-500 O 500 1000 1500 2000 2500 IO
3000
30C
Microstrain
Figure 1O: S&F beam tensile strain comparison [After (15)]
180 801
160 712
,140 623
3120 534
P
3 100 445
aa 80 356 zi
g.
60 267
*
40 178
20 89
O O
0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 O009
Strain
Figure 11: Fiber optic vs. resistive strain gauge comparison [After (16)]
@Seismicisolation
@Seismicisolation
206 Kachlakev
250 1112
-2 200 890
a 667
A 150
-2
d
3
100 445
2B
$ 50 222
O O
-3500 -3000 -2500 -2000 -1500 -1000 -500 O 500
Concrete Sûain at Top of Beam (Microsrrain)
Figure 12: Design vs. experimental S&F beam results
@Seismicisolation
@Seismicisolation
SP 209-12
Innovative Seismic Retrofit Scheme

for 215 Fremont Street Building,
San Francisco, California
by N. R. Amin, D. Figueira, and V. Wan
SvnoDsis: The 215 Fremont Street building in San Francisco, California was de-
signed by Albert E Roller Architect, San Francisco, and built in 1927. It was a 7-
story reinforced concrete structure, “L”-shaped in plan, with a 3 story tower locat-
ed over the elevator core. The structure is supported on individual spread footings
at the interior colums and continuous grade beams at the building perimeter. Dam-
aged extensively in the 1989 Loma Prieta Earthquake, the building was declared
unsafe. It remained unoccupied until 1999, when the property was sold “as is” to a
new developer (Fremont Properties LLC). The developer embarked on a seismic
retrofit of the existing building and the addition of two new floors, all on a build-to-
suit basis for a single tenant (Charles Schwab Inc.). This paper will describe in de-
tail the evaluation of the existing building, analysis and design of the retrofit
scheme, including the foundation, which meets the 1997 Uniform Building Code.
Keywords: building evaluation; seismic retrofit; Uniform

Building Code
@Seismicisolation
@Seismicisolation
207
208 Amin et al.
Navinchanàra (Navin) R. Amin, SE is a Principal at Middlebrook + Louie
(San Francisco). He has led the structural design of a number of important
structures, including the new International Terminal at San Francisco
International Airport, the base isolation seismic retrofit of the U.S. Court of
Appeals in San Francisco, and many high rise office towers and mixed use
complexes around the world. Most recently he led a team of engineers in
designing an 88-story tower in China,using a composite steelkoncrete system.
Construction on this building has been postponed.
Previous to joining Middlebrook + Louie, Mr. Amin was in charge of

structural design for Skidmore Owings & Merrill, in Chicago, and later in San
Francisco. While with SOM, Mr. Amin worked with the late Fazlur Khan on
the structural design of Sears Tower. Throughout his 30+ year career, Mr.
Amin’s structural concepts have been models of efficiency and creativity, and
have earned several prestigious awards.
David Figueira, PE and Vivian Wan are structural engineers with

+
Middlebrook Louie (San Francisco).
Mr. Figueira holds a Master’s Degree in Civil Engineering from the

University of California, Berkeley (1995) and a Bachelor of Science Degree in
Aerospace Engineering from UCLA. He joined Middlebrook + Louie in
1996.
Ms. Wan holds a Master’s Degree in Engineering from Massachusetts institute

of Technology and a Bachelor’s Degree in Engineering from the University of
+
Hong Kong. She joined Middlebrook Louie in 1999.
BUILDING DESCRIPTION
Located in San Francisco’s South of Market district, 215 Fremont Street was a
7-story, L-shaped office building consisting of large floor plates of about
46,000 sf / 4,300 sm. Total building area was approximately 320,000 gsf /
29,700 gsm. (Fig. 1).
The original structure was reinforced concrete with two-way flat slabs on a 20
ft. 6 in. (6250 mm.) square bay (Fig. 2). The slabs were 7% in. (197 mm.)
thick with a 2 in. (50 mm.) topping with column capitals/drop panels (Figures
3a and 3b).
The perimeter punched window wall façade consists of wingwalls (nominally 6

in. / 152 mm. thick) and spandrel beams perforated by a regular window
arrangement.
@Seismicisolation
@Seismicisolation
A combination of the punched window wall “frame” on the building perimeter
and frame action of the interior columns and floor slabs served as the original
lateral load system. The foundation system consisted of truncated pyramid-
shaped spread footings at interior columns and grade beam foundations with
strap beams along the building perimeter. The building’s underlying soil
conditions includes different regions of dense silty sands and stiff silts. Since
the original construction, the building had experienced considerable settlement.
The building suffered significant damage during the 1989 Loma Prieta
Earthquake, most notably extensive diagonal shear cracks in the perimeter
spandrel beam wingwall construction (Fig. 4).
BUILDING EVALUATION
Review of the original design drawings provided an initial assessment of the

building’s construction. The design drawings supplied information on column
sizes and reinforcement, slab reinforcing quantity and layout, size and location
of existing spread footings and grade beams, and thickness and depth of
perimeter spandrel beams. While the drawings proved to be useful, certain
desired information was either illegible or not included on the drawings.
To verifj the information provided and to determine the unknown structural

parameters, an engineering testing-inspection agency, Applied Materials &
Engineering, Inc. (AME) was retained to perform a physical and visual
inspection of the existing building (AME 2000). In addition, AME measured
member sizes and thicknesses and locally removed concrete cover to
verify/determine member reinforcements (Fig. 5).
Core samples were taken at various floor locations to determine slab concrete
strength and the density of the topping slab. Coupons of existing bar
reinforcing were taken to determine chemical composition and yield strength.
In addition, AME performed a series of loaddeflection tests at the seventh
floor to determine the load carrying capacity of the slab. Following the
guidelines of AC1 318-95 Chapter 20, the floor slab was incrementally loaded
with 55-gallon (7.35 cubic ft.) drums filled with water. The drums were
spread over the floor bay until maximum permitted deflection was achieved.
The load was removed and the subsequent “rebound” in the slab deflection
was measured. Based on AME’S investigations, the following properties were
determined and used in the analysis & design phases of the project:
Concrete Strengths: 2,570 psi (18 MPa) for columns, 3,140 psi (22
MPa) for slabs, 4,040 psi (28 MPa) for walls
Rebar Yield Strength: 44,000 psi (303 MPa)
Allowable superimposed loads on floor slabs: 100 psf (4788 Pa)
@Seismicisolation
@Seismicisolation
210 Amin et al.
During the initial survey it became evident that there was extensive cracking in
the existing perimeter walls, columns, spandrel beams, and floor slabs. The
owner’s testing and inspection agency, Consolidated Engineering Laboratories
(CEL) performed a survey of cracks in the building (CEL 2000). CEL
documented floor cracks with widths greater than 0.5 mm. The project’s
geotechnical engineer, Treadwell & Rollo, in conjunction with the building
surveys, performed standard testing for the soils including cone penetrometer
tests (Treadwell & Rollo, 2000).
BuildinP Engineering Evaluation
Based on the combined information acquired from the existing drawings and
field surveys, a computer model was generated to determine the performance
of the existing structure under gravity and seismic loading (Computers &
Structures, 1997). The model consisted of a series of Ordinary Moment
Frames in each direction comprised of the interior and exterior columns, the
floor slabs (as beam elements), and perimeter spandrel beams. The model
used the concrete strengths determined from the field surveys; gross section
properties were used in the analysis.
Structural Deficiencies
From the computer model, it was determined that the existing structure was
inadequate to resist the seismic forces. With the absence of bottom slab
reinforcement and insufficient top reinforcement at the end spans, the floor
slabs would not be able to resist the induced bending moments due to gravity
and seismic loads. The cracked section properties would further reduce
building performance. In addition the existing reinforcement in the lateral
frame members lacked ductile detailing.
BUILDING PROGRAM
In an attempt to maximize rentable space, the architectural scheme called for a

2-story addition on top of the existing concrete structure. To accommodate
this, the existing 3-story concrete tower over the elevator core was
demolished. In addition, the architectural design called for open “signature”
facades along the south and west elevations.
Structural Concept
With the vertical addition, the City and County of San Francisco required that
the structure be retrofitted to the design seismic force level of the 1997
Uniform Building Code. Since the existing structure would not be able to
accommodate the high seismic force due to large building weight and 1997
UBC seismic force coefficients, a new structural system needed to be
@Seismicisolation
@Seismicisolation
proposed. This system needed to be consistent with the architectural concept,
as well as the long spans required by the Owner for the new additional floors.
in addition to accommodating architectural façade considerations, attention was

given to formulating a structural system that would use the full length and
width of the building to resist seismic overturning moments thus minimizing
the effect on the foundation. Initial design concepts incorporated infill shear
walls (Fig. 6) in diagonal strut patterns along all exterior surfaces of the
building. While this scheme appeared viable from structural performance
considerations, the presence of infill walls on the south and west elevations was
not compatible with the architectural intent and had to be revised. An
alternative scheme of steel braces installed in a similar strut fashion along the
entire building perimeter was considered. This system lacked sufficient stiffness
to adequately relieve the existing concrete structure of the anticipated seismic
demands predicted from computer modeling. After additional iterations, the
final design chosen was a hybrid of the initial concepts: perimeter braced fiames
along the building perimeter with selected infill shear walls primarily in the
reentrant comers and in the interior elevator core (Figure 7). The design met the
architectural requirements of open exterior facades and maintaining accessibility
of the interior floor space.
An exposed façade was chosen for the west elevation (Fremont Street).
Consequently, the existing concrete punched window wall and adjacent 8 ft.
(2.44 m.) strip of floor was removed down to the ground floor level and
rebuilt with a steel braced frame and glass curtain wall. In the interim, the
floor slab was supported by a series of pipe shores extended to the ground
level and then supported by bracing to the basement columns and perimeter
wall.
New infill slabs had to be cast in place and doweled into the existing slab
(Figure 8). The remaining perimeter elevations consisted of connecting steel
braces to existing concrete columns through a “shear-block” connection. New
infill shear walls were installed along portions of the perimeter facades
(particularly in the faces at the reentrant comer adjacent to the 301 Howard St
Building) and at the building interior at the stairwell and elevator core. A new
foundation system consisting of pin-piles connected via a network of grade
beams to existing footings was constructed.
The vertical addition (consisting of two new floor levels and a mezzanine) is a
steel structure with braced frames along the south and west elevations and
building interior (Fig. 9).
@Seismicisolation
@Seismicisolation
212 Amin et al.
Design Parameters
Based on the geotechnical engineer’s report, the foliowing soil parameters

were used in the calculation of the building base shear:
Soil profile type: SD

Near Source Factors: Na= 1.0, Nv= 1.04
From the computer analysis, the building periods were determined to be 0.58
seconds and 0.55 seconds in the north-south and east-west directions
respectively. These periods compare with the calculated method A of (0.66
sec) per UBC eqn 30-8. Combined with R=5.5 (concrete shear wall/ordinary
braced frame), 1997 UBC eqn 30-5governs the design base shear value:
V = (2.5CaI/R) = 0.2W
In accordance with the 1997 UBC section 1630.1, the redundancy factors were
determined as 1.17 in the east-west and north-south directions.
ANALYSIS RESULTS
Shear Walls
As would be anticipated, a majority of the seismic shear force accumulated in

the shear wall elements, particularly along the faces that converge at the
reentrant comer. In order to utilize the existing structure as much as possible,
the shear walls were connected to existing concrete columns through horizontal
dowels installed at each floor level. The shear wail and existing structural
columns were designed as a composite unit, utilizing the column vertical
reinforcement as part of the total boundary element reinforcement. The
internal forces of the column and walls respectively were combined into a
single combination of shear, axial force and overturning moment. The
horizontal dowels between the columns and shear wails were designed to
transfer the shear at the interface calculated by elastic beam theory.
Wall boundary element reinforcing steel was added at the ends of the infill
wall, extended through floor slabs, and connected via type II mechanical
couplers. Wall vertical web reinforcement was doweled through each floor
slab and into the foundation. With the requirements of access to exit stairs and
equipment rooms, the resulting forces in the coupling beams are substantial
(Fig. 10).
@Seismicisolation
@Seismicisolation
In one particular wall, the standard diagonal shear reinforcement for the
coupling beam had to be replaced with a encased steel girder and studs in
order to resist the design bending moment and shear.
Shear Block
Part of the lateral system for this project is the perimeter braced frames. The
steel tube braces are connected to the concrete columns through the ‘shear-
block’ connections. The connection consists of bearing plates, Dywidag
threaded Gr. 150 bars, reinforcing bars, ties and epoxy grouted dowels (Figs.
11 and 12).
A finite element model of the shear brace was built for the analysis. Existing
column and the shear block were modeled with the steel bracing design forces
input at the shear block faces. Axial and shear stress results are plotted and
summarized (Fig. 13). These stress results were used to design the horizontal
Dywidag bars and the vertical reinforcement in the shear block. The bearing
plate is designed to develop its bending and shear capacity; the concrete was
checked against the bearing stress. In order to ease the congestion of the bars
in the shear block and for constructability, mechanical couplers and T-headed
bars were utilized.
DiaDhragm
Given the differences in stiffness between the shear walls and brace frames, a
substantial amount of shear transfer occurs through the diaphragm at various
floor levels. In order to have a more accurate understanding of the behavior,
floor finite elements were used to model the diaphragm flexibility. Seismic
forces were applied uniformly to the floor elements. With the extent of the
cracking in the floor slab there was concern regarding the ability of the floors
in their existing condition to transfer the shears predicted by the computer
model.
In order to accommodate the demands on the diaphragm, a two step procedure

was introduced. First, a crack repair program was devised and implemented.
The process utilized the Kemko Epoxy Sealer, a gravity-based infill system
where the epoxy was placed on top of the floor slab in a cracked region and
allowed to flow via gravity into the cracks. Procedures for the process were
provided by Gregg Haskell, S.E. and implemented by CEL. Cores were taken
to verify the penetration of the epoxy. In addition, slant shear tests were done
to verify the strength of the crack repairs.
Secondly, collector beams were formed and cast in place underneath the floor
slab at various locations. Figure 14 shows a typical chord/collector beam
details. The shear forces are transferred at the interface between new collector
@Seismicisolation
@Seismicisolation
214 Amin et al.
beam and existing slab by shear friction between concrete and the epoxy-
grouted dowels. The collector beams are ais0 attached to the infll shear walls
and shear blocks via horizontal dowels. The collector longitudinal bars and
ties were configured and suspended from the underside of the floor slab via the
epoxy dowels. Once the reinforcing steel was placed, form work for the
beams were placed and shored from the floor slab below. The contractor was
given the option of pumping the concrete through holes in the side wail of the
form work or through intermittent holes in the floor slab above the beam.
Foundation
From the Geotechnical Engineer?s report the sandy soils supporting the
shallow foundations were found to be susceptible to liquefaction.
Consequently, a new foundation system was developed.
The new foundation system is a combination of ?column load transfer blocks?,

grade beams, and pin piles. The building loads are transferred from the
coiumns/wails to the existing footings by the transfer blocks, which then
transfer the loads to the pin piles through the grade beams spanning between
footings (Fig. 15). At the east & west ends of the building footprint, a 4 ft. 6
in. (1.37 m.) thick mat footings are used in lieu of the grade beams to support
the high seismic overturning forces. New basement walls were also added to
aid the eccentric load transfer from the wails to the pin piles.
The existing footings are typically 11 ft. by 11 ft. (3.35 m. by 3.35 m.) in
dimension. The footings were chipped around the existing columns to allow
for the installation of vertical dowels (Figs. 16 and 17). The column loads are
transferred to the grade beams by these dowels. A safety factor of 1.5 was
used to design the size and number of dowels.
The typical grade beam is 5 ft. 10 in. (1.77 m.) wide by 3 ft. 6 in. (1.07 m.)
deep (Fig. 18). A minimum of four longitudinal reinforcing bars were drilled
and epoxy grouted through the existing columns and extended into the grade
beams with couplers. The balance of longitudinal reinforcement was placed
symmetrically on either side of the existing columns.
Circular pin piles were used to support the gravity and seismic loads (Fig 19).
The pin piles were approximately 67 ft. (20.5 m.) in length below grade. Four
pile load tests have been performed. Ultimate capacity of the pin piles averages
about 550 kips (2446 kN). The piles are designed to support a dead plus live
load of 400 kips (1780 kN) and a total load, including seismic loads of 475
kips (2113 kN) with average factors of safety of 1.4 and 1.1 respectively,
conforming to the geotechnical engineers? recommendations. Some of the
piles were post-grouted to meet the design capacity requirements.
@Seismicisolation
@Seismicisolation
A SAFE model was built to analyze the foundation system. All 245 pin piles
were modeled as spring supports with a spring constant of 700 kips/inch (123
kN/mm). The foundation system was designed to meet loads and 95 in. (12.7
mm) of differential settlement.
CONCLUSIONS
Given the damage the buiiding suffered in the 1989 Loma Prieta Earthquake
and the high seismic forces determined from the 1997 UBC, a new structural
system needed to be devised for 215 Fremont Street. This new structural
system needed to be sufficiently stiff to alleviate the induced internal forces in
the existing floor slabs and punched exterior walls.
In addition, the structural system needed to utilize the full length and width of
the structure to minimize the seismic overturning forces applied to the
foundation, while at the same time being compatible with the architectural
considerations for the building façade. A combination structural system
comprised of steel brace frame and concrete shearwalls was devised that
accomplished all the necessary requirements. This retrofit has led to the
creation of a unique connection between steel braces and concrete columns.
In conjunction with the added structural system, repairs and modifications to

the concrete slabs, a redesign of the foundation system, the addition of two
new floors, and the demolition and rebuilding of the Fremont St. elevation
were all implemented. (Fig. 20)
Acknowledgements
We would like to acknowledge the following team members: Hellmuth, Obata

& Kassabaum (Architects), Beacon Capital Partners (Developers), Danielson-
Whitehead (Development Managers), Webcor Interior Construction Group
(General Contractor).
REFERENCES
Applied Materials & Engineering, Inc. (2000) BuiMing Investigation, 215

Fremont Street, Sun Francisco, Oakland, California.
Computers & Structures, Inc. (1997) ETABS Version 6.2 (Computer program
for Dynamic & Static Analysis and Design of Building Systems), Berkeley,
California.
Computers & Structures, Inc. (1998) SAP 2000 (Computer program for Three
Dimensional Dynamic & Static Finite Element Analysis and Design of
Structures), Berkeley, California.
@Seismicisolation
@Seismicisolation
216 Amin et al.
Consolidated Engineering Labs Consulting (2000) 215 Fremont Street, A
Summary Report Slab and Wall Cracks for Beacon Capital Partners, Oakland,
California.
ICBO (1997) Uniform Building Code, Structural Engineering Provisions, Vol.

2, 1997 Edition, international Conference of Building Officials, Whittier,
California.
Treadwell & Rollo (2000) Geotechnical Investigation, Remodeling and Seismic

Strengthening, 215 Fremont Street Building, San Francisco, California.
@Seismicisolation
@Seismicisolation
-""-
Fig. 1 - 215 Fremont St. (1927-1998)
t i 1
Fig. 2 - Existing Plan
@Seismicisolation
@Seismicisolation
218 Amin et al.
Fig. 3a - Typical Interior Column
Fig. 3b - Typical Exterior Column
@Seismicisolation
@Seismicisolation
Fig. 4 - Diagonal Sheer Cracking at Window Wall
Fig. 5 - Exposed Column Reinforcement with Square Reinforcing Bars and Spiral Ties
@Seismicisolation
@Seismicisolation
220 Amin et al.
3
k
o
c
rn
@Seismicisolation
@Seismicisolation
Fig. 7 - New Structurai System in 3-D
Fig. 8 - New Infiü Slab
@Seismicisolation
@Seismicisolation
222 Amin et al.
Fig. 9 - New Steel Framing at Western

Elevation (Fremont Street)
@Seismicisolation
Fig. 1 O - Boundary Element, Coupling Beam Reinforcement
@Seismicisolation
Fig. 11 - Shear Block Detail
@Seismicisolation
@Seismicisolation
224 Amin et al.
Fig. 12 - Shear Block Connection
Fig. 13. - Shear Block Model

(first of two images)
@Seismicisolation
@Seismicisolation
.....
;!
_ i :
;
> "
.
. ., .,
i: :
Figure 13 - Shear Block Model

(second of two images)
DRILLED HOLES FILLED WITH

EPOXY ADHESIVE
ROUGHEN EXISTING SURFACE
CHORD/COLLECTOR BARS
Fig. 14 - ChordCoiiector Beam Details

@Seismicisolation
@Seismicisolation
226 Amin et al.
a
5
@Seismicisolation
@Seismicisolation
-oocRx;mrrrmauiuN
wncwceamwe
1 I I
Fig. 16 - Column Load Transfer Block Detail
Fig. 17 Column Load Transfix Block Construction

@Seismicisolation
@Seismicisolation
228 Amin et al.
M
.e
R
@Seismicisolation
@Seismicisolation
,eedings 229
r NEAT C E u O C " ' ~ 1
Fig. 19 - Pin Pile Detail
@Seismicisolation
@Seismicisolation
230 Amin et al.
Fig. 20 - 215 Fremont Street (2001)
@Seismicisolation
@Seismicisolation
SP 209-13
Seimic Strengthening of a
Non-Ductile Concrete Frame
Building-Adapazari, Turkey
by C. L. Thompson, B. T. Knight, and L. R. Redlinger
SvnoDsis: After the August 17,2000 Kocaeli, Turkey, Earthquake (Mw = 7.4) De-
genkolb Engineers sent a field reconnaissance team to observe earthquake related
building damage in Turkey. Observations were made in Adapazari, which is locat-
ed approximately 52 km northeast of the earthquake epicenter and 3 km directly
north of the North Anatolian Fault. In Adapazari, a range building performance for
the typical low-rise concrete frame residential building was observed. The building
performance varied from virtually no damage to complete collapse.
A four-story residential building in Adapazari that was observed to have
signficant structural damage was chosen for evaluation. The building was evaluat-
ed using a Tier Three evaluation in accordance with FEMA 3 10, Handbook for the
Seismic evaluations of Buildings-A Prestandard. As expected, the evaluation in-
dicated the building would not meet the Life Safety Performance Objective of
FEMA 310 for the 10%exceedance in 50-year earthquake.
Traditionally, buildings with Life Safety deficiencies would be strength-
ened to comply with current building code. Rather than strengthening the building
with a traditional code based upgrade, a conceptual strengthening scheme for Life
Safety Performance was developed using FEMA 356, Prestandard and Commen-
tary for the Seismic Rehabilitation of Buildings. The strengthening scheme, which
includes the addition of concrete shear walls, is presented. In addition, a compari-
son between the FEMA 356 lateral design force level requirements for the
strengthened building and current Turkish Building Code is presented.
Keywords: buiIding p e río rma n c e; con c rete frame

building; FEMA 310 evaluation; Life Safety performance;
strengthening
@Seismicisolation
@Seismicisolation
231
232 Thompson et al.
Chris Thompson is a Principal and Group Director of Degenkolb Engineers’
Portland office. He earned his B.S. degree at Tri State University and his
Master of Engineering degree at UC Berkeley. He is the Chairman of the
Oregon Seismic Safety Policy Advisory Commission and is a registered
structural engineer in Oregon, Washington and California.
Brian Knight is a Project Engineer for Degenkolb Engineers. He is a member

of AC1 Committee 374, Performance-Based Seismic Design of Concrete
Buildings. He received his B.S. from the University of Washington and his
M.S. from the University of Illinois at Urbana-Champaign. He is a registered
Structural Engineer in the State of California.
Leanne Redlinger is a Designer for Degenkolb Engineers. She received her

Civil Engineering B.S. and M.S. degrees from Oregon State University. She is
a licensed Professional Engineer in California and an active member of the
Structural Engineers Association of Oregon.
INTRODUCTION
A common type of residential building found in Turkey consists of a

low-rise, non-ductile concrete frame with unreinforced clay tile masonry infill
walls. These buildings are typically constructed with conventional reinforced
concrete columns, beams and one-way slabs and founded on either isolated
spread footings or a structured slab and grade beam system. The buildings
typically contain retail space at the first story and residential units in the upper
stories.
Based on the availability of materials, hollow clay tile (HCT) walls are
used extensively throughout the building construction industry in Turkey. The
HCT walls are commonly found in concrete frame buildings and.typically used
as inteiior partition wails and perimeter infill walls. These HCT wails are
typically placed in loose contact with the surrounding concrete frame (1).
@Seismicisolation
@Seismicisolation
The observed range of performance for the typical low-rise concrete
frame residential building in Adapazari varied from virtually no damage to
complete collapse. During the time spent in Adapazari, measurements and
construction drawings were gathered for a representative building. The
building is a four-story concrete frame residential structure that experienced
significant earthquake damage. Based on the information gathered, the as-built
condition of the building was evaluated for Life Safety performance using both
linear and nonlinear static procedures in accordance with FEMA 310,
Handbook for the Seismic evaluations of Buildings-A Prestandard (6). In
addition, a conceptual strengthening scheme using FEMA 356, Prestandard
and Commentaryfor the Seismic Rehabilitation of Building (2), was developed
to strengthen the lateral force resisting system deficiencies of the building.
The strength of the upgraded lateral force resisting system was compared to the
required lateral strength for new buildings using the 1997 Turkish Building
Code (4).
BUILDING DESCRIPTION
The building, located in Adapazari, Turkey, is a four-story concrete

frame building structure with total floor area of approximately 1,375 square
meters. The building is rectangular in plan and has approximate overall
dimensions of 15.9 meters by 22 meters. The spacing and orientation of the
rectangular building columns is irregular (Fig. i), which may be due to
architectural or gravity load carrying considerations (2). The building was
under construction, but substantially complete at the time of the Kocaeli
Earthquake. The building’s intended use was to include retail spaces at the
ground level (first story) and residential living quarters in the upper three
stories.
The total building height is 13.5 meters with a typical story height of 3
meters, except for the first story, which measures approximately 4.5 meters
high. The typical floor is constructed with a 11.4 cm conventional reinforced
concrete one-way slab supported by reinforced concrete beams and columns.
The foundation consists of a 30.5 cm reinforced concrete mat slab with.
inverted grade beams.
@Seismicisolation
@Seismicisolation
234 Thompson et al.
The lateral force resisting system consists of non-ductile reinforced
concrete moment frames with HCT infill wails in both the principle building
directions. At the upper residential stories, the building perimeter consists of
HCT infill walls. However, at the first story retail level, a combination of
glass storefront and HCT infill walls was used for the building perimeter.
Typically the HCT wails are finished with plaster, but not connected to the
structural frame. The orientation and size of the moment frame columns varies
across the building, which results in an irregular distribution of lateral strength
and stiffness
OBSERVED BUILDING PERFORMANCE
The building experienced significant structural damage as the result of

the Kocaeli Earthquake. The structural damage was consistent with column
shear failures due to insufficient transverse reinforcement in columns as well
as short column effects caused by the openings in the HCT infill walls. Other
reconaissance teams noted similar earthquake damage to other residential
buildings (3). Flexural and shear hinging of the building columns was
observed in both the strong and weak axes. These failures were most likely
due to inadequate splice lengths for the smooth longitudinal reinforcement and
insufficient confinement of the concrete.
SITE DESCRIPTION AND SEISMICITY
The building is located on a level site with the ground floor at grade.
The building appeared to be founded on relatively soft clay soils. Based on
FEMA 310, the seismic soil type SE was assurlied for the analysis of this
building. There was no observable evidence of liquefaction at the building
site.
The 1997 edition of the Turkish Building Code, Specification for

Structures to be Built in Disaster Areas (4), is modeled after the 1994 edition
of the Uniform Building Code (UBC) (5). Similar to the 1994 UBC,
@Seismicisolation
@Seismicisolation
seismicity zones are used to determine the spectral accelerations for any given
location within Turkey. Adapazari is located in seismic zone 1, the highest
region of seismicity found in Turkey. Using the normalized response spectrum
in the Turkish Building Code, the spectral accelerations for the 10%
exceedance in 50-year event for Adapazari were calculated and modified for
site soil classification and seismic zone factors. For Adapazari, the 0.2 second
period spectral acceleration, Ss, is l.Og, while the 1.0 second period spectral
acceleration, SI, is 0.92g. Information regarding the 2% in 50-year event was
not available.
SEISMIC PERFORMANCE EVALUATION OF THE EXISTING

STRUCTURE
The Adapazari building was evaluated for its expected seismic

performance using FEMA 3 10. The FEMA 3 10 methodology utilizes a series
of checklists that address possible seismic hazards. Checklists are included in
the Handbook for all of the major structural systems, nonstructural elements,
geologic and site hazards. The engineer addresses each statement and
determines whether it is compliant or non-compliant. Compliant statements
identify conditions that are acceptable. Non-compliant statements identify
conditions that are in need of further investigation or strengthening.
FEMA 310 provides a three-tiered process for the seismic evaluation of

existing buildings. The Tier One evaluation is designed to quickly identify any
potential seismic deficiencies, or “weak links”, in the lateral force resisting
system. Checklists are used to identify specific building deficiencies while
requiring minimal analysis effort. The Tier Two evaluation includes a more
detailed analysis of the building deficiencies identified by the Tier One
evaluation. The Tier Two evaluation may be used to demonstrate that
potential deficiencies do not require rehabilitation. The Tier Three evaluation
is a comprehensive study of the building, which includes incorporation of the
building’s nonlinear behavior.
The building was evaluated for Life Safety performance. Life Safety
performance may be defined as significant damage to the structural and
nonstructural components of the building with some margin of safety against
either partial or total structural collapse when subjected to the Maximum
@Seismicisolation
@Seismicisolation
236 Thompson et al.
Considered Earthquake. Some structural elements experience severe damage,
but life safety hazards, such as falling debris either within or outside the
building, are minimal.
Tier One Evaluation
Based on site observations and a review of the original construction

documents, a Tier One evaluation was performed and several lateral force
resisting system deficiencies were identified. The Life Safety deficiencies for
this building include the following:
Soft Story--Based on the retail spaces in the first story, a majority

of the HCT infill walls located at the second through fourth stories
are not present at the first story. Due to the discontinuity of these
infill walls, the elastic stiffness of the building at the first story is
significantly less than the stories above.
Mass--A large reduction of building mass at the first story is due to

lack of HCT infill walls.
Interfering
- Walls--The HCT walls are not isolated from moment
frame beams and columns.
Column Bar Splices--FEMA 356, Chapter 6.4.5, indicates that the

development length and splice lengths for smooth reinforcing bars
shall be 200% of the values specified in ACI-318 (8) for deformed
reinforcing bars. Based on FEMA 356, the vertical reinforcing bar
splice lengths are inadequate (for smooth reinforcing bar).
Beam Bar Splices--Inadequate longitudinal reinforcing bar splice

lengths (for smooth reinforcing bar).
Column Tie SDacing--Inadequatecolumn tie spacing.
@Seismicisolation
@Seismicisolation
Stirrup SDacing--hadequatebeam stirrup spacing.
Joint Reinforcing-Inadequate ties through beam-column joint.
The deficiencies noted are consistent with a building of this

constmction type in Turkey. Based on the observed building performance and
the identified lateral load resisting system deficiencies, a Tier Three evaluation
for Life Safety Performance was performed.
Tier Three Evaluation
A nonlinear static analysis (pushover) was performed on the existing

building using the target displacement procedure of FEMA 356. The target
displacement procedure uses the building’s initial elastic stiffness and the post-
yield stiffness to establish the expected displacement the structure will
experience during a major earthquake. The calculated displacement is known
as the “target displacement” in FEMA 356 and is the displacement at which
members are checked for strength and ductility. If the building can be pushed
to this displacement without significant global degradation of lateral force
resisting element capacities, then it is deemed to meet the performance
objective.
Material Properties--The material properties of the building are based on

site observations and reflect common construction techniques and materials
used in Turkey during the time of original building construction.
Concrete--After the Kocaeli Earthquake, several researches obtained

concrete core samples from various buildings located in areas that experienced
heavy structural damage. The quality and compressive strength of the in-situ
concrete varied from 10 to 15 MPa (3). Based on the observed condition of the
concrete at the building, a compressive strength of 13.8 MPa was assumed.
---The steel reinforcement used at the building consisted primarily

of smooth reinforcing bar. After the 1999 Duzce, Turkey Earthquake, testing
was performed on smooth reinforcing bar samples taken from earthquake
damaged buildings in Duzce, Kaynasli, and Bolu (7). The testing indicated an
@Seismicisolation
@Seismicisolation
238 Thompson et al.
average yield stress, fy, of 276 MPa with an average ultimate stress, fu, of 400
m a . The modulus of elasticity was reported as 200,000 MPa. Based on the
observed condition of the smooth reinforcing bars found in the building,
similar material properties were assumed.
Hollow Clay Tile Infill--The typical HCT walls were observed to be

finished with plaster, but not connected to the structural frame. Based on the
observed condition and published work from Turkish researchers (i), an
expected compressive strength, f‘m, of 2.4 MPa was assumed for the HCT
units.
Model Development
A three-dimensional computer model using SAP2000 of the building

was created. The beams and columns were modeled as “stick” elements
with offsets equal to half the column width. Based on the construction
documents, it appeared that the floor beam and column sizes vary from floor to
floor as they extend up the building. However, for modeling simplicity, only
two typical beam sizes and one column size were used for the building
analysis. In addition, several floor beams frame eccentrically into the building
columns. However, only a concentric framing alignment was used for
modeling the beams and columns.
The HCT infill walls were included in the building model to account
for their mass and elastic lateral stiffness. The HCT walls were modeled using
the strut method presented in FEMA 356, Chapter 7.5.2. The roof, floor and
foundation diaphragms,were modeled as rigid. The soil/structure interaction
was not modeled and the building base was assumed to be “fixed.”
A nonlinear, static pushover analysis was performed on the 3-

dimensional model for both building directions. Two displacement patterns
were used for the pushover analysis. The first pattern was proportional to the
inverted triangular first mode shape, while the second was a uniform
displacement pattern. Both displacement patterns were developed in
@Seismicisolation
@Seismicisolation
accordance with the provisions of FEMA 356.
Nonlinear Properties--The nonlinear member properties were explicitly
calculated for each component using both the principles of mechanics and the
FEMA 356 tabulated values. Based on the concrete and reinforcing steel
stress-strain relationships, moment-curvature diagrams were developed for
each beam and column element. The moment-curvature diagrams represent
the nonlinear force-displacement behavior of the components.
The column elements were modeled with flexural hinges at both the top
and bottom of each element. Based on the presence of smooth reinforcing bars
and inadequate lap splice lengths, the shear demand required to develop the
maximum column flexural capacity was less than the calculated column shear
strength. Based on FEMA 356, Chapter 6.4, the column flexural capacity was
reduced by 65% to account for smooth bars and inadequate lap slice lengths
observed at the building.
The deformation levels for each component was limited to the

maximum usable strain for concrete and reinforcing steel as specified in
FEMA 356. The moment-curvature relations were developed using a
maximum concrete compressive strain 0.003 and a limiting steel tensile strain
of 0.05. The stiffness and strength properties for the unreinforced clay tile
masonry infill struts were developed using Section 7.5 of FEMA 356.
AcceDtance Criteria
The component acceptance criterion is based on FEMA 356, Tables 6-7

through 6-9 for reinforced concrete beams, columns, and joints. The beam and
column elements were evaluated as primary elements using the Life Safety
performance criteria.
@Seismicisolation
@Seismicisolation
240 Thompson et al.
Results
Using the FEMA 310 Tier Three seismic evaluation, the modeled
building performance had good correlation with the observed building
performance. The evaluation indicated that the building could not achieve the
target displacement since the existing concrete frame columns lacked the
necessary strength and ductility. Results of the pushover analyses for both
building directions are shown in Figure 2. In general, the pushover curves
exhibit a limited yielding plateau followed by a stepwise degradation of
strength and stiffness.
Based on deformation compatibility, the unreinforced clay tile infill

walls resist a majority of the lateral force demands early in the pushover
record. However, the walls crack and lose strength and stiffness very quickly.
The lateral force demands are then resisted entirely by the concrete moment
frame. At the end of the yield plateau, the lateral force resisting system
degradation is due to the inadequate shear and moment capacity of the concrete
frame columns.
The target displacement, 6t, for the building was calculated as 21.2 cm,
which is relatively large for a four-story building. The large target
displacement is primarily the result of the tall first story height and the lack of
lateral strength and stiffness. As shown in Figure 2, the building displacement
capacity before degradation of the lateral force resisting systems begins is
approximately 2.5 cm.
Based on the degradation of the overall lateral force resisting system

occurring before the target displacement in all cases, the building is judged to
not meet the Life Safety Performance Objective on a global basis.
@Seismicisolation
@Seismicisolation
SEISMIC STRENGTHENING OF THE EXISTING STRUCTURE
A conceptual scheme was developed to add strength and stiffness to the

building. The primary objective of the conceptual scheme was to use only
materials and construction techniques that are readily available in Adapaza&
Turkey. The resulting conceptual scheme utilizes new concrete shear walls
placed on column grids in both building directions (Fig. 1). The new shear
wall elements are each 20.3 cm thick and extend full height of the four-story
building. The strengthened building was evaluated for Life Safety
performance using the nonlinear static analysis procedure of FEMA 356.
Material Properties
The material properties of the new shear walls were assumed to be

similar to those used for the existing building. The concrete compressive
strength was assumed to be 13.8 MPa. The reinforcing steel was assumed to
be smooth and have an average yield stress of 276 MPa.
Nonlinear Properties
The strengthening scheme assumes that the new shear walls will be
built with adequate lap splice lengths and satisfy current Turkish Building
Code ductile detailing requirements. Based on the aspect ratio and expected
flexural capacity of the new shear walls, the walls are shear controlled
elements and modeled with shear hinges.
Acceptance Criteria
The component acceptance criterion is based on FEMA 356, Tables 6-19 and
@Seismicisolation
@Seismicisolation
242 Thompson et al.
6-21 for reinforced concrete shear walls. The new shear walls were evaluated
as primary elements using the Life Safety performance criteria. The existing
concrete moment frame beams and columns were considered as secondary
elements since they do not appreciably contribute to the lateral force resisting
system of the strengthened building.
Results
Results of the pushover analyses for both building directions are shown
in Figure 2. In general, the pushover curves exhibit a yielding plateau
followed by a stepwise degradation of strength and stiffness. However, the
new concrete shear walls provide additional strength, stiffness and ductility to
the lateral force resisting system. The additional lateral force resisting stiffness
results in a lower target displacement and provides “protection” of the existing
concrete columns from excessive drift demands.
Based on the added stiffness to the lateral force resisting system, the
calculated target displacement is 4.4 cm and 4.9 cm for the East-West direction
and North-South directions, respectively. As shown in Figure 2, prior to
degradation of the lateral force resisting system, the target displacement is
achieved. Since overall degradation of the lateral force resisting system occurs
after the target displacement is reached, the lateral force resisting system is
judged to meet the Life Safety Performance Objective on a global basis.
In addition, Figure 2 indicates the base shear strength required by the

1997 Turkish Building Code for a new concrete moment frame building in
Adapazari. One traditional approach for seismically strengthening the building
would be to “bring it up to code”. However, as shown by the figure, a
considerably small (approximately 40%) design force level can be used when
using the FEMA 356 nonlinear static procedures for strengthening the
building. Rather than adding strength to the building based on a conservative
static analysis, the pushover analysis provides a method for adding an
appropriate level of strength and deformation capacity to the building for Life
Safety performance. Consequently, the reduction in design lateral forces
could correspond to a savings in construction dollars.
@Seismicisolation
@Seismicisolation
The conceptual strengthening scheme developed focuses primarily on
the addition of new concrete shear walls. However, strengthening of
additional elements would most likely be necessary to provide a complete
lateral force resisting system. However, for simplicity, the conceptual
strengthening scheme includes the following assumptions regarding the
existing lateral force resisting system:
0 The floor diaphragms posses adequate shear and bending strength

to resist the inertial seismic demands,
o The floor beams and girders are adequate to collect and deliver the
diaphragm forces to the new shear walls,
o The foundation mat slab is adequate to resist the shear and
overturning demands imparted by the new shear walls.
CONCLUSIONS
A seismic evaluation was performed for a four-story non-ductile

reinforced concrete moment frame building in Adapazan, Turkey. The
building was evaluated for Life Safety performance in accordance with FEMA
310 using a nonlinear static (pushover) analysis. Based on the seismic
evaluation, a conceptual strengthening scheme was developed to strengthen the
lateral force resisting system. Life Safety performance is achieved by the
addition of strength and stiffness with new concrete shear walls. The
strengthened building is capable of reaching the FEMA 356 target
displacement without experiencing overall degradation of the primary or
secondary lateral force resisting elements.
The results of the pushover analysis indicates that the required base
shear demands are significantly less than the required Turkish Code level
forces for a new building with the same lateral force resisting system. The
reduction in design force levels is proportional to a reduction of the direct
structural construction costs for a seismic strengthening project. The analysis
demonstrates that a historically poor performing lateral force resisting system
can be strengthened to meet Life Safety performance using less that code level
design lateral forces if nonlinear static analysis procedures are used.
@Seismicisolation
@Seismicisolation
244 Thompson et al.
REFERENCES
1. Yuksel, E, A. Ilki., F. Karadogan (1998). Brick Masonry Infill Walls Integrated

to the Peripheral RC Elements and Strengthening. Repair and Strengthening of
Existing Buildings, Second Japan-Turkey Workshop on Earthquake Engineering,
Istanbul Technical University, Istanbul, Turkey, 309-317.
2. Federal Emergency Management Agency (FEMA) (2000). Prestandard and

Commentav for the Seismic Rehabilitation of Buildings, prepared by the
American Society of Civil Engineers for the Federal Emergency Management
Agency. FEMA Publication No. 356. Washington, D.C.
3. Earthquake Engineering Research Institute (EERI) (2000). Kocaeli, Turkey,

Earthquake of August 17, 1999 Reconnaissance Report, Spectra. EERI
Publication No. 2000-03. Oakland, California, December 2000, p. 254.
4. Nuray, M., ed. (Amended 1998). English translation of the Specification for
Structures to be Built in Disaster Areas Part 111 Earthquake Disaster Prevention:
Chapters 5 - 13. Bogazici University, Istanbul, Turkey.
5. International Conference of Building Officials (ICBO) (1994). 1994 Uniform

Building Code Volume 2. Whittier, California.
6. Federal Emergency Management Agency (FEMA) (1998). Handbook for the

Seismic evaluations of Buildings-A Prestandard, prepared by the American
Society of Civil Engineers for the Federal Emergency Management Agency.
FEMA Publication No. 310. Washington, D.C.
7. Kurama, Y.C., B. Morgan, and Q. Shen (2000). Stress-Strain Behavior of

Turkish Reinforcing Bars from the November 12, 1999 Duzce Earthquake,
Technical Note, Department of Civil Engineering and Geological Sciences,
University of Notre Dame, Indiana, February 2000, 12 pp.
8. AC1 Committee 318, ?Building Code Requirements for Structural Concrete (AC1
318-99),?ACI 318R-99, American Concrete Institute, Detroit, 1999.
9. SAP2000 Ver. 6.1, Computers and Structures, Inc., Berkeley, California.
@Seismicisolation
@Seismicisolation
I\
m
9
@ )E(‘ \(E)
FLOOR BEAM, NP.
COLUMN, NP
Figure 1. Typical existing floor plan with new shear walls.
@Seismicisolation
@Seismicisolation
246 Thompson et al.
ao
-- I I~laadulg
Figure 2. Adapazari building pushover curve for the 10% in 50 year

earthquake.
@Seismicisolation
@Seismicisolation
S P 209-1 4
Quantitative Infrared Thermography

for Quality Control of Concrete
Structures Strengthened with
FRP Composites
by M. A. Starnes, N. J. Carino, and E. A. Kausel
Svnopsis: The finite-element method is used to carry out parametric analyses on

the thermal response of simulated defects in fiber-reinforced polymer (FRP)lami-
nates applied to a concrete substrate. The aim is to assess the potential for qualtita-
tive infrared thermography in not only detecting a flaw but also being able to de-
scribe its physical characteristics. Three parametric studies are presented, namely:
1) relationship between the thermal input, the maximum signal, and the maximum
surface temperature; 2) effects of flaw depth and the number of FRP layers; and 3)
effect of flaw width. From these simulations, procedures are established for select-
ing the thermal input and estimating the flaw depth and width.
Kewvords: fiber reinforced polymer composites; finite

e Iem e nt mode Iing; infra red the rmo g ra p hy;
nondestructive testing
@Seismicisolation
@Seismicisolation
247
248 Starnes et al.
Monica A. Starnes is a graduate student in civil engineering at the
Massachusetts Institute of Technology and a researcher at the National Institute
of Standards and Technology. Her interests include nondestructive testing,
structural rehabilitation, and application of high-performance materials in civil
engineering structures.
AC1 Fellow Nicholas J. Canno is a research structural engineer at the National

Institute of Standards and Technology. His research interests include
nondestructive testing of concrete, standard test methods, and high-performance
concrete. He is a former director of the AC1 and currently serves on the
Technical Activities Committee and chairs the Specifications Committee.
Eduardo A. Kausel is a professor in the Department of Civil and

Environmental Engineering at the Massachusetts Institute of Technology. His
research interests are in the areas of structural dynamics, soil dynamics, soil-
structure interaction, earthquake engineering, and nondestructive testing.
INTRODUCTION
The trend in civil engineering is toward the use of advanced materials such as
fiber-reinforced polymer (FRP) composites. Among other purposes, these
materials are being used to strengthen existing concrete and masonry structures
for rehabilitation purposes, which calls for the use of nondestructive evaluation
(NDE) techniques to ensure their correct application and performance.
The best NDE technique needed to detect and characterize a defect depends on
the critical size of the defect, the size of the structure being tested, and the
environment in which the inspection is carried out. The ideal NDE method in
civil engineering applications should be:
Accurate, reproducible, reliable, robust, and economical;
Able to inspect large areas as well as localized areas;
Able to detect critical defect sizes; and
Non obtrusive to the surrounding environment, and convenient to
the users and evaluator of the structure.
Nondestructive testing of thin FRF' laminates bonded to concrete or masonry

presents a variety of difficulties in the use of the traditional methods developed
for inspection of metals due to the anisotropy, variable and non-homogenous
composition, non-magnetic properties, and high ultrasonic attenuation of FRP
materials. Among the various available techniques, however, infrared
thermography offers the greatest potential as a global NDE method, and it has
been used successfully to detect defects in FRP laminates bonded to concrete
@Seismicisolation
@Seismicisolation
(Hawkins et al., 1998). Trial case studies performed by the Federal Highway
Administration (FHWA) and New York Department of Transportation
(Alampalli et al., 2001), among others, confirmed that infrared thermography is
a promising NDE method considering testing speed and ability to detect flaws.
Infrared thermography allows localized and global testing of structures, is non
obtrusive and convenient to the users. Technological advancements in uncooled
detectors are making infrared thermography a more accurate and economical
testing technique than it was in the past.
Current inspection techniques using infrared thermography, however, provide

only qualitative assessments of the state of the structure, but give no quantitative
information on existing defects. That is, they establish whether a subsurface
flaw exists within the FRP-substrate system, but not the depth of the defect or its
approximate volume. This qualitative nature of the results stems from the
complex relationships between the variables affecting the thermal response of
the bonded laminates. The need for defect characterization, not just defect
detection, has promoted further research in quantitative infrared thermography.
To enable widespread use of infrared thermography for quantitative assessment

of FRP materials applied to concrete and masonry structures, a standard test
method is needed. To develop such a standard, however, it is necessary to
develop a greater understanding of the factors affecting the thermal response of
FRP composites bonded to concrete. The National Institute of Standards and
Technology (NIST), in cooperation wi- the Massachusetts Institute of
Technology (MIT), is carrying out research to establish the scientific basis for
the development of a standard methodology for using infrared thermography in
nondestructive evaluation of concrete and masonry structures strengthened with
FRP composites. The goal in this development is to gain an understanding of
how the various parameters that may be encountered in infrared thermography
testing affect the measured response.
This paper presents the initial results of the NIST/MIT program. First the
principles involved in infrared thermography are summarized. Then, the finite-
element models used are described. Finally, the effects of the following
parameters on thermal response are discussed:
The amplitude and duration of the thermal input;
0 The depth of the flaw (air void); and
w The width of the flaw.
PRINCIPLES OF HEAT TRANSFER
Infrared thermography, as a tool for flaw detection, is based on the principle that
heat transfer in any material is affected by the presence of subsurface flaws or
any other change in thermal properties. Figure 1 is a schematic of a concrete
@Seismicisolation
@Seismicisolation
250 Starnes et al.
substrate with several FRF' laminates applied to the surface. There is a flaw
within the F W layer and this flaw affects the flow of heat into and out of the
FRP-concrete composite. The difference in heat flow through the flawed and
unflawed regions causes localized energy differences on the surface of the test
object, which can be measured using an infrared detector or radiometer.
Through data processing, the measured infrared radiation is transformed into a
surface temperature distribution and recorded in the form of thermograms
(isotherm plots). Anomalies in the thermogram indicate the presence of
subsurface flaws in the test object. The relation between surface temperature and
emitted radiation is based on the Stefan-Boltzmann (Eq. 1) and Wien
displacement (Eq. 2) principles (Schlessinger, 1995)
W = eoT4
b
4
l,-- -
T
where,
W = radiant intensity (W/m2),
e = emissivity of the test object (dimensionless quantity),
= Stefan-Boltzmann constant (5.67x10-*W/(m2-K4)),
T = absolute temperature (K),
arna= wavelength of the maximum radiation intensity (pm),and
b = Wien displacement constant (2897 p d ) .
Equation (1) indicates that the emitted radiation is a nonlinear function of

temperature, and Eq. (2) indicates that the wavelength of the peak intensity
decreases with increasing temperature. At near room temperatures, the
wavelength is in the infrared region of the electromagnetic spectrum. The
emissivity is a property of the surface and has a value between zero and one. An
emissivity value of one is found in perfect emitters such as a black body.
Flaw detection and characterization in civil engineering structures require active

infrared thermography, which generally involves transient heat transfer. In
addition, defect characterization requires the use of time-resolved infrared
thermography. Using this technique, the surface temperature of the test object is
monitored and analyzed as a function of time, instead of being monitored at only
one particular instant in time. To obtain transient heat transfer, an external
thermal stimulus, such as a heating lamp, must be applied to the test object.
Quantitative characterization of internal anomalies by using inflar&

thermography requires the study of transient heat transfer. Starting with the
general case, heat transfer in any material is governed by the theory of diffusion
(Lienhard, 1981), the following differential equation:
@Seismicisolation
@Seismicisolation
9 .K V T + Q = P C - aT (3)
at
where,
9. = divergence operator,
-
K
v =
= heat conductivity tensor,
gradient operator,
T = temperature,
Q = internal heat generation,
P = density,
c = specific heat, and
t = time.
If the change in the conductivity tensor is relatively small with respect to the
temperature of the material, Eq. (3) simplifies to
aT
K(v'T)+ Q = p c - (4)
at
where O* is the Laplacian operator. In the Cartesian coordinate system, the

Laplacian of the temperature is defined as
ô2T
V 2 T = -+- ô2T
ô2T + -
ax2 itv2 azz
When heat is applied suddenly to the surface of an object, transient heat flow
occurs until thermal equilibrium is reached. During transient heat flow, the
temperature at any point in the object changes with time. If the object has large
dimensions perpendicular to the direction of uniform heat input and area of
uniform heat input large enough compared to measurement depth, heat flows
parallel to the input, and the problem is reduced to one-dimensional heat flow.
One-dimensional transient heat flow theory states that the temperature within the
object changes in a nonlinear manner upon a step change in surface temperature,
as follows (Lienhard, 1981):
Td =T, +(q -T,)
where,
= temperature at any depth y in the object,
= applied constant temperature at the surface of the specimen,
= initial temperature of the solid,
= the Gaussian error function for ( ),
= time, and
= thermal diffusivity of the material.
@Seismicisolation
@Seismicisolation
252 Starnes et al.
Thermal diffùsivity of a material is defined as
where,
k = thermal conductivity,
p = density, and
c = specificheat.
Therma diffusivity affects how fast a material changes Zmperature under
transient conditions.
in summary, during transient heat transfer through an object, surface

temperature changes caused by internal anomalies in the matenal are time
dependent. Surface temperatures depend on the elapsed time and the type, size,
and depth of the defect. The surface temperature distribution is established by
measuring the emitted radiation using an infrared camera. Accurate
measurement of the surface temperature distribution depends on knowing the
value of the emissivity of the surface and on environmental “noise” such as
atmospheric attenuation and air movement.
ANALYTICAL MODELING
The complexity of transient heat transfer phenomena, involving space and time,
is a key obstacle to the development of quantitative infrared thermography.
From a theoretical perspective, solutions to transient thermal problems by direct
integration of the governing differential equations (Eqs. (4) and ( 5 ) ) is only
possible for the simplest of conditions. The alternative approach is to solve the
governing equations using numerical methods, such as the finite-difference
method or the finite-element method.
The flexibility of modem numerical algorithms allows the researcher to

investigate heat transfer under a variety of conditions, from the relatively simple
to the highly complex, provided the proper material properties and boundary
conditions are used to create the numerical model. A verified numerical model
allows for cost-efficient investigation of the effects of different factors on the
thermal response of an object, such as concrete with bonded FRP laminates.
The initial phase of the research is focused on the use of the finite-elemenî
method to investigate the parameters affecting the transient thermal response of
FRE’ composites bonded to concrete. in a sense, the finite-element method is
used to simulate thermographic testing of FRP laminates bonded to concrete
substrates, with and without defects. The results of three parametric studies are
presented in this paper. The objectives of the studies are:
@Seismicisolation
@Seismicisolation
0 Optimize heating time and intensity for maximum signal;
Investigate the effect of flaw depth and whether the depth can be
estimated from characteristics of the signal; and
Investigate whether flaw width can be estimated from the
characteristics of the signal and establish minimum detectable flaw
size.
Simulation Models
Numerical simulations are performed using a general-purpose, finite-element

analysis package. The simulated test object consists of a 100 mm wide by
20 mm thick concrete slab covered with several layers of 0.5 mm thick carbon
FRP (CFRP) laminates. To reduce computation time, two-dimensional models
are used, that is, the test object is infinite in the third dimension. Subsurface
flaws of varying depth and size are placed at the center of the model. Internal
flaws are simulated as air gaps, representing delaminations, debonds, or concrete
spalls depending on their location. Delaminations are gaps located between the
FRP laminates, debonds are gaps at the interface between FRP and concrete, and
concrete spalls are gaps in the concrete substrate. Because the flaw is located at
the centerline of the model, the model is symmetrical about its centerline, and
only one half needs to be analyzed. Figure 2 shows the physical model and the
model analyzed.
The material properties of the model are those of concrete for the substrate, air
for the defect, and CFRF’ for the bonded composite. The material properties
used in the simulations are presented in Table 1. For the FRP, the model
simulates a wet lay-up composite system. The material properties of the
constituent materials of the FRP (carbon fibers and epoxy resin) are combined to
simulate homogeneous but anisotropic laminates. Homogeneity is assumed
based on the scale of the problem, macroscopic instead of microscopic. As a
result, the material in direct contact with the concrete substrate is the FRP
composite. The CFRP laminate in direct contact with the concrete has the fibers
running in the x-direction, while each of the adjacent laminates has the fiber
direction rotated by 90 degrees.
The finite-element model is composed of four-noded quadrilateral elements.

Each node has one degree of freedom, namely the temperature. The model is
meshed using the “mapped meshing” feature of the program, which allows the
user to directly control the element size and type. The “global” element size is
set to 0.5 mm. Mesh refinement is applied to the thin CFRP laminates and at the
FRPIconcrete interface. In the y-direction, the thickness of each composite
laminate is subdivided into 4 elements while the thickness of the flaw is
subdivided into 2 elements. The rest of the concrete substrate is meshed using a
graded mesh, with a finer mesh at the top and a coarser mesh at the bottom of
the specimen. The average thickness of the concrete elements is 2.75 mm with a
@Seismicisolation
@Seismicisolation
254 Starnes et al.
“spacing ratio” of 10. The spacing ratio represents the ratio of the thickness of
the largest element to the thickness of the smallest element.
For comparison purposes, the same time stepping is used for all the problems.
The initial time step is 0.009 s. The automatic time stepping capability of the
finite-element software is used with minimum and maximum time steps of 0.008
s and 0.1 s, respectively. A sensitivity analysis was performed to determine an
appropriate maximum time step for this particular study. For this purpose, trial
simulations were performed using three different maximum time steps, namely
0.009 s, 0.1 s, and 0.5 s. The maximum time step of 0.5 s was found to be too
large to fully capture the thermal evolutions of these particular sets of
simulations. Maximum time steps of 0.009 s and 0.1 s successfully captured
the thermal behavior of the models. The maximum time of step of 0.1 s is
selected as a balance between computational economy and accuracy. The
simulation output (nodal temperatures) is recorded at every time step.
The analysis is defined as a transient heat transfer problem. A square heat pulse
of intensity q and duration t is applied to the top surface of the model (see
Fig. 1). Adiabatic conditions (dTldx = O and dTldy = O) are assumed for the
bottom and side surfaces of the model (Fig. 2). This assumption is realistic for
the short time durations used in the analyses and the geometry of the simulated
flawed test object. The initial temperature for all the simulations is 23 O C , which
represents ambient temperature. For simplification, uniform heating, no cooling
losses, and perfect contact between layers are assumed for the analyses.
Thermal Response Parameters

The surface temperatures at various points above the flaw as well as the
“background” temperature at a point where the flaw has no influence are
recorded (see Fig. 1) as the output of each analysis. The recorded nodal
temperature histones are then used to compute the following parameters:
Maximum surface temperature;
The thermal signal;
0 The thermal contrast;
0 The time to reach maximum signal; and
The time to maximum contrast.
The maximum surface temperature occurred above the flaw at the end of the
thermal pulse. The thermal signal AT is defined as
where,
Tde/ecr = temperature at the surface above the flaw, and
Tbackgmund = temperature at a point above sound matenal that is not
affected by the flaw.
@Seismicisolation
@Seismicisolation
The thermal contrast C is defined as
where,
AT = thermal signal,
Tbackgmund = temperature at a point above sound material that is not
affected by the flaw, and
TO = initial temperature of the test object.
The thermal signal is the most widely used parameter for detecting subsurface
flaws by infrared thermography. Some researchers, however, prefer using the
thermal contrast because this normalized value is independent of the amount of
heat applied to the specimen, but it depends on the details of the flawed test
object. The thermal contrast is analogous to a stress concentration factor in that
it indicates the effect of the flaw on magnifjmg the surface temperature above
the flaw.
The time to reach maximum signal ts and the time to reach maximum contrast tc
are used to investigate the effects of varying the depth of the flaw.
RESULTS
Effect of Thermal Input

The first parametric study evaluates the effect of the thermal input, which is
modeled as a square thermal pulse of amplitude q (W/m2) and duration z (s).
The values of q and z are varied over a wide range. The objective is to determine
whether a methodology can be developed for estimating the optimal thermal
energy for experimental configurations relevant to civil engineering
applications. The pulse durations vary from 0.05 s to 3 s and the applied heat
flux varies from 5,000 W/m2to 100,000 W/m2.
The test object consists of two CFRP laminates bonded to the concrete substrate
as described in the previous section. The subsurface flaw is a 0.2 mm thick
debond.
Analysis of the response parameters yields interesting results regarding the

maximum signal as well as the maximum surface temperature. It is found that
the effects of the thermal pulse can be summarized easily if the product of q ari'd
z are used. The product z*qrepresents the input energy per unit area, that is,
E=q*z (10)
@Seismicisolation
@Seismicisolation
256 Starnes et al.
It is found that the maximum signal can be expressed as a linear function of the
input energy per unit area, which will be called the “input energy density.” For
the particular test object used in this series of analyses, the best-fit empirical
relationship is found to be:
AT,, = 0.00024 E (1 1)
Figure 3 illustrates the maximum signal versus input energy density. This figure
provides a powerful tool for selecting the thermal input to achieve a desired
maximum signal. This tool, however, has a limitation because there are infinite
combinations of input heat flux and pulse duration for a specific maximum
signal. The differences achieved with the various combinations of q and are
the time when the maximum signal is obtained and the maximum surface
temperature that is attained.
The maximum surface temperature reached during thermographic testing of FRP

bonded to concrete requires special consideration. When the temperature of the
resin of the FRP increases above the glass transition temperature, Tg, the
mechanical properties of the matrix degrade. Glass transition temperatures for
CFRP used in civil engineering applications are often on the order of 50 “C
(Karbhari et al., 1998; Christensen et al., 1996). Thus the allowable maximum
surface temperature is a limiting factor in selecting the thermal input.
The results of the simulations are analyzed next to investigate the relationship
between the thermal input and the maximum surface temperature. The analysis
indicates that, for a given pulse duration, the maximum surface temperature is a
linear function of the input energy density, as follows:
T,, = S ( T ) E + 23
where,
6 ( T) = slope of the line for a given pulse duration.
The value of 6 is found to be a nonlinear function of ~ . pulse

e duration. For
short duration pulses, the slope S is large and surface temperature increases
sharply with the input energy density compared with longer duration pulses. It is
found that the slope Gcan be approximated by the following equation:
O. 783 + 0.192 - 0.0762

6(2)= -
I
Equations (12) and (13) allow the construction of Fig. 4, which shows the
maximum surface temperature as a function of input energy density and pulse
duration. Figure 4 illustrates that for a required minimum input energy density
(as established by the desired ATmnx)a series of different pulse durations may be
@Seismicisolation
@Seismicisolation
selected. The limiting maximum surface temperature determines the shortest
pulse duration that can be used to achieve the desired signal.
In summary, the results of the first parametric study lead to a simple approach
for selecting the thermal input to obtain the required maximum signal while
limiting the maximum surface temperature. For example, suppose we desire a
maximum thermal signal of dT,,=lO OC. According to Fig. 3, the required input
energy density is about 40 k.T/m2. Suppose we desire to limit the surface
temperature to 50 OC. Assuming that the initial temperature is 23 OC, Fig. 4
shows that the puise duration should not be less than 2 s. The required input heat
flux will be lower for a longer pulse duration. For a pulse duration of 2 s the
required input flux is 20 kW/m2, and for a 3 s pulse duration it is 13.3 kW/m2.
Of course, the relationships in Figs. 3 and 4 are only applicable to the specific
object used in these simulations.
Effect of the Depth of the Flaw
When applying FRP materials to strengthen concrete or masonry structures,

defects may arise between the FFV laminates (delaminations) or at the
FFVkoncrete interface (debonds). In addition, failure planes may occur in the
concrete if the shear stresses needed to transfer load to the FRP, exceed the
capacity of the concrete (spalls). Thus defects can occur at different depths
within the repaired object. It is, therefore, desirable to investigate how the depth
of a flaw affects the thermal response and to establish whether it may be feasible
to determine the depth of the flaw from the characteristics of the thermal
evolution. The term “thermal evolution” refers to the spatial and temporal
variation of the surface temperature.
Three test objects, which include 2, 3, or 5 CFRP laminates and flaws, are
analyzed. For each test object, the depth of the flaw is varied to include
delaminations, debonds, and concrete spalls at different depths. The results
reveal that the thermal response is a function of both the thickness of the FFV
layer and the depth of the flaw. For the purpose of estimation, a simple
procedure is developed to represent the maximum thermal signal and contrast as
a function of flaw depth, which is described by the following simple power
function:
Thermal response = yd“ (14)

where,
d = depth of the flaw,
y = empirical coefficient, and
E = empirical exponent.
The time for maximum signal and the time for maximum contrast are selected as
the thermal response parameters best suited for correlation with flaw depth. The
@Seismicisolation
@Seismicisolation
258 Starnes et al.
best-fitting power functions for the time of maximum signal ts and the time for
maximum contrast tc are presented in Table 2.
Based on the information obtained from this parametric study, a tentative

procedure for estimating the depth of the flaw consists of the following:
0 Estimate the thickness of the FRP layer (in mm) by some other
technique.
0 Estimate the coefficients and exponents (Eq. 14) for time for maximum
signal and for time for maximum contrast. For the time for maximum
signal, these values can be found from these empirically determined
relationships:
yo = 4.849(dFw)0.06g7
E~ = 0.0816dF, + 0.943 (16)
For the time for thermal contrast, the coefficients are:
(
y,, = 8.432 dFRp)'.03'
= 0.867dFw + 0.852
where,
dmp = thickness of the FRP layer,
= coefficient of the power function in Eq. (14), and
y
E = exponent of the power function in Eq. (14).
The coefficients and exponents determined in the previous step provide
the power function equations for the time for maximum signal and time
for maximum contrast for flaws at any depth, that is,
t, = yo d" (19)
where,
d = depth of the flaw,
ts = time for maximum signal, and
tc = time for maximum contrast.
Inversion of Eqs. 19 and 20 allows the estimation of the depth of the flaw
based on the measured thermal response of the test object.
@Seismicisolation
@Seismicisolation
Since these equations are based on the test objects used in the simulations,
experimental verification is needed to establish the reliability of the proposed
procedure.
An alternative approach for estimating flaw depth is to use previously

established plots of the time for maximum signal as a function of flaw depth, as
illustrated in Fig. 5. This procedure is simpler than using the above equations,
and it may provide an adequate estimation for the flaw depth. In the example
shown in Fig. 5, for a 10 s time for maximum signal, the estimated flaw depth is
1.8 mm for the case of a test object with 5 layers of CFRP, while it is 2.0 mm for
a test object with 2 plies of CFRP. For this case, the estimation of flaw depth
varies only by 0.2 mm, which is less than the thickness of an FRP layer. For a
longer time for maximum signal, say 15 s, the estimated flaw depth is 2.5 mm
for the case of 5 layers of CFRP and 2.9 mm for 2 layers of CFRP. The
estimated flaw depth varies by only 0.4 mm; again this is less than the thickness
of an FRP layer. Thus it may not be necessary to know accurately the thickness
of the FRP to estimate the flaw depth.
Effect of the Width of the Flaw
The size of subsurface flaws could be the determining factor on the proper
performance of an FRP composite bonded to concrete. With this concern, the
International Conference of Building Officials Evaluation Services (ICBO ES
2001) has introduced requirements for the allowable flaw size. Among the
conditions of acceptance, the ICBO ES states that small air flaws of diameters
up to 3 mm occur naturally in FRP systems. Flaws of this size and smaller do
not require repair. Flaws larger than 13 cm2in area, however, should be repaired.
The objectives of the third parametric study are as follows:

0 Understand the effect of flaw width on the thermal response;
Establish a procedure for estimation of flaw width; and
0 Determine the minimum width of detectable flaw.
For this part of the study, the assumed thickness of the flaw is 0.1 mm. Flaws
ranging from 3.0 mm to 25 mm wide (x-direction) are investigated. The models
include three different groups depending on the depth of the flaw: delaminations
@Seismicisolation
@Seismicisolation
260 Starnes et al.
0.5 mm deep, debonds 1.5 mm deep, and concrete spalls 2.5 mm deep. Three
layers of CRFP are used as shown in Fig. 6.
Since the investigation aims at estimating the width of the flaw, the surface
temperature as a function of distance (the x-coordinate) is considered. Figure 7
illustrates a typical temperature-distance plot. The location of the inflection
point illustrated in Fig. 7 is chosen as the basis for estimating the location of the
edge of the subsurface flaw. The location of the inflection point may be
computed by setting the second derivative of the temperature vs. distance curve
equal to zero
where
Ts = surface temperature,
X = distance from the centerline along the x-axis, and
wesrimBie = estimation of the distance of the edge of the flaw from the
centerline.
The surface temperature as a function of the distance is recorded at the time of

maximum signal for each simulation. The results of the estimation using the
location of the inflection point are presented in Table 3. The estimation error
computed as
error = "'estimate - "'actua, (24)
where
wact.,~ = actual width of the flaw,
is also presented in Table 3.
The results indicate that the flaw may be under or overestimated depending on
the size and depth of the flaw. For example, the width of large flaws tends to be
underestimated while the width of smaller flaws tends to be overestimated.
Further analysis indicated that the estimation mors could be expressed as

hyperbolic functions as shown in Fig. 8. The figure shows that the error
increases with the depth of the flaw, that is, the estimation of delamination width
is more accurate than the estimation of debonds or concrete spalls. If flaws
larger than 13 cm2in area (36 mm x 36 mm) are a concern (ICBO ES, 2001), the
width estimation error of such flaws is smaller than 6 %.
An additional issue that needs addressing is the minimum width of detectable

flaws. The width of detectable flaws is influenced highly by the thermal
sensitivity of the infrared detector or camera and the environmental noise,
among others. Thus, calculation of the minimum detectable flaw is based on the
required maximum thermal signal.
@Seismicisolation
@Seismicisolation
The minimum detectable flaw is estimated for three different cases: AT,, equal
to O. 1 O C , 1.O O C , and 2.0 OC. The first case, AT,, equal to O. 1 O C , corresponds
to the thermal sensitivity of most infrared cameras. Because the values reported
by the finite-element analysis are ideal, that is, with perfect layer interfaces and
without ambient noise, the cases with AT,, equal to 1.0 O C and 2.0 OC are also
considered.
Analysis of the FEM output reveals that the increase in AT,, can be expressed
as the following hyperbolic functions of the width of the flaw:
for delaminations (d = 0.5 mm),
O.40(w- 1.86)
N,, = 20.95
1+0.40(w-1.86)
for debonds (d = 1.5 mm),
O. 14 (w-1.90)
N,, = 5.95
I + O.14 (w-1.90)
and for concrete spalls (d = 2.5 mm),
0.1O (w-1.87)
dT,, =3.12
1 +O.1O (w-1.87 )
Thus, the minimum width of detectable flaws can be computed using Eqs. 28,
29, and 30. The results from the calculations are presented in Fig. 9.
Observation of the results presented in Fig. 9 indicates that as the resolution of

the signal increases (smaller AT,,), the minimum width for detectable flaws
decreases. Power functions can be used to express the minimum width of the
detectable flaw as a function of depth. For the case of a thermal resolution of
0.1 O C the minimum width of detectable flaw can be expressed as
where
Wmin = the minimum width of detectable flaws, and
d = the depth of the flaw in millimeters.
For the case of a thermal resolution of 1.0 O C the following power function can
be used to estimate the minimum width of detectable flaws:
@Seismicisolation
@Seismicisolation
262 Starnes et al.
wmin= 1.86 -+- 0.61d2'27
Finally, for the case of a thermal resolution of 2.0 O C , the following power
function estimates the minimum width of detectable flaws:
wmin= 1.68-k 1.33d"87
The results indicate that near-the-surface delaminations can easily be detected

up to widths of 2 mm. The minimum width required for detection increases with
increasing depth. For concrete spalls buried 2.5 mrn from the surface, the
minimum width varies from 2 mm with AT,, = 0.1 O C to 20 mrn with AT,, =
2OC.
CONCLUSIONS
The finite-elementmethod is used to evaluate the effect of several parameters on

the thermal response of subsurface flaws in concrete strengthened with FRP
composites. The parameters investigated are the thermal input required for flaw
detection, the depth of the flaw, and the width of the flaw.
The results indicate that the thermal input can be optimized. Given a specific
heat flux, the selection of the pulse duration is governed by the surface
temperature that can be tolerated and the required input energy density for the
desired signal. For a given flaw geometry, the maximum signal is a linear
function of the input energy density E (J/m2).The maximum surface temperature
is a function of the input energy density as well as pulse duration.
The investigation involving the effect of flaw depth leads to a simple procedure
for estimating the response parameters as a function of the thickness of the FRF'
and the depth of the flaw. The procedure developed for the estimation of flaw
depth is based on the time for maximum signal.
Finally, both the flaw depth and minimum flaw width required for detection are
considered. The results indicate that the minimum flaw width required to detect
a flaw increases nonlinearly with increasing depth. The determination of the
minimum flaw width is a function of the maximum thermal signal. The required
maximum thermal signal depends on the thermal resolution of the infrared
detector and the ambient noise due to convection losses among others.
The finite-element simulations indicate that infrared thermography has the

potential to be an effective NDT method for quantitatively determining the
quality of the bonding of FRP laminates to concrete structures. This analytical
research in combination with experimental trials in-progress may provide the
@Seismicisolation
@Seismicisolation
bases for standardization and implementation of quantitative infrared
thermographytesting in civil engineering structures.
REFERENCES
1. Alampalli, S.; Hag-Elsafi, O.; O’C~nnor,J.; Conway, T.; and k e f , A., 2001,
“Use of FRPs for Bridge Components and Methods of Performance
Evaluation,” Proceedings of the 2001 Structures Congress and Exposition,
ed. P.C. Chang, ASCE.
2. Christensen, J.B.; Gilstrap, J.; and Dolan, C.W., 1996, “Composite Material
Reinforcement of Existing Masonry Walls,” Journal of Architectural
Engineering, June, pp .63-70.
3. Karbhari,V.M. and Seible, F., 1998, “Design Considerations for the Use of
Fiber Reinforced Polymeric Composites in the Rehabilitation of Concrete
Structures,” NISTIR 6288. Proceedings from NIST Workshop on Standards
Development for the Use of Fiber Reinforced Polymers for the
Rehabilitation of Concrete and Masonry Structures, ed. D. Duthinh, January
7-8, Tucson, Arizona.
4. Hawkins, G.F.; Johnson, E; and Nokes, J., 1998, “Typical Manufacturing
Flaws in FRP Retrofit Applications,” NISTIR 6288, Proceedings from NIST
Workshop on Standards Development for the Use of Fiber Reinforced
Polymers for the Rehabilitation of Concrete and Masonry Structures, ed. D.
Duthinh, January 7-8, Tucson, Arizona.
5. ICBO Evaluation Service, Inc., 2001, “Acceptance Criteria for Inspection
and Verification of Concrete and Reinforcement and Unreinforced Masonry
Strengthening Using Fiber-Reinforced Polymer (FRP) Composite Systems,”
AC178, April.
6. Lienhard, John, 1981, A Heat Transfer Textbook, Prentice-Hall, Inc.,
Englewood Cliffs, New Jersey.
7. Schlessinger, Monroe, 1995, Infrared Technology Fundamentals, Marcel
Dekker, Inc., New York.
@Seismicisolation
@Seismicisolation
264 Starnes et al.
Table 1 -Material properties used in simulations
Concrete CFRP CFRP Air

(O degree layer) (90 degree layer)
p(kg/mq= 2400 1600 1600 I.2
c (J/kg.K)= 800 1200 1200 700
I k,(W/m.K)= I 1.5 1 7 I 0.8 I 0.024 ~
I
k,(Wlm.K)= 1.5 0.8 0.8 0.024
k, (W/m.K)= 1.5 0.8 7 0.024
Table 2 - Summary for set of FEM simulations involving the

investigation of flaw depth
No.of plies Thickness of FRP, cimp ts (a tc (9

5 2.5 rnrn 5.15 d1.15 8.72 d'.O*
3 1.5 mrn 5.03 d'.O6 8.58 d"'
2 1.0 mrn 4.82 d'.O3 8.42 ff 95
2.5 5.0 6.7 +1.7

2.5 3.0 6.0 +3.0
@Seismicisolation
@Seismicisolation
Heat flux, q Heat flux
4
111111111111111 2
I
Time, t
Time, t
Fig. 1 Schematic of the infrared thermography method to detect

presence of a flaw based on surface temperature differences
Heat flux Heat flux

i1111111111
-b
dTldx=O
I +
C.L. dTldy=O
Fig. 2 Example of test object used in numerical analysis; the physical

model is on the left and the analytical model is on the right
@Seismicisolation
@Seismicisolation
266 Starnes et al.
25
20
v
15
hf 10
O
O 20 40 60 80 100 120
E (kJ/m*)
Fig. 3 Maximum signal as a function of input energy density
1 O0
80
o 60
20
@Seismicisolation
@Seismicisolation
35.0
30.0
25.0
20.0
u)
v
15.0
10.0
5.0
0.0
0.0 1.0 2.0 3.0 4.0 5.0
Depth (mm)
Fig. 5 Example of estimation of flaw depth using a pre-established
relationship
12.5 mm 12.5 m m 12.5 mm
to 1.5 mm to 1.5 m m to 1.5 mm
E.5 mm 1.5 mm
1.0 m m
Z
C.L. C.L. %L.
Fig. 6 Summary of models used in simulations involving variations in flaw

width
60.0
'E
h
3
u
55.0
5 50.0
2
2 45.0
p>
o
E
2 40.0
v)
35.0 L I l I I I I
0.0 5.0 10.0 15.0 20.0 25.0 30.0
Distance from center/ine(mm)
@Seismicisolation
@Seismicisolation
Fig. 7 Estimation of width of flaw
268 Starnes et al.
n -e- Delaminations
E 3.0 o
I I
...........
(0.5 mm deep) -
E
W I I +- Debonds
k
O E
s -
a 3
'S g
25.0 ri
E 20.0
15.0
5 s 10.0
l:/;:;4- *T
......................
-
L
2*ooc 1..!.......i
I !
I / i
I
.......
....... ..... f .... i ..........

~
1 /
i/
I
1
...................... Li.. ....... ..;..........
/ I
E$
'-
'
se s8 5.0
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Flaw depth (mm)
Fig. 9 Minimum width of detectable flaw as a function of flaw depth
@Seismicisolation
@Seismicisolation
SP 209-1 5
Design of Pile Caps With

Strut-and=Tie Mode Is
by J. Dimig, E. Rogalla, and J. Duntemann
Svnopsis: The current AC1 Building Code (AC1 318-99) procedure for the shear
design of pile caps is the same approach used for two-way slabs. The procedure in-
volves determining a section thickness such that the concrete shear stress (@,) is
greater than the applied shear stress (vu) on the critical section. For footings sup-
ported on piles, the AC1 Code recognizes that these general provisions are not ap-
plicable as the depth of the footing increases and some of the pile loads fall within
the critical section. In deep pile caps, the critical section may even be located out-
side the footing, making it impossible to investigate shear at d or d/2. For these sit-
uations, the AC1 Commentary states that the designer should examine shear
strength at the face of the column and it refers to procedures outlined in the CRSI
Handbook (1996).
AC1 has recently published proposed revisions for the AC1 3 18-02 Code,
which promotes the use of strut-and-tie models as an alternative to the existing ACT
and CRSI procedures. The design methodology involves limiting the concrete
stresses in the compression struts and nodal zones to insure that the tension tie
(longitudinal reinforcement) yields before significant diagonal cracking develops
in the compression struts or crushing in the nodal zones. This paper explains the
existing ACT and CRSI procedures and the proposed ACT provisions for strut-and-
tie design.
Kevwords: foundation; pile cap; reinforced concrete;

strut-and-tie model
@Seismicisolation
@Seismicisolation
269
270 Dimig et al.
Jennifer Dimig is an AC1 member and a Project Engineer with Wiss,
Janney, Elstner Associates, Inc., Northbrook, Illinois.
Ernest A. Rogalla is an AC1 member and a Consultant with Wiss,

Janney, Elstner Associates, Inc., Northbrook, Illinois. Since Joining
WJE in 1986, Mr. Rogalla has been actively involved in the assessment
and repair of concrete and other structures.
John F. Duntemann is an AC1 member and a Senior Consultant with

Wiss, Janney, Elstner Associates, Inc., Northbrook, Illinois. He is a
licensed structural engineer with over 20 years of experience in the
assessment and rehabilitation of reinforced concrete structures.
INTRODUCTION
The current AC1 318 Building Code (AC1 318-99) does not contain any
specific provisions for the design of deep pile caps('). For footings
supported on piles, the AC1 318 Code recognizes that the general
provisions for shear design of footings become inapplicable as the depth
of the footing increases such that some of the pile loads fall within the
critical section around the column. For these situations, the Code refers
to design guidelines in the CRSI Handbook('). The Handbook contains
special requirements for deep pile caps, which involve investigating one-
way and two-way shear at the face of the column using increased
allowable shear stress values.
The most common practice in the U.S. is to design pile caps according to
the CRSI Handbook. However, there are varying opinions on whether
the CRSI method accurately models the behavior in deep pile caps. AC1
has recently published the proposed revisions for AC1 318-02 that
includes the use of strut-and-tie models in the design of concrete
members, such as deep pile caps, where the plane-section assumption of
beam theory does not apply(3). AC1 is proposing the strut-and-tie model
as an alternative to existing AC1 and CRSI procedures, and has added the
strut-and-tie provisions as an Appendix to the Code. These provisions
are similar to those prescribed by the Canadian Standards Association(4).
The design methodology involves limiting the concrete stresses in the
compression struts and nodal zones to insure that the tension tie
(longitudinal reinforcement) yields before significant diagonal cracking
develops in the compression struts, or crushing occurs in the nodal
zones.
@Seismicisolation
@Seismicisolation
CURRENT PRACTICE
The current AC1 3 18-99 procedure for the shear design of pile caps is the
same approach used for two-way slabs and conventional spread footings.
The shear design of footings requires consideration of both one-way and
two-way shear at critical sections located d and d 2 , respectively, from
the column face. The design procedure involves determining a minimum
section thickness so that the concrete shear stress (+vc)is greater than the
applied shear stress (vu)on the critical section. For footings supported on
piles, Section 15.5.3 provides general guidelines for calculating the
applied shear stress based on the location of the piles relative to the
critical section. The corresponding Commentary acknowledges that the
Code provisions for the shear design of footings are not applicable when
pile centers fall within the critical section. In deep pile caps, the critical
section may even be located beyond the footing perimeter, making it
meaningless to investigate shear at d or d 2 as prescribed for slabs and
footings. For these situations, R15.5.3 of the Commentary recommends
the following:
“When piles are located inside the critical sections d or d 2 from

face of column, for one-way or two-way shear, respectively, an
upper limit on the shear strength at a section adjacent to the face
of the column should be considered. The CRSI Handbook offers
guidance for this situation.”
Prior to the 1995 edition of the AC1 318 Code, the Commentary
suggested analyzing the footing for shear in accordance with the special
provisions of Section 11.8 for deep flexural members. However, these
provisions address only one-way shear in deep flexural members, and do
not readily apply to two-way pile caps.
The CRSI Handbook adopts the general design requirements in the AC1
Building Code for the design of pile caps, with additional requirements
for deep pile caps. When analyzing one-way shear, if the center of the
nearest pile is within d from the face of the column, CRSI recommends
investigating shear at the face of the column using a greater allowable
shear stress (v,) to account for deep-beam behavior. The upper limit on
the allowable shear stress for one-way shear in deep pile caps is io&.
When analyzing for two-way shear, if the center of the nearest pile is
within d 2 , CRSI recommends investigating a “pure shear” failure mode
at the perimeter of the column face using an increased vc, with an upper
limit of nJfC .
@Seismicisolation
@Seismicisolation
272 Dimig et al.
STRUT-AND-TIE MODEL
In recent years, there has been considerable discussion regarding the use
of strut-and-tie modeling over the current AC1 and CRSI methods to
more accurately model the behavior of deep pile caps. AC1 has
published an excellent bibliography on this subject(’). The CSA Code
adopted strut-and-tie modeling as a valid method for the design of pile
caps in 1984. The following discussion of strut-and-tie models is based
primarily on the proposed AC1 3 18-02 provisions.
General Concept of Strut-and-Tie Modeling
In strut-and-tie modeling the compression and tension fields in a

concrete member are modeled as axially loaded concrete struts and steel
ties. The struts and ties are connected at nodes to form a truss that
transfers the applied loads to the supports. The design methodology
involves limiting the concrete stresses in the compression struts and
nodal zones to ensure that the tension tie (longitudinal reinforcement)
yields before significant diagonal cracking develops in the compression
struts, or the nodal zones crush.
Limits on the Use of Strut-and-Tie Models
To discourage the application of strut-and-tie modeling in the design of

slender concrete members, the proposed Code provisions limit the use of
strut and tie modeling to members where the angle between the axes of
any strut and tie is at least 25 degrees (AC1 318-02, Section 15.5.3 and
A2.5). As this angle decreases, incompatibilities arise due to shortening
of the compression struts and lengthening of the tension ties occurring in
almost the same directions. For slender pile caps, the existing AC1
provisions for the shear design of footings still apply.
Development of the Strut-and-Tie Model for a Deep Pile Cap
In developing a strut-and-tie model, the compression struts and tension

ties are chosen to coincide with the compression and tension fields that
develop under the applied loads. Figure 1 illustrates a strut-and-tie
@Seismicisolation
@Seismicisolation
model for a typical four-pile cap. The axes of all struts, ties, and external
forces meeting at a node intersect the node. The forces in the struts and
ties may be determined by applying the equations of equilibrium at each
node.
When designing a deep pile cap using the strut-and-tie model, the struts,
ties, and nodal zones must all have adequate strength. For each strut, tie,
and nodal zone, proposed AC1 Equation A-1 requires that
@n 2 Fu
where Fuis the force in a strut or tie, or the force acting on one face of a
nodal zone due to the factored loads; F, is the nominal strength of the
strut, tie, or nodal zone; and 4 is the strength reduction factor.
The forces in the struts and ties depend largely on the depth of the pile
cap and the angle between the axes of the struts and ties. In turn, the
dimensions of the struts, ties, and nodal zones depend on the depth of the
pile cap and the angle between the axes of the struts and ties, and the
column and pile dimensions. Figure 2 illustrates the geometry of a
typical nodal zone at the pile bearing. Similar geometry is used for the
nodal zone at the column bearing.
The primary variable in designing a pile cap is the depth of the pile cap.
In most instances, the loads supported by the column determine column
and pile dimensions. The depth of the pile cap and the resulting
dimensions of the struts, ties, and nodal zones must be established to
meet the requirements of AC1 Equation A-1. The design is an iterative
process in which an initial pile cap depth is assumed. This may be
determined based on the general AC1 equations for one-way and two-
way shear in footings.
Compression Struts
The nominal strength of a compressive strut, F,,, is defined by AC1

Equation A-2 as
Fm = fcu Ac
where Ac is the cross-sectional area at one end of the strut, and&, is the
smaller of the effective compressive strength of the concrete in the strut
and the effective compressive strength of the concrete in the nodal zone.
The nominal strength must be calculated at each end of a strut. Table 1
lists the effective concrete compressive strengths assigned by AC1 3 18-
02 for various strut types. In the CSA Code, the compressive strength of
the strut depends upon the angle between the strut and tie and the tensile
strain in the tie, with an upper limit of 0.85f,.
@Seismicisolation
@Seismicisolation
274 Dímíg et al.
The compression struts in a deep pile cap strut-and-tie model may be
idealized as bottle-shaped, with the strut wider at mid-length than at the
ends, as illustrated in Fig. 3. The lateral spread of compression stresses
at mid-length of the strut produces transverse tension which may cause
longitudinal cracking. For this reason, the proposed AC1 provisions
assign a greater effective compressive strength for such struts when
transverse reinforcement satisfying Section A.3.3 is provided to resist the
transverse tension. However, Section 15.5.3 of AC1 318-02 states that
the effective concrete compression strength of the struts in pile caps shall
be determined based on a bottle-shaped stnit without reinforcement. The
Commentary explains that, in general, it is not practical to provide
transverse strut reinforcement in pile caps.
Nodal Zones
The nodal zones in the strut-and-tie model of a deep pile cap are
extended nodai zones, as shown in Fig. 2. The compressive strength of
each nodal zone, F,,, is given by AC1 Equation A-7 as
Ffin = f c d n
where fc. is the effective compressive strength of the concrete in the
nodal zone, and A,, is the area of the face of the nodal zone that F. acts
on or the area of a section through the nodal zone. Table 2 lists the
concrete compressive strengths assigned by AC1 318-02 for the concrete
in the nodal zones.
Tension Ties
The dimensions of the tension ties are established such that the
compressive strength at the nodal zones exceeds the compressive stress
on the nodal zone due to the factored force in the tie, in accordance with
AC1 Equation A-7. The reinforcement in the tension ties is determined
such that the design strength of the tie exceeds the factored force in the
tie. The nominal strength of the tension tie, Fni,is given in AC1 Equation
A-6 â~
Fnt = A s t f y
where A,, is the area of mild reinforcement, andfy is the yield strength as
defined in Section A.4.1. The reinforcement must be uniformly
distributed over the area of the tie, and the axis of the tie reinforcement
must coincide with the axis of the tie in the strut-and-tie model. In
addition, Section A.4.3 requires that the tension tie be anchored by
@Seismicisolation
@Seismicisolation
( Cancun Conference Proceedings 275
adequate development of the tie reinforcement at the point where it
leaves the nodal zone.
CONCLUSION
ACI’s adoption of the strut-and-tie model design procedure provides a

more rational design procedure than current AC1 and CRSI practice.
This procedure addresses the current unconservative application of
traditional flexural design procedures for beams and two-way slabs to
deep pile caps. The strut-and-tie procedure further abandons the practice
of identifiing a critical section at d o r d 2 , because the maximum bearing
stress at the base of the column or top of the piles is generally a better
indicator of strength than the stress on any prescribed critical section.
168879
REFERENCES
1. AC1 Committee 3 18, “Building Code Requirements for Structural

Concrete (AC1 318-99) and Commentary (AC1 318 R-99)”,
American Concrete Institute, Farmington Hills, Michigan.
2. CRSI Handbook, Eighth Edition, Concrete Reinforcing Steel

Institute, Schaumburg, Illinois, 1996.
3. AC1 Committee 318, “Building Code Requirements for Structural

Concrete (AC1 318-02) and Commentary (AC1 318 R-02)”,
American Concrete Institute, Farmington Hills, Michigan.
4. CSA Standard, “Design of Concrete Structures (CSA A 23.2-94)”,

Canadian Standards Association, Ontario, Canada.
5. AC1 Committee 445, “Strut-and Tie Bibliography”, AC1

Bibliography No. 16, American Concrete Institute, Farmington Hills,
Michigan, 1997.
@Seismicisolation
@Seismicisolation
276 Dimig et al.
Table 1 - Effective compressive strength of concrete
in strut (AC1 318-02, Section A.3.2)
Prismatic or uniformly tapered 0.85 (1 .O f c)

struts
Bottle-shaped struts
- w/ reinforcement 0.85 (0.75 f c)
- wio reinforcement 0.85 (0.60 f c)
Stnits in tension memberslflanges 0.85 (0.40 f c)
Table 2 - Effective compressive strength of concrete

in nodal zone (AC1 318-02, Section A.5.2)
Nodal zones bounded by struts and/or bearing O.SS(1.0 f c)

areas
Nodal zones anchoring one tie 0.85(0.8 f c)
Nodal zones anchoring two or more ties 0.85(0.6 fc)
Figure 1 Simple truss model for four-pile cap
@Seismicisolation
@Seismicisolation
z
a) One layer of steel.
II 'O I
1 iI
IC
b) Distributed steel.
Figure 2 Bottle-shaped strut
@Seismicisolation
@Seismicisolation
278 Dirnig et al.
a) Cracking of a bottle-shaped strut b) Strut-and tie model
Figure 3 Extended nodal zone
@Seismicisolation
@Seismicisolation
SP 209-16
Study of the Influence of Cement

Type, Cement Content, and Concrete
Cover Thickness on the Resistance
and Durability of Concretes
Subjected to Chloride Ion Aging
by A. Joukoskí, K. F. Portella, C. M. Garcia,
A. Sales, O. Baron, and J. F. de Paula
Svnopsis: Reinforced concrete structures constructed in coastal zones have con-

stantly been threatened by environmental damaging elements. The chloride ion is
known as one of the most aggressive of these elements, causing, among other dam-
ages, corrosion of the steel reinforcement and degradation of concrete. The goal of
this work was to determinate the influence of the cement type and cement content,
as well as the concrete cover thickness, in the resistance and durability of rein-
forced concrete elements exposed to aging in a 3.4% sodium chloride aqueous so-
lution. Many concrete mixtures were made using CPII-F32 (with filler), CPII-Z 32
(pozzolanic mixture) and CPV-AR1 RS (sulfur resistance) portland cements, with
contents of about 290 and 350 kg/m3 (490 and 590 lb/yd3), and with the concrete
cover thicknesses of 10, 15, 25 and 30 mm (0.394, 0.591, 0.984 and 1.18 in). The
evaluation of the concrete behavior was taken from the results of physical and me-
chanical tests of cylindrical concrete samples and electrochemical tests-mainly
the electrochemical impedance spectroscopy (EIS)-of small prismatic reinforced
concrete samples. The results are presented for each combination of cement type
and content, in terms of the aging time. Half cell potential measurements show that
concretes made with CPV-AR1 RS cement presented the best results, with longer
periods necessary to produce electrical change in the samples. The concrete made
with CPII-Z 32 cement and cement content of 288 kg/m3 (485 lb/yd3)was the mix-
ture with the worst durability, with some samples showing fracture after 110 days
of aging.
Keywords: chloride ion; corrosion of steel

re info rc e me nt; de g rad at¡ o n of concrete;
e Ie ctro c he rnic a I imp e da nc e spectroscopy
@Seismicisolation
@Seismicisolation
279
280 Joukoski et al.
A l a Joukoski is a Civil Engineer, Specialist in Metallic Materials
Engineering, and student of the Engineering Post-Gradution Interdisciplinary
Program (PIPE) with e m p k s in Materials Science and Engineering, at the
Federal University of Paraná State (UFPR), Curitiba, PR, Brazil. He works as
a research assistant at Institute of Technologyfor Development (L4CEC).
Kleber F. Portella works as a research scientist at the Institute of Technology

for Development, Curitiba, PR, Brazil. He received his Master of Science
degree from Aeronautics TechnologicalInstitute (ITA) in 1991,and his Doctor
of Science degree from Universityof São Paulo (USP) in 1997.
Carlos M. Garcia is a Doctor of Science, and works as a research scientist at
the Institute of Technology for Development, Curitiba, PR, Brazil. He is also
professor of the Engineering Post-Graduation InterdisciplinmyProgram at the
Federal University of Paraná State.
Almir Saía is a Doctor of Civil Engtneering andpro@mor of Civil Engneering

Depariment at Federal Universis,of ,530 Carlos (UFSCar), S& íklos, SP,Brazil.
ûrlando Baron is a Master of Science, and works at the Institute of Technology

for Development, Curitiba,PR, Brazil.
Juliano K de P a l a is a Electrical Engineer, and works at the Energy
Company of Sergipe State (ENERGIPE),Aracaju, SE, Brazil.
INTRODUCTION
The serious problem of service life reduction of reinforced concrete

structures located at coastal zones, caused by the corrosion of its steel
reinforcement and by the degradation of concrete, is already well-known by the
world technical and scientific community (i). Special care is required from
concrete components used for electric energy distribution and public
illumination systems, such as light poles and crossarms (2), due to the risk of
injuries or even death of pedestrians caused by failure of the deteriorated
structures.
Most cases of corrosion and degradation of reinforced concrete of coastal

structures are the consequence of electrochemical reactions started by the
presence of chloride ions (Ci-) (3). After its contact with the poles and
crossarms surfaces, the chloride ion can diffuse through the concrete cover and
reach the steel reinforcement before the end of the planned service life of the
structure (4).
@Seismicisolation
@Seismicisolation
Significant concentrations of chloride ion in the concrete can produce
depassivation of steel, a phenomenon which manifests itself in a polarization
curve as a strong increase in the electrical current, at the passivity potential
zone (4).
The corrosion of the steel reinforcement in concrete, as well as the methods

for its detection, are the subject of many studies all over the world, mainly
using non-destructive techniques. The electrochemical impedance spectroscopy
(EIS)technique is one of the most important and useful methods in evaluating
the behavior of many construction materials, as much for determining corrosion
speed as detecting the ingress of aggressive elements into the concrete to the
reinforcement steel (5).
EXPERIMENTAL PROCEDURE
Materíals Used
All materials used in the concrete mixtures were submitted to physical and
chemical analysis. The cements used were CPII-F 32 (addition of carbonate
filler), CPiI-Z 32 (addition of pozzolan) and CPV-ARI RS (high-early-strength
cement, with resistance to sulfate attack) types, and a few of its properties are
shown in Table 1. No chemical dteration was observed in the cement, and its
compressive strengths are in agreement with the requirements of the
correspondingBrazilian technical specification (6,7).
Both fine and coarse aggregates used are fiom the Curitiba region, in the
south of Brazil, and consist of crushed limestone with a maximum nominal size
of 19 mm (0.75 in), and natural sand (washed). Some results of the physical
analysis of these aggregates are given in Table 2. Potential reactivity test of the
aggregates was also made, and the results (Fig. 1) show that only the sample
with CPII-Z 32 cement had acceptable expansibility, according to the
specification ASTM C 1260-94 (8).
Concrete Mixtures
Six different concrete mixtures were made using the three cement types
described above, and using two cement contents for each type, as shown in
Table 3. The slump for all mixtures was fixed at 40 mm i 10 (1.57 in f 0.394).
No piasticizer, superplasticizeror air-entraining admixturewere used.
@Seismicisolation
@Seismicisolation
282 Joukoski et al.
SDecimens Castinq
For each mixhne, three types of specimens were cast:
Concrete Cvlindrical Specimens - 100 x 200 mm (3.94 x 7.87 inches) in

size, for determining compressive strength at 3, 14 and 28 days after casting,
and after 30 and 90 days of aging by immersion in a 3.4 % Sodium Chloride
(NaCl) aqueous solution.
1 ( - 150 x 300 mm (5.91 x 11.8 inches)

in size, for permeability tests and determination of specific density, absorption
and porosity of concrete (&er 28 days from casting).
Reinforced Concrete Prismatic Specimens - with two dimensions equal to

71 and 100 mm and a variable third dimension. Thus, it was possible to cast
samples with four different concrete cover thickness: 10 mm (0.394 in), 15 mm
(0.591 in), 25 mm (0.984 in) and 30 mm (1.18 in).
The reinforcement of the prismatic specimensconsisted of three bars of class

CA-50 steel (carbon-steel, with 500 MPa [72,519 psi] of tensile strength), 6.3
mm (0.248 inch) in diameter and 125 mm (4.92 in) long. Both extremities of
each bar were protected by a tar-based painting, resulting in only a centrai steel-
exposed region of the bar d - h about 35 mm (1.38 in) of length. Figs. 2,3 show
more details about the prismatic specimen.
After casting, all specimens were left covered in the casting room for 24
hours, then demolded and placed in a moist curing room, at 23 C f 2 (73.4 F i
4) and 100 % relative humidity, for 28 days.
Electrochemical Imoedance Snectroscow (EIS)
After 28 days of curing, the aging process of prismatic specimens by the

NaCl solution started, as well as the EIS testing. "Aging cells", consisting of
two PVC piping caps, were fastened to each specimen, as shown in Figs. 4,5.
The container at the side of the concrete cover was filled with the saline
solution, whereas the other one was filled with distilled water.
The steel bars of the specimen functioned as working and auxiliary

electrodes. A saturated calomel reference electrode was also used.
The corrosion potential measurements and the impedance tests were

performed, according to the orientations of ASTM C876-91 (9), after many
aging times. This standard test method also indicates the relative probability of
corrosion activity as function of the half-cell potential reading, for a Cu/CuSO4
reference electrode (Table 4).
@Seismicisolation
@Seismicisolation
The guidelines described in ASTM C876 provide basic principles for
evaluation of the reinforcing steel corrosion in concrete. These values are
referred to oxidation potentials, and to convert them to reduction potentials,
changing the signai is all is needed. The standard reduction potential versus that
of standard hydrogen electrode (SJB) of CdCuS04 and saturated calomel
(Hg/HgzClz) electrodes is +316 and +241 mV, respectively. Thus, the
difference of the two reference electrodes is 75 mV. So, for the calomel
electrode, the guidelines of the ASTM C876 become the values positioned at
the middle column of the Table 4.
Cornmessive Strenath
The cylindrical concrete specimenswith dimension of 100 x 200 m m (3.94 x

7.87 in) were tested for compressive strength at 3, 14 and 28 days of curing.
Additional specimens were immersed for aging in the saline solution for about
30 and 90 days, and then were also tested.
Other Tests or Anaivsis
Additional testing was performed, trying the evaluation of concrete behavior

to durability conditions. Besides absorption, specific gravity, porosity and
permeability dderminations, others techniques were applied, such as:
differential scanning calorimetry (DSC), thermogravimetric analysis (TGA), X-
ray difiatometry (XRD) and scanning electron microscopy (SEM).
Electrical Properties
The plots presented in Figs. 6 to 9 show the half-cell potentials (or corrosion
potentials) obtained fiom EIS measurements, for all six concrete mixtures, as
function of the aging time. Each plot is referred to a different concrete cover
thickness.
Comparing the curves of each plot, it’s easy to reaiise that specimens
representing concretes containing higher cement contents produced better
perfiormance, as expected. Concretes made with CPV-ARI R!3 cement
produced, on average, the lowest potentials values. The worst results were
obtained fiom the mixture made with CPII-2 32 cement type and with a cement
content of about 290 kg/m3 (490 lb/yd3), that is Mixture 3. The specimens with
cover thickness of 15 and 25 mm of Mixture 3 showed cracking, peeling and
scaling after approximately 120 days of aging.
@Seismicisolation
@Seismicisolation
284 Joukoski et al.
It isn't hard to notice the relationship between durability and surface quality
of specimensmade with certain cement types, mainly CPII-F32 and CPII-Z32
with lower cement contents (Mixtures 1 and 3). A more detailed observation of
the plots reveals that specimens with cover thickness of 10 mm showed,
comparatively, better electrical performance. This, in a first look, may sounds
contradictory, but can be explained. The side of these specimens related to the
cover thickness under analysis presented better surface quality, because it was
lefi at the bottom of the molds during the molding and vibrating processes.
Fig. 10 and Fig. 11 present, respectively, Nyquist and Bode plots obtained
from electrochemical measurements of specimens made with CPII-Z 32
cement, 288 kg/m3 (495 lb/yd3) cement content and concrete cover 10 mm
(0.394 in) thick, at four Werent aging times. It's interesting to observe that the
alteration of the curves are coincident with the potentials changing, which
happened between 27 and 34 days of aging, pointing to the start of a corrosion
process.
In Fig. 12 is shown the model (equivalent electric circuit) used in the fitting,
by non-linear least squares method, of the experimental results (10).
Compressive Strenatti Results
The results obtained fiom compressive st~engthtesting at 3, 14 and 28 days

of curing and &er 30 and 90 days on aging by a 3.4 % NaCl solution are shown
in Fig. 13. It can be easily seen that concretes made with CPV-ARIRS cement
type, as well as that ones made with higher cement contents, presenteú the
highest resistance values, as expected (I I).
Other Results
The results of specific gravity and porosity are in the normal range, and the
absorption values obtained are in agreement with the Brazilian standard NBR-
8451 (2).
CONCLUSIONS
The results obtained from the study of electrical properties and half-cell
potentials of concrete and reinforcement were presented. The variations
occurred in the electrochemical system under analysis as a function of the aging
by saline solution, that is steel bars corrosion. This was promptly detected by
the alterations noticed in the impedance curves and in its respective electrical
parameters. Thus, the electrochemical impedance spectroscopy technique is
very useful for evaluating the corrosion level and the mechanisms involved in
the degradationof concrete structuresand parts.
@Seismicisolation
@Seismicisolation
It was possible to technically demonstrate the benefit of a concrete specimen
containing good quality surface for durability, in relationship to its electrical
behavior by aging with a saline solution.
The advantages obtained in the durability and resistance of concrete by using

higher cement contents was obvious. The benefits of the use of CPV-ARI RS
(high-early-strength with sulfate resistance) cement type were also proved.
With this cement, it's easier to obtain concrete mixtures with higher mechanical
resistance and better durability, especially in cases of aggressive environmental
conditions.
ACKNOWLEDGMENTS
B i s work was sponsored by the Institute of Technologyfor Development

(LACTEC),the Energy Company Of Sergipe State (ENERGIPEJ,the Electric
Energy National Agency (ANEEL), and the National Council of Scient& and
Technological Development (CNPq). ïñe authors wish to thank the
Engineering Post-Graduation Interdmiplimmy Program (PIPE) at Federal
University of Paraná State (UFPR), and the Federal University of Sa0 Carlos
(UFSCar). í%anhare also due to engineer CintiaRibeiro and the stMof Civil
Structures Department (LACEC-DPEC), for their valuable help on
proportioning, testing and analysis of concrete, cement and aggregates.
REFERENCES
1. Neville, A. M. Propriedules do Concreto. Trad.: S. E. Giammusso. 2".

ed. PINí Ltda., São Paulo, 1997.
2. NBR 8451. "Postes de concreto armado para redes de distribuição de

energia elétrica - Especificação". Associação Brasileira de Normas Técnicas,
Rio de Janeiro, 1998.
3. Mehta, P. K.Concreto:Estrutura, Propriedas e Muteriais. Trad.: P. J.

M. Monteko. PINI Ltda, São Paulo, 1995.
4. Cascudo, O. O Controle Corrosão de Armaduras de Concreto. la.ed.

PINI Ltda, São Paulo; Ed. da UFG, Goiânia, 1997.
5 . Shi, M.; Chen, 2.;and Sun, J. "Mermination of chloride diffisivity in

concrete by AC impedance spectroscopy," Cement and ConcreteResearch, vol.
29, 1999, p. 1111.
6. NBR 11578. "Cimento Porîiand composto". Associaçb Brasileira de

Normas Técnicas, Rio de Janeiro, 1991.
@Seismicisolation
@Seismicisolation
286 Joukoski et al.
7. NBR 5737. "Cimentos Portland resistentes a sulfatos". Associagäo
Brasileirade Normas Técnicas,Rio de Janeiro, 1992.
8. ASTM C1260-94. "Standard Test Method for Potential Alkali Reactivity

of Aggregates (Mortar-Bar Method)". American Society of Testing and
Materials, West Conshohocken, Pa, 1994.
9. ASTM (376-91. "Standard Test Method for Half-Cell Potentials of

Uncoated Reinforcing Steel in Concrete". American Society of Testing and
Materials, West Conshohocken, Pa, 1999.
10. Feliu, V.; González, J. A.; Andrade, C.; and Feliu, S. "Equivalent circuit
for modelling the steel-concrete interface. I. Experimental evidence and
theoretical predictions". Corrosion Science, vol. 40, no 6, 1998, p. 975.
11. NBR 6118. "Projeto e execução de obras de concreto armado".

Associação Brasileira de Normas Técnicas, Rio de Janeiro, 1980.
@Seismicisolation
@Seismicisolation
Table I-Physical properties and chemical composition of cement
I CPII-F32 I CPII-Z32 ICPV-ARIRS
Aiuminum oxide (Alzo3) 4.1 6.9 6.7

Silicon dioxide (Sioz) 18.0 22.6 22.4
Fenic oxide (Feo3) 2.64 2.96 3.00
Calcium oxide (Cao) 59.6 52.6 54.7
Magnesium oxide (MgO) 5.42 4.54 3.66
Sulfur trioxide (SO,) 2.75 2.77 3.33
Sodium oxide (NaZO) 0.25 0.21 0.23
Potassium oxide &O) 0.86 0.77 0.84
Equivalent alkali ( N a 0 + 0.658 KzO) 0.82 0.72 0.78
Loss on ignition 5.63 5.75 3.81
specific gravity 3.10 2.94 I 3.00
3-daY 21.9 (3176) 19.7 (2857) 32.2 (4670)

7-daY 27.3 (3960) 25.2 (3655) 34.1 (4946)
28-day 3 1.3 (4540) 45.8 (6643)
Table 2-Physical properties of aggregates

Fine aggregate Coarse aggregate
specific gravis (dry) 2.61 2.68
Absorption, % 0.65 0.46
Maximum nominal size, mm (in) 2.4 (0.0945) 19.0 (0.748)
@Seismicisolation
@Seismicisolation
288 Joukoski et al.
Table 3-Mixture proportions and some prûperties of fresh concrete
hhihue 1 1 1 2 1 3 1 4 1 5 1 6
Cement type I CPII-F32 I CPIIZ32 I CPV-ARI Rs
Table “eneral guidelines for half-cell potential interpreWion
Haíf-ceii (oxidation) potential Half-celi (reduction) potential Corrosion

reading versus Cu/CuSO, readmg versus “ g z C l z adwîy
’ 90 % probability
Less negative thau -200 mV Less positive thim +275 mV
ofno cornion
I
Between -200 mV and -350 mV Between +275 mV and +425 mV a d ~ t v I Corrosion
@Seismicisolation
@Seismicisolation
1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 .
O 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Testing time (days)
Fig. I-Potential reactiviq of cement &$ne aggregate
TOP VIEW
Fig. 2-Prismdic specimen sketch.(Note: 1 mm = O. 0394 in).
@Seismicisolation
@Seismicisolation
290 Joukoski et al.
Fig. LPrimatic specimensphotograph.
R R R
th&ing rubber msspiece

bar o'ring
LATERAL VIEW FRONTAL VIEW
Fig. &Aging cell sketch.
@Seismicisolation
@Seismicisolation
Fig. S A g i n g cell photograph.
-VCPII-2 32 - 345 @/m'

+CPV-AR1RS - 292 kg/mf
-+-CPV-ARI RS - 344 kg/m'
O 40 80 120 160 200 240 280 320 360

Aging time (days)
Fig. bHa&cellpotential for specimen with cover thickness

of I O mm (O. 394 in). (Note: I kg/m3 = 1.69 1bijx.f).
@Seismicisolation
@Seismicisolation
292 Joukoski et al.
=
'-1I
1.o
i 0.81
I
œ
Fail
A
-
-v-
_... . __
-tCPV-ARI RS 292 kg/m'

-+-
CPV-ARI RS 344
- .. .-/Il?
-A- CPII-2 32 - 288 kg/Il?
-
CPII-Z 32 345 kg/tn'
-
- i
O 40 80 120 160 200 240 280 320 360
Aging time (days)
Fig. 7-Half-ell potentialfor specimen with cover thickness

of 15 mm (0.591in). mote: 1 kg/m3 = 1.69lb@).
-O- -
CPII-F 32 347 kg/m3 -
-
-A- CPII-2 32 288 kg/m3 .
-v-CPII-Z 32 - 345 kg/m3 -
-f-CPV-AR1 RS 292 k@m3. -
-+- CPV-AR1 RS - 344 kg/m3 -
O 40 80 120 160 200 240 280 320 360

Aging time (days)
Fig. 8-Half£ellpotential for specimen with cover thickness

of 25 mm (0.984 in). (Note: 1 kg/m3 = 1.69 lb/)&).
@Seismicisolation
@Seismicisolation
1.4
Cover thickness: 30 mm -m-c
1.2 -0- -
CPII-F 32 347 kglI??
-A- CPII-232 - 288 WI??
I.o -v- CPII-2 32 - 345 kglM
z3 0.8
+CPV-ARI RS - 292 k&f'
-+- -
CPV-ARI RS 344 kgln"
# 0.6
n
0.4
0.2
0.0
O 40 80 120 160 200 240 280 320 360
Aging time (days)
Fig. 9-Harf-cell potentialfor specimen with cover thickness

of 30 mm (1.18 in). (Note: 1 kglm3 = 1.69 lb@&).
-15000
-1m
-5000
O
O 5000 10000 15000
z
specimen made with
CPII-Z 32 cement.
@Seismicisolation
@Seismicisolation
294 Joukoski et al.
-100
I
-75
3
Q, -3
5
-25
O
-
102 10' Id IO' Id Id Id 16
Frequency (Ha
Fig. Il-Bode plotsfor specimen made with

288 kg/m3(485 lbbd) of CPII-Z 32 cement.
Fig. 12-Equivalent electric circuit used in thefining of results,
@Seismicisolation
@Seismicisolation
-+-CPV-ARI -
RS 349 ~g/m’
*--
CPII-2 32 - 346 Kg/rn’
+CPV-ARI RS - 294 w m 3
-.-
4- CPII-F 32
CPICF 32
- 350 Kg/m’
- 290 Kg/m’
O 10 20 30 40 50 60 70 80 90 I00 110 120 130

Time (days)
Fig. 13- Compressive strength of cylidical specimens.

(Note: I MPa = I45psi and I kg/m3 = 1.69 Ibíydy.
@Seismicisolation
@Seismicisolation
SP 209-1 7
Self-Consolidating Concrete Piles

Confined in FRP Tubes
by H. El-Chabib, M. Nehdi, and M. H. EI Naggar
Synoyds: Cast-in-place deep foundations such as drilled shafts and piers are often
subjected to two sources of problems. First, the integrity and uniformity of the
cross-sectional area of these structural elements cannot be assured using normal
concrete because of limited accessibility and visibility during construction. Cavi-
ties and soil encroachments leading to soil pockets can jeopardize their load-bear-
ing capacity. Second, corrosion problems of steel reinforcement in deep founda-
tions have been costly, requiring annual repair costs of more than $2 billion in the
US alone. To address these two challenges, a novel technology for the construction
of drilled shaft concrete piles is proposed in this study. Self-consolidating con-
crete, a material that compacts under its self-weight without vibration and without
bleeding or segregation, is used to assure the structural integrity and uniformity of
the cross-sectional area of deep foundations. The self-consolidating concrete is
cast into FRP envelopes, which provide corrosion-resistant reinforcement. This pa-
per presents results of a laboratory investigation on the mechanical performance of
these novel piles including the effect of using expansive cement and shrinkage-re-
ducing admixtures to enhance the FFV tube-concrete interfacial bond.
Keywords: bending; compression; concrete;

fib er- reinforced; piIes; p Iasti c; se If- c o nso Iidat¡ ng;
stress-strain
@Seismicisolation
@Seismicisolation
297
298 El-Chabib et al.
Hassan El-Chabib is a graduate research assistant in the Department of Civil
& Environmental Engineering, University of Western Ontario (UWO),
London, Ontario, Canada. His research interests include concrete
technology and artificial neural networks applications in civil engineering.
AC1 member Moncef Nehdi is an assistant professor in the Department of

Civil & Environmental Engineering at UWO. He is a member of AC1
Committees 225, Hydraulic Cements; 236, Material Science and E803,
Faculty Network. His research interests include modeling the behavior of
cement-based materials, recycling by-products and durability and repair.
M. Hesham El Naggar is an associate professor in the Department of Civil

& Environmental Engineering at UWO. His research interests include
analysis and design of deep foundations, machine foundations, earthquake-
resistant structures, and use of special concretes in geotechnical
applications.
INTRODUCTION
The ultimate load-bearing capacity of cast-in-place concrete piles is

significantly affected by the compressive strength of concrete, the pile
cross-section, the amount of steel reinforcement and various soil properties.
Both the compressive strength of concrete and the integrity of the cross-
section depend on the consolidation of fresh concrete during placement.
Generally, concrete is consolidated with mechanical vibrators, a process

that is operator-sensitive. In deep concrete piles where accessibility and
visibility are limited, consolidation is particularly difficult. Over-
consolidated concrete might segregate or destabilize soil particles around
the shaft wall causing it to collapse inside the concrete core. Under-
consolidated concrete can have voids and cavities, leading to air and soil
pockets and exposure of steel reinforcement to corrosion attack.
Consequently, the integrity and uniformity of the pile’s cross-section often
cannot be assured, concrete strength is adversely affected and the load-
bearing capacity of the pile is jeopardized.
To alleviate both the problem of air cavities and soil pockets in the concrete
and corrosion of steel reinforcement, a novel technology for the
construction of drilled-shaft concrete piles is proposed in this study. Self-
@Seismicisolation
@Seismicisolation
consolidating concrete (SCC), a material that consolidates under its own
weight with little or no vibration and without exhibiting segregation or
bleeding, is used instead of normal concrete to assure the structural integrity
and uniformity of the cross-sectional area of deep foundations. Self-
consolidating concrete is cast into a fiber-reinforced plastic (FRF') envelope,
which acts as stay-in-place structural formwork and provides a protective
jacket for concrete and steel reinforcement.
In addition, the FRP envelope acts as a protective jacket in aggressive

corrosive environments such as the case of marine piles (I) and piles
constructed in soils rich in chloride ions and sulfate salts (2), and may
provide total or partial replacement of longitudinal and shear steel
reinforcement. More importantly, the FRP tube provides a lateral
confinement to the concrete core in piles and significantly enhances the
compressive strength of concrete and its ductility (3), (4).
The confinement of concrete is the restraint of lateral deformation of

concrete under axial loading. Extensive research has been conducted on
concrete-filled steel tubes for use as piles and columns (5), (6). However,
FRP tubes provide several advantages over steel tubes when used as
confining protective jackets for concrete piles. In addition to their high-
strength to weight ratio, FRP materials are known for their high corrosion-
resistance. Unlike the confining effect provided by steel tubes, which
reaches a threshold value once the steel yields, FRF' tubes provide a
continuously increasing confining pressure until failure, which adds to both
the ultimate compressive strength and ductility of the concrete member (7).
The focus of this study is to investigate the confinement effect of glass

fiber-reinforced plastic (GFRP) tubes on the stress-strain behavior of short
self-consolidating concrete cylindrical piles under uniaxial compression and
transverse loading. Specimens made both of normal concrete (NC) and SCC
were tested. Moreover, because the interfacial contact between the concrete
core and the confining tube is extremely important to achieve a composite
behavior, especially under transverse loading, SCC and NC specimens
prepared using Type 1O Canadian ordinary portland cement (ASTM Type I)
and expansive cement (Type K) along with a shrinkage-reducing admixture
were investigated. Both ends of specimens tested under transverse loading
were monitored for any slippage between the concrete and the confining
tube.
@Seismicisolation
@Seismicisolation
A total of 24 cylindrical specimens were prepared and tested in this study.

Twelve 150 x 300 mm short cylindrical columns were tested under uniaxial
compression to investigate the confinement effect of GFRP on axially
loaded normal concrete and self-consolidating concrete piles made with
either ordinary portland cement (OPC) or expansive cement (EC). Twelve
150 x 1100 mm cylindrical beams were tested under transverse load to
investigate the confinement effect of GFRP tubes on normal concrete and
SCC piles under transverse loading and the effect of using expansive
cement and a shrinkage-reducingadmixture on the interfacial bond between
the concrete and the confining tube.
Short-column specimens included 4 plain concrete control cylinders (two

NC and two SCC) and eight concrete cylinders confined by GFRP tubes.
The confined specimens included 2 cylinders made of normal concrete
using OPC, 2 cylinders made of SCC using OPC and two cylinders of each
(NC and SCC) made using expansive cement. Similarly, the beam
specimens included 4 plain concrete control beams (two NC and two SCC)
and 8 concrete beams confined by GFRP tubes (2-OPC-NC, 2-OPC-SCC,
2-expansive-cement-NCand 2-expansive-cement-SCC).
Materials
Concrete specimens were prepared from four different concrete mixtures

proportioned to achieve similar 28-day compressive strengths. Proportions
of concrete mixtures and their corresponding compressive strengths are
shown in Table 1. Canadian Type 10 OPC (ASTM Type I) was used to
prepare the normal concrete (NC) mixture. The SCC mixture was prepared
using the same type of cement incorporating proportions of fly ash (FA) and
granulated blast furnace slag (GBFS) along with a superplasticizer and a
viscosity-modifying admixture (VMA). Two other concrete mixtures (NCE
and SCCE) having the same proportions as the NC and SCC mixtures were
prepared using expansive cement (EC) instead of Type 10 OPC in addition
to a shrinkage-reducing admixture. Anhydrous tetracalcium trialuminate
sulfate based Type K cement was used for the expansive cement along with
a commercially available shrinkage-reducingadmixture,
@Seismicisolation
@Seismicisolation
FRP tubes consisted of a filament-wound angle ply with bi-directional
glass-fiber at f 55' winding angle with respect to the longitudinal axis of
the tube. They had an inside diameter of 150 mm and a uniform thickness of
6 111111. The physical and mechanical properties of the GFRP tubes are listed
in Table 2.
SDecimen Preparation and Testing Procedures
Uniaxial Comtiression Test -- plain concrete and concrete-filled FRP tube

composite cylinders were prepared according to ASTM C 192-90a
(Standard Practice for Making and Curing Concrete Test Specimens in the
Laboratory) but no rodding or vibration was performed on SCC specimens.
All concrete mixtures were prepared using the same mixing sequence in a
laboratory rotary mixer. All specimens were moist-cured in a curing room
with 100% relative humidity for 27 days, then they were all capped with a
sulfur solution at both ends and stored at room temperature (about 22'C)
until testing at 91 days from casting. Table 3 shows the properties of all
cylinders used for the uniaxial compressiontest.
The average axial and circumferential strains at mid-height on the exterior

surface of each specimen were measured by means of strain gauges. The
axial deflection (axial shortening of the column) was measured using an
LVDT. Each specimen was instrumented with four 10-mm strain gauges
located at its mid-height and attached to its exterior surface. Two gauges
were attached in the axial direction at 180' apart to measure the average
axial strain and the other 2 gauges were attached in the circumferential
(hoop) direction at 180' apart and 90' from the axial strain gauges to
measure the average radial strain as shown in Fig. 1. The compression test
was performed using a 4600 kN MTS machine at a loading rate of 4.5
kN/sec and the axial load was applied to the entire cross-section of the
composite cylinder. Fig. 2 shows the typical test set-up of a specimen under
uniaxial compression.
Flexural Test -- all beam specimens were prepared in a similar manner to

the short-column specimens. However, they were moist-cured for 27 days
in a curing room at 100% relative humidity and tested at 28-days after-
casting. Table 4 summarizes the properties of all beam specimens used for
the transverse loading test.
@Seismicisolation
@Seismicisolation
All beam cylinders were instrumented with six 10-mm strain gauges
attached to the exterior surface of either the FRP tube in the confined
specimens or the concrete surface in the unconfined ones. Four strain
gauges were installed at the mid-span of the beam, two at the top section of
the specimen to measure the average compression strain, and two at the
bottom section to measure the average tensile strain. One strain gauge was
attached at a distance d2 from each support at the specimen’s mid-height
and at 45’ from the tube’s longitudinal axis to measure the average
maximum shear strain at that point, (d = the outside diameter of the
specimen). The maximum beam deflection was measured using an LVDT
located right under the bottom of the central mid-span point of the beam
(Fig. 3).
All beam specimens were subjected to four-point flexural loading. The

distance between the two supports (free span) was 1000 mm and the loading
points were 250 mm apart in the specimen’s mid-section. The applied load
was transferred to the specimen through a steel I-section beam resting on
two 25-mm diameter steel rollers located at the loading points. Two other
25-mm diameter steel rollers were placed at the supports. Four 10-mm
thick, 50-mm wide and 162-mm inside diameter half-steel rings having an
outside rectangular shape were manufactured and rubber coated to distribute
the load to the surface of the cylindrical tube at the loading points and
supports. Fig. 3 shows a schematic representation of the bending test Set-up.
The transverse load was applied using a 1300 kN hydraulic jack in a
displacement control mode at a rate of 6 mrdmin for composite beams and
0.5 mdmin for the unconfined concrete beams. Fig. 4 shows the laboratory
test Set-up of a beam specimen subjected to transverse load.
Uniaxial Comwession Test
The axial load-deformation curves of composite short-column specimens

under uniaxial compression can be characterized by three different regions.
Two linear stages (first and third regions) connected by a nonlinear
transition stage (second region). The first region is dominated by the
behavior of the concrete core and the third one is dominated by the behavior
of the FRP tube as shown in Figs. 5 and 6. All confined specimens behaved
similarly during the first and third regions of the loading process. At about
50 % of the ultimate load, cracking noises were heard indicating the
@Seismicisolation
@Seismicisolation
progress of failure of the concrete core, followed by a quite steady period
during the transition region. White lines started to develop along the fibers
at 55" with respect to the longitudinal axis of the tube at about 60 % of the
ultimate load, indicating the development of stress in the fibers as shown in
Fig. 7. The white colored area along the fibers increased with increasing
load to cover most of the middle third of the specimen followed by fracture
of fibers and a sudden failure. For all composite specimens, failure was
initiated within the middle third of the specimen height and progressed
towards its top and bottom ends. Fig. 8 shows a typical failure mode of a
concrete-filled FRP tube under uniaxial compression.
The test results for all short-column specimens are summarized in Table 5
and the axial load- deflection curves for NC and SCC specimens along with
a hollow FRP tube are presented in Figs. 5 and 6, respectively. Although the
behavior of NC and SCC composite cylinders were very similar, it is clear
from both Figs. 5 and 6 that the first region of the composite cylinder axial
load-deformation response coincided with that of unconfined concrete up to
30 % and 50 % of the unconfined concrete strength for NC and SCC,
respectively. However, in both cases (NC and SCC) the confinement effects
on concrete strength and axial deformation were similar and equal to 33%
(increase) and 13% (decrease), respectively. The confinement effect was
calculated as the difference between the concrete failure in confined and
unconfined cylinders. The failure of the concrete core in a confined
specimen was defined as the point at which the axial load-deformation
curve of the composite FRP-concrete specimen shifted from a linear (stage
1) to a non-linear (stage 2) behavior (Figs. 5 and 6). The ultimate axial-load
and axial-deformation of composite cylinders in both cases (NC and SCC)
were found to be about 2.5 and 12 times higher than those of the unconfined
cylinders, respectively.
The only significant difference between the confined NC and SCC behavior
was in the transition region. Transition between the first and third regions in
the response of NC was progressive, while for SCC the transition was more
sudden as shown in Fig. 9. This indicates that for SCC there was a short
time lag between the instant when the concrete core failed and when the
effect of the FRP was mobilized. This suggests that higher shrinkage may
have occurred in the case of SCC and made the FRP-concrete interface less
tight. This is confirmed by the fact that SCC made with expansive cement
and a shrinkage-reducing admixture did not show this type of transitional
phase in its axial load-deformation curve (Fig. 6).
@Seismicisolation
@Seismicisolation
The stress-strain response curves for plain NC and SCC specimens and
concrete-filled FRP composite specimens are plotted in Fig. 10, in which
the first region of the axial stress-strain response of the composite cylinders
coincides with the unconfined concrete response in both cases (NC and
SCC-filled FRP tubes). For the same axial deformation value, the higher
axial stress obtained in the case of SCC-FRP composite is attributed to the
fact that the 91-day compressive strength of plain SCC is slightly higher
than that of plain NC. For the axial stress-circumferential strain curves, Fig.
10 shows that the stiffness of the unconfined cylinders is slightly higher
than that of the confined ones in both NC and SCC members. This behavior
is likely due to the fact that the axial load was applied to the entire cross-
section of the composite cylinder, and because concrete is stiffer than FRP,
a higher value of circumferential strain is introduced at the outside surface
of the FRP tube under the same axial load. The effect of using expansive
cement instead of OPC in NC and SCC is demonstrated in Figs. 11 and 12
in which no significant change was observed in the confinement effect on
concrete strength and ductility under uniaxial compression. Again the
higher stresses in the response curves of NCE (normal concrete with
expansive cement) are attributed to the fact that the unconfined compressive
strength of NCE was slightly larger than that of the unconfined compressive
strength of NC.
Flexural Test
The load-deflection curves of concrete-filled FRP composite tubes

subjected to transverse load can also be characterized by three different
regions, as it is the case for concrete-filled FRP tubes under uniaxial
compression. However, the strength and stiffness of the FRP material
dominated the behavior of such beam specimens during the loading process
in all three stages. Since no steel was used to reinforce the concrete core in
the unconfined beam specimens, their ability to carry transverse load was
almost negligible. However, in the case of concrete-filled FRP composite
beams, the concrete core contributed substantially to the overall strength
and ductility of the composite specimens. Figs. 12 and 13 illustrate the load-
deflection responses of confined and unconfined NC and SCC specimens
along with the response of a hollow FRP tube, all subjected to transversal
load. It is shown that the capacity of the FRP tube in sustaining transverse
load increased by 175 % from 45 kN (ultimate load of hollow tube) to 124
kN (ultimate load of concrete-FRP composite member) and the maximum
deflection of the tube at mid-span was also increased by 133 % from 2 1 mm
in the case of a hollow tube to 49 mm in the case of an FRP-concrete
composite beam. SCC specimens behaved relatively in a similar manner to
@Seismicisolation
@Seismicisolation
NC specimens and no major differences were observed during the loading
process. Fig. 13 provides a comparison between the behavior of NC and
SCC specimens under transverse load.
Slippage between the concrete core and the FRP tube may compromise the
ultimate load capacity of the composite member, especially under transverse
load. Therefore, an attempt was made to strengthen the interfacial bond
between the two materials using expansive cement instead of OPC with the
addition of a shrinkage-reducing admixture. This type of concrete will
expand rather than shnnk and therefore can create an active hoop pressure
against the internal wall of the FRP tubes, therefore developing a better
bond between the concrete core and the confining tube. Since such
composite tubes are usually intended for use underground (saturated soil),
subsequent shrinkage is considered less significant. The shrinkage-reducing
admixture will reduce shrinkage strains, further enhancing this bond. The
behavior of specimens prepared with such type of concrete is described in
Figs. 13. It is shown that in the first and second stages of the response curve
and at the same deformation, the load capacity of a GFRP tube filled with
concrete using expansive cement is noticeably higher than that of a tube
filled with concrete made of OPC, indicating that the specimen made with
expansive cement was stiffer. However, the ultimate load and deflection of
such specimens at failure was slightly lower than those of specimens made
using OPC. This is believed to be due to the fact that the fibers in the FRP-
concrete composite tube made using expansive cement were already
prestressed due to the expansion of concrete before the loading process
started.
Another attempt was made to prevent slippage by anchoring the FRP tube
to concrete at 200 mm from each of its ends by inserting a 12-mm diameter
steel bar in each of two holes drilled through the FRP before casting the
concrete. It was later observed during testing that a local failure occurred in
the concrete around the steel bars due to stress concentrations, and no major
contribution to prevent slippage was noticed (Figs. 12). In fact, for all tested
specimens slippage between the concrete and the FRP at both ends of each
beam varied only between 1 mm and 3 mm. During the loading process, it
was observed that each time slippage occurred between the concrete and the
FRP, the load dropped, which explains the existence of harmonic events in
the load-deflection curves of Figs. 12 and 13. All concrete-filled GFRP
tubes shared the same failure mode. White lines along the fibers started to
form at mid-span of the bottom section of the FRP tube and progressed
towards the ends. Tensile cracks started to appear on the bottom section of
the tube and under both loading points and progressed towards the upper
@Seismicisolation
@Seismicisolation
section until a major crack was developed to cause a sudden failure as
shown in Fig. 14.
CONCLUSIONS
This study investigated the behavior of concrete-filled GFRP tubes under

both uniaxial compression and transverse loading for possible use in deep
foundations (drilled-shaft piles). Special focus was on: i) the axial load-
deformation response and axial stress-strain response of GFRP tubes filled
with either NC or SCC under uniaxial compression; ii) the load-deformation
behavior of such composite tubes under transverse load; and iii) the effect
of using expansive cement and anti-shrinkage admixtures in concrete on the
interfacial bond between the FRP tube and the concrete core. The following
conclusions can be drawn from this investigation.
1. SCC-filled GFRP tubes had a similar behavior to that of NC filled-

GFRP tubes under both uniaxial compression and transverse load.
2. The only significant difference between the behavior of NC and
SCC-filled GFRP specimens was in the transition .region of the
response curves, in which the shift from a linear to a non-linear
behavior in the load-deformation and stress-strain curves subsequent
to the failure of the concrete core was more sudden for SCC-filled
FRP specimens. This is believed to be due to autogenous and self-
dessication shrinkage, which is higher in SCC due to a large volume
of water and cement paste and lower volume of coarse aggregates.
However, the use of expansive cement and shrinkage-reducing
admixtures made the behavior of SCC-filled FRP tubes similar to
that of NC-filled FRP tubes.
3. The use of expansive cement in concrete delayed the occurrence of
slippage between the FRP tube and the concrete core, creating a
somewhat better bond between the two materials, but did not fully
prevent it. Likewise, the use of localized lateral steel bars placed
through the FRP tube and concrete core did not prevent slippage.
Shear connectors or ribs placed inside the FRP tubes may provide
better performance (8).
4. FRP tube confinement of concrete cylinders increased their ultimate
load by 2.5 times and their axial deformation by 12 times under
uniaxial compression. It also enhanced their ultimate load by 20
times and their mid-span deflection by 100 times under transverse
load.
@Seismicisolation
@Seismicisolation
REFERENCES
1. Sen, R., Shahawy, M., Rosas, J. and Sukumar, S., “Durabilio of AFRP
and CFRP Pretensioned Piles in Marine Environment,” Non-Metallic
(FRP) Reinforcement for Concrete Structures, Proceedings of the Third
International Symposium, Vol. 2, 1997, pp. 123-130.
2. El-Nawawy, O. and Kandeil, A., “A New Reinforcing Material for

Concrete in the Arabian Gulf;” Proceedings of the 2”dInternational
Conference on Deterioration and Repair of Reinforced Concrete in the
Arabian Gulf, CIRIA, Bahrain, 1987, pp. 185.
3. Mirmiran, A. and Shahawy, M., “Behavior of Concrete Columns

Confined by Fiber Composites,” Journal of Structural Engineering, Vol.
123, NO. 5, 1997, pp. 583-589.
4. Samaan, M. S., “An Analytical and Experimental Investigation of

Concrete-Filled Fiber Reinforced Plastics (FRP) Tubes,’’ Ph.D. Thesis,
University of Central Florida, Orlando, FL., 1997,220 p.
5. b o w l s , R. B., and Park, R., “Strength of Concrete Filled Steel Tubular

Columns,” ASCE, Journal of the Structural Division, Vol. 95, ST12,
1969, pp. 2565-2587.
6. Mander, J. B., Park, R.,and Priestly, M. J. N., “Theoretical Stress-Strain

Model for Confined Concrete,” Journal of Structural Engineering,
ASCE, 114 (8), 1988, pp. 1804-1826.
7. Samaan, M., Mirmiran, A. and Shahawy, M., “Model of Concrete

Confined by Fiber Composites,” Journal of Structural Engineering, Vol.
124, NO. 9, 1998, pp. 1025-1031.
8. Mirmiran, A., Shahawy, M., Samaan, M., El Echary, H., Mastrapa, J.

C., Pico, O., ‘‘Effect of Colomun Parameters on FRP-Confined
Concrete, Journal of Composites for Construction, Vol. 2, No. 4, 1998,
”
pp. 175-185.
@Seismicisolation
@Seismicisolation
El-Chabib et al.
NOTATIONS
=Winding angle of glass fibers in the FRF' tubes.

= Poisson ratio.
= Maximum axial deflection of unconfined cylinders at concrete
failure.
=Maximum axial deflection of confined cylinders at concrete
failure.
=Ultimate axial deflection of confined cylinders at failure.
=Maximum mid-span deflection of specimen subjected to pure
bending.
= Maximum axial strain of unconfined cylinders.
=Ultimate axial strain of confined cylinders.
ASHA = Anti-shrinkage admixtures.
C =Cement.
D = Inside diameter of FRP tubes.
E = Modulus of elasticity of FRP tubes.
EC = Expansive cement.
FA =Fly ash.
G =Gravel.
GBFS = Granulated blast furnace slag.
GFRP = Glass fiber-reinforced plastic.
L = length of specimen.
NC = Normal concrete.
NCE =Normal concrete with expansive cement and anti-shrinkage
admixtures.
OPC = Ordinary portland cement.
Pco = Maximum axial load of unconfined cylinder at concrete failure.
PCc =Maximum axial load of confined cylinder at concrete failure.
Puri = Ultimate axial load of confined cylinder at failure.
P,, = Maximum transversal load of cylinder subjected to pure bending.
S =Sand.
SCC = Self-compacting concrete.
SCCE = Self-compacting concrete with expansive cement and anti-
shrinkage admixtures
SP = Superplasticizer.
V M = Viscosity-modifying admixtures.
W =Water.
WíC = water to cement ratio.
fc = Compressive strength of unconfined concrete specimen .
f c u =Ultimate compressive strength of confined concrete specimen .
t =Thickness of the FRP tube.
@Seismicisolation
@Seismicisolation
Cancun Conference Proceedings
3 0 0 0
3 z g z
O O
1 2 1 2
n o m o
n o m o
c l m m m
> o m 0
3 2 2 2
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
Z?..
d W
w w w w
W
@Seismicisolation
@Seismicisolation
~ 162mm ~
150 m m
--....-Concrete
2 axial strain gauges
2 circumferential strain 4 (i 80' apart)
gauges (i 80' apart) i '
gi
O(
..-..FRP tube
O
mI
I
hydraulic jack
Fig. 1- Illustration of specimen used in uniaxial compression
Fig. 2- Test Set-up for uniaxial compression.
@Seismicisolation
@Seismicisolation
BEAM SPECIMEN
I
T
Free span 1000 mm

,I. *I I
1- Total length = 11O0 mm 'I
Fig. 3- Illustration of test specimen subjected to transverse load.
Fig. 4- Test set-up for flexural test.
@Seismicisolation
@Seismicisolation
2000
1600
1600
1400
E
U
1200
--g
~
1000
Confinement effect on SCC
3 Ia
600
600
400 . . . . . ........ ..............
200
O
O 2 4 6 8 10 12 14 16 16 20
Axial deflection (mm)
Fig. 5- Axial load-deformation respoi,,r of NC cylinders and FRP tube.
2000
1600
1600
1400
0 1200
o
-o 1000
Ia
3
I
800
600
E--
400
FRP hollow iube
2oc
o
O 2 4 6 8 10 12 14 16 16 20
Fig. 6-Axial load-deformationresponse of SCC cylinders and FRP tube.
@Seismicisolation
@Seismicisolation
Fig. 7- Highly stressed area (white lines) along fibers.
Fig. 8- Typical failure of a concrete-filled FRP tube under uniaxial

compression.
@Seismicisolation
@Seismicisolation
2000
O 2 4 6 8 10 12 14 16 18 20
Fig. 9- Axial load-deformation response of NC and SCC cylinders.
-0.025 -0.02 -0.015 4.01 -0.005 O 0.005 0.01 0.015 0.02 0.025
(Axial) Strain (Circumferential)
Fig. 10- Stress-strain response of NC and SCC cylinders under axial load,
(strain gauges failed at their maximum limit of 0.02).
@Seismicisolation
@Seismicisolation
r’
, -p
10
e
-0.025 -0.02 -0.015 -0.01 -0.005 O 0.005 0.01 0.015 0.02 0.025
(Axial) Strain (circumferential)
Fig. 11- Stress-strain response of NC and NCE cylinders under axial load,
(strain gauges failed at their maximum limit of 0.02).
O 10 20 30 40 50 60
Center deflection (mm)
Fig. 12- Effect of expansive cement and slippage reinforcement on load-

deflection response of NC specimens under transverse load.
@Seismicisolation
@Seismicisolation
. .-
i20
1O0
g
-2 80
V k
60
40
20
Unconfined NC or SCC
O
O io 20 30 40 50 60
Center deflection (mm)
Fig. 13- Load-deflection response of NC and SCC specimens under

transverse load.
Fig. 14- Typical failure mode of concrete-filled GFRP tube under transverse
load.
@Seismicisolation
@Seismicisolation
SP 209-1 8
Analysis of Shrinkage Reducing

Admixtures on Drying Shrinkage of
Concrete
by C. Videla and C. Aguilar
Synomis: The objective of this research was to verify the feasibility of reducing
the drying shrinkage of concrete made with portland pozzolan cement. Major vari-
ables included slump, shrinkage-reducing admixture brand, admixture dosages
and specimen dimensions. Tests were carried out on 132 prism specimens and
shrinkage strains were measured up to 360 days of drying. Also compressive
strength, splitting strength and static modulus of elasticity were determined.
Based on the results obtained, it was concluded that all of the five brand
admixtures evaluated reduce drying shrinkage. One of the experimental admix-
tures exhibited the best behavior and reduced up to 45% of the average free shrink-
age of concrete at 360 days of drying.
Results were compared with drying strain predictions of ACI, CEB, B3,
GL, Sakata and Sakata 2001 models. It was concluded that in general all the ana-
lyzed shrinkage models do not adequately represent the observed shrinkage behav-
ior of concretes made with portland pozzolan cement and shrinkage-reducing ad-
mixtures. Therefore, two prediction models to estimate drying shrinkage strains
were developed based on the AC1 drying-time function and Sakata’s ultimate
shrinkage strain. One of these models is for use at a design level when no informa-
tion about a particular shrinkage reducing admixture is known, and the other is ap-
plicable to concrete made with a particular admixture brand.
Kewords: admixtures; concrete; drying shrinkage;

extrapolation; mixture proportions; modulus of
elasticity; portland pozzolan cement; shrinkage
pred ictio n mo d e Is; s h rin ka g e - r e d uc ing ad mixtures;
statistic a I eva Iuatío n; updating; vo Ium e/su rf a ce rat¡o
@Seismicisolation
@Seismicisolation
319
320 Videla and Aguilar
Carlos Videla is Associate Professor of Construction Engineering and Project
Management, and Director of the Strength of Materials Laboratory at the
Pontifical Catholic University of Chile. He received his Ph.D. (civil engineering)
in 1989 from the University of Birmingham, UK. He is a member of AC1
Committees209,23 1,30 I , C6 1O and C630.
Carlos Aguilar is a Ph.D. candidate at the School of Civil Engineering of the

Pontifical Catholic University of Chile where he also obtained his BSc degree in
1995. Currently he is a Lecturer in Civil Engineering at the Universidad de
Santiago de Chile. His research interests include elasticity, shrinkage and
concrete cracking.
INTRODUCTION
It is recognized worldwide that drying shrinkage is a major cause of concern to

designers and engineers in concrete construction because deformations and
cracking due to this phenomenon can significantly impair the durability,
serviceability, structural integrity and aestheîics of a structure. Therefore
structural engineers are increasingly considering shrinkage strains in the design
process and therefore construction engineers are required to meet concrete
shrinkage specifications in addition to specified concrete strength specifications.
For many applications designers do not consider it essential that cracking be
eliminated (i), however cracks in concrete are objectionable from the standpoint
of durability and therefore consideration must be given to reduce cracking or to
control its consequences.
FOTthis reason research to prevent shrinkage cracking in concrete i s camed out

worldwide. Methods for controlling shrinkage cracking include selection of
suitable materials and construction procedures, detailing of joints to reduce
restraints, use of fiber-reinforced concrete and use of shrinkage-compensating
concrete. Reinforcement is use to preserve structural continuity despite cracking.
In recent years, a new type of admixture, called shrinkage reducing admixtures

(SRA), has been introduced into the concrete industry to reduce the drying
shnnkage cracking. Although numerous investigations have been performed to
study the behavior of these admixtures with ordinary portland cement concrete
(2,3), there is no information regarding the effect of SRA on portland pozzolan
cement concretes (PPC).
Finally, the development of an accurate procedure for estimating drying

shrinkage for concrete i s of paramount importance at the design and construction
stages. However, available methods to estimate shrinkage of concrete do not
@Seismicisolation
@Seismicisolation
explicitly recognize the influence of admixture (4, 5,6, 7, 8,9). Nevertheless, it
must be pointed out that shrinkage models cannot completely replace the testing
of concrete to be used in a particular project. Short-term testing is required to
calibrate a model to actual conditions and to reduce the inaccuracy of the
estimated short and long-term concrete deformations.
The objective of this research was to veri@ the feasibility of using SRA for
controlling the drying shnnkage of concrete made with portland pozzolan
cements. Concrete mixtures typically used in building construction, With a
specific strength of 30 MPa, were made using SRA in order to reduce free drying
shrinkage. Major variables included slump, shrinkage reducing admixture
brands, admixture dosages and specimen dimensions. Tests were carried out on
132 prism specimens and shrinkage strains were measured up to 360 days of
drying. Also compressive strength, splitting strength and static modulus of
elasticity were measured.
To determine the influence of SRA on hardened and drying shrinkage properties

of concrete, an extensive experimental program was carried out. This program
considered a total of twenty-two concrete mixtures with similar strength but
different slumps, brands and amounts of admixtures. The design of the
experiment is shown in Table 1.
Materials Characteristic
- Cement - Physical and mechanical tests were performed to characterize the

cement and the resuits are shown in Table 2, including the average chemical
properties reported by the cement manufacturer. According to ASTM C 595 (10)
and CEB (5) the portland pozzolan cement used in this research was classified as
type TP and N, respectively.
- -
Aggregates Standard tests were performed to characterize the properties of the
aggregates including unit weight, specific gravity, absorption and grading. The
proportioning of the aggregates was made by adjusting the grading of the
combined aggregate between the curves NO2 and NO3 proposed in Road Note
NO4 for concretes with maximum aggregate size of 40 mm (1 1). Also modulus of
elasticity tests of prepared rock samples were carried out using 75x150 mm
cylindrical cores. The test results are shown in Table 3.
- Admixtures - Five brands of SRA were used in this study: two commercial
(SR1 and SR2), and the remaining three are experimental admixtures. The
@Seismicisolation
@Seismicisolation
admixtures and their characteristics are summarized in Table 4.It must be noted
that the admixture dosage used for the fabrication of the concrete batches was
obtained considering the limits at third points of the range recommended by the
producers of the particular considered admixture.
Mixture Provoríions
For each theoretical concrete mixture, four preliminary trial mixtures were
prepared in order to adjust the water content of the mixture to satis@the slump
and strength requirements of the concrete. The mixtures were fabricated
according to the procedure specified in ASTM C192 (1 O) in a 0.25 m3 capacity-
vertical rotating mixer. The mixture proportions are shown in Table 5.
Testing Procedures
Several standard tests were carried out on every mixture. The slump, unit weight
and air content of the fresh concrete were determined in accordance with ASTM
C143, ASTM C138, respectively (10). Also the ambient and concrete
temperatures were recorded. The fresh concrete properties are shown in Table 5.
The following tests were performed on the hardened concrete: ASTM C39
compressive strength, ASTM C469 modulus of elasticity and ASTM C496
splitting tensile strength (10); two cylindrical specimens for each test were used
at 7,28 and 90 days. Also two 200-mm cubes were tested for compressive cube
strength at 28 days. All of the samples were maintained in a standard curing
room (T=2oI3 "C and RHSO%) until the age of testing. The measured
mechanical properties of hardened concrete are shown in Table 6.
Drying-shrinkage strain tests were measured according to ASTM C157 (10) up

to 360 days of drying. For each concrete mixîure, 3 prism specimens of
75x75~285mm and 3 prisms of 1OOx1OOx5OO mm were tested in order to
analyze the influence of the volume/surface ratio on shrinkage evolution and
magnitude. The samples were cured in a standard curing room for 1 day and
immersed in water saturated with lime at 20°C for 6 days then the samples were
dried in a special ambient conditioning room at 23 5 2 "C and 50 f 4 % relative
humidity. The shrinkage strains were measured periodically during the 360-day
drying period. Table 7 shows the average measured drymg-shrinkage strains of
concrete. Also, weight losses were measured directly on the shnnkage
specimensat the same drying time.
The drying shnnkage was measured using a dial gauge with a minimum reading
of 0.001 mm between two-gauge plugs inserted into the specimens. Weight loss
was measured with on automatic platform scale with a minimum reading of 1 g.
@Seismicisolation
@Seismicisolation
One mechanism thaî contributes to drying shrinkage is reduction of capillary

pressure. It takes place when a capillary pore of calcium silicate hydrate
skeleton loses moisture, forming a meniscus at the air-water interface. Surface
tension in this meniscus pulls the pore wall inward and the concrete responds to
these internal forces by shrinking. Recently, a type of admixture has been
developed to reduce the surface tension of water resulting in lower capillary
pressure. These admixtures are low viscosity, water-soluble liquids, and their
behavior on hardened concrete properties and shrinkage strains will be analyzed
in the following sections.
Effects of Shrinkage-ReducingAdmixture Brand on Mixture Prowrtions
From Table 5 it can be concluded that on average SRA concretes exhibit a 8%

lower water and cement content than plain concrete for a given slump and
similar strength. This could be a consequence of an increased flow of the
concrete due to the admixture properties (3). Therefore, a lower amount of water
and cement is required for a given slump and strength when SRA are used.
Consequently, it was concluded that the lower demand of water and cement
content of concrete made with admixture (therefore containing lower cement
paste which is the material that undergoes shrinkage), is a contributing factor to
the reduction in drying shrinkagein addition to the effect of the admixture on the
capillary tension.
Effects of Shrinkage-ReducingAdmixture on Hardened ConcreteProDerties
Fig. 1 shows the effects of the addition of SRA on hardened concrete properties
(compression strength, splitting tensile strength and modulus of elasticity). From
the daîa it can be concluded that,in general, the addition of SRA slightly reduced
the compressive strength and modulus of elasticiîy. However, this 8% average
reduction for compressivestrength and 3% for modulus of elasticity is ostensibly
minor compared to values reported by other researchers (2,3). Nevertheless, the
splitting tensile tests did not show a similar behavior; on the contrary the
addition of SRA resulted in a slight improvement(2%) in the splitting strength of
plain concrete. Therefore, fiom the point of view of cracking, SRA aid in
reducing the risk of cracking because the splitting strength increases and the
elasticity modulus decreases.
@Seismicisolation
@Seismicisolation
Anaivsis of Variance Test
To establish the effect of the independent variables on the drying shrinkage a

statistical analysis of variance was carried out, i.e. a MULTIFACTOR ANOVA.
The ANOVA decomposes the variability of drying shrinkage results into
contributions due to different variables. The contribution of each variable was
measured having removed the effects of all the other variables.
The results of the Statistical analysis performed over the normalized drymg
shrinkage strain evolution (12) and magnitude are shown in Table 8. The
normalized drying shrinkage strain evolution is defined as the ratio between the
measured drying shrinkage strain at any time and the last available measurement
(1 2). Also, the results were analyzed applying the weighting technique developed
by Bazant and Baweja (13) to avoid heteroscedacity of the data.
From the analysis of Table 8 it is observed that considering the normalized

evolution of drying shrinkage, the variables admixture brand, admixture dosage,
drying time and VíS ratio are statistically significant. On the other hand, it is
observed that the variable slump is not significant. It could be a consequence of
the differentwater content used on the mixtures to obtain a similar slump.
Considering the iduence of the variables on the drying shrinkage magnitude

obtained at 360 days of drying, it was noticed that the variables admixture brand,
admixture dosage and VIS ratio were also statistically significant.
Table 8 also presents the statistical significance of the variables grouped by

admixture brand. Although the overall results show that the variables affecting
the shrinkage time function and magnitude depend on the admixture brand, this
particular analysis shows different behaviors depending on the specific
admixture considered. Therefore each admixture brand will be analyzed and
compared considering concrete with and without admixture.
Effects of Shrinkage-ReducingAdmixture on Free Drving Shrinkage Strains
Fig. 2 shows the high impact of using SRA over the drying shrinkage of
concrete. The results show on average that a reduction of up to 45% can be
expected with respect to concrete without admixtures. This results can be
explained by the effect of these admixtures over the surface tension of the water
held in the capillary voids of the cement paste. The decrease of this surface
tension reduces the volumetric changes of the cement paste and therefore a
smaller h i n g shrinkage should be expected (2,3).
@Seismicisolation
@Seismicisolation
The results shown in Fig.3 highlight the large effect of SR 5 admixture with an
average reduction of 40% at 360 days of drying compared to concrete without
admixtures.
Moreover, Figs. 4 and 5 present the effect of admixture dosage on drying

shrinkage. Fig.4 shows all the results while Fig 5 presents the average shrinkage
obtained at 360 days of drying. It can be concluded from these data that the
larger the admixture dosage, the smaller the drying shrinkage strains of concrete,
i.e. increasing the amount of admixture provides the larger reduction of
shrinkage. The same result was previously obtained by Shah et al (3). An
exception to this behavior is observed at long term on the concrete containing SR
2, showing a slightly increased shrinkage with an increased admixture dosage.
Lastly, it should be kept in mind that a smaller value of drying shrinkage does
not necessarily ensure that concrete cracking will be avoided because several
additional factors affect cracking. These include the rate of shrinkage, creep
properties, modulus of elasticity, tensile strength of concrete and tensile strain
capacity. However, decreasing the drying shrinkage implies that the potential
risk of concrete cracking is diminished.
Relationshb between weight loss and drving shrinkage of concrete
Fig. 6 shows the relationship between the average results of weight loss and
drying shrinkage for the various admixture brands. Usually, the amount of drymg
Shrinkage has been assumed to be proportional to the amount of water lost.
However, Fig.6 shows that the use of SRA can reduce shrinkage significantly
compared with plain concrete for the same weight loss. Also the proportionality
appears to depend strongly on the type of admixture used. Therefore, at short
term (one year of drying) the effect of the type of admixture on concrete
shrinkage seems to be related to the reduction of the surface tension of the
mixing water such as other researches has previously concluded (2, 3), rather
than to water lost. It was concluded that for SRA concretes the proportionality
between weight loss and drying shrinkage depends on the type of admixture.
CornDarison between Measured and Calculated Drving Shrinkape with

PredictionModels
Drying shrinkage results were compared with strain predictions of ACI, CEB-
FP,B3, GL, Sakata and Sakata 2001 models (see Fig. 7) (4,5,6, 7, 8,9). It was
concluded that ACI, CEB, and B3 shrinkage models highly underestimated
shrinkage strains observed in this study and are not accurate enough because they
@Seismicisolation
@Seismicisolation
present major coefficientsof variation ( 0 ~ 2 0to) be considered adequate (7). On
the other hand, the GL and Sakata shrinkage prediction models overestimate
shrinkage strains obtained and also present high coefficients of variation.
Furthermore, it can be seen that the drying shrinkage time function of CEB, B3
and GL, models do not reflect the actual observed behavior.
Finally, it was found that the best prediction model was Sakata 2001. This model
presents a better shrinkage time function and also has the best correlation
coefficient 0.89 between measured and calculated shrinkage. Also, the
coefficient of variation (l6.8Y0) is considered adequate.
Udating of Shrinkage Predictions Models
With the purpose of improving the models estimate an updating of the models
was carried out with the technique proposed by Bazant and Baweja (6). Bazant
and Baweja proposed a procedure to update shrinkage prediction models using
the measured water loss (6). This procedure was applied, however the obtained
results were not satisfactory. The reasons could be that Bazant et al (6)
procedure assumes that shrinkage i s proportional to water loss and use an
approximate desorption isotherm equation to estimate the final water loss. From
previous analysis (see relationship between weight loss and drying shrinkage of
concrete) it was concluded that the relation between water loss and shrinkage of
SRA concretes depends on admixture brand. Similarly, Hansen (14) and Xi (15)
concluded that the shape of the desorption isotherm of concrete varies
considerably and the proposed equation appears not to be applicable to the SRA
concretes analyzed.
Therefore a statistical procedure was considered to improve the prediction of

drying shrinkage models allowing to fit the best shrinkage evolution and ultimate
shrinkage equations. The mechanism of shrinkage based on the diffusion theory
was not taken into consideration.
Proposition of a Drving Shrinkage Prediction Model for SRA Concretes
The used methodology was based on the normalization of the evolution curves
and later calibration of the time function and ultimate shrinkage strain equations
suggestedby proposed models on specializedtechnical literature (12).
The updating procedure used to achieve the outlined objective considered the
followinggeneral form of the available prediction models:
@Seismicisolation
@Seismicisolation
where E'sh (&,to) is the calculated drying shrinkage strain at time (t-to) of drymg,
with t the concrete age and to the time when drying begins; f(t,to) time function
representing the shrinkage evolution with drying time; &'hm is the estimated
ultimate or final shrinkage strain and ll K; is the cumulative product of
corrections factors Ki for different variables affecting shnnkage.
The modification of the prediction models considered in the study was

performed in two steps. First, by adjusting the time function by means of
calibration factors taking into account the effect of the statistical significant
variables on shrinkage evolution:brand of admixture, V/S ratio and drymg time.
Secondly, the adjustment of the ultimate shrinkage was carried out by
extrapolating the final drying shrinkage of each one of the test series from the
calibrated time function and the last available shrinkage measurement. Then by
calibration of the ultimate shrinkage equation with factors that minimizes the
differencesbetween the calculated and extrapolateddata.
Also the above mentioned procedure was applied at two levels. First, at a global
level, trying to characterize the overall behavior of SRA concretes, independent
of admixture brand and then for each particular admixture brand. Therefore
calibrations coefficients are provided for each case, allowing to apply the results
at the project design stage when there is no information available of the type of
admixture to be used and a the construction stage when a particular admixture
has been chosen.
From the analysis of the results it w a s concluded that the time function proposed
by AC1 best represents the behavior of SRA concretes. The adjusted evolution
equation corresponds to:
The coefficientsa,ß and y are shown in the Table 9.
Since it is not possible to obtain the actual value of the ultimate áqmg shrinkage
its magnitude was extrapolated from the last drying shrinkage measurement of
each tested concrete series. Its magnitude was calculated dividing the last
measured shrinkage result by the estimated value of the adapted time function at
the same age, i.e. :
@Seismicisolation
@Seismicisolation
The uitimate shrinkage strain equations suggested by different authors (4,5,6,7,
8, 9, 16, 17, 18, 19) were analyzed and it was found that the calibration of the
Sakata model proposed equation gives the best results with respect to the
extrapolateduitimate strain for local materials:
The coefficientsx, 6,4, cp, q and K are shown in the Table 10.
Fig. 8 and 9 show a comparison between the measured and calculated drying
shrinkage with the new model. Fig. 8 considers the global Calibration for SRA
concretes and Fig. 9 presents the results for each admixture brand. Clearly, the
proposed model presents a better coefficient of correlation (98%) and an
excellent (7)coefficient of variation (00=8.2%) than the results obtained from
the previously analyzed models.
SUMMARY AND CONCLUSIONS
Based on the experimental results presented in this paper, the following

conclusions on shrinkage of concrete can be drown:
- The addition of shrinkage-reducing admixtures reduces the compressive
strength and modulus of elasticity of concrete. However, this effet3 is
negligible on splitting strength results.
- Ail the admixtures evaluated reduce the drying shrinkage of concrete. It
appears that its main effect is over a period of time when the shrinkage
mechanism is governed by the capillary surface tension forces. At later ages,
the effect is primarily due to the reduction of water content of the mixture
because of improvement in the flow characteristics of the SRA concretes.
- Furthermore, the effect of the amount of admixturedepends on the admixture
brand, but in general the larger the amount of admixtures, the greater the
shrinkagereduction.
- The ACI, CEB, B3, GL and Sakata shrinkage prediction models do not
adequately represent the observed shrinkage behavior of concretes made with
shrinkage-reducing admixture and portland pozzolan cement. It was
concluded that the Sakata 200 1 model best represents the shrinkage behavior
of SR4 concretes and give adequate estimates be used as a first approximate.
- On the other hand, it is believed that the different observed behaviors
between shrinkage predictions by ACI, CEB, B3, GL and Sakata with
measured data, is mainly due to the properties of locally available materiais:
cement properties, aggregate type and admixture type. Therefore, the
variables included on existing prediction models should be verified and their
@Seismicisolation
@Seismicisolation
sensibiliîy adjusted by means of testing of typically used materials on a
region or specific construction work to assure the quality of the estimate and
to prevent potential negative effects over the structures.
- Using weight loss for updating and improving shrinkage prediction models
was shown to be not satisfactory.
- Based on the measured data, an updated shrinkage model was derived for
predicting the shrinkage strains of concrete for local materials. The model
requires variables that can easily be assumed at the construction stage, based
on previous local concrete construction experience or mixture proportions
methods. However, it is important to keep in mind the range of the
independent variables considered in the study for the application of the
adapted model: portland pozzolan cements, siliceous aggregates, admixture
brands and concrete with specified compressive strength between 30 and 40
MPa and slump between 60 and 120 mm.
Finally, it is expected that the results of this research will be a contribution to the
undersîanding of the influence of shrinkage-reducing admixtures on drying
shrinkage, and the proposed model can be used to evaluate deformations and the
risk of cracking of reinforced concrete elements made with SRA admixture and
portland pozzolan cements.
ACNOWLEDGEMENTS
Grateful acknowledgmentis made to the Chilean National Council for Scientific

and Technological Research for funding this study under FONDECYT Grant
NY980943 - 1998.Thanks are also due to the Chilean Cement and Concrete
institute for the support given to the project.
REFERENCES
1. Kheder, G.F., Al Rawi, R.S., and Al Dhali, J.K., “Study of the Behavior of
Volume Change Cracking in Base-Restrained Concrete Walls”, ACZ
Materials Journal, V.91,N”2, March-April 1994,pp. 150-157.
2. Shoya, M., Sugita, S. and Sugawara, T., “Jmprovement of Drying Shrinkage

and Shrinkage Cracking of Concrete by Special Surfactants”, Admixturesfor
Concrete - Improvement of Properties; Proceeding of the Intemationql
Symposium held by H U M ,Barcelona, Spain, 1990,pp. 484 - 495.
3. Shah, S.P., Kraguler, M.E. and Sarigaphuti, M., “Effects of Shrinkage -

Reducing Admixtures on Restrained Shrinkage Cracking of Concrete”, ACZ
Materials Journal, V.89,N”3, May-June 1992,pp. 289 - 295.
@Seismicisolation
@Seismicisolation
4. AC1 Committee 209, “Prediction of Creep, Shrinkage, and Temperature
Effects in Concrete Structures”, Designing for Creep and shrinkage in
Concrete Structures, SP-76, American Concrete Institute, 1982, re-approved
in 1992, pp. 193-300.
5. CEB-FE’, “Model Code for Concrete Structures MC90”, Bulletin

d’InformationW 199, Comité Euro-International du Beton, Thomas Telford,
London, U.K, 1991,575 pp.
6. Bazant, Z.P., and Baweja, S., “Creep and Shrinkage Prediction Model for
Analysis and Design of Concrete Structures: Model B3”, The Adam Neville
Symposium: Creep and Shrinkage - Structural Design Eflects, SP-194,
American Concrete Institute, 2000, pp. 1-83
7. Gardner, N.J. and Lockman, M.J., “Design Provisions for Drying Shrinkage
and Creep of Normal-Strength Concrete”, ACI Materials Journal, V.98, N”2,
MarCh-April2001, pp. 159-167.
8. Sakata, K.,“Prediction of Concrete Creep and Shrinkage”, Proceedings of

5th International RII.EM Symposium (ConcreepS), Barcelona, Spain, 1993,
pp. 649-654.
9. Sakaîa, K.,Tsubaki, T., houe, S. and Ayano, T., ‘Prediction Equations of

Creep and Drying Shrinkage for Wide-Ranged Strength Concrete”, private
communication (September 2001).
10. American Society for Testing and Materials, “Annual Book of ASTM
Standards, V.4.02, Concrete and Aggregates”, American Society for Testing
and Materials, West Conshohocken,Pa., 1998.
11. Road Research Laboratory, “Design of Concrete Mixes”, D.S.I.R. Road Note
No4, H.M.S.O., London, 2ndedition, 1950,16 pp.
12. Videla, C., Covarrubias, J. P. and Masana, C., “calibration of Drying
Shrinkage Prediction Models for Chilean Cement Concretes”, Revista de
Ingeniería de Construcción, Pontificia Universidad Católica de Chile, Vol.
16, NO 1,2001, pp. 5-17.
13. Bazant, Z.P., “Criteria for Rational Prediction of Creep and Shrinkage of
Concrete”, Materials and Structures, The Adam Neville Symposium: Creep
and Shrinkage - Structural Design Eflects, SP-194, American Concrete
Institute, 2000, pp. 237-260.
14. Hansen, W., ‘‘Drying Shrinkage Mechanism in Portland Cement Pastes”, J.

Am. Ceram. Soc., V.70, N”5, 1987,323-328.
@Seismicisolation
@Seismicisolation
15. Xi, Y., Bazant, Z.P., and Jennings, H.M., “Moisture Diffusion in
Cementitious Materiais: Adsorption Isotherms”, J. A&. Cement Based
Mater., 1994, V. 1.
16. Pickett, G., “Effects of Aggregate on Shrinkage of Concrete and a

Hypothesis Concerning Shrinkage”, Journal of the American Concrete
Institute, Proceedings, Vol. 52, No5, 1956, pp 581-590.
17.Hobbs, D.W., “Influence of Aggregate Restraint on the Shrinkage of

Concrete”, Journal of the American Concrete Institute, Vol. 71, 1974, pp.
445-450.
18. Hobbs, D.W., “The Dependence of the Bulk Modulus, Young’s Modulus,
Creep, Shrinkage and T h e d Expansion of Concrete Upon Aggregate
Volume Concentration”, Materials and Structures, Vol. 4, No20, 1971, pp.
107-1 14.
19. Fulton, F.S., “A Co-Ordinated Approach to the Shrinkage Testing of

Concretes and Mortars”, Magazine of Concrete Research, Vol. 13, No
39,1961, pp. 133-140.
Table I . Design of the expenmeni.
30 MPa
40
DI: low amount of admixture- D2:high amount of admixture.
@Seismicisolation
@Seismicisolation
Table 2. Cementproperties.
2.65
so3 I Yo I 2.52
Table 3. Aggregate properties.
@Seismicisolation
@Seismicisolation
Table 4. Admixture properties.
-: by weight of cement
0)).
Table 5. C o m efe mixturepropo~onsmrdjresh concrefepopemes
@Seismicisolation
@Seismicisolation
Table 6.Har&ned concreteproperties &@a)).
[ 22 I 34.2 I 21.0 I 32.8 1 42.3 I 2.1 I 2.9 I 3.7 122433 123497 127042
@Seismicisolation
@Seismicisolation
Table 7. Average meawed W n g shrinkage strains (mmostraim) at dflereiu drying times.
@Seismicisolation
@Seismicisolation
Table 8. Statistical signrficance of variàbles consiakred into the eXpenmental design.
Main Effects
Type of Period of time considered
Concrete
Shrinkage
(days)
Shrinkagetime function'
Admumire
Amount
+
I
Brand
+
lslmpl
Drying VIS
Time ratio
+
I+
Reducing
concrete Drylagshrinkage
magaitude"
+ + - NA +
+: Statistical significance variable; -: Non statistical significance variable.

NA: non-applicable.
Table 9. Calibration coeficients for the time finction.
- *
SRA Family 1.03544 50.0507 -0.00907736

L SR 1 0.993239 19.6963 0.0310824
SR 2 1.01838 21.3529 0.0521057
SR 3 1.02286 38.2625 -0.00289879
SR 4 1.06374 49.7259 -0.00757508
SR 5 I 1.14663 I 108.409 I -0.0422304
Table 10. Calibrationcoeflcientsfor Sakata ultimate shrinkage equation.
I Admixîm I Coefficients
@Seismicisolation
@Seismicisolation
O 7 i 4 21 28 35 42 49 56 63 70 77 84 91 98
Age of Concrete (days)
1.2
h
1.1
.-0 8
3 2 1.0
8 8
E .s
20.9
v
0.7
0.6
O 7 14 21 28 35 42 49 56 63 70 77 84 91 98
,. .
< .. . . .
.,, .... ..<.. . ... ... <. ... . .., .. < <
. .... .,.
< .
..
, ., .. <.. .. ,,.
. .<... ,.... ,....
,.. ... .. ... .. .. .. . .
.. . . . . .
. <
. . <
. .
.
.... .........o. ........ .
< .
........ .......... .i ........1...... ...........i.
i ...o; .....
: , , , , . . / A :
,.
,.. .., ,. ., .,. ,
,. .. ..
., .. ., ... :... :... : .. i . $

' X , . ... <
...
.
....
... . <.. . . . ..
i
....:... 9 . i
... .. .. ...............................................
,.
.. ,
.... ..,. .,.
..
. . . . . . ..
.. ... ... . . .
,
., ..
.. .. .. ..
.. .. .,. ,.
.
<
,. <
,. .
, .
, . .
. . i.....i......... i........ i. .....j.. .
.. .. <.
1 SR I concrete
j ..
. ~ ..
..
~ <
..
~
.
~
0 SR 2 Concrete
ASR,lc__j:
... ... ... ...............
............................. x SR 4 Concrete
..,. .. .. .. ..
.. .... <. j.. j. 0 SR5Concrete
,. .. <
< . .
... ... <.. .... ..., -Av eValues
,
.. ..
. <
.. .,. ... .
. .
. . . . .
O I 14 21 28 35 42 49 56 63 70 I I 84 91 98
@Seismicisolation
@Seismicisolation
Fig. i Effect of the addition of Shrinkage Reducing Admixture on average mechanical propeides of
concretes, a) compressive strength, b) splitting strength and c) modulus of elasticity.
1400
.sô
M
W
X
v1
2 1050 E sh SRZ-DI = sh Plain
2
E shSRZ-DZ= shPlain
u
d E sh SR3-D1= o.@*&sh Plain
41O 700
4
a
3 350
2
'E
II
O
O 350 700 1050 1400
Drying Shrinkage of Plain Concrete (AOd)
Fig. 2 Drying shrinkage of concrete with SRA versus plain concrete
@Seismicisolation
@Seismicisolation
1200
1000
%X
v
800
#J
600
VI
3 400
'8
n
200
O
O 28 56 84 112 140 168 196 224 252 280 308 336 364
Drying Time (days)
Fig. 3 Relationshipbetween average of drying shrinkageof concrete containing SRA and drying time.
@Seismicisolation
@Seismicisolation
Shrinkage Reducing Concretes

/
&shSRI-DZ=O.91*&
com = 0.94
&SRI-DI
. .==./
.
0
SR1
sR2
A sR3
SR4
x sR5
O 400 800 1200

Drying Shrinkage of Concrete with low content of Admixture @i)
Fig. 4 Effect of admixture content on drying shrinkage
@Seismicisolation
@Seismicisolation
~~~~

1.1
.....................................................................................................................................
....................
*
-.
w
B
0.8 ................................ .............
8
8
Q 0.7 .........
; -+ SR 2 Concrete
0.6 ......... ....................
W
a -.x--SR 4 Concrete
0.5 I I
O 0.5 1 1.5 2
Admixture content (YO)
Fig. 5 Relationship between drying shrinkage ratio and admixtures percentage.
4.5
4.0
3.5
3.0
h
s.
v
s” 2.5
ql 2.0
s 1.5
1.0
0.5
0.0
O 200 400 600 800 1000 1200
Drying Shrinkage (xi06)
Fig. 6 Relationship between weight loss and drying shrinkageof concrete
@Seismicisolation
@Seismicisolation
~
I
/
E sh Calculated ACI Model = 0 . 6 3 * ch
~ Measur#’
/
com = 0.88 /
/
/
coo AC1 = 33.1% /
/
/
/
/
/
/
/
Ø
OO
500
I ACI-209
O I I
O 500 1000 1500

Measured Drying Shrinkage Strains (x 1 O-6)
1500 /
E& Calculated CEB-FIP = 0.72*~& Measured //

/
/
com = 0.72 /
/
/
/
mo CEB-FIP = 3 1.1% /
/
1000 - /
/
O 500 1 O00 1500

Measured Drying Shrinkage Strains ( ~ 1 0 ~ )
Fig. 7 Comparison between measured and calculated drying shrinkage with various models.
@Seismicisolation
@Seismicisolation
1500 /
E sh Calculated B3 = 0.77*~sh Measured /
/
/
torr = 0.77 /
/
/
/
woB3 = 26.4% /
/
/
/
d
x O
o
500
1831
O
O 500 1 O00 1500
Measured Drying Shrinkage Strains ( x ~ O - ~ )
1500 I
/
E sh Calculated = 1 . 0 8 * sh
~ Measured /
/
corr = 0.64
O, GL = 23.5%
1 O00
O
O 500 1 O00 1500
Measured Drying Shrinkage Strains (xi Oa)
Fig. 7 Comparison between measured and calculated drying shrinkage with various modeis(wnt.)
@Seismicisolation
@Seismicisolation
~~ ~________

1500
O
O 500 1O00 1500
Measured Drying Shrinkage Strains (x 1O-6)
1500 /
/
/
E sh Calculated Sakata = 0 . 9 9 * ~
sh Measured //
/
com = 0.89
coo Sakata = 16.8%
O I l
O 500 1000 1500

Measured Drying Shrinkage Strains (x1Oa)
Fig. 7 Comparison between measured and calculated drying shrinkage with various models(cont.)
@Seismicisolation
@Seismicisolation
/'
E sh CalculatedModel Proposed = 1.0078*Ecb M e d 0
/
con = 0.95 /
/
/'
o,= 12.9%
o ,./: O
O 500 1000 1500

MeasuredDrying Shrinkage Strains (x10")
Fig. 8 Comparisonbetween measured and calculated drymg shrinkage with model proposed (SRA
family)
@Seismicisolation
@Seismicisolation
~~

1500 -
sh CalculatedModel Roposcd = l.02*&sh Measund

com = 0.98
4b o,= 8.2%
vi
1000
3 500
O
O 500 1000 1500
Measured Drying Shrinkage Strains ( x i 0 3
Fig. 9 Cornpanson between measured and calculated drymg shrinkage with model proposed (SRA
brand)
@Seismicisolation
@Seismicisolation
SP 209-1 9
Substitution of Fly Ash, Slag, and

Admixtures in FDOT Concrete
Mixture Proportions
by Y. Jin and N. Yazdani
Svnopsis: Florida Department of Transportation (FDOT) requires contractors to

submit proposed concrete mixture proportions prior to the production of any con-
crete. The contractor must use mixture proportions approved by FDOT. Substitu-
tions of ingredients, other than coarse aggregate, must be justified through trial
mixtures, and authorized in writing by FDOT Engineers. The substitutions of fly
ash, slag, air-entraining admixtures, and Types A, D and G water reducing admix-
tures were performed on two typical FDOT hot weather mixture proportions in this
study. The concrete properties considered were slump, air content, and compres-
sive strength. Test data for substitution mixture proportions were compared with
the data for the original mixture proportions on the basis of statistical criteria. Re-
sults show that the substitutions cause significant change in concrete properties for
the fly ash mixture proportions. The substitution of slag and air-entraining agent in
the slag mixture proportions does not cause significant change in the properties of
concrete.
Keywords: admixtures; fly ash; hot weather concrete;

slag
@Seismicisolation
@Seismicisolation
347
348 Jin and Yazdani
Yingqin Elaine Jin, EIT, is a structural engineer with Jacobs Sverdrup Civil, Inc.. She
received her M.S. in Civil Engineering from Florida State University, and B.S. in Bridge Engineering
fiom Tongji University, P.R. China. Her Master thesis focused on the substitution of fly ash, slag,
and chemical admixtures in hot weather concrete mixture proportions. Prior to her study in Florida
State University, she worked three years as a bridge engineer for a consulting practice in China.
AC1 member Nur Yazdani is a professor of civil engineering at the Florida A & M
University-Florida State University College of Engineering. He is a member of AC1 Committees 343,
Concrete Bridge Design-Joint ACI-ASCE; 123, Research; and 201, Durability. His research interests
include prestressed concrete members, bridge design and rehabilitation.
INTRODUCTION
Concrete is composed principally of aggregates, portland cement, and water, and may contain
other cementitious materiais such as fly ash, slag and chemical admixtures. Concrete mixture
proportions are the procedure of ascertaining the ingredient proportions involved in a particular batch
of concrete and to assure that various materials are combined and mixed together properly. A proper
mixture proportion satisfies the requirements of strength, durability, workability, safety, economics
and other specified elements.
Section 346-6.2 of the FDOT Standard Specifications for Road and Bridge Construction [i]
requires the contractors to submit a proposed concrete mixture proportions prior to the production of
any concrete. The contractor must use mixture proportions approved by FDOT for the purpose of
quality control. Any change on the approved mixture proportions must be authorized, in writing, by
the FDOT Engineer. In fact, concrete producers are increasingly finding altematives for concrete
ingredients for improvements, innovations and shortage of raw materials. They woul&-like to
substitute ingredients in FDOT project mixtures with the equivalent altematives for economic reasons.
According to FDOT 346-6.2 stipulations, such substitution must be justified through the re-testing of
approved mixture proportions. The only allowed exception is the substitution of coarse aggregate,
which is specified in Section 346-2.6. It states that the coarse aggregate can be substituted with the
@Seismicisolation
@Seismicisolation
same type of materials with similar physical and chemical properties to the original aggregate but
from different sources. However, if unsatisfactory results are obtained with the different source
aggregate, the contractor still needs to return to the originally approved aggregate source of supply.
This stipulation of substitution means that concrete producers at present may have to re-test many
FDOT mixture proportions that were previously approved for a particular producer, in case of product
substitution such as fly ash, cement, slag and admixtures. Even if one product is to be substituted, the
process is likely to be very costly in terms of manpower, time and material. In fact, when chemical
and other manufacturers who supply concrete products, replace one product with another similar
product, product substitution is necessary. Such substitution may affect quite a few concrete producers
and may require the re-testing of hundreds of mixture proportions.
This paper explores the possibility of saving the cost of re-testing approved mixture
proportions due to the substitution of such ingredients as fly ash, slag, and admixtures. An
experimental study was conducted to investigate whether the substitution of such similar ingredients
will result in negligible change in the properties of approved FDOT concrete mixtures. Two typical
FDOT approved mixture proportions were used as prototype for substitutions. The substitutions of fly
ash, slag and several common admixtures including Type A, Type D, and Type G admixtures were
investigated. The concrete properties considered are slump, air content, and 28-day compressive
strength. A parametric approach was followed, whereby one ingredient was substituted with products
from different sources, and other materials were kept constant as in the original mixture proportions.
BACKGROUND REVIEW
Fly ash has been widely used as a replacement of portland cement in concrete for aggressive
environments. Fly ash is a pozzolanic material obtained as a by-product from combustion of coal, and
categorized as class F and C according to the chemical composition [2]. Class F fly ash is normally
produced from burning anthracite or bituminous coal and has pozzolanic properties, while Class C fly
ash is normally produced from lignite or subbituminous coal and has pozzolanic properties as weil as
some cementitious properties. Because fly ashes from the same class stili exhibit significant variation
@Seismicisolation
@Seismicisolation
350 Jin and Yazdani
in their chemical and physical properties, and the type of cement and admixtures used also directly
influences the cement-fly ash reaction, trial mixtures are always suggested for fly ash concrete to
ensure that the desired characteristicsof the concrete are attained [1,15]. The incorporation of fly ash
into concrete affects fresh concrete in such aspects as workability, bleeding, pumpabiliîy, and time of
setting of concrete. The effects that fly ash has on the properties of hardened concrete include
temperature rise, strength and rate of strength gain, resistance to damage from freezing and thawing,
resistance to ingress of aggressive liquids and reinforcing bar corrosion, alkali-silica reaction,
resistance to chlorides and sulfates and a number of other properties. in addition, fly ash concrete
requires more air-entraining agents than regular concrete to obtain a set amount of air in the hardened
concrete. Variations in the ?loss on ignition?, a property related to the carbon content of fly ash,
require the amount of air-entraining agent to be varied to entrain a given amount of air.
Granulated blast furnace slag, also simply called slag, is the glassy granular material formed
when molten blast-furnace slag is rapidly chilled by immersion in water [SI. Like fly ash concrete,
slag concrete has also been widely used for aggressive environments. ASTM classifies slag in three
grades: 80, 100, and 120, according to the slag activity index. The effects of slag on properties of
fresh concrete vary with the replacement level, but generally include improved workability, decreased
water demand, and increased setting time. In reference to hardened concrete, slag can help to enhance
the durability of concrete by means of improving its resistance to chloride, sulphate, and alkali-silica
reactions, provided that the concrete is properly proportioned and cured. If cured properly, slag
concrete has better compressive strength after 28 days than regular concrete [ 6 ] .
Seven types of chemical admixtures are defined in ASTM C494-98 [7] based on their effects
on the properties of concrete made with these admixtures, and three of them were investigated herein.
They are: Type A-water-reducing admixture, Type D-water-reducing and retarding admixture, and
Type G-water-reducing, high range and retarding admixture. Air-entraining admixture is required by
FDOT Specifications [i] for all except counterweight concrete, so au-entraining admixture was also
investigated in this study.
@Seismicisolation
@Seismicisolation
FDOT Specifications [i] 346-8.7 defines hot weather concreting as the production, placing
and curing of concrete when concrete temperature at placing exceeds 30°C (86’F) but is less than 40’
C (104OF). Hot weather concreting may cause various difficulties including: increased water demand,
accelerated slump loss, increased rate of setting, difficulties in controlling entrained air, and critical
need for prompt early curing. For mixture proportions developed for use under hot weather conditions,
a special mixing procedure is specified in FDOT Specifications [i] 346-6.2 ( 5 ) . The trial mixture
should be prepared at a minimum temperature of 34 “C (93 ”F), and held in the mixer for 90 minutes
after completion of initial mixing. The extended mixing period follows the initial mixing. Dunng the
extended mixing period, the mixing drum should be turned intermittently for 30 seconds every 5
minutes, and covered with wet burlap or an impermeable cover material during the rest periods. The
mixture temperature at the end of the extended mixing period should not be less than 35 OC (95 OF).
MIXTURE PROPORTIONS AND TEST MATRICES
Two suitable FDOT mixture proportions containing ingredients to be substituted were
selected herein. The original materials used in the mixture proportions were substituted by similar
products, but different brand names. Concrete was mixed according to the original mixture
proportions and mixture proportions with substituted ingredients, and tests were performed on the
concrete. The chosen mixture proportions are presented in Tables 1 and 2. Both are Class IV,hot
weather concrete mixture proportions. The fly ash mixture proportion contains Class F fly ash, air-
entraining admixture, and Types A and G admixtures. The slag mixture proportion contains blast
h a c e slag, air-entraining admixture, and Type D admixture.
For each investigated ingredient, four different products with similar properties as the original
material were chosen fiom FDOT’s approved producers list. The term “similar properties” refers to
the same category in ASTM to which the products belong. The materials fiom different sources were
used to substiîute for the original ones. In order to simplie narrative, these materials will be called
“substitutes” as a comparison of the original materials. The original materials and substitutes are
listed for fly ash and slag mixture proportions in Table 3 and Table 4, respectively. The mixture
@Seismicisolation
@Seismicisolation
352 Jin and Yazdani
proportions were named by combining two or three letters and one number, for example S-0-1, and F-
A-RI. The first letter, “F” or “S”, represents fly ash mixture proportions or slag mixture proportions.
The second letter tells whether the mixture proportion is the original mixture proportion, or a
substitution mixture proportion and what kind of substitution. “ O stands for original mixture
proportion; “F” fly ash substitution mixture proportions; “E” air-entraining agent substitution mixture
proportions; “ A Type A substitution mixture proportions; “ G Type G substitution mixture
proportions; “S” slag Substitution mixture proportions; and “ D Type D substitution mixture
proportions. The third letter represents the name of the product or the manufacturer in the substitution
mixture proportions. The number following the third letter is used to differentiate the same original
mixture proportions or substitution mixture proportions. For instance, S-0-2 is the second mixture of
the Qriginal slag mixture proportions; F-G-SI is the first Type G substitution for the Ely ash
mixnire proportions, and the Type G substitute is supercizer 3.
in the process of substitution, the admixture producers were consulted to obtain appropriate
dosage of admixtures that are suitable for the investigated mixture proportions. However, for fly ash
and slag substitution, the dosage was kept identical to those of the original materials. The reason is
that producers provided considerably different recommended dosage rates for their admixture
products, while they usually do not provide such dosage rates for fly ash and slag. The dosage of fly
ash and slag is usually determined by the mixture proportions based on FDOT Specifications and the
desired effect on concrete’s properties, and trial mixtures. The dosage of materials used in this study
are presented in Tables 3 and 4.
TESTING PROCEDURE
As the control experiment, three batches of the original mixture proportions were made and
tested first. Then, substitutions of fly ash, air-entraining agent, Type A and Type G admixture were
performed on the fly ash concrete mixture proportions, and substitutions of slag, air-entraining agent
and Type D admixture on the slag concrete mixture proportions. While one ingredient was substituted,
@Seismicisolation
@Seismicisolation
other ingredients were kept the same as required in the original mixture proportions. In this way, the
effect of only the substituted ingredient was studied.
Absorption tests, ASTM C127 [SI and C128 [9], were performed to determine the absorption
rate of coarse and fine aggregate. Total moisture content test, ASTM C566 [lo], of coarse and fme
aggregate was performed every time prior to the concrete mixing. Subtracting absorption fiom total
moisture content gave the surface moisture content of coarse or fine aggregate, which should be
counted as part of the mixing water. Because the investigated mixtures are designated as hot weather
mixture proportions, concrete was made in accordance with both ASTM C192 [li]: “Making and
Curing Concrete Test Specimens in the Laboratory”, and FDOT Specifications [i] 346-6.2(5): “For
mixture proportions developed under hot weather concreting conditions”. A Gilson HM-224 170 1 (6
CU. Ft) portable concrete mixer was used to mixture concrete. All materials were added to the mixer
with the mixer running. Coarse aggregate, fine aggregate, and part of the mixing water were added
first. Then air entrainer solution was added, followed by the water reducer and/or retarder (Type A or
D). Cement, fly ash or slag, and the rest of the mixing water were added to the mixer at the end. If a
high-range admixture (Type G) was used, it was introduced to the mixer after other materials were
added and mixed for 2 minutes. The mixture was held at a minimum temperature of 34 “C (93 T) in
the mixer for 90 minutes after completion of 3-minute initial mixing. The mixture temperature at the
end of the 90-minute extended mixing period was maintained at not less than 35 OC (95 OF). During the
extended mixing period, the mixer was turned intermittently for 30 seconds every five minutes. The
concrete was remixed for 1 minute at the end of the 90-minute period. During the mixing process, a
kerosene garage heater was used to maintain the temperature of the mixture, and the temperature was
checked manually every ten minutes.
Tests performed on fresh concrete include slump, ASTh4 C143 [12], and air content;-ASTM
C173 [13]. Three 150 by 300 mm (6 by 12 in) cylindrical specimens were made and cured for each
batch of concrete in accordance with ASTM C192 [ l i ] for 28-day compressive strength test, ASTM
C39 [14]. All tests were performed at the Solid MechanicdStructuralLaboratory in the FAMU-FSU
College of Engineering and by the same research group throughout the length of the project. This
@Seismicisolation
@Seismicisolation
354 Jin and Yazdani
process is likely to decrease multiple operator differences in results. Figures 1 through 4 show
concrete mixing, slump text, air content test and compressive strength test Set-ups.
TEST RESULTS
The test results obtained from the fly ash and slag original and substitution mixture
proportions are listed in Tables 5 and 6, respectively. Three series of test results for each original
mixture proportions and one for each substitution mixture proportions were included in the tables.
Mean and standard deviation for the test results were calculated using the following equations and are
listed in Tables 7 and 8:
Mean: p=- cnx

Sample Standard Deviation:
n(n - 1)
ASTM E177 [i61 and ASTM C670 [15] define an index named “D2S”, “Two-Standard-
Deviation Limit”. Approximately 95% of all pairs of laboratory test results fÌom similar facilities
elsewhere may be expected to differ in absolute value by less than D2S. It can be calculated as
follows:
D2S = 1.96& (3)
where:
s = sample standard deviation
This index is also known as the 95% limit on the difference between two test results. The
difference between two individual test results would be equaled or exceeded in the long run in only
5% of the cases with the normal and correct operation method.
ASTM provides D2S values for most standard testing methods based on a large sample of test
results. Because the hot weather mixing method used in this research is not identical as the mixing
method described in ASTM C192 [li], it is not appropriate to use the indices obtained based on the
standard ASTM method. Therefore, the sample standard deviation for each test was obtained from the
@Seismicisolation
@Seismicisolation
test results for the original mixture proportions. D2S indices and the maximum ( p + D 2 S ) and
minimum ( p -D2S) acceptable values for the particular tests were calculated based on these standard
deviations, and are listed in Tables 7 and 8. The test data from each substitution mixture proportion
were compared with the acceptable range.
The results of the comparisons are shown in Tables 9 and 10. The tables show whether or not
an individual substitution test results fall within the acceptable ranges defined in Tables 7 and 8. It
may be observed from Table 9 that virtually all air content and compressive strength substitution,
resulting from the fly ash mixture, fall within the statistically acceptable range. However, most of the
slump substitution results are outside the range. For the slag mixture proportions, Table 10 shows that
100% of the slump and air content results are acceptable, while all but one compressive strength
substitution result are acceptable.
CONCLUSIONS AND RECOMMENDATIONS
The following conclusions can be made based on the fmdings of this study:
1. The slump of fly ash concrete varies significantly with the substitution of fly ash, air-
entraining agent, and Type A and Type G admixtures compared with that of the original
fly ash concrete mixture.
2. The air content of fly ash concrete is not affected seriously by the substitution of fly ash,
air-entraining agent and Type G admixture. In case of Type A admixture substitution, the
air content remained statistically unchanged in three out of the four substitutions.
3. The compressive strength of fly ash concrete is not affected significantly by the
substitution of fly ash. For the substitution of air-entraining agent, Type A admixture, and
Type G admixture, the compressive strength did not change significantly in three out of
four substitutions.
4. As for slag concrete, the changes in slump, air content and compressive strength caused
by the substitution of slag and air-entraining agent are not significant. For the substitution
of Type D admixture, the changes in slump and air content are negligible, and the
@Seismicisolation
@Seismicisolation
356 Jin and Yazdani
compressive strength remained statistically unchanged in three out of four Type D
substitute mixtures.
It is therefore recommended that for fly ash concrete the substitution of fly ash, air-entraining
agent, and Type A and Type G admixtures, be justified by trial mixtures. For slag concrete, the
substitution of slag and air-entraining agent may be allowed. Regarding to the substitution of Type D
admixture, precautions need to be taken and justification by trial mixtures is necessary if inconsistent
results were found. Further study of the influence of various materials kom different sources on fly
ash and slag concrete mixture proportions is suggested. More compatibility study of concrete
admixtures with products from a better variety of sources involved for fly ash and slag concrete is also
recommended.
ACKNOWLEDGMENT
The study reported herein was performed under a research grant from the Florida
Department of Transportation.
@Seismicisolation
@Seismicisolation
REFERENCES
1. Florida Department of Transportation, Standard Specifications for Road and Bridge

Construction, Florida Department of Transportation, 1999.
2. ASTM, Coal Fly Ash and Raw or Calcined Natural Pozzolan for Use as a Mineral
Admixture in Concrete. ASTM C 618-98. American Society for Testing and Materials,
1998.
3. Abou-Zeid, Mohamed Nagib, John B. and Stephen A. Cross, High Dosage Type C Fly
Ash and Limestone in Sand-Gravel Concrete. Transportation Research Record, No. 1532,
pp. 36-43.
4. Halstead, Woodrow J., Use of Fly Ash in Concrete. National Cooperative Highway
Research Program, @nthesis of Highway Practice 127. Transportation Research board,
1986.
5. ASTM, Ground Granulated Blast-Fumace Slag for Use in Concrete and Mortars.
ASTM C 989-97b. American Society for Testing and Materials, 1998.
6. Hooton, R. Doug, Canadian Use of Ground Granulated Blast-Fumace Slag as a

Supplementary Cementing Material for Enhanced Performance of Concrete. Canadian
Journal of Civil Engineering, August 2000, v 27 n 4, pp. 754.
7. ASTM, Chemical Admixtures for Concrete, ASTM C 494-98. American Society for
Testing and Materiais, 1998.
8. ASTM, Specrfic Gravity and Absorption of Coarse Aggregate. ASTM C 127.

American Society for Testing and Materials, 1998.
9. ASTM, Specific Gravity and Absorption of Fine Aggregate. ASTM C 128. American
Society for Testing and Materials, 1998.
10. ASTM, Total Moisture Content of Aggregate by Drying. ASTM C 566. American
Society for Testing and Materials, 1998.
11. ASTM, Making and Curing Concrete Test Specimens in the Luboratoy. ASTM C 192M-
95. American Society for Testing and Materiais, 1998.
12. ASTM, Slump of Hydraulic-Cement Concrete. ASTM C 143M-97 American Society for
Testing and Materiais, 1998.
13. ASTM, Air Content ofFreshly Mixed Concrete by the VolumetricMethod. ASTM C 173-
94a. American Society for Testing and Materials, 1998.
14. ASTM, Compressive Strength of Cylindrical Concrete Specimens. ASTM C 39-96.

American Society for Testing and Materials, 1998.
@Seismicisolation
@Seismicisolation
358 Jin and Yazdani
15. ASTh4, Bias Statements for Test Methods for ConstructionMaterials. ASTM C 670-87.
American Society for Testing and Materiais, 1998.
16. ASTM, Use of the Terms Precision and Bias in ASTMTest. ASTM E 177-86. American
Society for Testing and Matenais, 1998.
@Seismicisolation
@Seismicisolation
Table 1: Selected Fly Ash Mixture Proportions
Class Concrete: IV (37.9 MPa)

Hot Weather Concrete Mixture Proportions
Ingredient Source Details Proportions
Coarse Agg. Rinker Materials Grade 57, Cmshed Limestone 757.5 kg

~
Fine Agg. Florida Rock industries F.M.: 2.20, Silica Sand 493.5 kg
Cement Broco (Brooksville) AASHTO M-85 Type II 279.9 kg
Fly Ash Florida Mining & Materials ASTM C-618 Class F 61.7 kg
Air Entr. A d m i r e Daraex W. R. Grace AASHTO M-154 502.7 ml
1" Admixture WRDA 64 W. R.Grace AASHTO M-194 Type A 363.7 ml
Znd Admixture D a r m 100 W. R.Grace ASTM C-494 Type G 2188.2 ml

I I
Water 118.7kg
Table 2: Selected Slag Mixture Proportions
Class Concrete: IV (37.9 MPa)

Hot Weather Concrete Mixture Proportions
Ingredient
I'
I
Source
I
Details
I
Proportions
Coarse Agg. Rinker Materials Grade 57, Crushed Limestone 748.4 kg

I I I
Fine Agg. Florida Rock Industries F.M.: 2.18, Silica Sand 523.0 kg
Cement Tarmac (Medley) AASHTO M-85 Type Ii 127.9 kg
Slag Penn-Cen Pennsuco ASTM C-989 61.7-kg

C&S
4ir Entr. Admixîure Daraex W. R. Grace AASHTO M-154 147.9 ml
1" Admixture WRDA 64 W. R.Grace AASHTO M-194 Type D 1250.8 ml
Water 119.1kg
@Seismicisolation
@Seismicisolation
360 Jin and Yazdani
Table 3: Test Matrix for Fly Ash Mixture Proportions
Fly Ash
Original I Substitute
roducer
(ibslcu yd) (136) (136)

Air-Entraining Agent
Original I Substitute
F-0-1.2.3 I F-E-DI I F-E-M2 I F-EM3 I F-EM4
I roducer
I
I
II
W.R.Grace W.R.Grace
II I I1
Materials & Materials & Master Builder!
Technologies Technologies
roductName
bose (dm')
DAREX
657
DARAVIR
27 1
mo
657
1 329
MBAE90
475
(fl odcu yd) (17.0) (7.0) (17.0) (8.5) (12.3)
Type A Admixture
öripinal I Substitute
broduct Name
I
I
(fl odcuyd)
47 5
(12.3) 1 40;) 1 955
(24.7)
Type G Admixture
1 (f7:) 1 715
(18.5)
original Substitute
F-O-1,2,3 F-G-S1 I F-GP2 I F-GF3 I F-GB4
roducer 1 W.R.Grace 1 Frize
1 Master
Builders
1 Fox I Addiment
roduct Name
'y," SuperciZr3 pozzlLITH
440 FX-33543 BVS
2860 3494 3324 3494 5238
(74.0) (90.4) (86.0) (90.4) (135.5)
@Seismicisolation
@Seismicisolation
Table 4: Test Matrix for Slag Mixture Proportions
Substitute
S-E-M4
roducer
roduct Name
Dose (mum3) 193 97

MONEX
AIR30 I MONEX
AIR40 1 MBAE90
425 213 193

(fl odcu yd) (5.0) (2.5) (11.0)
I
I I
Type D Admixture
Substitute
S-D-RI I
I
S-D-D2 S-D-P3 S-D-P4
roducer W .R.Grace
mm I Master
Builders Builders
Master Builder:
kroduct Name
I WRDA 64 REDUXDP I DELVO pozzILITH
961R
I POZZILITHSC
P ose ( d m ’ )
(fl odcu yd) (42.3) (48.4) (17.6)
@Seismicisolation
@Seismicisolation
362 Jin and Yazdani
Table 5: Test Data for Fly Ash Mixture Proportions
Temperature Slump Air Content Compressive Strength

Mixture Proportions
(Cl (FI (m) (in) (“w Wa) (Psi)
Ply Ash
Substitution
Air-Entraining
Agent
Substitution
F-E-M4 40 (104.0) 173 (6.8) 3.2 37.7 (5472)

Type A F-A-RI 35 (95.0) 102 (4.0) 3.2 39.7 (5768)
Admixture
Substitution F-A-M2 37 (98.6) 190 (7.5) 7.5 28.6 (4144)
F-A-P3 37 (98.6) 127 (5.0) 5.0 35.4 (5139)
F-A-P4 35 (95.0) 140 (5.5) 3.0 38.1 (5534)

Type G
Admixture
Substitution
F-G-F3 (6985)
FG-B4 1.5 40.3 (5849)
@Seismicisolation
@Seismicisolation
Table 6: Test Data for Slag Mixture Proportions
Temperature Slump Air Content Compressive Strength

Mixture Proportions
(Cl (FI (m) (in) YW (Mp4 (Psi)
s-0-1 35 (95.0) 38 (1.5) 1.8 57.2 (8298)
s-0-2 37 (98.6) 46 (1.8) 2.0 55.5 (8060)
S-0-3 34 (93.2)l 127 (5.0) I 3.8 I 49.6 (7196)
s-s-s1 35 (95.0)l 13 (0.5) I 2.5 I 54.2 (7863)
S-S-M
S-S-N3 (8654)
S-S-R4 0.8 53.5 (7759)
S-E-D 1 ' 36 (96.8) 1 152 (6.0) 1 1.6 1 48.6 (7052)
S-E-M2 36 (96.8) I 63 (2.5) 1 2.8 I 48.5 (7037)
S-E-M3 36 (96.8) 76 (3.0) 2.8 46.4 (6727)
S-E-M4 34 (93.2) 44 (1.8) 2.1 58.9 (8554)
S-D-R1 34 (93.2) 13 (0.5) 1.8 43.4 (6299)
S-D-D2 34 (93.2) 127 (5.0) 4.0 52.0 (7540)
S-D-P3 34 (93.2) 46 (1.8) 2.5 55.8 (8098)
S-D-P4 34 (93.2) 83 (3.3) 2.8 52.9 (7681)
@Seismicisolation
@Seismicisolation
364 Jin and Yazdani
Table 7: Statistical Results for Original Fly Ash Mixture Proportions Data
Slump Air Content Compressive Strength

(mm) (i4 (To) (Mpa) (Psi)
Mean 70 (2.8) 2.9 41.3 (5995)
Standard Deviation 11 (0.4) 1.0 2.6 (373)
Acceptable Range 39 - 100 (1.5 -4.0) 0.1 - 5.8 34.2 - 48.4 (4961 - 7028)
Table 8: Statistical Results for Original Slag Mmture Proportions Data
Slump Air Content Compressive Strength

(mm) (in) W) (ma) (Psi)
Mean 70 (2.8) 2.5 54.1 (7851)
Standard Deviation 49 (1.9) 1.1 4.0 (580)
Acceptable Range O- 207 -

(0.0 8.1) 0.0 - 5.6 43.0 - 65.2 (6244 - 9459)
@Seismicisolation
@Seismicisolation
Table 9: Acceptability of Substitutions for Fly Ash Mixture
Compressive
Mixture Proportions Slump Air Content Stren%h
~~
Fiy Ash F-F-B1 NO YES YES

Substitution
F-F-V2 YES YES YES
F-F-I3 NO YES YES
F-F-P4 NO YES YES
Air- F-ED 1 NO YES NO
Entraining
Agent F-E-M21 NO I YES I YES
Substitution
F-E-M3 NO YES YES
F-E-M4 NO YES YES
Type A F-AR1 NO YES YES
Admixture
Substitution F-A-h42 NO NO NO
F-A-P3 NO YES YES
F-A-P4 NO YES YES
Type G FG-S 1 NO YES NO
Admixture
Substitution FG-P2 YES YES YES
~ ~~~
F-G-F3 NO YES YES

FG-B4 NO YES YES
@Seismicisolation
@Seismicisolation
366 Jin and Yazdani
Table 10: Acceptability of Substitutions for Slag Mixture
@Seismicisolation
@Seismicisolation
Figure 1: Concrete Mixing
Figure 2: Slump Test
@Seismicisolation
@Seismicisolation
~ _ _ _ _
368 Jin and Yazdani
Figure 3: AU Content Test
figure 4: Compressive Strength Test
@Seismicisolation
@Seismicisolation
SP 209-20
Biaxial Tensile Behavior of the

Reactive Powder Concrete
by O. PIé, E. Astudillo de la Véga, G. Bernier
and O. Bayard
SvnoDsis: Numerical and experimental studies of Reactive Powder Concrete

(RPC), reinforced by short steel fibres, under biaxial tensile loading are reported.
A semi-analytical model using the Eshelby’s method of inclusion to describe elas-
tic fields perturbation is employed to predict crack nucleation under biaxial tensile
strain. A criterion of crack nucleation is investigated theoretically by considering a
locally oriented mass of fibres embedded in an homogeneous matrix. An original
biaxial cruciform specimen is designed and fabricated by a systematic testing pro-
gram guided by the results of a numerical simulation. A prototype machine names
ASTREE (developed by SCHENCKLMT) is used in the biaxial test of cruciform
specimens. An adaptive control test method was designed, and digital image corre-
lation is employed to obtain the displacement field and the microstructural stress
concentration. These preliminary observations support our theoretical analysis
based on Eshelby’s inclusion and aims at a better understanding of the mechanisms
involved in the RPC damage process.
Kevwords: biaxial tensile behavior; digital image

correlation; fibres; inclusion; reactive powder concrete
@Seismicisolation
@Seismicisolation
369
370 PIé etal.
Olivier PIé is an associate professor at the Ecole Normale Supérieure de
Cachan, France. He received his PhD in materials science from Grenoble
University. His research, in the Mechanic and Technology Laboratory (LMT),
interests include the numerical modeling of fiber-reinforced and high-
performance cement-based composites. He is actually the sub-director of the
civil engineering department.
Esteban Astudillo de la Véga is a research scholar currently pursuing his

Phû degree in the Mechanic and Technology Laboratory (LMT) at the Ecole
Normale supérieure de Cachan, France. His research interests include
experimental and analytical evaluation of reactive powder concrete reinforced
by fibers under biaxial tension.
Gérard Bernier is an associate professor at the Ecole Normale Supérieure de

Cachan, France. He received his PhD in civil engineering Fom the same
institution in 1981. His research, in the Mechanic and Technology Laboratory
(LMT), interests include expenmental and analytical evaluation of fiber-
reinforced and high-performance cement-based composites.
Olivier Bayard is a research scholar currently pursuing his PhD degree in the
Mechanic and Technology Laboratory (LMT) at the Ecole Normale
Supérieure de Cachan, France. His research interests include the
computational modeling of reactive powder concrete reinforced by fibers
under tension and computational failure mechanics. He is actually the head of
the ductal department from Etudes et Préfabrication Industrielle
(EPVBOUYGUES,France).
INTRODUCTION
Classical uniaxial tests are often insufficient for correct description of

the behaviour of materials in their real applications. It is often necessary to
interpolate the results from uniaxial tests to predict biaxial response.
Generally, design codes rely only on the knowledge of behaviour under simple
loading, whereas, concrete is rarely subjected to uniaxial loading. Few studies
have been done on the strength characteristics of steel fibre reinforced
concrete subject to biaxial loading, particularly for those cases involving
@Seismicisolation
@Seismicisolation
tension (i), Actually, only Uniaxial tension tests on Reactive Powder Concrete
(RPC) have been realised and modelled (2).
This paper describes an onginal process designed to test steel fibre
reinforced concrete under biaxial loading. The experimental results - are
accurately reproduced by means of an original model of inclusion by
considering a locally oriented mass of fibres embedded in an homogeneous
matrix of concrete. With this technique is then possible to describe elastic field
perturbation in the internal morphology of the material. This numerical model
and the experimental results are compared to predict the shape of the yield
surface of crack nucleation in this kind of material under biaxial tensile
loading.
CRUCIFORM TEST SPECIMEN
In biaxial plate tests, loads are applied to the sides of a cruciform plate
specimen. The concept for this type of test can be explained as follows:
1. If we observe the photoelastic fringe pattern on a square plate loaded
biaxially through holes situated near the boundary, the uniformity of
deformations is disrupted by the lateral deformations of the compressed zone.
To create a field of uniform stresses we can introduce slots into arms to reduce
the constraining stresses in the central part of the specimen (3).
2. A uniform thinning down of the middle zone of the cruciform
specimen shows (by simulations using the finite element method) a
homogeneous strain field in this part.
3. The testing region should be sufficiently large to be considered as a
plate zone and must be connected to the loading actuators of the machine by
means of arms.
Thus, the design of a suitable specimen can be performed by
decomposing the test sample into three zones:
1. The central zone (gauge zone) where the level of loading has to be
measured precisely. This level of loading must also be higher than in the rest
of the specimen to assure failure.
2. The fixing zone, where the specimen ends are connected to the
loading actuators of the machine.
3. The connecting zone between the two previous zones, whose main
function is to avoid the mechanical coupling of the external loads which are
applied in two orthogonal directions. The design of this intermediate zone-
must be obtained without undesirable effects due to the geometry.
The selected geometry of the specimen corresponds to a cross with
slots and reduced thickness in the central zone. As there is no standardised
method for this type of testing, it is necessary to optimise the geometry by
considering the machine and the material itself (4,5).
@Seismicisolation
@Seismicisolation
372 PIé et al.
The Machine and the Size of the SDecimen
In order to explore the behaviour of materials under realistic loading

conditions, the LMT has purchased a machine for performing multiaxial tests.
Named ASTREE, this machine developed by LMT/Schenck, is characterked
by three directions of loading (Fig. 1).
The specimen size has to be selected according to the machine to be
used. ASTREE has a maximum distance between loading actuators of 900
mm. Under these conditions, it is necessary to limit the maximum length of
each specimen to 800 mm (in order to accommodate the connection between
the machine and the sample). We propose a central zone (gauge zone) of 100 x
100 mm (large enough with regard to the size of fibres, length = 13 mm and
diameter = 0.16 mm). The fibre length determines also the thickness of the
central zone which will be taken equal to 20 mm.
Shave Optimisation
The initial geometry is a cruciform specimen, containing a flat square

central zone and loading combs. The specimen thickness is proposed to be five
times as wide as the central zone. This ratio of thickness should favour the
concentration of stresses in the central zone. Between these two parts the
transition is obtained with a track geometry based on ellipses.
In a subsequent step, a finite element model of this initial geometry has
been constructed, with stresses determined by the finite element code
CASTEM 2000. For this first calculation, a 3D elastic model of the specimen
has been considered with a Young's modulus equal to 50 O00 MPa and a
Poisson's ratio equal to 0.18. A biaxial state of tensile stress was analysed (GI
= q)with the same conditions in two orthogonal directions. The analysis of
the results from successive 3D calculations allows us to optimise the geometry
in relation with the specimen fabrication.
A numerical simulation in the final geometry, with an elastic-plastic
behaviour, shows the stress level contours. The stresses are uniform
throughout the central part of the specimen and higher than in any other zone
of the sample (Fig. 2).
SpecimenFabrication
Wood moulds were manufactured for casting the samples. The moulds
fully define the outline and the arms of the specimen, and the central part is
@Seismicisolation
@Seismicisolation
finished with the help of a grinding machine. In order to obtain the maximal
performance in term of resistance and durability, specimens are cured at 90°C
for 48h, two days after casting.
TEST SETUP
The specimen is placed in the horizontal plan of the machine and

cemented to the actuator through pairs of metal angle sections, as shown in
Fig.3. Great care is taken in positioning the specimen to minimize any
parasitic flexural effects.
Deformation Measurements
The surface strains were measured in the middle part of the specimen
by means of three strain gauges (30 mm length) placed at O O , 45 O and 90 O.
We used also a full field displacement measuring system based on digital

image correlation (DIC) with an appropriate calibration in order to measure
and visualise the central part of the sample (100 x 100 mm2). Before cracking,
the strain gauges give some indications on the bitensile deformation that
occurs in the middle part (30 x 30 mm2). After cracking the evolution is
followed by means of digital image correlation, with this optical device
mounted perpendicular to the observed facet (Fig. 3).
Equal biaxial tensile loads (F1 = F3) is realised with a coupled

displacement control of the actuators. The loading rate of every actuator is
always constant (about O.Olmm/s). The coupled displacement is interrupted
for a few seconds to permit a video recording of the central part of the
specimen (Fig. 4).
Measurement of Strains Usina Image Correlation
The system includes a camera CCD Kodak (Charge Coupled Device),

coupled with a video card allowing to take and to store images on the hard
disk of a computer. The images of 1360 x 1024 pixels (stored in TIF format,
signal coded on 12 bits, 4096 levels of grey) correspond to the full central
@Seismicisolation
@Seismicisolation
374 PIé etal.
zone of the specimen (100 x 100 mm'). To create a random texture, this zone
is covered with an artificial roughcast made by means of black and white
painting. The displacement plan field is obtained by correlation between two
chosen images.
The tool computer allowing the digital image correlation (DIC)
developed at the LMT is named CORRELI 2D (6). When the crack begins, the
measurement from the strain gauges is stopped, and the DIC measures and
visualizes crack evolution.
Generally, the DIC consists of comparing two chosen images of the
observed surface for two different levels of load, first image is called
"reference" and the second "deformed". The process consist of a comparison
between two images in terms of level of grey (Fig. 5).
The employed algorithm uses the Fast Fourier Transform (FFT).
Strains varying from l o s to are measurable with a very good accuracy.
EXPENMENTAL RESULTS
The figure 6 shows the evolution of the report of efforts FlIF3. We

observe symmetric efforts until a maximal value of 48 kN. After that a fall of
efforts appear indicating the beginning of the damage process.
The figure 7 shows strains measured by strain gauges before crack. We
note, in spite of symmetric efforts, that strains are not symmetrical (EI =
0.00017 and = 0.0008). We could explain this difference by an asymmetry
of rigidity between the specimen and the machine due to cemented
connections.
The evolution of the displacement field is obtained by means of
CORRELI 2D; the measure of strains is also possible. A way of perceiving the
accuracy of DIC in the strain measurement consists of calculating strains in
the zone corresponding to strain gauges. Figure 8 shows complete load-strain
curves. Points A and B both correspond to strains measured with CORRELI
2D from images A and B. Figure 9 shows the displacement field obtained
from the image A corresponding to 41 kN with gauge strains EI = 0.0001 and
E) = 0.000076. The square in the middle of the figure indicates the zone where
the strain gauges are placed. Figure 10 shows the displacement field obtained
from the image B, the last image taken before cracking, the value of the load
was 45 kN and calculated strains were EI = 0.00017 and ~3 = 0.00008.
To try to show from images A and B the zone where the first crack
could appear, we correlated these two images by taking image A as reference
image. The result of this calculation is given in the figure 11 where we note a
disturbance of the displacement field in the right lower comer. The figure 12
shows the evolution of the displacement field after crack. We note that the
crack appears exactly in the disrupted zone.
@Seismicisolation
@Seismicisolation
Maximum strains measured before cracking ( about 0.00017 ) are
lower than those measured in uniaxial tensile load for this kind of material (
about 0.00025 ) (2). However these strains are superior to those of the normal
strength concrete subjected to the same type of loading ( about 0.00008) (3)
and to those of high strength concrete ( about 0.00012) (7).
In the figures 13 and 14 our experimental values of the maximum
stresses and strains before cracking are compared with those from the
literature. These results allow construction of a biaxial tensile strength
envelope for this type of building material.
DEVELOPMENT OF NUh4EFUCAL MODEL
The material (matrix and fibres) is heterogeneous. From this

heterogeneity of the mechanical characteristics, the elastic fields within
matenal before damage are disturbed by the presence of fibres. These fields
within the matrix, as well as the transfers of load, which operate between
fibres and the surrounding matrix, are not only functions of the mechanical
characteristics of each component, but also a function of the local percentage
of fibres as well as local orientation of the fibrous reinforcement. In other
words, the criterion of nucleation is dependent, with local scale (micro-scale),
on the local percentage of fibres and their orientation.
The nature of the fibrous suspension is that of a solution known as
concentrated. That induces, as a consequence, a local alignment of fibres
between them under effect of strong gradients of shearing and a regrouping of
fibres by cluster, flow being no more continuous but being done per block (8).
Size of the clusters that we observe in experiments is approximately 20 to 40
mm according to the mode of casting. This results in a particular configuration
for the distribution of fibres within the structure. If we regard a material as
W C ,its usual field employment is that of the industry of structural elements
with low dimensions and strong streamlined At this scale, the size of the
structure is close to the size of the cluster of fibres. We are dealing with a
composite consisting of two materials, each having its own laws of behaviour.
For this study the authors propose to show the influence of an unspecified
fibre distribution on the mechanical behaviour of a structure. With this
intention, it was selected to break up the structure into "directed" cells (Fig.
15). We will be interested in this part of our article to describe and model the
behaviour of such a cell (assembly will not be discussed).
@Seismicisolation
@Seismicisolation
376 PIé etal.
CELL UNDER TENSION IN ELASTICITY
A Locallv Oriented Mass of Fibres
Let us consider a directed cell isolated from the remainder of the

structure (Fig. 15). It will be supposed that fibre cluster is an inclusion of
factor-form r, = c/a (Fig. 16). This fibre cluster being isolated fkom the other
clusters, we consider it within a matrix of RPC (M) without fibre (f). This is the
problem of an ellipsoidal inclusion, modelled as a transversely isotropic
medium (9, 10) embedded in an isotropic infinite medium of RPC. The
calculation of the rigidity of inclusion defined in the base {1,2,3} by the
formulas of Christensen (1 1) was programmed in computation software named
SARAH, worked out on a Matlab basis. We suppose a volume fraction of
fibres in the cell, f equal to 2%. With the set of parameters { EM= 50 GPa, V M
= 0.18, Ef= 210 GPa, vf= 0.3) we obtain the rigidity for inclusion as:
5.63 1.13 1.30 O O O

1.13 5.57 1.30 O O O
1.30 1.30 6.00 O O O
R~ = l o 4
O O o 4.44 o O
O O O o 4.44 o
The formulations with this type of problem were presented within the
framework of preceding articles (12, 13). The calculation of Eshelby’s tensor
for various ellipsoidal forms was programmed in SARAH. For a factor-form r,
= 1, we obtain:
41.5 0.61 -0.72 O O O

-0.72 - 0.72 6.9 O O O
s = 10- 2
O O O 16.5 O O
O O O O 16.5 O
@Seismicisolation
@Seismicisolation
Localisation
A biaxial loading according to the direction (I) and (III) from the basic
reference is applied to the insulated cell. The strain applied to infinite(far away
from the inclusion) is noted Eœ pig. 16). The strains in the inclusion and in
the matrix at the inclusion-matrix interface (h = O) are given in the global
reference frame. Determination of the principal strains for each point M of
space (in and out the inclusion) is done with help of the relation:
det(s - AI) = O (3)
with A the eigenvalue of a diagonalized system.
These three eigenvalues (EI(M), EZ(M),E~(M))are called principal

strains. The associated eigenvectors are called principal direction and give
information on the direction of cracking. The determination of the principal
strains was camed out for various angle values (a= O", 30" and 45") included
between the direction (3) of fibres and direction (I) or (III) of loading.
The fibre concrete being a pseudo-fragile material having certain
ductility, it seems preferable to approach a criterion of nucleation in strain. We
will take the criterion of Mazars in equivalent strain:
where (E, )+ is the deformation in extension according to direction i.

We can determine for each point M of space, the equivalent strain
E,~(M).With the set ofparameters { EM= 50 GPa, V M = 0.18, E f = 210 GPa, vf
= 0.3, f = 2%, = O", 45" 2 a 1 O"}, the maximum tensile principal equivalent
strain occurs in the inclusion at the fibre-matrix interface (V f) and occurs in
the matrix at the inclusion-matrix interface ( h= O, V 9.
Thus for various cases of orientation a and for a percentage out of
fibres, f = 2%, we consider the distribution of the equivalent strains at
inclusion-matrix interface and in the inclusion at the interface of a fibre and
the matrix. The point of nucleation, Meritchecks a maximum equivalent strain
(noted Eeq). At this point, it is then possible to determine the principal strains,
as well as maximal extension (Fig. 17).
@Seismicisolation
@Seismicisolation
378 PIé et al.
Crack nucleation criterion
A microcrack is assumed to form when the local equivalent strain (14)

reaches the local critical equivalent strain. Restricting the analysis to cracks
which form at the grain-matrix interface and in the inclusion (at the interface
of fibre and matrix), the proposed criterion, as derived from equation (3) and
(4), is:
We can represent the evolution of this criterion with the angle ct (Fig.
18). For an inclusion of factor-form r, = 1, with f = 2% in the cell, the first
crack appears when a = 45". In these conditions the mark of the equivalent
strain in the plane (I), (III) is equal to about 0.67 high enough to nucleate a
crack if the critical equivalent strain of the matrix is equal to (15). This
illustrates and quantifies the localisation of stresses and strains that exist at the
interface between the mass of fibres and the matrix of RPC and in the
inclusion between fibres and matrix.
CONCLUSION
This present analysis is restricted to linear behaviour and focuses on

the formation of a crack in reactive powder concrete (RET) containing short
steel fibres. A first scale modelling provides a description of the micro
structural stress concentration and a model for crack initiation based on stress
intensity factor is obtained. A second scale analysis is investigated
experimentally on specific samples of RPC (see above). An original biaxial
cruciform specimen is achieved and tested under biaxial tension. An adaptive
control test method was designed and digital image correlation is employed to
obtain the displacement field and the microstructural stress concentration. Our
tests show the development of a pseudo-strain-hardening behaviour associated
with a given orientation of fibres. With all other parameters being the same,
we show experimentally that the beginning of damage process induced by
anisotropy can be initiated by the first crack and control by the bridging fibre
action in relation with a local orientation of fibres. These observations support
our theoretical analysis based on Eshelby's inclusion and also give some
indication of the relative influence of the direction of fibres on the mechanical
behaviour of RPC. Despite the restrictions that result from the preliminary
nature of this research, this shows the interest of taking great care in studying
@Seismicisolation
@Seismicisolation
the specimen for fundamental analyses of the processes involved in the
deformation and damage of reactive powder concrete.
ACKNOWLEDGEMENTS
This work is supported by the National contract Réactif / LMT / Bouygues /

Mécanique / BME / 151097. The technical help of BOWGUES was greatly
appreciated. We wish to thank Dr S . Calloch and Dr F. Hild for helpful
comments and discussions on the experimental procedure as well as an
anonymous referee.
REFERENCES
1. Demeke, A.; and Tegos, I., "Steel Fiber Reinforced Concrete in Biaxial
Stress Tension - Compression Conditions," ACI Structural Journal, Sep-Oct
1994, pp. 579-584.
2. Behloul, M., "Définition d'une loi de comportement du BPR," Annales de

Z'ITBTP, No. 532, 1995, pp.122-127. (in French).
3. Kupfer, H.; Hilsdorf, H.; and Rusch, H., "Behavior of Concrete under
Biaxial Stresses," AC1 Journal, Proceedings V. 66, No. 8, Aug. 1969, pp. 656-
666.
4. Richard, P. and Cheyrezy, M., "Reactive powder Concretes with high

ductility and 200-800 MPa compressive strength", ACI Spring Convention,
San Francisco, 1994.
5. Cognard, J.Y., Feuardent, V., Virely, J.M., "Optimisation of a structure for

biaxial mechanical tests," in Integrated Design and Manufacturing in
Mechanical Engineering, P. Chedmail et al. eds. Netherlands, 1997. pp.495-
504.
6. Hild, F., Périé J.N., Coret, M., "Mesure de champs de déplacements 2D par-.
corrélation d'images numériques : CORRELI 2D," Intern report LMT No.
230, december 1999, 101 pp. (in French).
7. Hussein, A.; and Marzouk, H., "Behavior of High-Strength Concrete under

Biaxial Stresses," ACI Materials Journal, Jan-Feb 2000, pp. 27-36.
@Seismicisolation
@Seismicisolation
380 PIe etal.
8. Quemada, D., Rhéologie Phénoménologique des dispersions concentrées,
Rheology of Heterogeneous Fluids, Groupe Français de Rhéologie, 2leme
Colloque Annuel, Strasbourg 1986, pp 1- 13.
9. Eshelby, J.D. 1957 "The determination of elastic jìeld of an ellipsoidal

inclusion and related problems", In. Proc. Roy. SOC.Lond., Vol. A241, pp.
376-396.
10. Mura, T. 1982. "Micromechanicsof Defects in Solids", Mariinus Nijhoff

Editor.
11. Christensen, R.M. 1979. "Mechanics of composite materials", John Wiley

and Sons Editors.
12. Bayard, O., Plé, O., An analysis of crack nucleation in Fibre-Reinforced

Concrete, Simulation and Testing in Micro Materials, 3, B. Michel T. Winkler
M. Werner and H. Fecht editors, Druckhaus Dresden GmbH publisher, (2000)
pp7 18-721.
13. Plé, O., Bayard, O., Preliminary study of multiscale analysis in fibre
reinforced concrete, Material and Structures, RILEM Publications, 2001, (in
press).
14. Mazars, J., Application de la mécanique de l'endommagement au

comportement non linéaire de la rupture du béton de structure, Thèse d'état,
Univ. Paris 6,1984.
15. Richard, P and Cheyrezy M. 1994. "ReactivePowder Concrete with high

ductility and 200-800 MPa compressive strength", AC1 Spring Convention,
San Francisco.
@Seismicisolation
@Seismicisolation
Fig. I-ASTREE machine with three directions, only axis (I) and (III) are used in biaxial
tension.
Fig. 2 4 u m e r i c a l simulation of the stresses in the final geometry.
@Seismicisolation
@Seismicisolation
382 PIé et al.
Fig. 34onjìguration of the test and the sample.
O 20 40 60 80 100
Time (sec)
Fig.4-îomplete biaxial displacements - time curve.
@Seismicisolation
@Seismicisolation
- 5 4 e a s u r e m e n t of deformations using image correlation, A) subimage for

Fie.
measurement, B) subimage displacement.
IQ
a
4.5
a)
35
3
25
20
15
10
5
O FI (W
o 5 m i ~ m u ~ u a ) a a
Fig. 64volution of the report of efforts in each direction.
o a,m5 opml &mois opmi
Fig. 7 S t r a i n s measured by strain gauges in each direction.
@Seismicisolation
@Seismicisolation
384 PIé et al.
15
10
5
O
O O,oWO25 0,00005 O,ooOo75 0,0001 0,000125 0,00015 0,000175
StraLi
Fig. 8<omplete load - strain curves.
J-
Fig. 9-DisplacementJield from image A.
Fig. 1O-Displacementfield from image B.

r'3
JI r .....................
.....E
.....................
.. .. .. .. .. .. .. .. .. .. .................
. ..... . . . . . . . . . . . . . .
.. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ..
.. .. .. .. .. .. .. .. .. .. .. ............
..........
:If
....................
:. .: :. :. . j ; ; ; 1 ; ; ; ; :
.. ....... . . . . . . . . . . . .
1:
-r
3
.. .. ....... .. .. .. .. . . . . . . .
......... :.::
Fig. I l A r n a g e A vs image B. Fig. 1 2 4 v o l u t i o n after crack.
@Seismicisolation
@Seismicisolation
Fig. 134xperimental values of the maximum stresses before crack.
-.-N m I Stnigth Concwe (35 MPa)

WEI<
et al i969
1.8
- - BW (loo wo)
HVSKIN C i I. xm
Fig. 144xperimentaI values of the maximum strains before crack.
B A
...
.. .......... .......... ....
.........
I
Fig. 1 5 4 i b r e distribution shape, A) complete shape: assembled cells in intëraction, B)
partial shape: isolated cell.
@Seismicisolation
@Seismicisolation
386 PIé et al.
A d3
/
B
isotropic plan
4/ cell wiîh fibre volume

üaction f=2 %
...........................................................................................
.. 3 ,i
....;:.
...... i
,/
,..'
......
,..'. .
!
!
............ .......,........
.......
I^......................
i "I .
\
matrix of reactive
powder concrete (M)
E"
Fig. ló-hclusion model, A) inclusion ofjìbres with the symmetry axis 3, B) the inclusion
model in the global frame ('IIJII).
@Seismicisolation
@Seismicisolation
I
a5 1
&li&=
Fig. 1 7 4 a r k of the maximum equivalent strain for different values ofa.
Si/&
Fig. 18-Mark of the crack nucleation criterion in equivalent strain.
@Seismicisolation
@Seismicisolation
SP 209-21
Behavior of Concrete Liquid

Containing Structures Subjected
to Seismic loading
by M. R. Kianoush, W. K. Tso, and M. Hamidi
SvnoDsis: The behavior of liquid containing structures (LCS)subjected to seismic

excitations is reviewed. The major parameters affecting the response of concrete
circular tanks for LCS are discussed. Existing codes and standards related to seis-
mic design of LCS are reviewed. With the aid of a design example, results of the
various design standards are compared. The effects of earthquake load on the be-
havior of reinforced concrete (RC)tanks are also investigated through a detailed
example.
Kevwords: circular tank; concrete; design code;

earthquake; liquid containing; seismic; water retaining
@Seismicisolation
@Seismicisolation
389
390 Kianoush et al.
AC1 member, M.Reza Kianoush is a Professor of Civil Engineering at Ryerson
University, Toronto, Canada. He received his Ph.D. from the University of Alberta in 1986.
He is a member of AC1 Committee 350 as well as the Chair of Education Subcommittee.
His research interests include analytical modeling of reinforced concrete structures as well
as behavior of concrete liquid containing structures under seismic loading.
W.K. Tso is Professor Emeritus of Civil Engineering at McMaster University, Hamilton,

Canada. He obtained Masters and Ph.D. degrees from California institute of Technology.
H. Hamidi is a Research Associate at Western University, London, Canada. He obtained

his Ph.D. from Sharif University of Technology, Tehran, Iran in 2000. His research inierests
include analytical modeling and design of struciures with a particular emphasis on seismic
performance.
INTRODUCTION
Liquid containingstructu~s (LCS) for environmental engineering concrete (EEC) structures

are considered as lifeline facilities that require careful design and detailing. The
performance of LCS during earthquakesis more important than the economic values ofthe
tanks and their contents. It is important that utility facilitiesremain operationalfollowing an
earthquake to meet the emergencyrequirementssuch as tirefighting water or meet the public
demands as a source of water supply. On the other hand, the containment of hazardous
waste is important to meet the public safety after a seismic event. For these reasons,
serviceability(crack and leakagecontrol)becomes the prime design consideration in most of
these structures. Cracking in these structures should be limited to micro-cracks.
There has been considerableresearch carried out on the seismic behavior of steel circular tanks.
However, the behavior of concrete circulartanksfor LCS during earthquakesis quite different
from that of steel circular tanks. The most common mode of failure in a circular steel tank is
by buckling ofthe wall near its base that is commonlyreferred to as "elephant's foot" buckling.
The main cause of this type of failure is due to large overturning moments caused by
hydrodynamic effects. However, in concrete tanks, large hydrodynamic forces could develop
in the highly stressed regions of the tank which could lead to severe cracking and leakage
which may eventuallylead to failure of the tank.
Fig. 1 (a) shows the model of a circular concrete tank under consideration. Under horizontal
seismic excitations, the hydrodynamic forces in the tank can be induced due to two effects.
First, the bottom portion of the liquid in the tank accelerates in unison with the walls of the
tank. This mass is shown as Mi and is called the impulsive mass. This system of structural
response behaves in a high frequency (short period) mode of vibration. Second, the upper
portion of liquid in the tank behaves as if an oscillating mass, &,flexibly connected to the
walls. This is known as the convective mass. This system of response behaves in a low
frequency (long period, T > 3.0 sec) mode of oscillation with damping of the order of 0.5%.
@Seismicisolation
@Seismicisolation
The impulsive component of the mass generates pressure and suction, which acceleratesinto
the fluid and away from the fluid, respectively. These effects will produce the force PI which
acts at a distance hi from the bottom of the tank and is evenly divided on the water side of the
walls as shown in Fig. l@). The convective pressures on the tank wall induce the resultant
force Pc,which acts at a distance of b from the tank bottom. Since the impulsive and
convective masses oscillate at different frequencies, it is reasonable to decouple the two
components and treat them separately.
The mathematical model described above was originaily developed by Housner [i] that
assumes that the tank is rigid. This model was later modified by Veletsos [2] by assumingthat
the tank-liquid system behaves as a single degree of freedom system. The analytical study by
Veletsos and Yang [3] showed that the impulsive forces in flexibletankscould be considerably
higher as compared with rigid tanks. However, the convective components were not affected
due to the flexibility of the tank.
The tank model described above is valid when the tank is relatively shallow ( i e . H / R < 1.5 ).
However, for deep tanks (ix. H/R > 1.5), a slightly different model needs to be adopted.
Housner [i] showed that for deep tanks, generation of liquid velocity relative to the tank is
essentially restricted to a height of 1.5 R from the free surface. The mathematical model for
these types of tanks is described by Epstein [4]. Most of the ground tanksbuilt in Canada for
EEC structures are of shallow type. For this reason, the work described in this paper is
restricted to shallow tanks. However, minor modifications will make it applicable to deep
tanks.
in this paper, the major parameters that affect the response of concrete circular tanks for LCS
subjected to earthquake loading will be investigated.The existing documentsrelated to seismic
design of LCS shall be reviewed. Based on such a review, recommendationsshall be made on
the seismic design criteria for LCS.
Response of Tank-Liquid System
Tank parameters
Based on the Housner’s tank model described above, the impulsive and convective weights (W,
and WC) and the equivalent heights (hr and b)are given by:
@Seismicisolation
@Seismicisolation
392 Kianoush et al.
CS] h, = 0.375 HL
Gosh ß - 1
[4] h, = (1 - - )HL.
ß sinhß
where ß = 1.84 (HL

/ R) and WLis the total weight of liquid in the tank.
The impulsive and convective forces (PI and Pc) produce bending moments in the tank wall
just above the base. However, to determinethe overturningmoment at a sectionjust below the
base of the tank wall, the dynamic pressure distribution on the bottom of the tank should be
considered. For this purpose new equivalent heights must be used. These equivalent heights
are defined in the TID-7024 report [5].
Assuming that the tank wall is rigid, the maximum horizontal force exerted against the tank
due to the impulsive and convective weights of liquid are:
where Ügis the maximum ground acceleration and Ü is the maximum acceleration of the
sloshing(convective)mass. However, if it is assumed that the tank wall is not rigid, the effect
of the tank flexibility needs to be considered in equation 5. For this purpose, the period of
vibration of the tank-liquid system has to be established.
Period of Vibration of Tank-Liquid System
imuulsive comDonent
A tank wall can vibrate in several different modes if assumed that it behaves elastically.
Representativemodes of vibration for a cantilever circular tank wall are shown in Fig. 2. The
axially symmetrical mode corresponding to breathing responie (n=O) is exited by vertical
motion only and produces hoop stresses in the tank wall. Of interest here is the bending
response(n=l) of the tank,which is excited by horizontal motion. The frequency of this mode
is affected by the impulsivemass of the liquid and the geometric and material properties of the
tank itself.
The natural frequenciesof circular tanks can be obtainedusing closed form solutions. Yang [6]
used Flugge’sshell theory to determinethe natural modes of vibrationof cylindrical steeltanks.
The effect of the contained liquid was approximated by assuming that a portion of the mass of
the liquid was attached to the shell. Based on height over radius ratio (i.e. H / R), the behavior
@Seismicisolation
@Seismicisolation
of cylindricaltanks was described in terms of four simple types of structural actions that are
summarized in Table 1.
For concrete circular tanks under consideration, i.e. H / R-d.5, the behavior will be mostly
similar to type 1 or 2 action. The fundamental natural frequencies for this mode of behavior
(type 2) as given by Yang [6] are:
Empty Tank
Full Tank
Where. t is the wall thickness, E and v are the modulus of elasticity and the Poisson’s ratio of
concreterespectively.pL,ps and EL, ESare the mass density and the virîuai mass coefficients for
the liquid and the tank respectively [6].
A similar expressionfor the natural frequency of fluid-tank system has also been developed by
Haroun and Abou-Izzadin [7]. This is given in terms of non-dimensionalparameter ?5 that is
based on a best-fit formula as follows:
[lo] G = [O.O00335(H/R)’ -0.000021(M¿)2 - O.O16361(WR) + 0.0655981
Convective ComDonent
As mentioned earlier, the flexibility of the tank wall has little or no effect on the convective
forces. However, to determine seismic forces, the frequency of the convective component of
the liquid has to be established.
The frequency of the mass-spring system shown in Fig. 1 is given by:
1.84gtanhß
[ l i ] wz =
R
@Seismicisolation
@Seismicisolation
394 Kianoush et al.
Design of Tank-Liquid System
Review of existing codes and standards
Currently, there are few provisions in the North American Building Codes related to seismic
design of LCS. The Uniform Building Code, UBC [8] addresses seismic requirements for
liquid holding tanks in a very general manner. Recently the AC1 Committee 350 [9] has
developedan independentcode for seismicdesign of LCS for EEC structures. in addition,this
Committeehas developeda separate document [101in a mandatoryform for seismic design of
LCS. Also, there are some standardsthat provide guidelines for determiningseismicforcesfor
LCS. The US. Atomic Energy Commission Report, TID-7024 [5] and the NSF report [111are
most widely used in practice. It is believed that tanks that have been designed based on these
two.standardshave performed well under earthquakes. In these two standards, there are no
specific statementsrelated to concretetanks. The only comprehensive code otherthanAC1 350
[9] that is related to concrete LCS is the New Zealand Standard, NZS 3106 [12].
in most current codes and standards,the determinationof tankparameters is based on Eqs. 1 to

4 and 11. Table 2 shows a summary of a comparison of some of the current codes and
standards for determination of the seismic base shear. in this table, some of the symbols are
slightly adjusted for better comparison.
Both the TID 7024 report and the NSF report are based on the rigid tank assumption. The TID-
7024 method is based on average velocity response spectrum, which requires informationfor
the particular site. The NSF report is based on the provision of the ATC-3-06 [131. Based on
this report, two alternative design approaches can be used, namely the working stress design
(WSD) method and the strength design (SD) method. The coefficients, CWand CS,control the
performancecriteria of the tankfor WSD and SD, respectively. For examplefor CW=0.8, shut
down of the facility is acceptable up to 6 months and for Cw = 2.4, continuous operation is
essential. In both reports, the total base shear is determined by simply adding the forces
obtained from the impulsive and the convective components.
The NZS [121 is also based on rigid tank assumption for circular tanks. The effect of tank
flexibility is considered only in rectangular tanks.A risk factor, R, is used which is based on
the performance criteria of the tank and the annual probability of exceedance (APE) of the
design eadhquake. For example, R = 1.O is for tanks which should remain functional in the
restoration period for a moderate earthquake based on APE of 0.01 and R = 2.0 is for tanks
containing highly valuable or hazardous contents based on APE of 0.0004. The impulsive and
convective components of forces are combined by the square root of the sum of the squares
(SRSS) rule. Since the two components oscillate at two different frequencies, the use of this
nile is quite valid. In the proposed AC1 350 method, the flexibiliîy of the tank wall is
considered using the CI parameter.:,The basic equation for the seismic base shear is similar to
the NSZ standard. However, different structural system factors are used for impulsive (Rw)
and convective (Rwc) components. Rw varies from 2.0 to 4.5. For example, anchored
@Seismicisolation
@Seismicisolation
flexible-basetanksand nontlexible-basetanks(fixed or hinged) are assigned RWIvalues of 4.5
and 2.75 respectively. The value of RWCis set to be 1.0. This is because the convective
component of the liquid has a high natural period of vibration (T > 3 seconds) in which the
system will respond in the displacement sensitive region of the response spectrum curve. In
this case, the maximum deformationwill be the same as the deformation of the linear system
and therefore Rwc = 1 .O. The importance factor I = 1.O is used for tankswhich may undertake
repairable structural damage and I = 1.25 are for tanks that must remain usable with slight
structural damage. For tanks containing hazardous materials I is assigned a value of 1.5.
Desirn example
Consider an open top circular concrete fixed-base unburied tank having the following
dimensions:
R = 15 m, H = 6 mand t = 0.3 m.
The tank is assumedto be full of water of density= 1O00 kg/m3. Concrete has a mass density of
2400 kg/m3 and a modulus of elasticity of 2x10'' N/m2. Furthermore, it is assumed that
importancefactor for the reservoir is I=l .O, the type of soil under the tank is rock (S=l .O) and it
is located in a high-risk seismic zone (24.4).
The design forces according to four design standards described earlier are computed through
formulas in Table 2, and the design forces are summarizedin Table 3. The comparisonbetween
different design methods indicates that all of the methods give forces greater than that of
ACI-350 (they are 40 to 60%more conservative than ACI). It is generally accepted that the
reservoirs designed according to TID or NSF are considered as very conservative designs.
Earthquake Load Effects on RC Tanks
To practicallyunderstandthe earthquakeload effect on the RC circulartankwalls, a parametric

study is conducted to study the effect of the ratio of earthquake load to the hydrostatic load.
This study was done using the AC1 350 Code (2001).Earthquakemoment MEQ,static moment
Ms, earthquakebase shear VEQ,static base shear Vs are calculated based on the code procedure.
Variables used in this study include the height, HL (3 to 10m) and the thickness, tw(300 to
6 0 0 m )of the tank wall and the seismic zone factor Z. The radius of the tank in this study is
considered as a constant parameter equal to 20 meters.
The results of the study are shown in Figs. 3 to 5 . Detailed comparison of the results leads to
the following observations:
1. The effect of the earthquake moment MEQis more significant than that of earthquake
shear VEQComparisonof Figs. 3 and 4 shows that the ratio Of M E Q Mis~consistently greater
than VEQNSfor different tank wails. Therefore, the effect of the moment due to earthquakeon
the structure is more critical than that of shear to the structure.
@Seismicisolation
@Seismicisolation
396 Kianoush et al.
2. The earthquake effect is more critical for shallower tank walls (3 to 5m) than that of
taller tanks (6 to 1Om). As shown in Figs. 3 and 4, ME&S and VEQNSdecrease as the height
of the wall increases. Therefore,it can be concludedthat for a relatively shallow tank (3 to 5m)
the earthquakeload effect is more significantthan that of static load. This is because static base
shear, VS,increases with the parabolic of the water height and the static moment MSincreases
with the cubic of the water height, while the earthquake base shear, VEQ,and moment, MEQ,
increase fairly linearly. Thus, for taller tanks, (e.g. wall height of 10m) the static shear and
moment, Vs and Ms, are the major parameters that affect the response of the tank.
3. While the earthquake load effect is significant for designing in seismic Zone 4 (with
Z=0.4), the earthquake effect also needs to be considered in designing in seismicZone 2B (with
Z=0.2). However, as indicated in Fig. 5 , the effect of the static moment is more critical than
that due to earthquake load effect.
4. The effect of tank wall thickness on the earthquake load is negligible. Figs. 3 and 4
show that the MEQ/MSand VEQNSare not so sensitiveto wall thickness changes. The effect of
wall thickness becomes more significant only in very tall tanks.
Conclusion
The simplified approach, developed originally by Housner for analyzing liquid containing
structures is adopted by most of the design standards to estimate the design seismic forces.
Based on the usage and importance of the tanks and its content, these standardsuse greater or
smaller coefficientsfor the design. As the example in this paper showed, significantdifferences
between different standards exist.
A parametric study on the effects of the earthquakeon the circular tanks showed that:
- In designing of the tank walls, earthquake induced bending moment is more critical
than shear.
- Earthquake load dominates the design of shallow water tanks, but for taller tanks,the
static load becomes the dominating factor.
- Earthquake load is not affected significantlyby the variation of wall thickness.
- Although, in the seismic Zone 2B, the static load dominatesthe design of the tank,but
the seismic load cannot be disregarded, due to its effect as compared with static loads.
References
1. Housner, G.W. 1957. Dynamic pressures on accelerated fluid containers, Bulletin of the
SeismologicalSociety of America, 47(1): 15-35.
2. Veletsos, A.S. 1974. Seismic effects in flexible storage tanks, proc. int. assoc. for
@Seismicisolation
@Seismicisolation
earthquakes eng., Rome, Italy, pp. 630-639.
3. Veletsos, AS., and Yang, J.Y. 1977. Earthquake response of liquid storage tanks. Proc. of
2nd Annual ASCE EngineeringMechanicsSpecialtyConference,North Carolina StateUniv.,
Raleigh, N.C., pp. 1-24.
4. Epstein, H.I. 1976. Seismic design of liquid storage tanks. ASCE Journal of Structural
Division, 102 (ST9): 1659-1673.
5. TiD-7024,1963. Nuclear reactors and earthquakes, Prepared by Lockhead Aircraft Corp.

and Holmes and Nurver, Inc. U.S. Atomic Energy Commission.
6. Yang, J.Y. 1976. Dynamic Behavior of Fluid-Tank System. Ph.D. Thesis, Rice University,
Texas.
7. Haroun, M.A. and Abou-Izzeddine, W. 1992. Parametric study of seismic soil-tank

interaction I: Horizontalexcitation, ASCE Journal of Structural Engineering, 118(3): 783-797.
8. Uniform Building Code, 1995. International Conference of Building Officials. Whittier,

California.
9. AC1 350-01,2001. Code requirements for environmental engineeringconcrete structures and

commentary,American Concrete Institute, Detroit, Michigan.
10. AC1 350.3/350.3R-01,2001. Seismic design of liquid-containing concrete structures and

commentmy. American Concrete institute, Detroit, Michigan.
11. NSF Report, 1980. Earthquakedesign criteriafor water supply and wastewater systems. A
National Science Foundation Report, Environmental Quality Systems Inc., Rockville,
Maryland.
12. New Zealand Standard, 1986. Code of practice for concretestructuresfor storage of liquids.
NZS 3 106:1986,New Zealand.
13. ATC-3-06, 1978. Tentative provisions for the development of seismic regulations for
buildings, Applied Technology Council, National Bureau of Standards Special Publication -
510,National Science Foundation Publication 78-8.
@Seismicisolation
@Seismicisolation
398 Kianoush et al.
Table 1. Effect of H/R Ratio on Structural Action
Type WR Structural Action

1 < 0.2 Bending and shearing vibrations. Each vertical strip of tank wall
behaves as a cantilever beam in either bending or shearing
vibration
2 4.0 Extentional vibration of a half circular arch. Each cross section
of the tank behaves as a circular arch with angle of opening
of 180' fixed at both ends.
3 Medium H/R Shearing vibrations of a cantilever beam mode shape which is
values say similar to a half sine curve.
1.5
4 1 > 5.0 I Bending vibration of a cantilever beam.
@Seismicisolation
@Seismicisolation
Table 2. Lateral force requirement for liquid containing structures.
rID - 7024 NSF (1980) NZS (1986) AC1 350-01

(1963) (2001)
.~
Base Shear ü
"W,
1.2CA, ZISC,
WI WI
impulsive (VI) g 2! RWI
Base Shear A, ío2 1.2c c ZISC
-WC wc ccwc d ... W c
convective (VC) g 2R Kwc -
Base Shear VI -k vc Jy' t vc'

Total
Numerical ï)= circular

Coeflicients kequency of
dosh liquid
üo=
max. A, = 0.05 to A, = 0.4R Z = 0.075 to
yound accl. 0.4 0.4
Zone Factor 0.3R
A, = 0.05 to 0.2R
4, = max.
iisplacement 0.4
Cl =Ao CI =s
2.75
Seismic or
Response 1.25
Cl =-
Factor 2SA, S 2.7A0 J=3
Cl =- Cc =-
Tc" TY
6
Cc =z
TC
C=C,=l.O
Importance 2.0
3.0 I = 1.0
Factor
c = c, = 0.8 to
1.6 R = 2.0 1.5
2.4 1.6
1.3
1.o Rw = 2.0
Structural R = 2.0 to
0.7
System 4.5
Factor
Rwc = 1.0
Soil Factor --- S = 1.0 to 1.5 --- s = 1.0 to 2.0
@Seismicisolation
@Seismicisolation
400 Kianoush et al.
Table 3. Comparison of Design Parameters
Base Shear (kN)

Base n ~ m e n t Coefficients
w-m)
Impulsive Convective Total
TID-7024 5475 2083 7558 18840

(1963)
NSF (1980) 5257 1521 6779 17776 C w = 1.6
NZS (1986) 7121 2598 7580 19403 R = 1.3
ACI 350-01 4560 1372 4762 11621 Rw = 2.75
, (2001)
@Seismicisolation
@Seismicisolation
L -
-
-
Kf2
1% 1 L
HL
-=+Y 1 s hc
4
T A -
(a) TankModel
0.5PC o*pc 0.5Pc

0 . p ~
__t ~ p c + p P - - =
40.5PI 0.5Pqcp* OSP,

4
Fig. 1. Housner's Dynamic Tank Modei
@Seismicisolation
@Seismicisolation
402 Kianoush et al.
n= 1
o o
n=û
n=2 n=3 n=4
Fig. 2. Modal Response of Flexible Circular Tanks
@Seismicisolation
@Seismicisolation
-tiw = 300 mm
-i-îw=400mm
+iw=500mm
3 4 5 - 6 7 8 9 1 0
Height of the tank wall (meter)
Fig. 3. Variation of M@4s with the Height of the Tank
-F
.
3
U
1 +tw=400rnm
+tw = 500 rnm
2 05 -H-tw=600mm
O ?
3 4 5 6 7 8 9 10
Height of tank wall (meter) I
-~ ~~
Fig. 4. Variation of VEQNSwith the Height of the Tank
@Seismicisolation
@Seismicisolation
404 Kianoush et al.
2.5
.-co 2
i 1.5 +Zone,2B
I
-U 1 +Zone 4
al
0.5
O I I
3 4 5 6 7 8 9 10
Height of tank wall (meter)
I
Fig. 5. Seismic Zone Effect on M E ~ S
@Seismicisolation
@Seismicisolation
SP 209-22
Three Misunderstandings in
Uniaxial Tension Test of Concrete
by H. Akita, H. Koide, M. Tomon, and S. M. Han
Svnopsis: The best way to investigate the tension softening process is by applying
a uniaxial tension force directly on a concrete specimen, because it can measure
both tensile strength and the tension softening curve from an identical specimen.
However, no standard tests have been adopted to provide a direct measurement of
the tension softening curves of concrete. There are three misunderstandings in in-
vestigating the tension softening process, concerning the effect of secondary flex-
ure, boundary conditions and notches. Because of these, many inadequate test pro-
cedures were proposed until recently. In this paper, the misunderstandings are
discussed and clarified in detail with theoretical and experimental considerations.
A test procedure for the uniaxial tension test of concrete is proposed with success-
ful test results.
Kevwords: concrete; notch; secondary flexure; tensile

strength; tension softening; test procedure; uniaxial
tension
@Seismicisolation
@Seismicisolation
405
406 Akita et al.
AC1 member, Hiroshi AKITA is a Professor in the Department of Civil
Engineering at Tohoku Institute of Technology. He obtained his D. Eng. From
Tohoku University in 1992. His research interest relates to the fracture
mechanics of concrete.
AC1 member, Hideo KOIDE is an Associate Professor in the Department of

Civil Engineering at Tohoku Institute of Technology. He obtained his D.Eng.
From Tohoku University in 1990. His research interest relates to the fracture
mechanics of concrete and the reliability analysis of concrete structures.
AC1 member, Masanao TOMON is a Professor in the Department of Civil

Engineering at Tohoku Institute of Technology. He obtained his D. Eng. From
Tohoku University in 1992. His research interest relates to the fracture
mechanics of concrete.
San Mook HAN is a post doctoral fellow in the High Technology Center at
Tohoku Institute of Technology. He obtained his D. Eng. From Tohoku
University in 2001. His research interest relates to the utilkation of
cementitious materials and recycling of concrete.
INTRODUCTION
The knowledge of the tension softening process of concrete is essential to

clarifying the fracture mechanism in analyzing fracture behavior and in the
evaluation of the properties of concrete. By using tension softening curves
with fictitious crack model proposed by Hillerborg (i), nonlinear finite element
analysis enables accurate prediction of deflection, crack width, bond transfer
and shear transfer in concrete structures. The best way to investigate the
tension softening process is by applying a uniaxial tension force directly on a
concrete specimen, because it can measure both tensile strength and the tension
softening curve from an identical specimen. These advantages are not found in
other test methods such as 3 point bending, compact tension and wedge splitting.
NO standard tests have been adopted to provide a direct measurement of the
tension softening curves of concrete. There are three misunderstandings for
the test procedure; namely the effect of secondary flexure, boundary conditions
and notch. Because of these, many inadequate test procedures have been
proposed over the past 30 years, and some of them have adopted quite opposite
test conditions. For example, a secondary flexure was completely eliminated
by Carpinteri, et al. (2) and the authors (3), was left to develop freely (4)-(6), or
@Seismicisolation
@Seismicisolation
attempted unsuccessfully to prevent (7)-(8). Such misunderstandings led to
some serious arguments and prevented the establishment of a standard test
procedure. in order to remove obstacles to the establishment of a satisfactory
method, the misunderstandings are discussed and clarified in detail with
theoretical and experimental considerations. A test procedure for the uniaxial
tension test of concrete is proposed with successfultest results.
SECONDARY FLEXURE
When a concrete specimen is subjected to tensile loading, two types of

unexpected flexure occur. One comes from load eccentricity and the other
from the heterogeneity of concrete. The latter flexure, to be denoted as
secondary flexure, is caused by the local arrests of crack propagations by
aggregates. It generates strain or deformation gradient along the cross section
and consequently decreases the measured peak load significantly. The
knowledge of the peak load reduction is essential in estimating the tensile
strength of concrete for the uniaxial tension test, because tensile strength is
usually calculated by dividing the peak load by the ligament area. As the peak
load reduction is significant and unpredictable, it is a serious misunderstanding
that no flexure occurs when only the load eccentricity is excluded in the uniaxial
tension test.
In order to investigate the actual reduction of peak load or tensile strength, the
experiment in which a secondary flexure was left to develop freely was
performed using prismatic specimens of 100x100x400 mm. In Fig. 1, the
experimental and analytical results are shown with respect to tensile strength.
The true tensile strength, ftmo in the experiment was assumed as the average
from 4 or 5 specimens being tested by complete elimination of the secondary
flexure as described later in the “RECOMMENDED TEST PROCEDURE”
section. As two orthogonal deformation gradients are obtained from two
combination within 4 side deformations 61, 6 2, 6 3 and 6 4 when the secondary
flexure is left to develop freely in the experiment, the maximum deformation
gradient, grmax,is then calculated as follows for the test specimen:
where 61 and 63, for example, are mutually opposite side deformations, and 100.
indicates width and depth of the prismatic specimen. The magnitude of the
experimental deformation gradient produced by the secondary flexure cannot be
predicted beforehand and Fig. 1shows that the deformation gradient is scattered
widely. The figure also shows that tensile strength reduction due to secondary
flexure easily exceeds 10% and sometimes exceeds even 20%. These serious
@Seismicisolation
@Seismicisolation
408 Akita et al.
reductions of tensile strength are the reasons why the elimination of a secondary
flexure is significant in uniaxial tension test of concrete.
BOUNDARY CONDITIONS
In order to prevent the secondary flexure, fixed boundaries were proposed and
specimen ends were glued to loading platens in some studies. However, it is
also a misunderstanding that the fixed boundary can be realized by gluing
specimen ends to the platens. AS the rotation of loading platens allow a
deformation gradient, even small rotation about lo5 radian will destroy the
fixed boundary and easily produce the secondary flexure (See Fig.1). In fact,
the deformation difference of mutually opposite faces of the specimen can be
observed at the peak load in such experiments (8). The deformation difference
was also observed even when the rotation of both platens was restricted by using
a guiding system (7). Thus, the fixed boundaries cannot be realized by gluing
the specimen ends to the loading platens, and the secondary flexure cannot be
eliminated by this procedure.
When a specimen is stiffened by steel bars at both sides, the k e d boundary is

realized. However, relatively smail rigidity of specimen cannot prevent the
secondary flexure when the specimen is fairly long, as reported by Hordijk, et al.
(7). Whereas, a short specimen cannot guarantee uniform stress field in the
cross section near the middle height. However, only by maintaining equal
mutually opposite side deformations, the fEed boundary can be produced and
the secondary flexure can be eliminated.
NOTCH EFFECT
The last misunderstanding is that the strong stress concentration due to notches
disturbs the observation of an exact tensile strength. However, stress
concentration occurs only in the elastic range and diminishes when the notched
section becomes softened. This was reported by Zhou (9), and consistent
results were also reported by the authors (10). Fig. 2 shows the stress
distributions of notched section in the prismatic specimen of 100x100x400mm
when the notch depth is 5 mm. Each stress distribution is expressed in terms of
tensile stress, o, versus tensile strength, fi. P means applied load and Pt,,
means the true peak load that is realized when the entire ligament is subjected.
uniformly to tensile stress equal to tensile strength. Because of a strong stress
concentration caused by notches, stress level near the notch tip, even at low
loading such as when P/Pm, = 0.35, reaches the tensile strength. AS the load
increases, a fictitious crack, that models the softening zone, develops and
expands from the notch tip toward the center. The size of crack opening
@Seismicisolation
@Seismicisolation
displacement (COD) within the fictitious crack is so small that each cohesive
stress within the zone reduces little from tensile strength. When PP,,
becomes 0.79 , the stress distribution consists of an expanded plateau near the
notch and a concave in the middle. When the crack tip reaches the specimen
center, P/Pmc becomes 0.99 and a single plateau across the whole cross section
is created, resulting in softening of the entire cross section. The peak load is
reduced little, compared with the true peak load in spite of the existing notches.
These are the reasons why almost exact peak load or consequent tensile strength
is obtained by using a notched specimen in uniaxial tension test. On the
contrary, unnotched specimens give inadequate results, because multiple cracks,
as shown in Fig. 3, and accompanying secondary flexure in each cracked section
signjñcantly reduces the measured tensile strength.
RECOMMENDED TEST PROCEDURE
Fig. 4 shows an example of the prismatic specimen used for the uniaxial tension
test. Two kinds of notches, named primary notch and guide notch, were
applied in the center of the specimen. The role of the primary notch is to
prevent multiple cracks, and that of the guide notch is to prevent overlapping
cracks (Fig. 3). The recommended depth of the primary notch is 10 mm, while
that of guide notch is 5 mm for the proposed specimen size. A 90-mm
extensometer was attached on ali four side faces aligning at the center. Four
deformations obtained from the extensometers were used for both the
elimination of the secondary flexure during the test and load-deformation curve
after the test. The average of the four deformations is reflected to load control
by the so-called closed-loop loading machine. Fig. 5 shows the experimental
set up. Fig. 6 illustrates one of the adjusting gear systems that is attached at the
horizontal end of the steel arm of all four faces. The gear system works to
eliminate both the secondary flexure and the flexure caused by load eccentricity.
This elimination was executed in such a way that the more elongated side was
given sufficient contraction to reach a proper balance by tightening the
concerned adjusting gear system during the test. The adjusting gear system is
tightened manually with a handle (Fig. 6), or automatically by the computer
controlled DC motor (Fig. 5). The both types of adjusting gear systems were
s p e c i ñ d y designed by the authors (3).
RESULTS AND DISCUSSIONS
Similar results were obtained for more than 2 hundred specimens; the
representative results are presented in this paper. Fig. 7 shows typical histories
of the machine load and the largest adjusting force within the four steel rods.
After peak load, the machine load decreased almost to zero during a hundred
@Seismicisolation
@Seismicisolation
410 Akita et al.
minutes. This means that a stable fracture was produced by this test procedure.
The magnitude of adjusting force is about 3% of machine load, indicating that
the secondary flexure is easily eliminated by relatively small forces.
Fig. 8 shows load-deformation curves (P-6 curves) from the same specimen,
using the average of the four side deformations as 6. In the figure,
“uncorrected” means simple machine load and “corrected” means applied load
which is the machine load subtracted by all adjusting forces, because those
forces give compression to the specimen. Both curves coincide with only a
smail difference, depending on the small adjusting forces (Fig. 7).
Fig. 9 shows P-6 curves, not for average deformation but for each individual
deformation, where ch-2 and ch-4 express the deformations of the two primary
notched faces. Both curves coincide, showing that any secondary flexure has
been eliminated. In addition, these curves also coincide to the “corrected”
curve in Fig. 8.
From the P-6 curves, tension softening curves (o-wcurves) are derived. Crack
opening displacement, w, is obtained from the following equation:
PL
w = ô - -
EA
- 6,
where 6: observed deformation, P: applied load, L measuring length, E: Young’s

modulus, A: average of cross sectional areas of whole and ligament, and 6r:
residual deformation when the load decreases to zero. Fig.10 shows 5 tension
softening curves obtained from 5 specimens of an identical concrete. They
show almost similar curves and scattering is small. In addition, all curves but
one are obtained until almost zero load. These facts show the reliability of the
test procedure developed by the authors.
CONCLUSIONS
Three misunderstandings that are argumentative problems in uniaxial tension

test of concrete were clarified theoretically and experimentally. To eliminate
the previous problems, a test procedure was proposed of which reliability was
c o n i h e d by experimental results.
@Seismicisolation
@Seismicisolation
REFERENCES
Hillerborg, A., ?A model for fracture analysis?, Report TVBM-3005, Lund

Inst. Tech., Lund (1978)
Carpinteri, A. and Ferro, G., Size effects on tensile fracture properties: A
unified explanation based on disorder and fractality of concrete
microstructure?,Mater. Struct. 27 (1994) 563-571.
Akita, H., Koide, H., Sohn, D. and Tomon, M.: A testing procedure for
assessing the uniaxial tension of concrete, AC1 Int. SP-201 (2001) 75-91.
Li, Z. and Shah, S . P., ?Localization of microcracking in concrete under
uniaxialtension, AC1 Mater. J. (1994) 372-381.
Li, Q. and Ansari, E ?High-Strength concrete in uniaxial tension?, AC1
Mater. J., 97 (1) (2000) 49-57.
Zheng, W., Kwan, A. K. H. and Lee, P. K. K., ?Direct tension test of
concrete?,AC1 Mater. J., Vo1.98, No.1, pp.63-71, 2001.
Hordijl, D. A, Reinhardt, H. W. and Cornelissen, H. A. W. ?Fracture
mechanics parameters of concrete from uniaxial tensile tests as influenced
by specimen length?, Fracture of Concrete and Rock, Soc. Exp. Mechanics,
(1987) 138-149.
van Mier, J. G. M., Schlangen, E. and Vervuurt, A., Tensile cracking in
concrete and sandstone: Pari 2 - Effect of boundary rotations?, Mater. Struct.,
29 (1996) 87-96.
Zhou, E P., ?influences of notch size, eccentricity and rotational stifEness on
fracture properties determined in tensile tests?, in: Wittmann, E H. (ed),
?Fracture Mechanics of Concrete Structures? (AEDIFICATIO Publishers,
Freiburg, 1995) 65-74.
10 Akita, H., Sohn, D. and Ojima, M., ?Simulation study of secondary flexure
versus fracture behavior of concrete under uniaxial tension loading?, Proc.
6th Int. Symp. Brittle Matrix Composites, (2000) 371-378.
@Seismicisolation
@Seismicisolation
412 Akita et al.
olft
1
1
'.-__------__-______
P/Ptrue=O.ig
1.1 -
i
i
0.5
i
'. ---------------
P/Ptrus=0.35
O 0.25 0.5 0.75 1.0 1.25

-
O- 1'0 210 o; io o;
grmm(XI 0.~1
Fig.1 Tensile stress reduction * =- x(mm) :
Multiple cracks Overlapping cracks
Bottom face Guide notch
Extensometer
400
Unit: mm
Fig.3 Multiple and overlapping cracks Fig.4 Prismatic specimen
@Seismicisolation
@Seismicisolation
n Grip end
Il I Aluminum date
Fig. 5 Experimental set-up

Unit: mm
U Grip end
Fig.6 Adjusting gear system
O 30 60 90 120 150
time (min) 6 (mm)
Fig.7 Machine load and adjusting force Fig.8 Load-deformation curves
@Seismicisolation
@Seismicisolation
414 Akita et al.
O 0.1 0.2 0.3 0.4 ( 5

6 (mm) w (mm)
Fig.9 Individual load-deformation curves Fig.10 Tension softening curves
@Seismicisolation
@Seismicisolation
SP 209-23
Evaluation of Residual Strength of

Reinforced Concrete Members
Subjected to Reversed
Cyclic Torsion
by N. Kawaguchi, H. Kuga, and T. Yoda
SvnoDsis: The study is aimed at obtaining basic data for evaluating the residual
strength and deformability of damaged reinforced concrete members.
The behavior of members that suffered damage during cyclic loading such
as cracking that penetrate into the cross section, plastic deformation or partial fail-
ure was compared with that of member tested under simple torsion. Reversed
cyclic torsion was applied for loading because few studies have been reported in
open literature although it has been estimated to cause the most brittle failure.
Tests and analyses identified the effect of reversed cyclic torsion with vary-
ing numbers of cycles and with varying amplitudes had on the reduction of resid-
ual strength and torsional rigidity. It was also confirmed that a closed joint that en-
circles longitudinal reinforcement at the ends of transverse reinforcement was
stronger and more resistant to deformation than a lapped joint.
Kevwords: closed joint; lapped joint; residual strength;

reversed cyclic torsion; torsional rigidity
@Seismicisolation
@Seismicisolation
415
416 Kawaguchi et al.
Naotaka Kawaguchi: Professor in the Department of Civil Engineering,
Kokushikan University, Tokyo, Japan.
Hidetatsu Kuga: Research engineer in the Department of Civil Engineering,

Kokushikan University, Tokyo, Japan.
Teruhiko Yoda: Professor in the Department of Civil Engineering, Waseda

University, Tokyo,Japan.
INTRODUCTION
Brittle failure of concrete members under torsion has come to be of major

concern after the Hyogoken-Nanbu Earthquake (1995, Japan). Even when the
cross section of structural members are partially damaged or ruptured by the
effect of seismic motion, it is possible that partial capacity can be maintained
by residual or potential strength.
The residual strength and deformation capacity of reinforced concrete

members subjected to reversed cyclic torsion were investigated both
experimentally and analytically. This paper reports the results of tests
conducted to examine the behavior of concrete members reinforced with
conventional mild steel when subjected to various levels of cyclic torsion
before the ultimate strength has been reached.
This study also paid attention to the anchorage of the ends of transverse
reinforcing bars. Hoops and stirrups are generally equipped with hooks at their
ends and are anchored to each other to encircle longitudinal reinforcement. For
construction reasons, however, simply lapped joints are sometimes used for
@Seismicisolation
@Seismicisolation
end-anchorage. Application of torsional forces may reveal varying behavior of
members according to anchorage methods. A study was therefore made with
varying anchorage methods to study the effects on the strength and
deformation behavior of members.
This study is thus aimed at obtaining basic data on the behavior and
design of reinforced concrete members subjected to reversed cyclic loading.
TEST METHOD
Specimens and M a t e O u a u
The 200 cm long specimen had a cross section of 200 mm high and 150
mm wide (Fig. 1). For reinforcement, four longitudinal reinforcing bars of a
nominal diameter of 13 mm and transverse reinforcing bars of a nominal
diameter of 10 mm at spacings of 35 mm were used. These specificationswere
common to all the specimens.
' h o variations of specimens were used with different modes of anchoring

the ends of transverse reinforcing bars. In one variation, the ends of transverse
reinforcing bars were simply lapped (lapped joint type) (Fig. 1). The other had
135 degrees hooks encircling the longitudinal reinforcement (closed joint type)
(Fig. I).
The target strength of concrete was 50 N/mm2. The mixture proportions

are listed in Table 1. All the reinforcing bars were made of mild steel. The
results of strength tests of the materials are shown in Table 2. The joint type is
specified for each specimen in Table 2.
@Seismicisolation
@Seismicisolation
Loadin? M
W-t Items
Fig. 2 shows the test apparatus. One end of the apparatus is fixed and the
other end can be moved in the circumferential direction. Direct hydraulic
application of reversed cyclic torsion is possible. A pair of cylindrical devices
attached on the specimen is an inclinometer for measuring torsional angles.
Torsional angles and strain in the longitudinal and transverse reinforcements
were measured.
The test was conducted in the following procedure.

(a) The maximum (failure) strength was obtained by monotonic torsional
loading.
(b) The levels of load to be applied in reversed cyclic loading were
determined. In this test, 6û%, 70%, 80%, 83% and 85% of the maximum
strength were applied.
(c) The number of reversed cycles of loading: ten and twenty reversed cycles
of loading.
(d) The specified torsional moments were first applied cyclically for the
specified number of times, and subsequently torsional moment was
applied up to failure.
The test aimed at comparing the failure strength and deformation behavior
obtained by (a) with those obtained when the specimen failed after reversed
cyclic loading according to (b) through (d).
Loads were applied cyclically as described above because the following

information had been obtained in the tests the authors had conducted earlier.
(i) When the number of reversed cycles of loading was less than twenty
times and the load was less than 50% of the maximum strength, no
reduction was observed in strength.
@Seismicisolation
@Seismicisolation
(ii) It was noted that under torsional moment, load of about 80% of the
maximum strength could cause yielding of longitudinal reinforcement and
that members may behave considerably differently before and after the
yielding of longitudinal reinforcement.
Structural members are said to be subjected to a dozen or so seismic

waves before failure during a severe earthquake. The number of reversed
cycles of loading was therefore set at a low level in the test.
TEST RESULTS
Examples of cracks at the time of failure are shown in Fig. 3. The process
of crack development was similar to traditional torsional behavior regardless
of whether reversed cyclic loading was applied or not. The first diagonal crack
was found on the longer side of the cross section where shear stress was
largest. Then the crack progressed toward the shorter side, cracks occurred in
other cross sections, and all the cracks spiraled. The spacings of cracks were
nearly identical to the spacings of transverse reinforcement. The members
reached their maximum strength following the yielding of longitudinal
reinforcement.
Typical examples of relationships between torsional moment and torsional

angle are shown in Figs. 4 and 5 . The measurement results are summarized in
Table 2. The test results are briefly described below by item.
As shown in Table 2, no reduction in maximum strength was observed
@Seismicisolation
@Seismicisolation
regardless of the number of cycles or the mode of anchoring the ends of
reinforcing bars when the reversed cyclic torsion was 60% or 70% of the
maximum monotonic torsional strength. Unless the longitudinal reinforcement
has yielded due to reversed cyclic torsion, original torsional strength could be
maintained when the number of cycles is small as in the test conducted in this
study.
On the other hand, when the reversed cyclic torsion exceeded 80% of the
maximum monotonic torsional strength, the maximum strength decreased 10%
to 20%. The higher the cyclic loading amplitude in preloading, the more
decreased maximum strengths. In the case of yielding of the longitudinal
reinforcement due to reversed cyclic torsion, therefore, strength may decrease
when a failure finally occurs. The degree of strength decrease was larger than
in the tests that the authors conducted using different combinations of loads (i),
(2).
Difference in a d o ring of transverse rgmf

' orcement
The difference in anchoring method had a pronounced effect under

reversed cyclic torsion of equivalent to 80% of the maximum monotonic
torsional strength or greater. Table 2 and Figs. 4 and 5 indicated the following
tendencies.
(i) No significant difference in maximum monotonic strengths was observed.

(ii) The post-cyclic-loading strength decreased about 20% for the lapped joint
type, and about 10% for the closed joint type (Table 2). The rate of
decrease was higher for the lapped joint type.
(iii) Deformation after the maximum strength was reached was more moderate
for the closed joint type (Figs. 4 and 5). The closed joint type suffered less
strength decrease after the maximum strength was reached, and proved
@Seismicisolation
@Seismicisolation
effective even under reversed cyclic torsion.
As shown in Fig. 6, spalling of concrete cover and opening of joints were

observed in the specimens with lapped joints when the strength almost reached
the maximum value or during the application of reversed cyclic torsion. This
may have weakened the confinement of internal concrete as have been
discussed.
In numerous countries, it is stipulated that ends of transverse reinforcing

bars should be anchored to each other so as to encircle longitudinal
reinforcement. The stipulation should be adhered to especially when the
member is subjected to torsion.
Ten or twenty reversed cycles of loading were applied before the member
was finally subjected to failure. As shown in Table 2, the maximum strength
decreased more when the specimen suffered 20 cycles of reversed loading.
Some specimens failed before the planned number of cycles was applied.
General discussion of the effects of the number of cycles on the development
of the maximum strength is, therefore, still difficult based on the results of the
test. Identifying a general tendency requires the collection of basic data on
low-cycle fatigue strength under the yield condition in torsion.
Torsional a
The measurements of torsional rigidity shown in Table 2 were calculated

as described below. First, the relationships between torsional moment and
@Seismicisolation
@Seismicisolation
torsional angle shown in Figs. 4 and 5 were approximated as straight lines by
making models (Fig. 7). The gradients were obtained as the measurements of
torsional rigidity. JO through J, in Fig. 7 correspond to the following
conditions.
J,: Initial rigidity.

J,: Cracking of concrete due to torsion has progressed into the section or other
cracks occurred.
J,: All the major cracks have occurred, but no yielding of the both longitudinal
and transverse reinforcement has yet occurred.
J,: The maximum strength has almost been reached.
As is obvious from Table 2, the measurements varied considerably except

the initial rigidity. Although not shown in the table, the coefficient of variation
in torsional rigidity was much larger than that in strength or other terms for
specimens subjected to the same conditions. No generalization was, therefore,
possible, but a general declining tendency of torsional rigidity was noted.
Torsional rigidity decreased to about 10% to 20% of the initial rigidity once
cracking occurred. With deformation, it finally decreased to a few percents.
The rate of decrease seems higher than the rate of reduction in the flexural
rigidity in bending. The tension stiffening as observed in bending was not
observed in the test. Torsional cracking hardly induced the stiffening effect.
ANALYTICAL CONSIDERATION
Prediction of the response of reversed cyclic torsion requires modeling of

the effects of cracks on the constitutive law. Although the spatial orientation of-
the cracks within the damaged members may be assumed to be uniform in
space to render the member anisotropic, the reversed cyclic torsion behavior is
hard to anticipate due to the three dimensional behavior. In this analytical
@Seismicisolation
@Seismicisolation
study, the secant stiffness method has been introduced for the monotonic
response of reinforced concrete members subjected to pure torsion. The
application of the present method for calculating the secant stiffness of
members under pure torsion is closely related to the damage theory for micro-
cracked materials. As an example, the analytical results are compared with the
experimental results for the members with closed joint type as shown in Fig. 8.
Slight difference can be seen at the initiation of cracking. On the other hand,
softening behavior can be estimated using pertinent deteriorating models.
In Fig. 8, the secant stifíñess is characterized by the following four

models:
(1) Elastic Model

In the elastic region, the conventional relationship between the torsional
moment M, and torsional angle 8 is'given by
where GJ,, is the torsional rigidity of the form
GJ,, = C,Gd3b(C, = 0.187) (2)
where C, : a constant, G : shear modulus, d: depth of the cross-section,b :

width of the cross-section.
(2) Smeared Cracking Model

In this case, a prismatic beam will be replaced by a thin-walled beam
whose torsional rigidity is expressed by
GJ,? =C,G- 4A2 (C, =1/15) (3).

2b + 2d
t
where A: cross-sectional area (=bd), t: thickness of the thin-wall (double
of the cover thickness, 60 mm in this paper).
@Seismicisolation
@Seismicisolation
Then torsional moment is given by
M, =GJd20 (4)
(3) Space Truss Model
According to the Current Japanese Standard Specifications for Concrete

Structures, design torsional strength Mt is denoted by
in which
where A , : cross-sectional area of longitudinal reinforcing bar, Ai,:

cross-sectional of one transverse reinforcing bar, fld: design yield stress of
longitudinal reinforcing bar, f w i design yield stress of transverse reinforcing
bar, s: spacing of transverse reinforcing bar, u: total length of a transverse
reinforcing bar.
Torsional rigidity GJ will be written as
d3
(4) Deteriorating Model

This damage model is characterized by shear yielding of longitudinal
reinforcing bars. In this case, torsional moment is expressed by
M, = gA, (b+d)
where 7 = f,,/& , A,: cross-sectional area of one longitudinal

reinforcing bar.
@Seismicisolation
@Seismicisolation
Torsional rigidity GJd4will be written as
A
GJd4= G L (b2+ d2) (10)
4
CONCLUSION
Residual strength was examined in reinforced concrete members

subjected to reversed cyclic torsional moment with a small number of cycles
and a high-level amplitude. Tests and theoretical analyses produced the
following results.
At a cyclic loading level below 70%of the monotonic strength, no decline

in strength was observed.
At a cyclic loading level that caused the yielding of the longitudinal
reinforcement (higher than 80% of the monotonic strength), strength
decreased by 10 to 20%.
Difference in anchoring of transverse reinforcement had the following
effects.
(a) No significant difference in maximum monotonic strength was
observed.
(b) Difference was observed when the member was preloaded by cyclic
loading. The maximum strength decreased about 20%for the simply
lapped joint type. When transverse reinforcing bars were anchored to
each other SO as to encircle longitudinal reinforcement, the maximum
strength dropped about 10%.
Relationships between torsional moment and torsional angle are
characterized by the four models depending on the degree of damage:
elastic model, smeared cracking model, space truss model, deteriorating
model in order.
@Seismicisolation
@Seismicisolation
References
(1) Kawaguchi, N. "Ultimate Strength and Deformation Characteristics of

Concrete Members Reinforced with AFRP Rods Under Combined Axial
Tension or Compression and Bending", Fiber-Reinforced-Plastic
Reinforcement for Concrete Structures, International Symposium, aci SP-
138,pp. 671-683, 1993, American Concrete Institute
(2) Kawaguchi, N., et. al.: "Residual Capacity of Concrete Members

Reinforced with AFRP and CFRP Rods under Reversed Torsion or
Bending", Proceedings of the Second International Conference on
Concrete Under Severe Conditions, Vo1.2, 1489-1498, 1998, E & FN
SPON
Symbols
E, : modulus of elasticity of concrete

E, : modulus of elasticity of steel
f, : compressive strength of concrete
f, : tensile strength of concrete
f, : yielding strength of steel
f, : tensile strength of steel
J, : measured initial torsional rigidity
J,, J, J, : measured torsional rigidity at any stage of loading
M,: torsional moment
0 : torsional angle
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
O
C
c1
rd
s
VI
ce,
Yd
o
.-
'p
.-C
_I
M
.-c
m-
O
-x
.-o
o
a2
L
'CI
C
1
%
Y
63
Y
.. .. .. . ,
*
* **
Y *
Y * * *
@Seismicisolation
@Seismicisolation
4
35 X 25 = 875 b
4
1O00
Lapped joint type Closed joint type
Fig. 1-Dimensions of specimen
@Seismicisolation
@Seismicisolation
Fig. 2-Torsional loading test apparatus
Fig. 3-An example of crack at the time of failure
@Seismicisolation
@Seismicisolation
O
u3
O
v
O
m
O
cv
O
-c
O
7
I
O
CiJ
O
9
O
7-
O
u?
@Seismicisolation
@Seismicisolation
"
J
O
*
O
O
m
h O
cv
O
Y
O
Y
I
O
I
O
I
I O
Y
O
I
@Seismicisolation
@Seismicisolation
Fig. &Opening of a lapped joint
JO
Torsional angle e
Fig. 7-Approximation of relationship between torsional moment
and torsional angle as straight lines
@Seismicisolation
@Seismicisolation
n
.-c n
.-L .-S
I
UI
7
(v
s
I
W
O
c
ci
O
c
"-I
> .-L
.-o
>W
.-cn
Cu
I
!?
I- w -
O
'5
C
d
- 1
II
@Seismicisolation
@Seismicisolation
SP 209-24
Extreme Concrete
by K. W. Awad, H. Mazen, and R. Fakih
SvnoDsis: The durability and permeability of concrete are of special concern in the
Arabian Gulf due to the prevailing aggressive weather conditions. Very few practi-
cal examples showing the ability of the local ready-mixed industry to adapt to ex-
treme hot weather, characteristic of this region of the world, have been exposed
and analyzed.
The following paper describes extreme concrete properties in two particu-
lar projects in Saudi Arabia demonstrating how the local concrete industry has
used specific measures to produce high-performance concrete in very hot weather.
In the first project concrete was hauled for 8 hours in hot weather using
high dose of standard retarders and superplasticisers and following rigorous mea-
sures to control the mix temperature.
In the second project high-strength high-performance concrete was hauled
for 45 minutes at an ambient temperature of 40" C and achieved 24 hours strength
in excess of 50 MPa.
It is concluded that controlling the temperature of fresh concrete in hot
weather should be primarily based on parameters related to cost, workability re-
tention and long-term strength. Besides, the cited examples demonstrated how the
use of admixtures in hot weather could be very beneficial even if the dosage con-
sidered did not comply with ASTM C 494 requirements and the manufacturer stan-
dard recommendations.
Kevwords: admixture dosage; concrete temperature;

heat generation; heat loss; high early strength; hot
weather concrete; long haul; retarder; setting time;
superplasticizer; temperature rise; W/C; workability
retention
@Seismicisolation
@Seismicisolation
435
436 Awadetal.
1- AC1 Fellow Khaled Awad is the President of Advanced
Construction Technology Services (ACTS), a material testing
and consulting firm based i n Beirut, Lebanon and operating
i n several countries of the Middle East. He is a member of
several AC1 committees and also a member of ASTM, and the
Institute of Concrete Technology in the United Kingdom. He
was the manager of a multinational ready mix concrete firm
operating in the UAE, Egypt and Saudi Arabia.
2- AC1 member Hazim Mazen,PhD is the General Manager of

Arab Consulting Group (ACG), a design firm based in Cairo,
Egypt, specializing in value engineering applications for large
construction projects. Previously, he was the manager of the
structural department of a large Saudi construction firm,
operating in Asia and the Middle East.
3- AC1 member Rabih Fakib, MBA, is the Laboratory Manager

of Advanced Construction Technology Services (ACTS),
Beirut, Lebanon. He oversees the operation of several
laboratories for ACTS in Lebanon and other countries of the
Middle East.
Overview
The weather in the Arabian Gulf is considered to be one of

the world's most aggressive environments for mixing and
placing concrete and meets typically all the hot weather
definitions of Section 1.1.1of AC1 305(1):
High Ambient Temperature: The ambient temperature

during the summer season (between April and October)
can reach 5OoC (122OF).
Wind speed in desert areas can exceed 40 km/hr-

during some periods of the year.
@Seismicisolation
@Seismicisolation
Evaporation rates are high in arid regions like Riyadh
city in Kingdom of Saudi Arabia or the city of AI Ain in
the United Arab Emirates.
Solar radiation levels are very high, and considered

one of the highest in the world.
Besides, the relative humidity in the coastal cities of the Gulf

(Dubai, Abu Dhabi, Dhahran, Bahrain and Doha) can reach
100% and therefore combines with the high temperatures to
create a very favorable environment for corrosion
acceleration .
Due to the nearness of these cities to the ocean, the high

relative humidity and the high concentration of air borne
salts necessitate special attention to concrete permeability
and durability.
To cope with these difficult weather characteristics, concrete

producers follow systematically a series of measures
including:
1. Consistent use of retarders, with variable dosages

throughout the year.
2. Consistent use of water reducers and often High-Range

Water Reducers to control w/c, that is predominantly
below 0.50 for reinforced concrete.
3. Cooling the water with water chillers and using cooling

towers in arid areas.
4. Replacing up to 50% of the water with flake ice during

the hot period of the summer (June-September).
@Seismicisolation
@Seismicisolation
438 Awad et al.
5. I n some areas like Kuwait city, concrete is still batched
dry and truck-mixed on site as the control of slump
loss is very difficult.
The adoption of such measures has permitted the local
ready-mixed industry t o produce high quality concrete,
including high strength concrete up to 80 MPa for the
columns of high-rise buildings extending above 330 meters
(Emirates tower in Dubai) and high-performance concrete
with ternary systems (Portland cement, slag or fly ash and
silica fume).
No ready-mixed concrete in the Arabian Gulf is produced

without admixtures. As cement (and concrete) consumption
per capita in the Arabian Gulf countries remain the highest in
the world (3105kg/capita in the United Arab Emirates), the
extensive use of chemical admixtures have lead to thorough
understanding and specialized experience in handling,
dosage control and appropriate utilization of these
admixtures.
Under this very aggressive weather condition of the Arabian

Gulf, producing special types of weather defiant concrete can
be considered extreme.
Extreme Concrete
Two particular projects reflecting the "extreme" use of

concrete in this region are described herein:
@Seismicisolation
@Seismicisolation
The first project involves the use of a high dosage of ASTM
C494 (2) Type B retarder to allow the hauling of ready-mixed
concrete for more than 8 hours in hot weather, and under a
high evaporation rate.
The second project involves the production, hauling and

placing of high-early strength concrete in hot weather and
under high relative humidity.
AT&T communication towers on Taif-Riyadh road
The project consisted of building 20 communication towers

for AT&T between Taif and Riyadh in the period between April
and May. Taif is a city located in the western part of Saudi
Arabia around 870 Km from Riyadh, the capital of the
Kingdom, at 1800 m altitude. During this period of the year,
the ambient temperature measured in the shade in early
afternoon reached 45OC.
The towers were 35 k m distant from each other. Two

different types of concrete foundation were designed for
these towers, one requiring 28 m3 and the other 38 m3 of 28
MPa concrete. The specified maximum W / C ratio was 0.50.
The contractor first evaluated the setting up of a batching

plant in the middle of the road but due to the small overall
quantity, the use of ready mix concrete was much more
eco nom i cal.
The second consideration was shipping concrete dry from Taif

and mixing it on site using water tankers and/or the water
tanks available on the trucks. This option was more
economical than setting up a batching plant for this small
project but had its own several disadvantages:
@Seismicisolation
@Seismicisolation
440 Awad et al.
1- More truck loads were needed by pour as each would
be carrying less in order to achieve proper mixing and
therefore the cost per cubic meter will increase.
2- Control of W/C could not be very accurate as it will

depend on the calibration of the meters of the water
tanks.
3- The control o f concrete temperature, which was

imperative in order to handle and place the concrete,
would be extremely difficult as it implicated either
having ice water at the mixing site or sending ice water
in insulated water tankers. The first option was too
expensive and impractical to arrange as it involved
transporting mobile cold stores with ice blocks and
adding these ice blocks in the water tankers and truck
mixers then waiting for them to melt. Subsequently,
the control of slump and W/C would have also been
much more difficult. On the other hand, if ice water
was sent in the water tankers, its temperature would
still increase sharply by the time i t reaches site.
Additionally, proper insulation of the water tankers
would have been also very expensive.
4- Special attention would be required for site handling

and sequence of addition of admixtures especially
superplasicizers (e.g. time of addition during mixing,
retarders to be added first, possibility of adding type G
during mixing and then type F before placing), which
were required anyway for the control of W/C. A higher
dosage of superplasicisers than needed could also be
required t o counterbalance potential mixing problems.
5- Each placement would require at least 2 to 4 hours to

account for mixing and placing instead of a maximum
placing time of one hour had the concrete been pre-
mixed.
@Seismicisolation
@Seismicisolation
This also implied increase in concrete temperature and
therefore more difficult control o f setting time and
worka biI it y retenti on .
6- Not enough water will be available for cleaning the

trucks after discharge if the trucks water tanks were
used. This implicated potential concrete buildup inside
the drums as the trucks had t o travel back sometimes
for more than 8 hours. Concrete buildup would require
daily cleaning of the trucks and since the concrete
would not have a high dose of retarders if mixed on
site, the cleaning would have possibly required jack
hammering. This was another sizeable cost to be
considered.
7- Alternatively, given the expensive cost of water in

Saudi Arabia the use of water tankers and the
coordination between water tankers and concrete
trucks would also drive the cost up.
Based on all the above factors, the contractor wanted t o

evaluate the option of delivering pre-mixed concrete.
The nearest plant that could produce high quality concrete

was located in Makkah, situated around 90 Km from Taif and
therefore farther from all towers by an additional 90 Km. The
plant was also equipped with a flake ice plant, water chillers,
shaded aggregates bins and semi-insulated cement silos
wrapped with burlap, which permitted a very good
temperature control of concrete.
The contractor required mockup trials in the plant to verify if

the retardation will be controlled so that the concrete arrives
to the site with the right slump range and starts setting
within maximum 4 hours after placing thereby causing no
delay in the removal of the forms.
@Seismicisolation
@Seismicisolation
442 Awad et al.
Mixture requirements:
The mixture proportioning criteria were set as follows:

Compressive Strength: Minimum 28 MPa after 28 days
Maximum w/c: 0.50
Minimum delay in setting time over control: 6-10 hours
depending on the location of the tower.
Maximum delay in setting time over control: 10-14 Hours
depending on the location of the tower.
Slump range at placing point: 100-150mm
Maximum concrete temperature: 38'C.
Mixture considerations:
Based on these requirements it was decided to identify a

retarder meeting ASTM C 494 Type D specifications and
evaluate the effect of overdosing this retarder on workability,
strength and setting time.
I n addition, two types of compatible superplasticisers had to

be used in conjunction with the retarder. The first
superplasticiser was to be used at the mixing plant, and
should therefore meet ASTM C 494 type G definition, and the
second one was to be used at site just before placing for
slump control and hence should meet the requirements of
ASTM C 494 Type F.
It was also decided to keep very high the initial slump a t the
plant in order to reduce friction during transport and reduce
the quantity of superplasticisers t o be used at site.
@Seismicisolation
@Seismicisolation
Trials to determine range of retarder dosage:
Two different retarders were tried at 4 different dosages

(0.3%, O.6%, 0.9% and 1.2% by cement weight). The trials
consisted of checking the effect of mixing on initial setting
time using a standard concrete penetrometer. For each
dosage, a portion of the concrete was sampled initially and
tested for initial setting time, then another portion sampled
after 2,4,6 and 8 hours and tested for the same.
Even a t the very high dosage of admixtures, compressive
strength results at 7 and 28 days were all above specified
requirements.
Table 1 summarizes the compressive strength obtained with
different dosages of these 2 retarders as a percentage of the
compressive strength of the control mixture without
admixture.
The trials showed that the use of different types of retarders

yielded different effects on setting times. Retarder 1,
meeting ASTM C 494 type B requirements, was a pure
hydroxyl-carboxylic base, and showed reduced setting time
with increased mixing while Retarder 2, meeting ASTM C494
type D requirements had traces of sugar and trisodium
phosphate, and increased the setting time with increased
mixing.
Figures 1 and 2 summarize the effect of delayed mixing on

initial setting time for the two tested retarders without the
use of superplasticisers.
Other trials were made using different combinations of

Retarder 1 and a type G superplasticiser t o check the effect
of various dosages of these two materials on the initial
setting time and strength gain.
A t the required initial setting time, the trials indicated a drop

of around 20% in 7 days and 5 O/O in 28 days strength.
@Seismicisolation
@Seismicisolation
444 Awad et al.
This was due to an increase of 1.5% in air entrainment over
the mix with using Retarder 1.
The larger effect on 7 days strength was attributed to a
secondary increase in initial setting time caused by the Type
G admixture. Accordingly, the addition of a type G
superplasticiser allowed the reduction of the dosage of
Retarder 1 by around 0.15O/0 by cement weight.
Table 2 summarizes the 7 and 28 days results of the trial

made with a combination of 1.8% Type G with 1.2% Type B
(Retarder 1)as a percentage of the strength achieved with
Retarder 1only.
Measures for batching, mixing, transporting and placing:
The following measures were implemented in order to meet

the mixture proportioning criteria:
I n order to achieve better workability retention and

very low initial m i x temperature, the use of flake ice to
replace more than 50% o f the chilled water was
considered but due t o the effect on losing control over
the initial slump and the possibility of increasing
friction with low initial slump, which in turn could
increase the temperature, it was decided to limit ice
replacement t o 15-20% of the 4OC chilled water. A t this
replacement level, all ice was melting in the truck at
the plant and initial slump could therefore be checked.
Spraying water on coarse aggregates 4 hours before

batching allowed cooling by evaporation as the relative
humidity was less than 75%. This measure maintained
the aggregates temperature below 36OC.
Cement was stored for 72 hours to allow cooling as

incoming cement had temperature in excess of 55OC.
@Seismicisolation
@Seismicisolation
o Type B retarder (Retarder i) was used, the dosage
varying depending on the location of the tower but well
exceeding the maximum recommended dosage by the
manufacturer. The dosage used varied between 0.55%
and 1.OS0/o by cement weight, and this was well in
excess o f the manufacturer dosage of 0.3% by cement
weight.
o 1.8% by cement weight of the Type G superplasticiser

was added at the plant, well in excess of the
recommended manufacturer dosage of 1% by cement
weight while Type F superplasticiser was added on site
to increase the slump if needed. The rate of addition of
the type F depended on the slump of the concrete
delivered on site.
Agitating speed was maintained a t strictly minimum

(around 1 rpm) t o reduce friction and heat rise.
The drums were wrapped with wet burlap to provide

humid environment and prevent sharp loss of slump.
The truck water tanks had 700 liters capacity, and

were filled with chilled water and wrapped with
insulating material. This measure was intended t o
make available re-tempering water on site if needed.
o Batching started consistently at 2 : O O am as at that

time the aggregates had the lowest temperature.
Maximum of one foundation tower per day was placed.
A back-up pickup accompanied the trucks to assist in

the event of mechanical breakdowns or tire puncture.
@Seismicisolation
@Seismicisolation
Fine Aggreqates 0-5 mm 780 Kg Natural screened
Coarse Aggregates 5-10 m m 365 Kg Gneiss
Coarse Aggregates 10-20 710 Kg Gneiss
l I I I
!
mm
Ilement, 1v (wc)
370 Kg ASTM C 150 Type
Free Water (Ww) 185 Kg

Water for Absorption (Wwa) 22 Kg
Retarder 1.6-3.8 ASTM C 494 Type
L B
Superplasticiser 5-6.5 L ASTM C 494 Type
G
The control of the concrete temperature was done using the

formulae suggested by Appendix A of AC1 305R (2) which
proved to be consistently accurate.
The following lowest temperatures were recorded at the plant

at 02:OO am:
Cement temperature (Tc): 44OC
Aggregates Temperature (Ta): 36OC
Chilled Water Temperature (Tw): 4OC
According to AC1 305, and using the above mentioned data,

the initial mixture temperature was expected to be:
T= O.ZZ(TaWa+TcWc)+ TwWw+ TaWwa +T=28OC

O. 22(Wa+ Wc) + Ww +Wwa
As the trip would take in some instances around 8 hours, the

increase in the concrete temperature was expected to be in
the range of 10°C and accordingly i t was decided to maintain
the initial concrete temperature at a maximum of 25OC.
@Seismicisolation
@Seismicisolation
Accordingly, the level of ice replacement was decided using a
rearrangement of the second equation in AC1 305 and was
calculated as follows:
wi= 0.22 (TaWa+TcWc) + TwWw +TaWwa -

+
Tconcrete(O.22( Wa Wc) +Ww+Wwa)
Tw+79.6
+ Wi= 30 Kg
ResuIts:
One hundred and two truck loads consisting of 711 m3 of

concrete were pre-mixed and delivered to the towers
locations.
Hauling time ranged from 4 hours for towers close to Taif up

to 8 hours for the farthest towers. Towers located less than
300 Km away from Riyadh were supplied from a Riyadh
batching plant.
The measured slump on site ranged from 20mm to 140mm.

The wide range was attributed to variable slump loss caused
by variable temperature and evaporation rates. Two to six
liters of the Type F superplasticiser were added.
Concrete temperature ranged between 36OC and 38OC.
Compressive strength results after 28 days were all higher

than 35 MPa, with a n average of 38.3 MPa and a standard
deviation of 2.65 MPa. The 28 days results were lower than
normally expected after over-retarding the concrete due t o
the probable effect of "grinding" (3) caused by extended
agitating time in the truck.
The compressive strength results after 7 days were all higher

than 25 MPa.
@Seismicisolation
@Seismicisolation
448 Awad et al.
The removal of forms occurred consistently 24-30 hours after
placing.
Curing compound was used immediately after placing

concrete in the foundation and the walls and as soon as wall
forms were removed.
One case of over retardation occurred and the concrete did

not set for 3 days. I t was due to a breakdown in the
admixture dispenser at the plant that was not spotted by the
batch plant operator. The error was discovered after the
concrete had n o t set and the concrete, still not set after 3
days, was removed as the final setting time was not known
and the forms had been slightly loosened after 24 hours and
that caused intolerable surface defects.
High-Early Strength for a precast operation in Jeddah
I t was required t o build an Emergency Center Facility

at a certain location in Saudi Arabia. The client
intended t o inaugurate the facility within a very short
time, leaving very limited time for the construction
activities.
The designer choice was to go to a prestressed /

precast solution so that all floors would be cast in a
precast factory owned by the contractor, while
foundation works and main concrete frame works could
be going on in parallel at the site.
The floor plan module was 12x12 meters while the

expected live loads were of a heavy nature
@Seismicisolation
@Seismicisolation
Accordingly, a prestressed, double tee slabs flooring
system was agreed on. The placing was planned to
commence in early May and end in the first week of
June t o meet project schedule.
The speedy nature of this project required intensive

planning of resources, manpower, erection crews,
handling, transportation and quality control measures.
Due to the compacted time schedule, the available
steel moulds and erection lines of the factory were
another fixed parameter and there was not enough
time neither to import material t o fabricate new
moulds for the beams nor to assemble new production
lines.
Only 4 moulds for the designed section were available

whereas the number of beams needed was 80. Accordingly
the moulds turnaround time must be less than 24 hours.
Other factories producing the proposed section were

far from the building site and transportation of units
was hence impractical.
Accordingly, the target was t o speed up the casting activity

and increase the production rate of units to cope with this
very tight time schedule.
Consequently, it was decided t o use a high-strength

concrete mixture that attains a very high early
strength to reach the transfer stage in a rather quick
time frame, and hence to be able to strip and remove
the units and reuse the moulds.
Many problems were addressed such as heat generation and

its control, deflection control, handling stresses and required
intensive curing to avert thermal cracks(4).
@Seismicisolation
@Seismicisolation
450 Awad et al.
The issue o f temperature rise and heat generated inside the
concrete was particularly addressed to avoid thermal shock if
demoulding was t o occur a t night.
The following equations were used for guidance ( S ) , (6)
Temperature Rise=Heat Generated+Heat Gained-Heat lost to

su Trou ndings
Thermal Capacity
Thermal Capacity= Volume x Density x Specific Heat
Specific Heat of Concrete= 0.8 to 1.2 kJ/kg°C
I n order t o control the temperature rise of the concrete it

was necessary t o address the effects of the section depth,
the insulating properties o f the formwork, the ambient
temperature, the concrete placing temperature, the mixture
proportions and the type and source of cement and
cementitious materials.
As the section of the moulds was relatively thin and the

moulds used were principally made of non insulating
material, the heat loss to the surrounds was sizeable and
therefore the temperature of the concrete inside the moulds
was not expected to largely exceed the already elevated
a rnbient tem peratu re.
The precast plant which was awarded the

manufacturing of the beams had a semi-automatic
batching plant not capable of producing concrete with
stringent requirements. The precast factory was
situated at around 45 k m from Jeddah, the large city of
the western region of the Kingdom. I n May, the
ambient temperature exceeded 4OoC during afternoon
hours.
@Seismicisolation
@Seismicisolation
Based on the above considerations, it was decided to use
ready-mixed concrete for this precast operation from a plant
located around 45 Km from the factory.
Design considerations:
The required design stresses were f'c = 5 MPa after 1 day

I
while the required f'c after 28 days was 50 N/mm*.

Accordingly the prevailing design factor was the one day
strength. A maximum W / C of 0.45 was also specified.
The concrete placing temperature was specified as a

maximum of 35OC.
The concrete was expected to be flowable as the forms

were narrow and congested with steel reinforcement.
The slump required was minimum 150 mm. Flowability
and cohesiveness were also of great importance.
The placing of concrete in each mould was expected to take

around one hour and therefore relatively extended
worka bi Iity retenti on was required.
Stress increase monitoring was of great importance to

control all subsequent activities. I n addition to field curing of
cylinders near the pre-stressed double tee beams, hourly
monitoring with rebound hammer t o assess strength gain
was also specified.
Mixture proportion ing :
The requirements for this project were somehow conflicting.
On one hand two factors necessitated delay in setting time:
@Seismicisolation
@Seismicisolation
452 Awad et al.
i- Ready-mixed concrete was specifiéd to achieve high-
early strength as the precast factory batching plant
was not geared for such high requirements.
2- Workability retention was necessary for placing.
On the other hand, very high-early strength was

required which narrowed the window for the setting
time of the concrete.
The proportioning approach was to use a pure
hydroxyl-carboxylic or sodium glucanate based
retarder in an amount well below the minimum dosage
specified by the admixture manufacturer. The use of
this retarder at a minimal dosage would act as a
hydration control and enhance workability retention for
a limited time.
Water t o cementitious ratio was kept at 0.32 and

hence the use of a high range water reducer enhancing
workability and early strength was a must.
Several trials were made to check the following

factors:
1- Initial setting time designed to be minimum 2

hours and not to exceed 3 hours.
2- Workability retention designed for a slump of

minimum 150 mm for 90 minutes.
3- Strength gain over 24 hours as compressive

strength at 24 hours was designed to exceed 45
MPa.
@Seismicisolation
@Seismicisolation
The use of a naphthalene based superplasticisers with a very
low W/Cm (water to cementitious ratio) produced
segregation while the use of a pure melamine based
superplasticiser produced plastic and cohesive concrete.
The dosage of the melamine based admixture was increased

to around 4.3% in order t o achieve the desired workability.
Although the melamine-based products are known to have
very short workability retention, the combination of a high
dose of this admixture with a retarder extended the
workability retention of concrete.
A redose at the precast plant with the melamine

superplasticiser was necessary. I t varied between 6-10
additional liters for a 7 m3 truck.
As the solids content of the superplasticiser was 25%,

necessary adjustments to W/C were made to account for the
additional water that was contributed by the superplasticiser.
The cement content was increased to 430 Kg/m3 in order to

increase the total amount o f free water and enhance the
cohesiveness of the mix.
I n order to limit the cement content at 430Kg/m3, it was

decided to use silica fume.
The use of 5% by cement weight of silica fume also

contributed immensely to the flowability of the concrete and
permitted the big reduction of the water to cementitious ratio
to the very low level of 0.32.
All placements started a t 1 7 : O O in the evening to allow heat

loss in the forms during the night.
@Seismicisolation
@Seismicisolation
454 Awad et al.
The ready-mixed concrete plant observed strict temperature
control by using 5OC chilled water, shaded aggregates bins
and cement stored for 2 days. Temperature control was not
as critical as in the first project but was similarly intended t o
optimize workability retention.
Transportation time was in the range of 40 minutes.

Necessary precautions at the ready-mixed plant were
taken to ensure quick dispatching of the trucks.
The precasting plant was informed that the maximum

allowable placing time was 30 minutes per truck and
therefore necessary manpower and equipment for placing
must be available.
The mixture proportions that yielded the desired

requirements were as follows:
* Total water was 144 Kg/m3 but 17 liters were contributed by the
Superplasticiser
** Sodium Gluconate
*** Melamine based, 25% solids content
@Seismicisolation
@Seismicisolation
ResuIts :
Concrete mix temperature at placing: 35-36OC

Slump after 45 minutes: 180 m m
Slump after 60 minutes: 120 mm
Slump after 80 minutes: 60 mm
Compressive strength after 3 hours: 3-5 MPa
Compressive strength after 26 hours: consistently> 50
M Pa
Compressive strength after 3 days: 55-59 MPa
Compressive strength after 7 days: 69-70 MPa
0 Compressive strength after 28 days: 73-87 MPa
The design of lifting points was intended t o cover many

situations and scenarios. Every unit was checked for
compressive strength after 3 hrs, 6 hrs, 12 hrs, 24 hrs,
3 days, 7 days and 28 days. The early age (up to 3
days) compressive strength were monitored using field
cured cylinders near the moulds whereas 7 and 28
days strength were tested according to standard curing
procedu res.
Rebound hammer readings consistently showed 70-100%

increase in strength between 6 and 12 hours and 50-60%
increase between 12 and 18 hours.
Due to the high-early strength, the precast units were

removed from the moulds after 20-26 hours, which
was according to initially predicted figures using
Harrison equation (6), and the turnaround cycle for the
moulds was according to schedule.
@Seismicisolation
@Seismicisolation
456 Awad et al.
Extreme attention was paid to curing as there was
great concern with temperature rise and rapid setting.
Curing compound was applied on the surface as soon
as placing of the concrete ended. After demoulding,
the units were subjected to additional curing for 3 days
a t the precast plant using tepid water before they were
transported to the project site.
The use of this system resulted in an extra cost of

around 25% compared t o the normal cost due mainly
to the increase in the cost of the ready-mixed concrete
but for such a project need, i t was vital t o follow such
an approach.
Conclusion
0 The completion of these two projects proved that concrete

can be successfully placed in extreme conditions of hot
weather.
0 The manufacturers recommendations concerning the

range of dosage of admixtures in concrete is often based
on the requirements of ASTM C 494. l h e use of dosages
different than those recommended by the manufacturer,
and therefore not meeting ASTM C 494 requirements,
produced satisfactory results in the case of these two
projects.
0 I n the case of the first project, the use of a high dose of

retarder allowed extended hauling times despite extreme
hot weather conditions. Sodium gluconate or Hydroxy-
carboxyl base retarders proved to be particularly useful
for the control of setting time as well as extended
workability retention when used with appropriate
superplasticisers.
@Seismicisolation
@Seismicisolation
I n the case of desired extended setting time, trials done
for the first project showed that overdosing with different
retarders had conflicting effects on the initial setting time
o f concrete.
For the high-early strength project, the use of a very low

dosage of a retarder allowed extended workability
retention without affecting the setting time.
0 The effects of overdosing or underdosing admixtures on

shrinkage, creep, and other durability properties of
concrete should be researched.
@Seismicisolation
@Seismicisolation
458 Awad et al.
References:
1. AC1 305R-99, Hot Weather Concreting, Reported by

Committee 305, Section 1.2.1
2. ASTM C 494/C 494M-99a Standard Specification for
Chemical Admixtures for Concrete
3. Neville, A,Concrete Properties, Fourth Edition
4. Bamforth P.B., Concrete Society Digest No.2, Mass
Concrete
5. Harrison, T A. Early-age thermal crack control in
concrete, Report 91, 1992, Construction Industry
Research and Information Association (CIRIA)
6 . Harrison, T A. Formwork Striking Times- Criteria,
Prediction and methods of assessment. Report No 136,
1995, CIRIA
@Seismicisolation
@Seismicisolation
Compressive by cement weight
o/o o/oby cement weight
: Strength O/O Retarder 1 Retarder 2
: of control
0.3% 0.6% 0.9% 1.2% 0.3% 0.6% 0.9% 1.2%
7 days 90 83 72 58 95 84 77 62
28 days 121 115 108 97 133 129 115 102 -
Table 2: Effect of the addition of a type G superplasticiser on

the compressive strength of concrete made with a
hydroxycarboxilic base retarder (Retarder 1).
I Compressive Strenath
- 1 1.2% bcw 1 1.2% bcw Retarderl + l . 8 % bcw 1
I 0/0 of control I Retarderl I Type G Superplasticiser
7 Days 158 1 46
28 Days I97 1 93
Note: The slump of the control mix was below 50mm.
@Seismicisolation
@Seismicisolation
460 Awad et al.
___
6 - - i
t0.3%
1 2 3 4 5 bcw
Sampling Cycle
Fig.1: Effect of continued mixing on initial setting time with

variable dosages of Retarder 1 in percentage by ceme
weight
Retarder 2
~. +1.2%
l5 1 tXW
-0.9%
tXW
*0.6%
bCW
t 0.3%
I 5 1 7 * , v- I
bCW
1 2 3 4 5
1 . _. _ ~ _ _Sampling
_ _ _ _ _Cycle
______
Fig. 2: Effect of continued mixing on initial setting time with
variable dosages of Retarder 2
Note: Sampling cycle represents around 2 hours of continuous mixing at

the end of which concrete samples were taken for testing initial setting
time.
@Seismicisolation
@Seismicisolation
SP 209-25
Creep of Concrete with Substitution

of Normal Aggregate by Recycled
Concrete Aggregate
by J. M. V. Gómez-Soberón
Svnopsis: This study presents the experimental results on properties of concrete

with replacement of natural aggregate by recycled concrete aggregate (RCA). Ex-
perimental data on the creep behavior of concrete mixtures (basic and drying
creep) was obtained. The replacement factor of natural aggregate by RCA were
O%, 1S%, 30%, 60% and loo%, and the test conditions were SO% RH and 20°C.
The results of these trials were used to provide a comparison with results of tests
on the reference concrete, for ages up to 270 days. The creep coefficient data (in-
stantaneous, basic and drying) presented, along with the maximum strain and the
specific creep data. The results reveal considerable increase in creep when is in-
crease replacement of natural aggregate with recycled concrete aggregate. The dry-
ing creep, especially shower more significant increase when compared to the refer-
ence concrete.
Kevwords: basic creep; creep; drying creep; recycled

aggregate; recycled concrete aggregate
@Seismicisolation
@Seismicisolation
461
462 Gómez-Soberón
José M. V. Gómez-Soberónis Civil Engineer of the Autonomous University of
Puebla in Mexico and studied the M.Eng. (Construction) at the National
University of Mexico. Presently he is doing Ph.D at the School of Civil
Engineering of the Technical University of Catalonia, Spain. He has published
numerous papers on the area of Recycled Concrete in International Journals and
Conferences and works as consulting engineer. jmvl15@hotmail.com
INTRODUCTION
There has been little research hitherto into the long-term properties of recycled
concrete aggregate (RCA). Amongst these, creep is of considerable importance
if this type of aggregate is to be used for making structural concrete. Some
studies with RCA indicate higher creep coefficients than are found in ordinary
concrete containing natural aggregate. The creep coefficients generally tend to
be 30% to 60% higher than the reference concretes mixtures. In extreme cases
with total substitution of RCA for natural aggregate (NA), the creep coefficient
may be up to 300% higher [1, 2, 3,4, 5, 6 , 7, 81. In studies carried out by Sato,
Kamai and Bada [9], the reported increases of creep coefficients are about 2.0,
2.2 and 3.4 times higher than with the reference concretes mixtures. According
to Nishibayashi and Nishobajasi [lo], it seems that these increases gradually
become more marked with the passage of time.
Amongst the probable causes responsible for increase in the creep in concrete
made with recycled concrete aggregate (CRCA) are low density and high
porosity of the RCA, and the possibility that this type of concrete contains more
free water. In addition, lower strength and lower static modulus of elasticity of
this type of concrete [ l l , 12, 13, 14, 15, 16, 17, 18, 19, 20, 211 also contribute
to high creep [22, 23, 241. The above characteristics of RCA are attributed to
the presence of old mortar adhering to surface of aggregate particles. It is
known that, with ordinary concrete, the mortar matrix contributes to high strain,
while natural aggregate particles tend to restricting it. Expressions have even
been proposed showing the correlation between the creep phenomenon and the
amount of matrix in concrete [23, 241
In the case of CRCA, the above-mentioned premises is valid, it may be possible

to estimate creep as a function of the total content of matrix in the concrete (the
newly-mixed matrix plus the old matrix adhering to the natural aggregate).
However, there are reports stating that creep undergoes a great increase when
the aggregate (in the fine fraction of a concrete) are substituted by RCA [25]
@Seismicisolation
@Seismicisolation
If all the above is taken into account, the predictions and formulations of creep
coefficient will need to consider these particularities in the properties both of
the RCA and the concrete mixtures made with this type of aggregate (CRCA)
[26]. This paper discusses what the implication might be of an increased
presence of an RCA in the behavior of different of different samples of CRCA,
in particular with respect to creep properties.
EXPERIMENTAL DETAILS
At this stage of our research into the creep properties of CRCA, the materials,
variables, procedures and techniques were identical to those discussed in a
companion paper “Shrinkage of concrete with replacement of aggregate with
recycled concrete aggregate” [27]. The provisos and exceptions at this stage,
which are exclusively devoted to the study of these creep properties of CRCA,
are discussed in detail below.
Following the 28-day curing, two out of every four samples established for each
of the r factors under study (two for basic creep and two for total creep) were
coated and protected according to the specifications established in the above-
mentioned study, so that these samples could then be placed inside the climatic
chamber and positioned in the loading frames to begin the loading process. The
samples were maintained in these environmental conditions until the end of the
test period (t = 270 days)
Figure 1 shows 12 of the 24 samples used at this stage of the study. These
samples were arranged in pairs and according to r factors in each one of the
groups of twelve loading frames used. Figure 2 presents the details of the
complete configuration for each of the loading frames used. In each frame, this
configuration included a Freyssinet jack, a nitrogen gas accumulator, a
manometer, a swivel for each end of the frame, a nozzle for introducing oil, and
distribution plates and supports.
The beginning of the CRCA strain measurements was to = 28 days of age while
the established level of stress for all the concrete samples was oC/f‘,= 0.35.
Table 1 presents the experimental values off‘, for each r factor of the study and
the level of applied stress.
Creer, of Concrete Containing RCA
The strain values given below are the result of an average of the strain of two
identical samples for each of the variables. To calculate the strain for basic creep
(E, basic) and drying creep (E, drying), the principle of superposition of effects was
@Seismicisolation
@Seismicisolation
464 Gómez-Soberón
applied. This considers in its calculations the instantaneous strain (E,) and total
strain ( E tad)
~ in each case.
Similarly, the coefficients of basic creep (%=ic) and drying creep (q*ing), were
obtained in simplified form from the creep strain (&), i.e., by considering them
as proportional to the instantaneous strain (Q) at the constant level of stress. The
equation used was: E , (t ,to)= &)Fi (to)=c p ( t ) A . Here, q(t) represents the
E(W
creep coefficient (relation between the stress caused by creep at a particular age
and the elastic stress), o, = constant stress applied to the concrete and E =
modulus of elasticity for the instant to [181
Finally, for the calculation of specific creep, the level of stress proposed in this
experimental project was considered as being the result of oJYC= 0.35 in each
case. Then, the following equation was applied: Spec@ creep = Eys , with
EP = total creep and S = oJYCstress measured in the concrete [ 10,281
Table 1 shows the instantaneous creep strain for the basic properties (&i basic) and
properties associated with drying (eming)[27]; in both cases, these values were
obtained at the moment the load was applied in the respective loading frames.
These values, their variations and the limit values obtained (including those of
the reference concrete) are summarized below:
1. In the case of E, basic, a maximum of 0.1645 mmím was obtained for the
factor of r = 0.30, and a minimum of O. 1350 m d m for the factor of r =
0.15, with a variation of only 0.0295 mmím. In this group of values, an
average e b a s i c of 0.1453 mmím is calculated.
2. In the case of Q drying, a maximum of O. 1600 mmím was obtained for the
factor of r = 0.00, and a minimum of O. 1350 m d m for the factor of r =
0.15, with a variation of only 0.0250 mmím. In this group of values, an
average drying of 0.1488 mm/m is calculated.
The above results lead us to deduce that the approximations in the initial stress
levels established may be considered as uniform for all the samples. This means
that future validity is guaranteed in the comparison of the behavior of these
samples with respect to common initial conditions and constant conditions
throughout the period of study.
As for basic creep (&basic) in the CRCA samples under study, we shall say that
those with a factor r 2 0.15 presented overall a higher increase in strain than
that obtained in the reference concrete (factor r = 0.00). Thus, it may be said
that the presence of only small quantities of RCA was sufficient to alter the
strain of the CRCA samples (at least with respect to &+,=ic). Nonetheless, it is
@Seismicisolation
@Seismicisolation
also necessary to bear in mind that the range of maximum variation obtained lies
between 0.1580 mmím for the factor r = 0.00 (to = 28 days), and 0.2135 mmím
for the factor r = 0.00 (t = 270 days). This is only 0.0555 m d m , so that it
should not be considered as a parameter that would not be applicable to test
concrete. The above may be seen in graphic form in Figure 1, and also in Table
2, which presents a summary of all results obtained, and calculated for the
periods of 28,90, 180 and 270 days of testing.
In the case of calculation of coefficients of basic creep ((&,sic) obtained on the

basis of the foregoing material, it may be said that they present a tendency to
rise with an increase in the factor r , while the factor of concrete age is also
congruent with increases in these coefficients. The above may be seen in Table
2 and in Figure 5 (specific creep), where the coefficients of <Phasic are presented
in a solid line for the different r factors, at 28, 90, 180 and 270 days of testing.
In comparative terms, and taking into account the limit values of the coefficients
of <Pbsic presented, we are speaking of increases of the order of 30% more when
an CRCA with a factor of r = 1.00 is used, in comparison with the reference
concrete.
As for the total creep of the CRCA samples shower in Figure 4, it may be noted
that the behavior of these CRCA shower a gradual increase, where as the
differences increase over the testing period, as well as with the factor r . It is of
particular importance to note that, beyond factors of r 2 0.30, the total increases
in strain are corisiderably higher than they are below this factor.
The drying creep coefficients (qdrying) of the CRCA samples (see Table 2 and the
dashed lines in Figure 5 for specific creep) increase to the order of 24% in the
least of cases (factor r = 0.15), and up to 47% when the factor r = 1.00. As
may be seen, the proportions of increases in drying strain with variation in the r
factors are considerably more marked that in the case of properties attributed to
basic strain.
Finally, Figure 5 presents the behavior of CRCA samples with respect to

specific creep. For variables in basic properties (solid line), it is important to
note that practically all the r factors may be considered as constant for equal
ages of comparison while, for variables of properties in the case of drying
(dashed line), the curves are horizontal until a factor of r = 0.30, after which all
the curves increase in their upward slope. This is the case almost from the very
beginning of the tests through to their end.
@Seismicisolation
@Seismicisolation
466 Gómez-Soberón
CONCLUSIONS
Based on the results presented in this paper, the following conclusions are
reached:
Specific conclusions:
1. With respect to instantaneous strain (ci): when the CRCA samples when
submitted to equivalent levels of stress shower approximately equal
strain.
2. With respect to the basic creep (&basic) of the CRCA samples studied:
factors of r 2 0.15 presented overall a higher increase in strain than that
obtained in the reference concrete. This suggests that the presence of
small quantities of RCA in the concrete is sufficient to raise the basic
creep of the concrete.
3. The coefficients of basic creep obtained showed a tendency to

rise with an increase in the r factor, while the factor of the age of the
concrete is also congruent with an increase in these coefficients. These
increases in the coefficients of basic creep were around 30% higher
when a factor of r = 1.O0 was used.
4. The total creep in the CRCA shower a gradual increase, while the
difference increase according to testing period time and the r factor.
5. Beyond factors of r 2 0.30, the increases in total strain were

considerably higher than they are below this factor. It would seem that
an RCA content below 30% was safe in the sense that this RCA content
draws the line between the behavior of this types of concrete and normal
aggregate concrete.
6. The coefficients of drying creep (qdiying) in the CRCA samples increased

to the order of 24%, and up to 47%, higher. Here, the proportions of
increases in drying strain, with variation in the r factors, were
considerably greater than in the case of properties attributable to basic
strain.
7. The specific creep in the CRCA samples was practically constant for all
r factors, while for variables of properties through drying, the curves are
horizontal until a factor of r = 0.30, beyond which all curves increase in
their upward slope.
@Seismicisolation
@Seismicisolation
General conclusions:
1. The use of RCA in structural concrete mixtures is feasible if the

parameters and increases in creep coefficients are correctly taken into
account with respect to their behavior.
2. The development of the mechanical properties of CRCA is similar to

that of ordinary concrete, although the strength levels are lower.
3. The increase in creep properties in CRCA can be explained on the basis

of RCA properties.
4. The phenomenon of creep is of great importance in concrete mixtures

containing RCA; the characteristics and composition of RCA affect the
concrete behavior in the short and long terms. This brings about
variations in the creep coefficients, therefore the use of RCA will be
restricted if these coefficients are either overlooked or assumed as being
identical to those of conventional concrete.
ACKNOWLEDGEMENTS
The author would like to express their gratitude to the company Cervezas Finas
de Cd. Serdán, S. A. de C. V., Puebla, Mexico by the partial financing of this
investigation; to the Technical University of Catalonia, Barcelona, Spain by the
use of their facilities and to the Dr. L. Agulló F. and to the Dr. E. Vázquez R.
REFERENCES
1. Environmental Council of Concrete Organizations. http://www.ecco.org
2. Hansen, T. C., and E. B@egh. (1995) “Elasticity and drying shrinkage of

recycle-aggregate”. AC1 Journal, No. 5. Pp. 648 - 652, September -
October.
3. Hansen, T. C. (1992) “Recycled of Demolished Concrete and Masonry”

Report of Technical Committee 37-DRC Demolition and Reuse of
Concrete. Part one. Pp. 1 - 160.
4. Bairagi, N. K., K. Ravande and V. K. Pareek. (1993). “Behavior of

concrete with different proportions of natural and recycled
@Seismicisolation
@Seismicisolation
468 Górnez-Soberón
aggregates”. Resources, Conservation and Recycling. Elsevier
Science Publishers B.V. Vol. 9. Pp. 109 -126.
5. Sakata, K.and T. Ayano. (2000). “Improvement of concrete with recycled

aggregate”. CANMET/ACI. Durability of Concrete. Proceedings
Fifth International Conference. Barcelona, Spain 2000. Vol. II. Pp.
1089 - 1108. Edited by V. M. Malhotra.
6. Gomez, Jose M., E. Vasquez y L. Agullo. (2001). “Strength and

deformation properties of recycled aggregate concrete”. Fifth
CANMET/ACI International Conference on Recent Advances in
Concrete Technology. Supplementary Papers. Pp. 103 - 120.
Singapore. August 2001. Compiled by Venturino, M.
7. Gomez, Jose M., L. Agullo y E. Vasquez. (2001). “Physical and

mechanical properties of the recycled concrete aggregates.
Application in concrete”. Technology and Construction. Magazine
of the Mexican Institute of the Cement and the Concrete, A. C.
(IMCYC). Vol. XlIi, Num. 157, ISSN 0187-7895. Pp. 10 - 22. July
2001. Mexico D. F., Mexico. (In Spanish).
htm
http://www.imcyc.com/cyt/junio/cualidades.
8. Henrichsen, A. (2001). “Creep and shrinkage in concrete with recycled

aggregate -an- attempt to review -the present state-of-the-art”.
Danish Recycled Cooperation. Dansk Beton Teknik A / S . (personal
conference)
9. Sato, R., K. Kawai and Y. Baba. (2000). “Mechanical performance of

reinforced recycled concrete beams”. Intemational workshop on
recycled concrete. JSPS 76 Committee on Construction Materials.
Pp. 127 - 146
10. Nishibayashi, S . and K. Yamura. (1988). “Mechanical properties and

durability of concrete from recycled coarse aggregate prepared by
chousing concrete”. Demolition and Reuse of Concrete and
Masonry. Vol. 2, Reuse of demolition waste. Pp. 652 - 659,
November. Edited by Kasai, Y.
11. RILEM Recommendation. (1994). “121-DRG Guidance for demolition

and reuse of concrete and masonry. Specifications for concrete with
recycle aggregates”. Materials and Structures, No. 27. Pp. 557 - 559
12. http://www.imcyc.com/revista/l998/junio/rilem.htm
13. Fujii, T. (1988). “Strength and drying shrinkage behavior of concrete

using concrete crushed aggregate”. Demolition and Reuse of
@Seismicisolation
@Seismicisolation
Concrete and Masonry. Vol. 2, Reuse of demolition waste. Pp. 660 -
669, November. Edited by Kasai, Y.
14. Wainwright, P. J., A. Trevorrow, Y. Yu and Y. Wang. (1993).

“Modifying the performance of concrete made with coarse and fine
recycled concrete aggregates”. Demolition and Reuse of Concrete
and Masonry, Guidelines for Demolition and Reuse of Concrete and
Masonry. Pp. 3 19 - 330, October. Edited by Lauritzen, E. K.
15. Gómez, José M., L. Agulló and E. Vázquez. (2001). “Repercussions on

concrete permeability due to recycled concrete aggregate”. Third
CANMET/ACI International Symposium on Sustainable
Development of Cement and Concrete. ISBN 0-8703 1-041-0. SP
202-13. Pp. 181-195. September 2001. San Francisco, USA. Edited
by V. M. Malhotra.
16. Gomez, Jose M., L. Agullo y E. Vasquez. (2001). “Relationship between

porosity and concrete properties with natural aggregate replacement
by recycled concrete aggregate”. 200 1 Second Intemational
Conference on Engineering Materiais. ISBN 1-894662-00-8. Vol. 1.
Pp. 147 - 156. August 2001. San Jose California, USA. Edited by
Nagataki, S., A. Al-Manaseer and K. Sakata.
17. Gomez-Soberon, Jose M., E. Vasquez and L. Agullo. (2001). “Shrinkage

and creep of recycled concrete interpreted by the porosity of their
aggregate”. Intemational Congress - Challenges of Concrete
Construction. Dundee University. Scotland, RU. (Accept)
18. Gomez, Jose M., E. Vasquez and L. Agullo. (2001). “Concrete with
aggregate recycled. A guideline for the material”. Monograph M60-
2001. ISBN: 84-89925-80-1. Pp. 1 - 137. Barcelona, Spain. Edited
by International Center of Numerical Methods in Engineering
(CMNE). (In Spanish)
19. Wainwright, P. J. and J. G . Cabrera. (1994). “Use of demolition concrete

to produce durable structural concrete”. Environmental Aspects of
Construction with Waste Materiais. Pp. 553 - 562. Edited by
Goumans, J. J. J. M., H. A. van der Sloot and Th. G., Aalbers.
20. Uchikawa, H. and S. Hanehara. (1996). “Recycling of concrete waste”.

Concrete for Environment Enhancement and Protection. Pp. 163 -
172. Edited by Dhir, R.K. and T. D. Dyer.
21. Roos, F. (1998). “Verification of the dimensioning values for concrete

with recycled concrete aggregate”. Uses of Recycled Concrete
@Seismicisolation
@Seismicisolation
470 Gómez-Soberón
Aggregate. Sustainable Construction. Pp. 309 - 319. Edited by K.
Dhir, R., N. A. Henderson and M. C. Limbachiya.
22. Mark, G. A. (1996). “Aggregates and the deformation properties of

concrete”. AC1 Materiais Journal. Pp. 569 - 577, November-
December.
23. Neville, A. (1989). “Elasticity, shrinkage and creep”. Technology of the

Concrete. Pp. 209 - 289. (In Spanish)
24. Neville, A. (1964). “Creep of concrete as a function of its cement paste

content”. Mag. Concrete Res. 16. No. 46. Pp. 21 - 30. March.
25. Ayano, T. and K. Sakata. (2000). “Improvement of concrete with

recycled aggregate”. Intemational seminar on recycled concrete.
Sponsored by Niigata University and Japan Concrete Institute (JCI)
Pp. 79 - 86
26. Lopez Javier. (1996). “The new Euro-code 2 Part 1-4: Concrete of
aggregate light of closed texture”. Ready Concrete, No. 27. Pp. 38 -
42, September. (In Spanish)
27. Gomez-Soberon, Jose M. V. (2001). “Shrinkage of concrete with

replacement of aggregate with recycled concrete aggregate”. AC1
5th International Conference. Cancun, Mexico.
28. Smadi, Mohammad M., F. O. Slate and A. H. Nilson. (1987). “Shrinkage

and creep of high, medium and low strength concretes, including
overloads”. AC1 Materiais Journal. Pp. 224 - 234. May-June.
@Seismicisolation
@Seismicisolation
Table 1 Test mixtures and conditions for the cree? test.
Test Test Age of beginning
Factor r, ac,r,o,,f', of (to), Ec instantaneous Ec instantaneous
MPa MPa (days) basic (mdm.) drying (mmim.)
r= 1.00 -34.50 12.08 0.1370 0.1580

r =0.60 -35.80 12.53 0.1470 0.1530
r =0.30 37.00 12.95 28 O. 1645 0.1380
r =0.15 38.10 13.34 0.1350 0.1350
r = 0.00 -
38.80 13.58 0.1430 O. 1600
oc 38.40 13.44 200 0.1180 O. 1265
lote: Y =percentage of natural aggregate replaced by recycled aggregate concrete. OC =
original concrete.
Table 2 Creep of test mixtures containing recycled concrete aggregate.

I 1
Specific
Test Ec basic
Factor (mmim.) creep
duration (I/MPa)
Note: r =percentage of natural aggregate replaced by recycled aggregate concrete. OC =

original concrete.
@Seismicisolation
@Seismicisolation
472 Gómez-Soberón
Figure 1 Specimens in climatic chamber.
suppon dsc I : Embedded -leadm
Figure 2 Supported load frame.
@Seismicisolation
@Seismicisolation
0.35
h
.-
E
v1
0.20
0.15
0.05
o.1o
0.00 4
1
0 1500 3000 4500
Time (hours)
Figure 3 Basic creep for different recycled concrete.
0.40
oc
0.20 - Time (hours)
0.004
0 1500 3000 4500 m
Figure 4 Total creep for different recycled concrete.
@Seismicisolation
@Seismicisolation
474 Gómez-Soberón
0.10
0.09
0.08
0.07 2
0.06 3 z
P
0.05
U
0.04 8
0.03 4
0.02
0.01
I
0.00
I = 0.00 r = 0.15 r = 0.30 I = 0.60 I= 1.00
Factor r
Figure 5 Specific creep for different recycled concrete.
@Seismicisolation
@Seismicisolation
SP 209-26
Shrinkage of Concrete with

Replacement of Aggregate with
Recycled Concrete Aggregate
by J. M. V. Gómez-Soberón
SvnoDsis: In this paper we present the experimental analysis of samples of con-

crete where portion of the natural aggregate were replaced with recycled aggregate
originating from concrete (RCA). Experimental analysis to obtain the shrinkage
properties (basic and dried) of the concrete containing recycled concrete aggregate
(CRCA) was performed. The percentages of replacement of natural aggregate with
RCA were O%, 15%, 30%, 60% and 100% with test conditions of 50% RH and
20°C. The results of these trials are compared with reference concrete tests, at an
age of 270 days. The results demonstrated an increase in the shrinkage of the
CRCA that is proportional to the amount of RCA used as a replacement for the
natural aggregate. When compared to the reference concrete, the drying shrinkage
showing significant changes; however, their evolution over time is similar to stan-
dard concrete.
Keywords: basic shrinkage; recycled concrete;

recycled concrete aggregate; shrinkage; shrinkage by
drying
@Seismicisolation
@Seismicisolation
475
476 Gómez-Soberón
José M. V. Gómez-Soberónis Civil Engineer of the Autonomous University of
Puebla in Mexico and studied the M.Eng. (Construction) at the National
University of Mexico. Presently he is doing Ph.D at the School of Civil
Engineering of the Technical University of Catalonia, Spain. He has published
numerous papers on the area of Recycled Concrete in International Journals and
Conferences and works as consulting engineer. jmvl15 @hotmail.com
INTRODUCTION
It is known that the use of recycled concrete aggregate (RCA) in concrete

involves an increase in the mortar content in the resulting concrete, and
therefore, potential increases in shrinkage strain. Parallel to this, Kikuchi,
Yasunaga and Ehara [i], together with Tavakoli and Soroushian [2], report the
high importance of RCA quality with respect to the increase in strain due to
shrinkage (Eh); or, stated in another way, the importance of the wlc ratio of the
concrete which produced the RCA, its service history and period of hardening
[3, 41, together with the process used in its production (a high level of
refinement or elimination of the old mortar adhering to the natural aggregate
translates into a reduction of shrinkage) [4]
In addition to the above, the purity of the employed RCA should be taken into
account as a factor for the increase in shrinkage reported in the concrete
containing recycled concrete aggregate (CRCA). The type and amount of
contaminants that might be present in these aggregates have been reported by
Yanagi, Nakagawa, Hisaka and Kasai [5] as another of the causes for the
differences (they can generate increases of 129% to 150%) in the shrinkage
behavior of the CRCA. In other papers on CRCA [6, 7, 8, 9, lo], the authors
also discuss the shrinkage increases, which are usually found and establish
typical ranges of increases, between 10% and 25% more compared to reference
concretes. When ai1 of the natural aggregates (both coarse and fine fractions) are
replaced by RCA, increases in shrinkage may even reach 100%. Other research,
also involving CRCA [I. 11, 12, 13, 141, has shown that the increase in coarse
aggregate replacement correlates with the increase in shrinkage; this increase
appears to be more plausible when the RCA exceeds a threshold percentage of
the total aggregate content (2 30%) [13, 14, 151
In addition to these relationships, as with ordinary concretes, the CRCA algo

presents the correlations known in all concretes; an example of this is that
existing between the wlc ratio of the concrete and the shrinkage increase [2,4]
Finally, with respect to drying shrinkage (&h dry) of CRCA only, the papers
@Seismicisolation
@Seismicisolation
published up to now may be summarized as follows:
1. The increase in drying shrinkage of CRCA is imputable to the old mortar

adhering to the natural aggregate, and also due to the content,
interconnection and distribution of pore size [16, 17, 181.
2. The values may range from 20% to 70% more than the reference
concrete, and when 100% of the aggregate is replaced, this increase may
exceed 70% [19,20, 21, 22, 23, 241, even reaching values of up to 263%
more [25]
3. Attenuation of the increases in drying shrinkage of CRCA may be

feasible through suitable use of shrinkage-reduction agents [26]
The main purpose of this paper is to explain on and provide practical

information concerning the properties of drying shrinkage of CRCA, for which
confirmation is sought in the scientific basis of their behavior and thus, to
provide a solution to current application requirements.
EXPERIMENTAL DETAILS
Original Concrete
In this study 4 m3 of original concrete (OC) was used. The concrete was made in
a mixer and was placed into wooden formwork frames measuring 0.40 x 0.20 x
0.10 m. Fifty cylinders measuring 0.15 m in diameter x 0.30 m in height and
four cubes O. 10 m were also used to study the porosity and mechanical behavior
of oc.
Twenty-four hours after casting, the specimens were removed from the
formwork and placed in the curing regime for 150 days (see Table 1, where the
specific characteristics of this concrete are given). The specimens were then
passed once through a semi-fixed roller grinder with % inlet width of 0.45 m
and a maximum outlet size of 0.025 m. Finally, the resulting material was
classified into sizes: 0-5, 5-10, 10-20 and 20-25 mm. The 5-10 and 10-20 mm
fractions were used as RCA in this work.
Recycled Aggregate and Natural Aggregate
The designation used by sizes was: for RCA, gravel 10-20 mm and fine gravel 5-10
mm; and for the natural aggregate (NA), gravel 12-20mm and fine gravel 5-12 mm.
@Seismicisolation
@Seismicisolation
478 Gómez-Soberón
The criterion used for this fit was the compacted maximum density (which
reduced the possible influences of different particle size). These were:
> For RCA the combination was 55% gravel and 45% fine gravel.
> For NA the combination was 70% gravel and 30% fine gravel.
Table 2 shows the properties of the aggregate used. The RCA used in this study
can be considered as being within the RILEM recommendation for TYPE II
RCA (absorption 2 10% and D, 2 2000 kg/m3); for the Belgian recommendation
they are GBSBII (absorption e 9% and D, > 2100 kg/m3); and in the Japanese
case they comply with the absorption requirement (absorption 5 7% and D, 2
2200 kg/m3) in the fractions used [27, 28, 29, 301 Consequently, the RCA
employed in this study may be used in both plain and reinforced concrete if its
application and factors of behavior are taken into account.
Mixture of Recvcled Concretes
Due to the difficulty in determining the real w/c because of the high variation of
absorption in the RCA, it was decided to use basic AC1 21 1.1 and AC1 21 1.2
mixture concepts proportioning with the following criteria:
1. The substitution of RCA for NA was done using equal volume fractions
with the following condition:
(0.005r _< 1.00)
Where: r = percentage of NA replaced by RCA, by volume; RCAc0,, =

55% recycled gravel + 45% recycled fine gravel; NA,,,,, = 70% natural
gravel + 30% natural fine gravel.
The aggregate replacement percentages of the five samples of CRCA

were: r = 0.00, 0.15, 0.30,0.60 and 1.00. The fine aggregate used was,
100% crushed natural limestone sand from the Garraf quarry, Barcelona.
2. The RCA showed an increase in absorption proportional to the time

spent in water. An immersion time of 20 minutes was used for the
concrete mixtures, with up to 97% fine gravel and 77% gravel, and
compared with the 24 hour absorption in all cases.
3. The amount of water absorbed by the aggregate was taken into account
separately, in addition to its surface moisture before mixing and the free
@Seismicisolation
@Seismicisolation
water formed part of the mixture water. The above aspect is justified in
criteria that were emphasized in previous publications of the authors [ 13,
14, 18, 31, 32, 331
With the established mixing time and the required amount of water, the order of
mixing the materials guaranteed (as far as possible) the immobility of the water
and an improvement in the transition zone. The following sequence was
adopted: (a) all of the coarse aggregate and water was introduced in the mixer;
(b) these were mixed for 2 minutes; (c) the mixer was switched off for 3
minutes; (d) stages b and c were repeated twice; (e) the cement was introduced
and mixed for 3 minutes; and (f) the sand was added and mixed for another 3
minutes.
The mixtures obtained using the above criteria are given in Table 3. The
variation in consistency and volumetric proportions for the different percentages
of aggregate replaced are within tolerable limits (slump 0.1 i 0.03 m and
concrete with volumetric proportion normal)
ProDerties of the Concretes
Tests were conducted on the different concretes included in the study to

determine physical properties such as absorption, density, and porosity; and
mechanical properties such as compressive and tensile strength, and Young’s
modulus. Table 4 shows the results of the tests, in which the Spanish Norm were
used. For additional details on similar work, see previous publications by the
authors [13, 14, 18, 31,32, 331
Shrinkwe of the Concretes Çonîainine RCA
Shrinkage tests were conducted on four cylinders measuring O. 15 m in diameter

x 0.45 m in height for each of the proposed variables (28 in total). The strain
measurements were made with embedded MM series EGP-type gauges fitted in
the center of the concrete specimens. The specimens were kept in a curing
chamber for 28 days (T = 20°C 2 2 and RH = 90% 2 5), after which they were
moved to a climate chamber (T = 20°C and RH = 50%) until the end of the test
period (270 days). The initiai deformation measurements (to) were made 24
hours after the specimens had set. When the specimens were moved to the
climate chamber, the surfaces of two of the samples for each variable under
study (for basic shrinkage) were sealed with paraffin (3-mm-thick layer) and
finally wrapped in three sheets of aluminium foil to prevent seepage of water
from the chamber environment.
@Seismicisolation
@Seismicisolation
Gómez-Soberón
Figure 1 shows four samples during the preparation phase before commencing
the drying process in the climatic chamber. From left to right, the first is a test
specimen that is used to determine total shrinkage (un-coated surface) and the
remaining three are test specimens used to determine basic shrinkage (the
middle two were coated in paraffin, and the last was also wrapped in sheets of
aluminum foil). Figure 2 show 12 of the 24 test specimens used in this
experimental project during study in the climatic chamber.
Figures 3 and 4 show the behavior of the specimens during the initial curing
chamber process and later climatic chamber drying. In all the figures, the curves
represent the mean value for two test specimens, in which the expansion strains
are expressed arbitrarily with a negative sign (-), whereas a positive sign (+) is
used for shrinkage effects. It can be seen from these figures that the OC samples
undergo greatest expansion (also known as swelling) during the curing chamber
process. This is explained by the fact that, unlike the other CRCA samples, the
OC samples undergo a prior period of ageing (150 days of curing conditions and
15 further days under laboratory conditions).
This ageing period, especially the 15 days of drying under laboratory conditions,
leads to the loss of a large percentage of the water retained in the OC capillary
pores, and when the test specimens are again placed in an environment with a
high relative humidity, the expansion process is accelerated.
With respect to the CRCA samples, previous research work papers [13, 14, 34,
35, 36, 37, 381 have discussed the existence of certain differences concerning
the expansion increase, together with the similarity in behavior with light or
porous concretes. For our own study, we will say that the behavior of the CRCA
samples shows an agreement between the replacement factor r and the
expansion experienced during this first curing chamber phase, although it is
important to note that there is also a notable separation of the CRCA curves into
two well differentiated groups:
1) The curves for the variables r = 0.00, r = 0.15 and r = 0.30 are
grouped together in the low or medium range (expansions in the order of
-0.045 m d m )
2) The curves for the variables r = 0.60 and r = 1.00 are grouped together
in the range of maximum or high expansion (expansions in the order of -
0.109 m d m , which, in comparative terms, represents an increase of 2.5
times greater expansion).
After the time spent by the test specimens in the climatic chamber, there is-a
clear difference in behavior between those with free water loss (total shrinkage
‘‘Gh totd”) and those which were sealed with paraffin wax and wrapped in
aluminum foil (basic shrinkage ‘‘esh
basic)’)
@Seismicisolation
@Seismicisolation
With regards to the basic shrinkage curves (Figure 3), it should be stated that the
variables r = 0.60 and r = 1.00 rapidly recover part of the expansion
undergone in the first stage and can even reach the other variables under study.
The reference concrete is reached at approximately 4,500 hours (this recovery
being approximately 50% of the strain caused by expansion), and finally, these
variables demonstrate the greatest levels of expansion recovery (around 68% of
the expansion undergone).
The remaining variables under study ( r = 0.00, r = 0.15 and r = 0.30) show
slight recovery of the expansion strain at the beginning of the second period,
after which they remain practically constant until the end of the study period
(the expansion recovery of this set of variables only reached 22% of the
expansion strain undergone).
For those test specimens showing total shrinkage during the climatic curing
stage (Figure 4), it can be said that their behavior is notably similar for all
samples at a young age; however, over time, the increase in the r factor reveals
a behavior that is more proportional to the strain suffered by total shrinkage,
which in turn is even more accentuated with time. The order of magnitudes of
total shrinkage for CRCA is quite significant when compared with the reference
concrete; with extreme values of 0.561 m d m for r = 1.00 compared to 0.394
m d r n for r = 0.00. These values of total shrinkage translate into an increase of
1.4 times more total shrinkage at an age of 270 days, these values being
obtained from the accumulated behavior-fromthe-curing chamber stage.
With respect to the OC variable, referring to total shrinkage and for the climatic
chamber stage, it can be said that its curve rises rapidly at the beginning, then
continues to rise very steeply until it reaches a final total shrinkage value of only
0.065 m d r n . This can be explained by the fact that this concrete possesses a
much greater level of maturity than the other samples studied, so that any
possible increase in total shrinkage is predictably much less.
In order to obtain the real strain values for each of the properties in this study
(only during the climatic chamber period), and thus make them useful for the
design coefficient calculations, it was n e c e s s q to eliminate the values
corresponding to the first stage of strain (curing chamber). In order to do this,
the drying start time in the climatic chamber was set as zero (in other words to =
28 days became to = O days) and the corresponding strains were also set as zero,
with the result that shrinkage (to = 28 days) = O m d m . Commencing with this
standard, the other time and strain measurements were recalculated in relation to
these points, and it was then a simple matter to obtain the real strain values-for
the climatic chamber stage.
Finally, effect superposition criteria were applied to each pair of total (&h
and basic (&sh basic) shrinkage values in order to obtain the real drying shrinkage
values (Gh dry).
@Seismicisolation
@Seismicisolation
482 Gómez-Soberón
Figures 5 and 6 show the curves of basic shrinkage and total shrinkage after
having carried out these modifications. In the first (Figure 5), it can be seen that
the previously described grouping criterion continues to be applicable. On the
one hand, the variables r = 0.60 and r = 1.00 can be located in the zone of
maximum strain, whereas the variables r = 0.00, r = 0.15 and r = 0.30 are
located in the zone of low strain.
It is important to take note of the fact that these last three variables (in the same
way as OC) still express the strain in the expansion zone; however, their values
should be considered as being the product of the basic shrinkage phenomenon.
This can be explained by the possibility that the hydration of the cement
particles may be restricted not by lack of water, or by having exhausted the
cement’s hydration potential, but by the chemical bonds existing prior to this
period in the climatic chamber. This results in a blocking of the space available
for the cement particles to form new hydrated compounds. On the other hand,
when dealing with concrete having a high content of recycled concrete
aggregate (RCA), its high porosity and specific surface area, together with the
possibility of contributing non-hydrated cement particles [ 13, 14, 18, 31, 32,
331, will allow the hydration to be prolonged for more time and thus achieve a
better hydration process.
In quantitative terms, Table 5 contains the calculated values for the three types
of strain (&sh basic, &sh total and &sh d i y ) which were obtained from the climatic
chamber stage. These values correspond to the usual age points that enable
references or comparisons to be made with the mechanical properties previously
above. In absolute values, the mean value of basic shrinkage for r = 1.00 and r
= 0.60 is 0.050 m d m , whereas for the group r = 0.30, r = 0.15 and r = 0.00
this is only 0.007 mm/m at an age of 270 days; in other words, CRCA with high
RCA content may reveal a value of basic shrinkage which is 7 times that of the
reference concretes, with the disproportion being a direct function of concrete
age.
With regards to total shrinkage (see Figure 6 ) , the proportions calculated as a

function of the reference concrete ( r = 0.00) are: 1.03 times more for r = 0.15,
0.98 times for r = 0.30, 1.25 times for r = 0.60, 1.25 times for r = 1.00, and
0.48 for OC at an age of 270 days. Finally, the reported drying shrinkage (see
Figure 7 and Table 5) indicates that the shrinkage rate is high at early ages, and
then slows down with time (after 180 days, for practical purposes the strain
increase due to drying becomes a low constant rate). In addition to this, in the
same way as the other properties mentioned above, replacement factors of RCA
by natural aggregate above 30% ( r 2 0.30) cause a rapid increase in the values
of drying shrinkage.
@Seismicisolation
@Seismicisolation
CONCLUSIONS
Based on the research and results presented in this paper, the following
conclusions are reached:
Specific conclusions:
1. With regards to the expansion reported for the CRCA, we would say that
this expansion is correlative with the factor r ; and that in addition, the r
2 0.60 factors show expansion increases, which are 2.5 times greater
than the r c 0.60 factors.
2. As far as the expansion recovery process is concerned, the factors

contained in the range of r 2 0.60 require approximately 187 days to
reach the other variables, ending in the recovery of an average of 68% of
the expansion undergone at an age of 270 days, whereas the factors
falling in the range of r I0.30 were only able to recover an average of
22%of the expansion undergone at the same age.
3. With respect to the basic shrinkage, the factors within the range r 2 0.60
show an average strain of 0.050 m d m against an average of 0.007
m d m for the factors in the range of r I0.30 (which is the equivalent of
7 times less).
4. For the property total shrinkage, the strain proportions seen as a function
of the reference concrete are: r = 0.15 of 1.03 time more, 0.98 for r =
0.30, 1125 for r = 0.60, 1.25 for r = 1.00, and 0.48 for OC at an age of
270 days.
5. Finally, the separation of the variables into two groups as stated above is
also consistent with drying shrinkage; this threshold in the RCA content
percentages leads to considerable CRCA strain increases.
General conclusions:
1. The use of RCA for the creation of CRCA is a viable, profitable

technique provided that suitable conditions are employed in its
implementation.
2. The shrinkage phenomenon is crucial in CRCA, given the characteristics

and composition of the RCA, since these affect both the short and long-
@Seismicisolation
@Seismicisolation
484 Górnez-Soberón
term behavior of this type of concrete. This causes variations in their
behavior coefficients and restricts their application if they are omitted or
compared to conventional concretes.
3. The CRCA properties differ from those of ordinary concrete, including

considerable increases in shrinkage. These increases correspond to the
high porosity and permeability of the RCA, since they are only partly
constituted (part old mortar and part natural aggregate).
4. The increase in the availability of water and hence the possibility of

movement until water balance is attained with the surrounding
environment cause both the drying shrinkage and its rate to increase in
the CRCA.
5. In the case of basic shrinkage, the variations may be small in quantitative

terms; however, there is a clear correlative increase with the factor r .
6. It therefore appears that an RCA content which is less than 30% in the
CRCA is both safe and feasible, provided that shrinkage inhibitors are
not employed
ACKNOWLEDGEMENTS
The author would like to express their gratitude to the company Cervezas Finas
de Cd. Serdán, S. A. de C. V., Puebla, México by the partial financing of this
investigation; to the Technical University of Catalonia, Barcelona, Spain by the
use of their facilities and to the Dr. L. Agulló and to the Dr. E. Vázquez.
REFERENCES
1. Kikuchi, M., A. Yasunaga and K.Ehara. (1993). “The total evaluation of

recycled aggregate and recycle concrete”. Demolition and Reuse of
Concrete and Masonry, Guidelines for Demolition and Reuse of
Concrete and Masonry. Pp. 367-377, October. Edited by Lauritzen,
E. K.
2. Tavakoli, M. and P. Soroushian. (1996). “Strengths of recycled aggregate

concrete made using field-demolished concrete as aggregate”. AC1
Materials Journal. Pp. 182-189, March -April.
3. Tazawa, E. and S . Miyazawa. (1993). “Autogenous shrinkage of concrete

and its importance in concrete technology”. Creep and Shrinkage of
@Seismicisolation
@Seismicisolation
Concrete. Proceeding of the Fifth International RILEM Symposium.
Edited by Bahnt Z. P. and I. Carol. Pp. 159-168
4. Fujii, T. (1988):“Strength and drying shrinkage behavior of concrete using

concrete crushed aggregate”. Demolition and Reuse of Concrete and
Masonry, Vol. 2, Reuse of demolition waste. Pp. 660-669,
5. Yanagi, K., M. Nakagawa, M. Hisaka and Y. Kasai. (1988). “Effect of

Impurities in recycled coarse aggregate upon a few properties of the
concrete produced with it”. Demolition and Reuse of Concrete and
November. Edited by Kasai,Y.
6. Sri Ravindrarajah, R. and C. T. Tam.(1988). “Methods of improving the

quality of recycled aggregate concrete”. Demolition and Reuse of
Concrete and Masonry, Vol. 2, Reuse of demolition waste. Pp. 575-
7. Hansen, T. C., and E. Bwgh. (1995). “Elasticity and drying shrinkage of

recycle-aggregate”. AC1 Journal, No. 5. Pp. 648-652, September -
October.
8. Nixon, P. J. (1979). “Recycled of the concrete”. Technology and

Construction. Magazine of the Mexican Institute of the Cement and
the Concrete, A. C. (IMCYC). Vol. 17. No. 102. Pp. 35-44, Mexico
D. F., Mexico. October. (In Spanish)
9. Yanagi, K., M. Hisaka and Y. Kasai. (1993). “Physical properties of

recycled concrete using recycled coarse aggregate made of concrete
with finishing materials”. Demolition and Reuse of Concrete and
Masonry, Guidelines for Demolition and Reuse of Concrete and
Masonry. Pp. 379-390, October. Edited by Lauritzen, E. K.
10. Ujike, Isao. (2000). “Air and water permeability of concrete with
recycled aggregate”. International workshop on recycled concrete.
JSPS 76 Committee on Construction Materials. Pp. 95-106
11. Matsushita, H.,H. Tsuruta and Takao Chikada. (2000). “Experimental

studies on compressive strength and drying shrinkage of recycled
aggregate concrete”. International seminar on recycled concrcte.
Sponsored by Niigata University and Japan Concrete Institute (JCI).
Pp. 69-78
12. Hasaba, S., M. Kawamura, K. Toorii and K. Takemoto. (1981). “Drying

shrinkage and durability of the concrete made of recycled concrete
@Seismicisolation
@Seismicisolation
486 Gómez-Soberón
aggregates?. Transactions of the Japan Concrete Institute. Vol. 3. Pp.
55-60
13. Gómez, José M., E. Vázquez y L. Agulló. (2001). ?Strength and

deformation properties of recycled aggregate concrete?. Fifth
CANMET/ACI International Conference on Recent Advances in
Concrete Technology. Supplementary Papers. Pp. 103-120.
Singapore. August 2001. Compiled by Venturino, M.
14. Gómez, José M., L. Agulló y E. Vázquez. (2001). ?Physical and

mechanical properties of the recycled concrete aggregates.
Application in concrete?. Technology and Construction. Magazine
of the Mexican Institute of the Cement and the Concrete, A. C.
(IMCYC). Vol. Xm,Num. 157, ISSN 0187-7895. Pp. 10-22, July
2001. Mexico D. F., Mexico. (In Spanish).
http://www.imcyc.corn/cytíjunio/cualidades.htm
15. Kashino, N. and Y. Takahashi. (1988). ?Experimental studies of

placement of recycled aggregate concrete?. Demolition and Reuse of
Concrete and Masonry, Vol. 2, Reuse of demolition waste. Pp. 557-
16. Nishibayashi, S. and K. Yamura. (1988). ?Mechanical properties and

durability of concrete from recycled coarse aggregate prepared by
chousing concrete?. Demolition and Reuse of Concrete and
17. Sakata, K. and T. Ayano. (2000). ?Improvement of concrete with

recycled aggregate?. CANMET/ACI. Durability of Concrete.
Proceedings Fifth International Conference. Barcelona, Spain 2000
Edited by V. M. Malhotra Vol. II. Pp. 1089-1108.
18. Gómez, José M., L. Agulló y E. Vázquez. (2001). ?Relationship between

porosity and concrete properties with natural aggregate replacement
by recycled concrete aggregate?. 200 1 Second International
Conference on Engineering Materials. ISBN 1-894662-00-8. Vol. 1.
Pp. 147-156. August 2001. San José California, USA. Edited by
Nagataki, S., A. Al-Manaseer and K. Sakata.
19. Environmental Council of Concrete Organizations. http://www.ecco.org
20. Hansen, T. C. (1992) ?Recycled of Demolished Concrete and Masonry?.

Report of Technical Committee 37-DRC Demolition and Reuse of
Concrete. Part one. Pp. 1-160.
@Seismicisolation
@Seismicisolation
21. Bairagi, N. K., K. Ravande and V. K. Pareek. (1993). “Behavior of
concrete with different proportions of natural and recycled
aggregates”. Resources, Conservation and Recycling. Elsevier
Science Publishers B.V. Vol. 9. Pp. 109 -126.
22. Vries, P. (1995). “Recycled materials for concrete. A report on the Dutch
experience”. Journal Article. Quarry Mabagement. Vol. 22. No. 12.
Pp. 23-26, December.
23. Ajdukiewicz, A. B. and A. T. Kliszczewicz. (1996). “Properties of

structural concrete with rubble aggregate from demolition of RCPC
structures”. Concrete for Environment Enhancement and Protection.
Pp.73-80. Edited by Dhir, R.K. and T. D. Dyer.
24. Ravindrarajah, R., M. Stewart and d. Greco. (2000). “Variability of

recycled concrete aggregate and its effects on concrete properties: A
case study in Australia”. International workshop on recycled
concrete. JSPS 76 Committee on Construction Materials. Pp. 9-25.
25. Henrichsen, A. (2001). “Creep and shrinkage in concrete with recycled

aggregate -an attempt to review the present state - of - the - art”.
Danish Recycled Cooperation. Dansk Beton Teknik NS. (personal
conference)
26. Yamato, T., M. Soeda and Y. Emoto. (2000). “Physical properties of

recycled aggregate and the utilization as concrete aggregate.
International Seminar on Recycled Concrete. Sponsored by Niigata
University and Japan Concrete Institute (JCI). Pp. 59-68.
27. RILEM Recommendation. (1994). “121-DRG Guidance for demolition

and reuse of concrete and masonry. Specifications for concrete with
recycle aggregates”. Materials and Structures, No. 27. Pp. 557-559.
28. Hendriks, Ch. F. (1994). “Certification system for aggregates produced

from building waste and demolished buildings”. Environmental
Aspects of Construction with Waste Materials. Edited for Goumans,
J. J. J. M., H. A. Van Der Sloot and T. G. Pp. 821-843.
29. Vyncke, J. and E. Rousseau. (1993). “Recycling of construction and

waste in Belgium: actual situation and future evolution”. Demolition
and Reuse of Concrete and Masonry, Guidelines for Demolition afïd
Reuse of Concrete and Masonry. Edited for Lauritzen, E. K,
October. Pp. 57-69.
30. Kasai, Y. (1993). “Guidelines and the present state of the reuse of
demolished concrete in Japan”. Demolition and Reuse of Concrete
@Seismicisolation
@Seismicisolation
488 Gómez-Soberón
and Masonry, Guidelines for Demolition and Reuse of Concrete and
Masonry. Edited for Lauritzen, E. K, October. Pp. 93-104.
31. Gómez, José M., E. Vázquez and L. Agulló. (2001). “Concrete with
aggregate recycled. A guideline for the material”. Monograph M60-
2001. ISBN: 84-89925-80-1. Pp. 1- 137. Barcelona, Spain. Edited
by International Center of Numerical Methods in Engineering.
(CIMNE). (In Spanish)
32. Gómez, José M., L. Agulló and E. Vázquez. (2001). “Repercussions on

concrete permeability due to recycled concrete aggregate”. Third
CANMET/ACI International Symposium on Sustainable
Development of Cement and Concrete. ISBN 0-8703 1-041-0. SP
202-13. Pp. 181-195. September 2001. San Francisco, USA. Edited
by V. M. Malhotra.
33. Gómez-Soberón, José M., E. Vázquez and L. Agulló. (2001). “Shrinkage

and creep of recycled concrete interpreted by the porosity of their
aggregate”. International Congress - Challenges of Concrete
Construction. Dundee University. Scotland, RU. (Accept)
34. Mulheron, M. (1988). “The recycling of demolition debris: Current

practice, products and standards in the United Kingdom“.
Demolition and Reuse of Concrete and Masonry, Vol. 2, Reuse of
demolition waste. Pp. 510-519, November. Edited by Kasai, Y.
35. López Javier. (1996). “The new Euro-code 2 Part 1-4: Concrete of
aggregate light of closed texture”. Ready Concrete. No. 27. Pp. 38-
42, September. (In Spanish)
36. Miyazawa, Shingo, T. Kuroi and R. Sato. (2000). “Fatigue behavior of

reinforced concrete beam with recycled coarse aggregate”.
International workshop on recycled concrete. JSPS 76 Committee on
Construction Materials. Pp. 147-156
37. Sato, R., K. Kawai and Y. Baba. (2000). “Mechanical performance of

reinforced recycled concrete beams”. International workshop on
recycled concrete. JSPS 76 Committee on Construction Materials.
Pp. 127-146
38. Yamato, T., Y. Emoto and M. Soeda. (1998). “Mechanical properties,

drying shrinkage and thawing of concrete using recycled aggregate!’.
AC1 Recent advances in concrete technology. Pp, 105-122.
@Seismicisolation
@Seismicisolation
List of Tables
Table 1 - Mixtures used for original concrete.

Table 2 - Properties of recycled and natural aggregate.
Table 3 - Mixtures used for recycled concrete.
Table 4 - Mechanical and physical properties of recycled concrete.
Table 5 - Shrinkage of recycled concrete.
List of Figures
Figure 1 - Specimens preparation for trial.

Figure 2 - Specimens in climatic chamber.
Figure 3 - Deformation by swell and basic shrinkage for different recycled
concrete.
Figure 4 - Deformation by swell and total shrinkage for different recycled
concrete.
Figure 5 - Basic shrinkage for different recycled concrete.
Figure 6 - Total shrinkage for different recycled concrete.
Figure 7 - Shrinkage due to drying for different recycled concrete.
@Seismicisolation
@Seismicisolation
490 Gómez-Soberón
Flakiness index 6
15 _____ 8 19 _____ ~
- Modulus of fineness 7.2 6.2 3.8 6.9 5.0 3.3

Sand equivalent (%) ----- _____ 93.6 _____ _____ 93.8
Particles < 2041 (%) 0.06 0.29 9.85 0.50 2.46 9.24
Sand (0-5)(kg/m') 67 I 662 1

WIC I 0.52
Coarse A I Fine A (Vol.) I 1.53
"CEM 152.51 UNE 80 301 96 RU97 ('O) Limestone aggregate, Garrafquarry, Barcelona
Table 4 Mechanical and physical properties of recycled concrete.

I Tensile I Compressive I Young's Modulus I E I I -1
D
-=' 172, 193,255 and 345 days of age
@Seismicisolation
@Seismicisolation
M
35
D o
3- "b
u
'3
0
5
W
.g
V
3
j g
5 -
W
u
.-
eI
e
W
s
V
M
3
D
3- %
u
'3
n
3
M
i
c
w"
3
B
3
o
"o
.-u
D
I
@Seismicisolation
@Seismicisolation
492 Gómez-Soberón
Figure 1 Specimens preparation for trial.
Figure 2 Specimens in climatic chamber.
@Seismicisolation
@Seismicisolation
O.30
v
E
M
0:
&
0.20
0.10
0.00
-0.10
-ICuring
chamber Climate chamber
(+) Shrinkage
m
-0.20
r = 0.60
-0.30
-0.40 J
O 1500 3000 4500 m
Figure 3 Deformation by swell and basic shrinkage for different recycled
concrete.
Climate chamber
W. N
Figure 4 Deformation by swell and total shrinkage for different recycled

concrete.
@Seismicisolation
@Seismicisolation
494 Gómez-Soberón
(+) Shrinkage
0.15 -
0.10 -
0.05
4
3
!.
0.00 .30
.IS
$.s 4 -0.05
-0.10
oc
-0.15 -
(-) Expansion
Time (hours)
Figure 5 Basic shrinkage for different recycled concrete.
0.80
1 1
0.70
0.60
- 0.50
'5 0.30
cz
0.20
0.10
0.00
O 1500 3000 4500 6OOo
Figure 6 Total shrinkage for different recycled concrete.
@Seismicisolation
@Seismicisolation
0.60
0.50
0.40 2
0.30 P-.-
E
0.20
e
;;
+28 days 4%-

90 days +180 days 3t270 days 0.10
0.00
oc r=0.00 r=0.15 r=0.30 r=0.60 r=1.00
Factor r
Figure 7 Shrinkage due to drying for different recycled concrete.
@Seismicisolation
@Seismicisolation
SP 209-27
Comparative Study of the

Mechanical Properties and
Performance, Under Accelerated
Aging, of Four Repair Systems for
Concrete Hydraulic Structures
by A. C. M. Korrnann, K. F. Portella, P. N. Pereira,
R. P. Santos, and A. Joukoski
Synopsis: The performance of conventional, polymer-modified, steel-fiber, epoxy,

and silica-fume mortars subjected to a high velocity water flow is described. Be-
sides the mechanical requirements and wearing resistance of concrete hydraulic
structures, especially spillway surfaces, the high solar radiation and the environ-
mental temperature contribute significantly to decreasing in the lifespan of repairs.
Such contributions are particularly evident on surfaces repaired with polymer-
modified concrete and epoxy mortar, mainly at the interfaces between the substrate
and the environment. This paper presents the results of a comparative study to de-
termine mechanical properties, adherence, and accelerated aging of four repair
systems. Two systems exhibited higher performance: the epoxy and silica-fume
mortar. Steel-fibers, due to good adherence to the substrate and good mechanical
performance, are appropriate for use as an intermediate layer for underwater re-
pair. This system exhibited an apparent corrosion at the steel-fibers on the surface
after aging and underwater abrasion tests, which should be considered prior to be-
ing used on exposed spillway slab surfaces.
Keywords: a brasion; concrete repair; dam hydraulic

structures; epoxy mortar; polymer-modified concrete;
silica fume; spillways; steel-fiber
@Seismicisolation
@Seismicisolation
497
498 Kormann et al.
Aline C M.Kormann is a Civil Engineer, and student of the Civil Engineering
PostGraduation Program (PPGEC),at the Federal University of Paraná Siate
(UFPR)),Curitiba, PR, Brazil.
Kleber F. Porcecl<r works as a research scientist at the Institute of Technology

for Development (LACEC), Curitiba, PR, Brazil. He received his Master of
Science degree from Aeronautics TechnologicalInstitute @TA)in 1991, and his
Doctor of Science degreefrom University of Sa0 Paulo (USP) in 1997.
Pedro N. Pereira is a Civil Engineer at Duke Intemational Geração

Paranapanema, São Paulo, SP, Brazil.
Rui i? dos Santos is a Civil Engineer at Duke International Geração

Paranapanema, São Paulo, SP, Brazil.
Alex JouRoski is a Civil Engineer, Specialist in Metallic Materials

Engineering, and works as a research assistant at Institute of Technology for
Development.
WTRODUCTION
Repairs in exposed and underwater hydraulic surfaces of concrete structures

such as dams must be made by considering characteristics of cost, pdormance,
áurability, workability, and application time of the repair materiais as well as
their compatibility with the reference concrete (RC). In addition, it is of
fundamental importance to evaluate the existing conditions where the materials
are going to be applied, such as: solar radiation, temperature, corrosive agents,
and aggressive chemicals, due to the wide range of price and availability of the
materials in the market (1,2). Considering the concrete structures of
hydroelectric power plants, it is essential to use materials and/or systems that
are compatible with the time available for repairs. Access to the repair area is
usually dictated by power production schedules.
Indirect and direct costs like loss at energy production because generators
are shutdown for making the repairs must be considered. Similar factors must
be considered in the repair of spillway concrete surfaces and downstream
sidewalls of the concrete structures.
The present work was focused on the hydraulic d a c e s of the concrete

structures of the Hydroelectric Power Plant of Capivara, which is located in the
Paranapanema River, in the state of São Paulo, in southeastern Brazil.
@Seismicisolation
@Seismicisolation
Technical reports, together with inspections carried out locally, showed that
the main deterioration in the analyzed concrete surfaces was abrasion and
erosion damages caused by the water flow from the reservou on the hydraulic
surfaces, mainly in the spillway region, on the upstream faces of the sustaining
pillars of the floodgates, and in the water outlet works (3-5).
Usually, the additives that stand out in the technical literature for making
concrete and mortar include silica-fume or other mineral, organic or metallic
fibers, and polymer materials such as epoxy, and methylmethacrilate, among
others (i, 6-10).
EXPERIMENTAL INVESTIGATION
In the experimental investigation the selected repair materials, properties and

testing techniques adopted in the work are presented.
Rei3air Materials Mixtures and Concrete Substrate
The repair materials were chosen based on an extensive review of the

literature considering desirable characteristics of application, maintenance,
durability and time of execution, including hardening.
Four repair materials were studied: epoxy mortar (EM), polymer-modified

cement mortar (PM), silica-fume mortar (SF) and concrete reinforced with
steel-fibers (ST).
The repair material mixtures were varied and studied in laboratory; initially
starting with proportions in the literature, modified based on results of material
properly tests and information given by the manufacturers (1-10).
The RC was developed based on the concrete mixture used in the slab of the
spillway rapid of Capivara Power Plant, which used CPIi-Z 32 portland cement
(pozzolan-modified) whose average compressive strength was 40 MPa (5800
psi) in 28 days. The average mixture proportions were 1 : 1.61 : 2.99 : 0.376,
with a cement content of 425 kg/m3 (716 lb/yd3).
Silica-fume mortar íSF) - Fine aggregate was replaced by silica-fume (10%
by weight of the fine aggregate). The average mixture proportions were 1 : 4.0 :
0.45: 0.1 (cement : sand : w/(c+sf) : silica-fume).
EDoxV mortar EM) - A commercially avaiIabfe product was used, the base
material consisting of resin, hardener and quartz sand. Since the final product
was appropriate for application on sloping surfaces, such as spillway slabs,
quartz sand at different contents (up to 50 % more) was used in the final
mixture, even though a reduction in compressive strength was expected.
@Seismicisolation
@Seismicisolation
500 Kormann et al.
polvmer-modified mortar (I'M) - A commercially available product with an
initial water content of 18 % was used in accordance with the manufacturer's
recommendation. After the tests, the optimized water content was altered to
11 %.
Steel-fiber concrete íST) - This mixture was proportioned by adding steel-

fiber to @e RC at a rate of about 12 % by weight of the cement as a replacement
for the coarse aggregate. The resultant mixture proportions were 1 : 1.61 : 2.94 :
0.45 : 0.2. Due to a lack of homogeneity in the mixture there was a need to add
10 % more fine aggregate, which resulted the following mixtUre proportions: 1 :
1.98 : 2.77 : 0.45 : 0.11.
-
Tests
The tests were conducted in the laboratoxy on concrete and mortar

specimens under simulated environmental conditions, which are usually found
to have significant effect on mechanical properties and aging, such as solar
radiation and humidity.
The primary tests included resistance to abrasion and erosion, flexural

strength, compressive strength, elastic modulus, adherence resistance, in
accordance with the Standards (11-14), accelerated aging in a UV-ray chamber
and humidity and permeability tests as well as physicochemical characterization
tests of all materials.
& - All of the constituent

materials, cement and aggregates (fine and coarse) and some of the repair
materials (silica-fume and polymer-modified mortar) were characterized by
physicochemical analysis, previous to use in the mixtures.
Abrasion and erosion - The underwater abrasion and erosion resistance was
determined in accordance with ASTM C1138-97 (1 i). Specimens with diameter
of 300 mm (11.8 in) and 100 m m (3.94 in) thick were produced by using
substrate mixtures. A 200 mm x 50 mm (7.87 in x 1.87 in) void was formeú in
the center of the specimens for application of the repair materiais.
Flexural s t r e n d - Flexural strengths were determined, &er 60 days from

molding, on concrete beams with 100 mm x100 mm x 350 mm (3.94 in x 3.94
in x 13.8 in) in accordance with ASTM C78-94 (12).
Compressive strength - Compressive strengths were determined in

accordance with ASTM C39K39M-99 (13).
@Seismicisolation
@Seismicisolation
Permeability - The methodology used in this test is based on percolating
water under sufficient pressure to force water passage though the exposed
surface and concrete.
Accelerated aGnp test - Specimens containing the repair materials were

tested for 1600 hours in W-ray chambers and for 400 hours in chambers of the
Wheater-O-Meter type. Flexural strength tests were conducted to evaluate the
effects of aging on the materials in the given exposure period. Substrate beams
with 15 mm x 30 mm x 100 mm (0.591 in x 1.18 in x 3.94 in) which were 15
mm (0.591 in) high, were cast for application of the SF, PM, and EM. For the
repair with the steel-fiber concrete, beams with dimensions of 20 mm x 40 mm
x 100 m m (0.787 in x 1.57 in x 3.94 in) and 20 mm (0.787 in) high were cast.
The average results of the compressive strength tests indicate that, with the
exception of the polymer-modified mortar, strengths were compatible with of
the RC. The compressive strength of the PM was less than expected. The
average results are shown in Table 1. Ali the results were compared and the
analysis showed a correlation between higher compressive strengths and
improvementsin other properties.
The results of the abrasion and erosion tests on the repair systems were
generaily compatible with the results obtained in compressive strength tests
(Figs. 1,2,5). The majority of the mass loss in most of the systems occurred in
the interfacial region between the repair and the RC,as shown in Fig. 3 and in
the three-dimensional schematic image of PM test specimen (Fig. 4). in the
abrasioderosion test on EM, there was little or no mass loss (Fig. 5). The mass
loss for the test specimen was generaily confined to the surrounding reference
concrete (Figs. 6,7).
in the direct tensile tests, no fractures were observed in any of the tested
repair materials. The performance of the repair materials followed a decreasing
order: EM > ST > SF > PM, with tensile strength of 3.0, 2.0, 1.5, and
0.95 MPa (435, 290, 218, and 138 psi), respectively. Therefore, the bond
strength between the RC and the repair materiais is considered good (Fig. 8).
The coefficients of permeability of the cement-based repair materials, SF

and ST,remained between 60 % and 90 % of the coefficient exhibited by the
RC,a result of lo-'' m/s (3.28 x Ws),that is about what would be expected
for concrete with a wlc of about 0.45 (2).
The tangent elasticity modulus of the tested materials can be found in

Table 2.
@Seismicisolation
@Seismicisolation
502 Kormann et al.
Results of the flexural strength tests on concrete and mortar are shown in
Table 3. The EM system exhibited the superior flexural tensile strength,
followed by the SF and ST systems. The PM system exhibited a somewhat
poorer performance. The results of flexural strength tests on the repair materials
showed an increase of 30 % to 60 % after being exposed for 800 hours to W
light. This increase is considered normal for cement-based materials, that is, the
time of exposure to radiation was not sufficient to cause irreversible damage,
especially on the organic structures (polymer materials), which are more
susceptibleto this kind of deterioration.
Despite the generally good performance of ST material long to humid
environments resulted in significant corrosion of the exposed steel-fibers
(Fig. 9), that is considered a disadvantage for this application. There is a need
for a finishing touch to the surfwe of this material to cover the superficial fibers
so that corrosion will be minimized. The cracks generated by the corrosion of
metal inside the concrete might become worse because of the abrasion effect.
Overall, information obtained by testing of repair systems indicated that
their relative performance could be rated, in decreasing order, as follows: EM,
ST,SF and PM. Each repair system was used in an on-field application at two
spillway slabs at the Capivara Power Plant. The systems were placed in similar
positions, along the spillway floodgate line. The procedures for repair at the
surface concrete followed the standards recommended in the literature.
The performance of the studied systems was assessed, also considering the
results of tests on the field specimens and the constructability of the repair
systems. When the epoxy mortar mixture recommended by the manufacturer
was used, the material was self-leveling and therefore not recommended for
sloping surfaces. To improve its constmctability, the quartz sand content of the
mixture was increased by up to SO %. As a result, the compressive strength was
reduced by 10 YO; however, this did not significantly affect the performance
expected. Results of compressive strength tests on field-molded specimens are
shown in Table 4.
The performance of the four repair systems subjected to exposure conditions
imposed by the dam operating procedures will be evaluated periodically by the
Duke-Energy Geração Paranapanema team.
CONCLUSIONS
In this work, the erformance of four repair systems applied to concrete

P
containing 425 kg/m (716 Ib/yd3) of class CPIi-2 32 (pozzolan-modified)
porfland cement was evaluated. Based on the results of this investigation, it is
concluded that the relative performance rating of the materials is, in decreasing
order: epoxy mortar > steel-fiber concrete > silica-fume mortar > polymer-
modified mortar.
@Seismicisolation
@Seismicisolation
In addition, the following can be concluded:
The EM mixture recommended by the manufacturer, despite its good

resistance to abrasion, presented a certain difficulty in application due to
its self-leveling properties, which is not recommended for sloping
surfaces. To minimize this problem, it was necessary to increase the
quantity of quartz sand in the mixture by about 50 %. This alteration in
the mixture resulted in a 10 % decrease in compressive strength;
however, the strength was still superior to that of the RC. The primary
advantage of this material is its rapid strength gain, a desirable property
for an emergency repair.
The exposed fibers on the surface of the ST concrete exhibited a high

corrosion rate when undergoing aging in the humidity chamber. This
corrosion is a disadvantage for this material and may contribute to
accelerated abrasion loss.
The SF is an excellent repair material, exhibiting properties that were

compatible with those of the RC. The material was also easy to handle
and apply. The only restriction being that it requires a longer period for
its complete hardening compared to the EM.
Given the unsatisfactory performance of the PM mixture and considering

the 18 % water in the original mixture, additional tests should be
conducted to evaluate the possibility of further reductions in water
content from 11 to 8 %.
ACKNOWLEDGMENTS
The authors would like to thank the Institute of Technologyfor Development

(ZACTEC) &Duke Energy International - Geração Paranapanem for their
cooperation and providing of the injtastructure neededfor accomplishment of
this work. Special thanh are due to the Civil Structures Department (ZACEC-
DPEC) by the analysis contribution.
BIBLIOGRAPHIC REFERENCES
1. International Concrete Repair Institute, 1999 - Concrete Repair Mimual,

AC1 International, 1999.
2. Neville, A. M. Propriedades ab concreto. Translation: S. E.

Giammusso. 2". edição, Editora PINI Ltda., São Paulo, 1997. 828 p.
@Seismicisolation
@Seismicisolation
504 Kormann et al.
3. Leite, R. C. A. de P.; Abbud, A. R. "Considerações sobre o método
construtivo e a durabilidade da estrutura em concreto do vertedouro da UHE
Capivara, da CESP". In: Anais da 38" REIBRAC, IBRACON. Ribeirão Preto,
SP. vol. 1,507-514, (1996).
4. Salles, F. M.; Almeida Jr., W. "Recuperagão do concreto do rápido do

vertedouro da UHE Capivara". In: Anais do 42" Congresso Brasileiro do
Concreto. Fortaleza, CE, IJi-E-006, (2000).
5 . Usina Capivara Relatório. Duke Energy, b a g o Paranapanema, ( 1 9 3

p. 57-70.
6. Carneiro, E. F.; Andrade, J. R. A.; Almeida Jr., W. "Avaliação do

desempenho dos reparos executados na soleira do vertedouro da UHE de Porto
Primavera". In: Anais da 38" MBRAC, IBRACON. Ribeirão Preto, SP. vol. 1,
393-395, (1996).
7. McDonald, J. E.; Liu, T. C., "Repair of concrete surfaces subjected to

abrasion erosion damage". In: Annual Convention, Milwaukee, Wisconsin,
1979.
8. Ghafoori, N.; Diawara, H. "Abrasion resistance of fine aggregate replaced

silica-&me concrete". ACZMaterials Journal, sep-oct, 555-567 (1999).
9. Cabrera,G.; Al-Hasan, A. S. "Performance properties of concrete repair

materials". Construction and Building Maferials, vol. 11, no 5-6, 283-290
(1997).
10. Abdel-Fattah, H.; El-Hawary, M. M. "Flexural behavior of polymer

concrete". Constructionand BuildingMateriais, vol. 13,253-262 (1999).
11. ASTM C1138-97. "Standard Test Method for Abrasion Resistance of

Concrete (Underwater Method)". American Society of Testing and Materials,
West Conshohocken,Pa, 1999.4 p.
12. ASTM C78-94. "Standard Test Method for Flexural Strength of Concrete
(Using Simple Beam with Third-Point Loading)". American Society of Testing
and Materials, West Conshohocken, Pa, 1994.3 p.
13. ASTM C39/C39M-99. "Standard Test Method for Compressive Strength

of Cylindrical Concrete Specimens". American Society of Testing and
Materials, West Conshohocken,Pa, 1999. 5 p.
14. ASTM 469-94. "Standard Test Method for Static Modulus of Elasticity
and Poisson's Ratio of Concrete in Compression". American Society of Testing
and Materials, West Conshohocken, Pa, 1994.4 p.
@Seismicisolation
@Seismicisolation
Table ?-Average Compressive strength of concrete and mortar
specimens
Age Compressive Compressive

Material
(days) strength (MP a) strength (psi)
RC 28 44 6381.6
ST 28 38 5511.4
EM 28 91 13198.4
PM 28 22 3190.8
SF 28 44 6381.6
Table 2-Average modulus of elasticity test results
Material Age Etans Etaog

(days) (GPa) (kpsi)
RC 28 50 7,252
ST 28 39 5,656
EM 28 21 3,046
PM 28 41 5,947
SF 28 38 5,511
@Seismicisolation
@Seismicisolation
Age Flexurai Strength Flexural Strength
Material
(days) (MPa) (psi)
ST 28 6 870
EM 28 57 8,267
PM 28 3 43 5
SF 28 6 870
Compressive Strength Compressive Strength

Material (MPa) (psi)
7 days 28 days 7 days 28 days
ST 29 34 4,206 4,93 1
EM 67 84 9,718 12,183
PM 13 20 1,886 2,901
SF 24 31 3,481 4,496
@Seismicisolation
@Seismicisolation
'~o
0.99 o~
2 0.94
-v-- PM
0.93 --e-- EM
0.92
O 10 20 30 40 50 60 70 80
Testing time (h)
Fig. I-Results of abrasion and erosion tests.
o io 20 30 40 50 60 70 ao
Testing time (h)
Fig. 2-Mass loss YS. testing time for PMspecimen subjected to abrasion test.
@Seismicisolation
@Seismicisolation
508 Kormann et al.
Fig. 5-Polymer-modijìed mortar (PM) specimen a3er abrasion testing.
Fig.&Schematic of the PMspecimen surface after abrasion testing.
@Seismicisolation
@Seismicisolation
O 10 20 30 40 50 60 70 80
Testing time (h)
Fig. 5-Mass loss vs. testing time for Hkíspecimen subjected to abrasion test.
Fig. &Epoxy mortar (Ekf) specimen @er abrasion testing.
@Seismicisolation
@Seismicisolation
510 Kormann et al.
Fig. 7-Schematic of the EMspecimen surface @er abrasion testing.
Fig. P Tsical direct tensile test.
@Seismicisolation
@Seismicisolation
Fig. %Typical exposed steel-Jber corrosion on surface of specimen.
@Seismicisolation
@Seismicisolation
SP 209-28
Strut and Tie Models for Strength

and Deformation of RC Members
by S. G. Hong and S. 6.l e e
Svnopsis: This paper presents strut and tie models that can be used to determine
strength and deformation limits for RC members. Conventional strut and tie mod-
els have been used for strength design, but they cannot appropriately address the
ductility of members that is important to insure safety under severe cyclic loads.
This proposed strut and tie model is constituted to address the end region of flex-
ural members, including plastic hinge regions, and will be the basis for the calcu-
lation of deformation and corresponding shear strength as well. The behavior of
each component of the strut and tie model is evaluated based on the material prop-
erties and stress paths. The elongation of the tie element depending on crack spac-
ing and width is obtained from the bond-slip relationship. The contraction of strut
element is simply determined from the properties of concrete. The member defor-
mation is obtained by combining the truss deformation determined from compo-
nents behaviors with joint rotation. The strut and tie model in this paper will pro-
vide useful tools for both the design and evaluation of ductility-required RC
members.
Kevwords: deformation analysis; limited ductility; RC

column; strut and tie model
@Seismicisolation
@Seismicisolation
513
514 Hong and l e e
Sung Gu1 Hong is an assistant professor in the department of architecture at
Seoul National University, Seoul, Korea. He received his BS and MS degrees
from Seoul National University and his PhD degree in 1994 from Lehigh
University, Bethlehem, Pennsylvania. His research interests include steel bar
anchorage to concrete, shear friction, and design method for RC structures.
Soo Gon Lee is a PhD student in the department of architecture at Seoul

National University, Seoul, Korea. He received his BS and MS degrees from
Seoul National University.
INTRODUCTION
Most current codes for seismic design have relied on strength-based design
with selection of appropriate system ductility. To meet the requirement for
strength and ductility, most building codes, such as AC1 318[1], provide the
methodology for the dimensioning of members and detailing of reinforcement.
Recent researches on earthquake engineering, however, are turning toward
performance-based design so as to overcome the disadvantages of the current
strength-based design concepts, which have no direct relation between seismic
load and performance of the structure such as strength and ductility. Accordingly,
appropriate member design methods, which can estimate behavior of the
members, are required for the newly developed seismic design methodology
such as displacement-based design methods.
According to capacity design philosophy [2], plastic hinges are supposed to

form at the bottom end of columns at the ground story and at the ends of beams
to produce desirable failure mechanisms. It is necessary to ensure sufficient
deformation capacity of such members. Brittle members such as beam-column
joints should remain elastic at ultimate to ensure proper seismic performance of
the structure. For this purpose, it is important to determine the deformation limit
of members that require sufficient ductility. in particular, the rotation limit of the
@Seismicisolation
@Seismicisolation
ends of the bottom columns should be determined; this may control the lateral
displacement capacity of the whole system.
Strut-and-tie models depict internal force flows represented by discrete

compression and tension elements joined together at nodes. In terms of the
effective strength of each element, these models have been successfully applied
to the design of D- regions of RC members, where the strain and stress are not
linearly distributed. These models, however, provide only required strength of
members for seismic design. To overcome this limitation, Marti et a1.[3]
suggested a tension chord model to calculate deformation of the tension tie
element, which is applied to determine the deformation limit of flexural
members and extended to deal with shear problems. As another application of
strut and tie models for seismic loading design, To et a1.[4] suggested a strut and
tie computer model to analyze nonlinear behaviors of RC knee joints by truss
elements within the Drain-2DX program.
The objectives of this paper are to define the behaviors of components of

the strut and tie model and to apply them to the deformability estimation of RC
members. The strut and tie model to be presented is to calculate the ultimate
deformation and its corresponding shear strength. In this model, a simplified
bond-slip relationship is applied to the behavior of tie elements, and concrete
strut elements employ the constitutive equations of cracked concrete available in
the published literature.
STRUT A N D TIE MODELING
Fig. 1 (a) shows a RC flexural member subjected to shear and axial force.
The model represents a typical RC intermediate and short column required to
have limited ductility.
The shear force V and the axial force P act on the inflection point of the
member. The length of the member between the inflection point and the end
face is denoted by L. The deformation of the member is mainly controlled by the
@Seismicisolation
@Seismicisolation
516 Hong and Lee
deformation of the end region including the plastic hinge. Thus, the strut and tie
model shown in Fig. 1 (b) and (c) is selected to focus on the behavior of the
region, while the other region above the end region is assumed as a rigid body.
Element forces T and C represent tie elements and strut elements,

respectively. Subscript 1 denotes association with the longitudinal element, and
subscripts tr and d, transverse and diagonal element, respectively.
The failure sequence of RC member is selected such that flexural

reinforcement at the end yields first and subsequent failure of any other
component causes member failure. Thus, two models are presented: strut and tie
models in elastic state (initial yielding state), and after yielding state (ultimate
state), as shown in Fig. 1 (b) and (c) respectively.
Diagonal strut angle 19is assumed as the inclined crack angle at the yielding
of main bars at the bottom end:
where E, is cracking stress of concrete; and s is spacing of transverse

reinforcing bar.
FORCES ACTING ON EACH COMPONENT
Truss models may be treated as determinate structures assuming yielding of

components at the ultimate state. Then, element forces are calculated by
equilibrium equations. The force of each component is expressed in terms of
external shear force V and axial force P , as summarized in Table 1. After
yielding, the strut and tie model shown in Fig. 1(b) cannot explain the increasing
of shear force dV that is carried by longitudinal strut in the end region. Fig. i(c).
shows a changed strut and tie model after yielding, which can address additional
shear transfer mechanism by the change of the strut angular orientation. The
forces acting on each component by the additional shear are presented in Table 2.
@Seismicisolation
@Seismicisolation
FORCE - DEFORMATION RELATIONSHIPS OF COMPONENETS
Longitudinal Tie element
The longitudinal tie element is subjected to uniaxial tension with transverse

cracks, used in the flexural tension region. The area of the tie element is
assumed as the longitudinal steel bar area As/, and the element length 1, is
(jdcotû)/2 (in I;, ) and j d c o t û (in TZ , I;3 ). Since the member
deformation is mainly dependent upon the extension of the tie element, their
calculation models need to be precise. The deformation of the tie element can be
determined with crack spacing and crack width, which are estimated by the
bond stress-slip relationship between bar and cover concrete.
Minimum crack spacing is estimated as follows:

s =- A,,,f;
min nad, f b
where Sh, is minimum crack spacing; Ac,effis effective concrete area; f, is
tensile strength of concrete; n is number of bars in tension chord; db is bar
diameter; and h is average bond stress. For the convenience of calculation,
the effective concrete area A , , f is simply selected as A d 3 , where A, denotes
gross sectional area. Assuming that f,/f; = 2 and S = 1.5Srni,gives a simple
estimation of the crack spacing along the longitudinal tie element as follows:
s/ =-4nnd,
A,
(3)
The equilibrium and compatibility conditions for a differential element as

shown in Fig. 2 give a differential equation for the bond - slip relationship,
[51J61:
where S is relative displacement between bar and surrounding concrete; and x

is distance from non-slip point along the bar.
@Seismicisolation
@Seismicisolation
518 Hong and l e e
Fig. 3 (a) shows the forces acting ona longitudinal tie element between
two adjacent cracks. Distribution of local bond stress fb is assumed as a
constant value that equals local bond strength, as shown in Fig. 3 (b). Equating
of the sum of local bond stress to the vertical components of the diagonal strut
forces gives x, ,the length between a crack and a non-slip point:
where u = V/(nnd,,jd) is a global bond stress developed by shear force; and

f, = f ;( 2 - L l / f y , ) is a local bond strength that is derived from the bond
strength depending on steel strain.
Using the boundary condition, 6 = O at x = O and E, =f , / E , ,E, = O

at x = xo, Eq. (4) results in:
f, 1
6 ( x o )=-xo --K fx?
Es 2
A slip in the opposite direction should be calculated in the same way so as
to obtain a crack width. Accordingly, the slip is distributed as shown in Fig. 3
(c), and the crack width is calculated by adding two of the end slips:
Note that the second term of the right side of the above equation expresses the
reduction of deformation by a tension stiffening effect. Using the relations of
E/ = w /S, and AT, = lecl,the force-deformation relationship for a longitudinal
tie element in elastic state is determined as follows:
For yielding conditions, bond failure as well as bar yielding is considered:
At bar yielding condition : I;/ + A,Ify,

At bond failure : u + f, or x,, + S,
The behavior after yielding is assumed to be perfectly plastic. Thus, the
@Seismicisolation
@Seismicisolation
deformation after yielding should be determined by compatibility with that of
longitudinal strut element.
Longitudinal Strut Element
A longitudinal strut element is defined as a flexural compression element

subjected to uniaxial compression force. To define the load deformation
relationship of the element, a linear stress-strain relationship is assumed with the
Young’s modulus as half of the real elastic modulus of concrete, E, / 2 as
shown in Fig. 4.
The area is assumed as 213 of bc, an elastic triangular compression block

area at initial yielding of the tension steel bar:
L
A, =-bc (9)
3
where the concrete block length c is determined by the compatibility condition
between the compressive strain of concrete triangular block and the tensile
strain of steel bar at the initial yielding:
+df, + Pl + JASlfy, + Pl’ + W& + P)dEybEc

C= (10)
A distance between longitudinal tie and strut elementj d is calculated as follows:

jd=d--C
3
After yielding, lever arm length j d is changed to cany additional shear dV,
which is expressed by equilibrium of longitudinal strut element in terms of lever
arm lengthj d at yielding and changed lever arm lengthjd’:
dV =
j d cot e
Furthermore, the deformation of longitudinal tie element after yielding depends

on that of longitudinal strut element and changed lever ami length. Fig. 5 shows
the compatibility condition between the longitudinal elements. Thus, the
@Seismicisolation
@Seismicisolation
520 Hong and Lee
deformation of longitudinal tie element increased by additional shear is
Calculated as follows:
3j d '- 2d
41, =
3(d - j d ')
&,
Transverse Tie Element
Transverse tie elements are supposed to carry the member shear force.
According to the current AC1 design equation for shear strength of RC flexural
member [i], the shear force is carried by the contribution of transverse
reinforcement and that of web concrete:
v,=J(+v, (14)
Fig. 6 (a) shows the shear mechanism in a diagonal crack at shear failure. It
is assumed that all of the transverse bars yield. The deformation of the
transverse tie element is expressed in terms of diagonal crack spacing and width
using the equation = w, /S, ,as follows:
where wd is diagonal crack width, and S, is diagonal crack spacing

approximated as follows:
SISI
s, = S,sinQ+S,cosQ (16)
where S, = bs/(4nndb,) is the horizontal component of diagonal crack spacing.

Neglecting the bond effect between the transverse reinforcing steel and concrete,
the steel stress can be determined in terms of crack width.
The shear force carried by concrete is due to fiction along the crack face.
This force acting on the diagonal crack is calculated using the equation
proposed in the MCFT(modified compression field theory) by Vecchio et al. [7]
If there is not the effect of flexural deformation, the crack width at the
maximum shear resistance, wdY, can be determined ffom yield stress fyh for
@Seismicisolation
@Seismicisolation
the transverse bar stress L h . Therefore, the strength and corresponding
deformation of the transverse tie element is calculated as follows:
It is assumed that the relationship between force and deformation before

the deformation in Eq. (18) is linear7 neglecting large shear stiffness before
cracking.
The shear strength is considerably affected by member flexural behavior.

Flexural deformation components in a crack width are depicted by tensile
deformation at longitudinal tie 6, = .c6 S, /cos 8 , and compressive deformation
longitudinal strut 6, = ~ ~ , S ~ / c o,s ûwhere and E,-, denote the
longitudinal strain .cT2 , sq3 and .cc,, E ~ , , in the cases of T,, and T,, ,
respectively. The longitudinal deformation components give the additional crack
width so that all of the transverse reinforcing bars at the crack face yield. Figs. 6
(b), (c) and (d) show the change in yield crack width.
Note that the deformation of the tie element is treated as a positive value in
tension and that of the strut element is positive in compression. Using this crack
width at maximum shear resistance, Eqs. (17) and (18) give transverse tie
element behavior with flexural deformation.
Diagonal Strut Element
A diagonal strut element is a discrete representation of a diagonally cracked

compression field subjected to uniaxial compression. The strength of the
compression strut decreases as the transverse tensile strain increases. The
constitutive equation proposed in the Modified Compression Field theory by
Vecchio et al. [7] is used for the stress strain relationship of the diagonal strut:
@Seismicisolation
@Seismicisolation
522 Hong and Lee
where A,,, = f,’ Ifc(MPa); u, and denote compressive stress

0.8+170~,
and strain respectively; E,’ = 0.002 ;and /

= A?, j d .
The area of element C, is bj d coso. The area of C,, , representing the

strut on a fan-shaped region, however, cannot be obtained from the stress field.
For convenience of a simple analysis, the area of diagonal strut C, is assumed
as (bj d cosû)/2.
MEMBER DEFORMATION
Member deformation is obtained by combining truss element deformation

with joint rotation. The truss deformation is represented by the lateral
displacement and the rotation at point of (3 j d cot 8)I2 from the bottom end,
as shown in Fig. 7 (a):
Member deformation must include the effect of a joint rotation determined

by the slip of the anchored steel bars and concrete block contraction. By
assuming a joint region as an elastic state at failure, the rotation is simply
estimated. The joint deformation is dominantly affected by the shear mechanism,
as shown Fig. 7 (b). At the member face of the joint, however, this deformation
can be expressed by end rotation due to the extrusion of anchored tension bar
and the shortening of the compression concrete block, as shown in Fig. 7 (c). By
@Seismicisolation
@Seismicisolation
assuming the joint to be elastic, the joint rotation is calculated as follows.
Therefore, the member deformation at inflection point is obtained as below.

A,=A-+(L- 3 j d cot 0 )OW+LO+,,
(24)
In addition, the plastic hinge rotation O, can also be determined by dividing

AT by member length L.
ANALYSIS PROCEDURES AND VERIFICATION
A set of procedures for deformation analysis is proposed as follows:

Step 1: Assuming the yielding of longitudinal bars in tension, determine the
geometric properties of the strut-and-tie model, 0 , c and j ú from Eqs. (i),(1O)
and (11) respectively.
Step 2: Calculate the yield strength V, , the shear forces derived from,'"i
in Table 1 by substitutinghl forf,l.
Step 3: Calculate all the element forces at V, ñ-om Table 1, and all the
element deformations from the force - deformation relationships.
Step 4: Calculate the member yield deformation from Eqs. (21)) (22), (23),
and (24).
Step 5: As the lever arm lengthjd' increases gradually, find the additional
shear force dV and the state of failure of the other element. The longitudinal tie
element extends while maintaining its strength before any of the other elements
yield. Then, calculate the member deformation at that state by repeating Step 4.
Note that this deformation is the limited ductility of members.
Step 6 : After yielding of shear component at Step 5, the descending branch
of load-deformation curve can be obtained. Repeat Step 4 and Step 5 with
increasing the deformation of the transverse tie element. This represents stren8h
degradation after shear failure.
The results based on the proposed model are compared with those of prior
@Seismicisolation
@Seismicisolation
524 Hong and lee
experimental programs. The data include two rectangular columns of author’s
own study, and three circular and four rectangular columns of Priestley et al. [8]
[9] All of the specimens are intermediately short columns having the shear span
ratio of 1.5 to 2. Most of the test results show the shear failure mode after the
yielding of the longitudinal main bars. The comparison of strength and
deformation shows a good correlation between the two results as shown in Fig.
8.
CONCLUSIONS
This paper presents strut and tie models for ductility required members
under shear force combined with axial force and moment. Each component of a
strut and tie model is defined by its force-deformation relationship and strength.
The member strength is determined at the state of the failure of any other
element after yielding of longitudinal tie element, while deformation is
calculated by combining the deformation of all elements with joint rotation. On
the basis of the model, the following conclusions are drawn:
(i) The deformation of the longitudinal tie element obtained from the
bond-slip relationship gives an accurate estimation of elastic behavior. After
yielding, the location change of the longitudinal strut element and perfectly
plastic tie element give reasonable results of ultimate strength and deformation
estimations.
(2) Yielding of transverse tie elements or crushing of diagonal struts

represents a decisive stage in determining the shear strength. The force -
deformation relationship of a transverse tie element is dependent on the diagonal
crack spacing and crack width. Thus, the flexural deformation influencing
diagonal crack width should be considered to define behaviors of transverse ties
and diagonal struts, which in turn permits the limited ductility deformation of
columns in shear to be estimated. Based on these models, unfavorable failure
can be identified in the design of shear critical members.
@Seismicisolation
@Seismicisolation
(3) The proposed deformation model needs refinement of the load-
deformation relationship of shear components (transverse tie and diagonal strut)
in order to obtain a more accurate limited ductility deformation. Future study
will refine the model for fan-shaped strut regions. The degradation of the
strength and the stiffiess by cyclic load effect also will be considered.
ACKNOWLEDGEMENTS
This work was partially supported by the Brain Korea 21 Project and
funded by the Korea Earthquake Engineering Research Center (KEERC) under
project No. 2001-G0303, sponsored by KOSEF.
REFERENCES
[i] AC1 Committee 318, Building Code Requirements for Structural Concrete
and Commentary (AC1 3 18-991 AC1 3 18R-99), American Concrete Institute,
Farmington Hills, Michigan, 1999.
[2] Paulay, T., and Priestley, M. J. N., Seismic Design of Reinforced Concrete
and Masonry Buildings, John Wiley & Sons, Incs., 1992, pp. 38-46
[3] Marti, P., Alvarez, M., Kauíìnann, W. and Sigrkt, V., Tension Chord Model
for Structural Concrete, ETH, Zürich, Swiss, 1998
[4] To, N. H. T., Ingham,J. M. and Sritharan, S., Cyclic Strut & Tie Modeling of
Simple Reinforced Concrete Structures, 12th World Conference on Earthquake
Engineering, Auckland, New Zealand, 1249,2000
[5] Abrishami, H. H., and Mitchell, D., Analysis of Bond Stress Dishbutions in
Pullout Specimens, Journal o f S tructural E ngineenng, ASCE, 1 996, Vol. 122,
NO.3, pp. 255-261.
[6] Kankam, C. K., Relationship of Stress, Steel Stress, and Slip in Reinforced
Concrete, Journal of Structural Engineering, ASCE, 1997, Vol. 123, No. 1, pp.
79-85.
[7] Vecchio, F. J. and Collins, M. P., Modified Compression Field Theory for
@Seismicisolation
@Seismicisolation
526 Hong and l e e
Reinforced Concrete Elements Subjected to Shear, AC1 Structurai Journal,
American Concrete Institute, 1986, Vol. 83, No.2, pp. 415-425.
[8] Priestley, M. J. N., Seibie, E, Xiao, Y., and Verma, R., Steel Jacket
Retrofitting of Reinforced Concrete Bridge Columns for Enhanced Shear
Strength - Part 1: Theoretical Considerations and Test Design, AC1 Structurai
Journal, American Concrete Institute, 1994, Vol. 91, No.4, pp. 394-405.
[9] Priestley, M. J. N., Seible, E, Xiao, Y., and Verma, R., Steel Jacket
Retrofitting of Reinforced Concrete Bridge Columns for Enhanced Shear
Strength - Part 2: Test Results and Comparison with Theory, AC1 Structurai
Journal, American Concrete Institute, 1994, Vol. 91, No.5, pp. 537-551.
@Seismicisolation
@Seismicisolation
Table 1. Element forces by equilibrium before yielding
Tie ElementForce I Strut 1 Element Force
T 2
V Visine
Table 2. Element forces afie vrielding

Tie I Element Force Strut Element Force
Y
y
(
jd
I
2jd’
jd ,te).’
2
C , ~ ~
hor. dV
‘d I y, J 1+-
~
‘d 2 (Y, +dV)lsinB
@Seismicisolation
@Seismicisolation
528 Hong and Lee
(a) (b) (4
Fig. 1 Strut and tie modeling (a) forces acting on the column (b) strut and tie
model before yielding (initial yielding state) (c) strut and tie model after
yielding (ultimate state)
(4 (b)
Fig. 2 Differential tension chord element (a) force acting on a tension chord
length of dx (b) equilibrium condition
t’
(a) (b) (4
Fig. 3 Bond stress and slip relationship within a single crack (a) force acting
on a crack spacing (b) assumed bond stress distribution (c) slip distribution
@Seismicisolation
@Seismicisolation
Ltp
o c
; ew
Real curve
Fig. 4 Stress and strain relationship of longitudinal strut concrete element
After yielding
Initial yielding
4, Ja
jd'
(4 (b)
Fig. 5 The relationship between the deformation of longitudinal tie element and
that of longitudinal strut element (a) deformation mechanism (b)
compatibility condition between the longitudinal elements
(4 (b) (cl (4
Fig. 6 Force-deformation relationship of transverse tie element (a) equilibrium
and deformed shape (b) without longitudinal element deformation (b) crack
width change by flexural rotation (c) crack width change by axial deformation
of member
@Seismicisolation
@Seismicisolation
530 Hongand l e e
Ili
(a) @I (4
Fig. 7 Member deformation (a) deformation by truss deformation (b) joint
shear mechanism (c) calculation of joint rotation
Fig. 8 Comparisons with test results, Strength and deformation at failure
@Seismicisolation
@Seismicisolation
SP 209-29
Seismic Performance of a RC
Building with Columns
Rehabilitated with Steel Angles
and Straps
by S. M. Alcocer and R. Durán-Hernández
SvnoDsis: An ongoing study investigating the response of a reinforced concrete

building rehabilitated after the 1985 Mexico City earthquake is presented. As part
of this program, the behavior under cyclic loads of reinforced concrete columns re-
habilitated with steel jackets made of angles and straps is being experimentally as-
sessed. First results on the response of the building and on the behavior of rehabil-
itated columns are shown. So far, Building BL has responded elastically, with no
damage under recent events. It is apparent that BL response follows the soil funda-
mental frequency. A first look at the behavior of steel jackets made of angles and
straps through two column units, indicates that strength and energy dissipation Ca-
pacities can be improved, especially in undamaged columns.
Kevwords: confined concrete; instrumentation;

jacketing; reinforced concrete; repair; shear strength;
she a r stress; strength en ing
@Seismicisolation
@Seismicisolation
531
532 Alcocer and Durán-Hernández
Sergio M. Aicocer, FACI, is a research professor at the Inst. of Engng. of the
Natl. Univ. of Mexico (UNAM) and Director of Research at the Natl. Center for
Disaster Prevention (CENAPRED). He is a member of AC1 Committees 318,
Building Code Requirements, and 374, Performance Based Seismic Design of
Concrete Buildings, as well as a member of Joint ACI-ASCE Committee 352,
Joints and Connections in Monolithic Concrete Structures. His research interests
include earthquake-resistant design of reinforced concrete and masonry
structures.
Roberto Durán-Hernández is a Research Engineer at CENAPRED. He holds

B.Sc. and M.Sc. degrees from UNAM, where he is a Ph.D. candidate at present.
His research interests include the seismic behavior and rehabilitation of
INTRODUCTION
On September 19,1985, an 8.1 magnitude earthquake hit Mexico City. Although

the focus was 400 km WSW from the city, unique earthquake characteristics,
such as frequency content, duration and ground motion intensity, were evident in
the city. From the total building inventory of the city, only 1.4% collapsed or
sustained serious damage. Most damaged and collapsed buildings were
reinforced concrete (RC) fiames and waffle slab structures. The most vulnerable
proved to be those having 7 to 15 stones. The relationship between number of
stones and vulnerability was explained in terms of the shape of the acceleration
response spectra and the fundamental period of vibration of the buildings (1).
in RC frame structures, damage generally involved column failures due to high

axial and flexural forces, shear failure in short columns, shear distress in beams
due to lateral bending moments or settlement of foundations and joint distress
due to inadequate confinement or poor layout of connected members. in RC slab
structures, damage fi-equentlywas associated to shear distress near to columns or
at the edge of the capital over the column, and to punching shear and flexural
distress in columns (2). In many cases, inadequate detailing to achieve ductile
performance was common; widely-spaced ties, 90-deg hooks at hoop ends,
bundles of three, and even, four bars, and insufficient bar anchorage were
observed.
Of the buildings that suffered collapse or severe damage, 42 percent were

buildings at comers of a city block. Most of these had masonry walls on two
perpendicular sides and wide open facades to the street. in some cases, torsion
was due to an asymmetric layout of masonry infills. Weak ñrst-stow failures
@Seismicisolation
@Seismicisolation
wére also due to an irregular distribution of masonry infills along the height,
leaving almost bare frames at the ground floor.
According to building officials, more than one-thousand buildings were

rehabilitated (3). Repair and strengtheningefforts began almost immediately after
the event. A substantial number of buildings which suffered little or no damage
were strengthened. In many cases, strengthening was undertaken because similar
buildings had collapsed or were heavily damaged. All designs of repair and
strengthening schemes had to meet the emergency building regulations issued a
few weeks immediately after the quake, and subsequently, the Mexico City
Building Code of 1987 (4). Both regulations demanded higher lateral forces and
more stringent requirements to attain ductile behavior.
Techniques used for retrofitting were intended to both strengthen and stiffen the
structures (5 & 6). Although in most cases, economic factors dictated the choice
of the rehabilitation schemes, least-cost construction was not always the most
important consideration. The predominant rehabilitation techniques were
concrete and steel fiame jacketing, addition of shear walls or diagonal bracing,
replacement of damaged elements, and removal of top floors.
It is important to note that at the time, only qualitative guidelines were available
so that a great deal of engineering judgment and intuition were involved in the
decisions regarding damage evaluation and analysis, as well as design and
detailing of rehabilitation schemes.
Due to the large number of buildings rehabilitated and the frequent occurrence of
events along the Pacific coast, Mexico City provides the unique opportunity of
studying the performance of the different rehabilitation techniques used. Aimed
at studying the performance of such buildings, and at assessing the adequacy of
the analysis, design and construction considerations made for retrofitting, the
National Center for Disaster Prevention (CENAF'RED) has launched a long-term
research program (7). As a part of the program, selected buildings will be
thoroughly evaluated and, if possible, will be instrumented to record future
earthquakes. Buildings are to be selected based on a simple and symmetrical
structural layout, typical rehabilitation scheme, availability of structural
drawings, and importance of the structure.
Building BL is being studied. The primary objective of this project is to

determine the expected performance of the rehabilitated building BL under
@Seismicisolation
@Seismicisolation
different earthquake intensities, and to learn about the level of safety of the codes
and assumptions used. It was decided to assess the response through a series of
structural analyses of varying complexity. To calibrate the mathematical models
of the structure, BL was instrumented with seven accelerometers. It was
hypothesized that during the project at least some subduction earthquakes of
moderate intensity could be recorded, and therefore could be used to calibrate
such models. It is the aim of this paper to discuss the preliminary results of the
first motions recorded, as well as to present the overall cyclic behavior of
reinforced concrete columns rehabilitated (repaired andíor strengthened) by steel
jackets made of steel angles and straps. This technique is being assessed because
little experimental evidence is available on its performance when applied to
frame members (8, 9 & 10); data will be useful to evaluate the inelastic response
of Building BL, because perimeter columns were rehabilitated with this type of
scheme.
DESCRIPTION OF BUILDING BL
Building BL is an eleven-story reinforced concrete flat-slab structure, designed

under the 1966 Mexico City Building Code (Fig. 1). According to this code, the
seismic coefficient that corresponded to the site and type of the structure was
0.06. The building is regular in plan, consisting of three bays in both of the two
orthogonal directions. The flat slab floor system is 400-mm thick at the first level
(above the ground level), and 300-mm thick at the upper levels. To reduce the
weight of the slabs, sand-cement blocks were left embedded. The floor system is
supported on rectangular RC columns. Several bays of the structure were infilled
with brick masonry walls.
The structure is located in the lake zone in downtown Mexico City, and rests on a
partially compensated foundation that is 2.5-m deep. The box is supported on
point bearing piles driven to a 27-m depth.
Although the masonry infills were intended to work as partitions, that is, they
were not explicitly considered as structural elements for increasing the lateral
stifiess and strength of the building, walls were built against the fiame
members. Evidently, under lateral force demands, infills contributed to the
stiffness and strength of the lateral force resisting system.
In the aftermath of the 1985 earthquakes, as expected, damage in BL was

concentrated in the masonry infills, which were almost completely destroyed
throughout the entire building. Structural damage was recorded in few columns at
the ground story, characterized by moderate inclined cracking at column mid-
@Seismicisolation
@Seismicisolation
height. No distress was observed in or around the slab-column connections.
Based on the response of other similar buildings, it could be safely stated that had
the infills not been provided, consequently not dissipating energy, damage in the
structure would have been quite serious.
-eP . .
Once the large lateral flexibility of the building was identified as the primary
source of the damage observed, it was decided to stiffen the structure laterally
and to increase the strength of the foundation. Diagonal steel braces on the four
facades of the structure were provided (Fig. 1). This solution was selected
because the large distance between the edges of the braced fiame reduced the
axial forces and moments transmitted to the foundation, and therefore, the
number of piles that had to be installed. Design of the rehabilitation complied
with reference 4.
Steel diagonals were connected to the foundation and to slabs and columns. At
columns, steel diagonals and horizontal collectors were welded. Collectors were
also connected to the slabs through adhesive anchors. To improve the axial
strength and to increase the concrete confinement, columns of the braced frames
were strengthened. Columns were jacketed with steel skeletons made of four
steel angles placed at the comers along with horizontal steel straps (flat bars)
welded to the steel angles at 300-mm spacing. Comer columns were jacketed
throughout the building height, whereas interior columns of the perimeter fiames
were strengthened along the first six floors (Fig. 1).
Three fiction piles were driven at each corner to a total depth of 27 m. To

facilitate construction, segmental piles were installed through previously
perforated boreholes. New grade beams and pile caps were constructed. Further
details can be found in reference 6 .
The intended performance objectives of the rehabilitation scheme were: a) to

stiffen the structure laterally; b) to carry all the earthquake-induced forces to the
foundation; c) to reduce the fundamentalperiod of the structure to about 1 s; d) to
control interstory drift ratios to a maximum of 0.006; and e) to reduce stresses in
the interior fiames.
The building was instrumented with seven triaxial force balance accelerometers,
interconnected in a master-slave arrangement. Accelerometer layout was
conceived to measure the overall dynamic behavior of the building, especially
mode frequencies and shapes, dynamic amplification, and modal participation. At
@Seismicisolation
@Seismicisolation
each the ground story and roof level, three accelerometers were installed, one at
the center and two in opposite comers. The seventh accelerometer is at the center
of the 8th floor. The instrument at roof center acts as the master, which triggers
the other recorders. Due to space limitations around the building, it was not
possible to install a free field instrument. To overcome this situation, records
ffom the Liverpool station, operated by Centro de Instrumentación y Registro
Sisrnico (CIRES) and located some 50 m from BL, are being used. In this paper,
this station is referred to as Free Field.
Since 1999 up to date (April 2002), six seismic events of distinct intensity have
been recorded. Main seismological characteristics and maximum accelerations
are summarized in Table 1. Data from ambient vibration tests are also included.
Despite of its largest epicentral distance, the September 30, 1999 Mw 7.5
earthquake produced the highest acceleration demands onto the structure.
Detailed damage evaluations after such events have indicated that BL has
remained unscathed.
Transfer functions Roof Center / Ground Story Center for all signals recorded in
the horizontal directions are shown in Fig. 2. It is apparent that the shape and
location of peaks for all records are very similar. It can be noted that the
fundamental fiequency of the building in the horizontal directions (EW and NS)
are comparable and are around 1.2 Hz.
The variation of the fundamental periods of vibration in the EW and NS

directions is presented in Fig. 3. The event numbering in the abscissas
corresponds to the order in which events are summarized in Table 1. The
calculated periods from a standard modal analysis considering the building mass
at present are also shown. It is clear that largest periods are measured for the
most intense seismic motions. However, changes are too small to suggest any
significant damage in the structure, such as cracking or yielding. This is
consistent with the post-earthquake evaluations performed.
To evaluate the effects of soil dynamics on the building response, Roof Center -
to - Ground Story Center and Roof Center - to - Free Field transfer functions
were calculated. In Fig 4, transfer functions are presented for the EW and NS
components of the September 30, 1999 Mw 7.5 earthquake. These ratios show a
peak at about 1 Hz, thus implying that the building response is closely affected
by the fundamental period of the soil. This assumption can be verified by means
of response spectra calculated for the EW and NS components of the signals
recorded at the Ground Story Center and Free Field during the same event (Fig.
5). The similitude between the response spectra is evident. Two distinct peaks
can be distinguished at around 1 and 2 Hz. BL response is most sensitive to the
fundamental frequency as it can be noted in Fig. 4.
@Seismicisolation
@Seismicisolation
EXPERIMENTAL ASSESSMENT OF THE BEHAVIOR OF
REHABILITATED RC COLUMNS THROUGH STEEL JACKETS MADE
OF ANGLES AND STRAPS
To develop a model for nonlinear dynamic analyses of the building, it is

necessary to calibrate an appropriate hysteresis rule of all elements likely to
exhibit inelastic response under moderate-to-large earthquake excitations. Due to
the lack of enough experimental evidence on the hysteretic behavior of columns
rehabilitated with steel jackets made of angles and straps, it was deemed
necessary to carry out a series of tests aiming at: a) obtaining evidence on the
adequacy of this technique as a rehabilitation scheme; b) developing analysis,
design and construction guidelines; and, c) calibrating, or developing, a
hysteresis model of rehabilitated columns.
A series of four large-scale reinforced concrete columns were constructed (Fig.

6). The specimens corresponded to a 0.8-scale model of columns in building BL.
Two experimental variables were studied, i.e. detailing of the reinforcement, and
the level of damage in the specimen before the test. The first column was tested
to failure (test C-99); this column satisfied the detailing requirements of modem
RC standards for seismic design, such as those in Chapter 21 of the AC1 3 18-99
Building Code Requirements for Structural Concrete (1 1). The second specimen
(C-66) was intended to emulate a column at the ground story of BL; this column
was also tested to failure. The longitudinal and transverse reinforcement of C-66
satisfied the requirements of the 1966 Mexico City Building Code (12). The third
specimen (C-66-R) was similar to C-66 and was first tested to a drift angle to 1%
(test C-66*). Then the column was rehabilitated with steel jackets and was
retested to failure (test C-66-R). It was considered that damage in this type of
columns at 1% drift was similar to that observed in BL during the 1985 event,
and was still repairable. Specimen C-66-S was similar to C-66 and was
undamaged prior to jacketing to assess the effect of a damaged core on the
response. All columns were tested under a constant axial compression load
equivalent to O. 15f’cA,, which was found to be the axial load level at the ground
story in Building BL.
Column dimensions and reinforcement details are shown in Fig. 6. Columns

were square with 500 mm sides and 2,000 mm high. Longitudinal reinforcement
was continuous throughout the column height and was anchored inside concrete
blocks by means of 90 deg hooks. For specimens C-66, C-66-R and C-66-S;
longitudinal bars were placed in bundles of three bars and the tie hooks were bent
at 90 deg, as it was usual in 1960‘s construction.
Jackets of C-66-R and C-66-S were the same as those provided in Building BL.
@Seismicisolation
@Seismicisolation
Horizontal steel straps were welded to comer angles through fillet welds.
Similarly to the jacketed columns in BL, jacketed specimens had wider steel
plates welded onto the angles at the column ends (Fig. 6).
Columns were built with ready-mix concrete, with 19 mm maximum size coarse
aggregate. The concrete compressive strength at time of test was 27.9 MPa on the
average. The specified yield stress of column reinforcement was 412 MPa; steel
jackets were made with A36 steel. Column jacketing was constructed in a similar
fashion for the two columns rehabilitated. The only exception was C-66-R, which
required the removal of loose concrete from test C-66* before jacketing. First,
angles were placed and secured to the column comers. Afterward, the steel flat
bars (straps) and end plates were fillet welded to the angles. Finally and like in
columns of BL, the gap left between the concrete column and the steel flat bars
(straps) and end plates was filled with a high-strength grout.
Columns damaged in BL were repaired with epoxy resin injected into the cracks
prior to jacketing. in this series of tests, it was decided not to repair the cracks of
specimen C-66-R to avoid introducing other experimental variable whose effect
on the column would be difficult to differentiate.
As it has been mentioned, mild steel bars were used in the experiments. in
contrast, the reinforcement in BL consists of high-strength cold-twisted bars.
This type of steel reinforcement was commonly used in Mexico Ci!y prior to the
1985 earthquakes, and is not longer being fabricated nor used in the country.
Such reinforcing bars have a specified yield stress of 589 MPa. Although this
type of bars is available in other countries and could have been imported for the
project, it was not possible to build the columns with this type of bars because of
economical constraints. It is recognized that this fact may be a shortcoming of the
test program, especially with regards to columns with shear-dominated behavior.
It can be argued that for columns in which shear-bond failures are expected, the
type of reinforcement (grade and deformations), are likely to affect the inelastic
cyclic behavior.
The specimens were mounted on the reaction floor to avoid rotation of the lower
end of the column. To restrain the rotation of the upper concrete block, a set of
hydraulic jacks and steel pinned struts were used (Fig. 7). All columns were
tested using the same bi-directional cyclic load history, shown in Fig. 7. The load
history was displacement-controlled based on the interstory drift and represented
a severe load condition for a column. Nine levels of deformation were applied.
Most cycles were applied in the primary (EW) direction only. Unidirectional
loading in the NS direction occurred in cycles 7 to 9. Bi-directional cycles are-
identified in Fig. 7. Three cycles at same deformation levels were applied. The
loading history was modified from that proposed in reference 13.
@Seismicisolation
@Seismicisolation
The EW lateral load versus interstory drift curves and the final crack patterns for
all specimens are presented in Figs. 8 and 9, respectively. The ultimate story
shears corresponding to column shear failure and column hinging (at the bottom
and at both ends) are also shown in the figure. The story shear corresponding to
shear failure of columns C-99 and C-66 was based on the shear strength
recommended by reference 11. The story shear to produce column hinging was
calculated from measured material properties and dimensions, considering an
equivalent rectangular stress block for the concrete and assuming that plane
sections remain plane. For specimens C-66-R and C-66-S, the contribution of the
steel jacket to shear strength was added to the shear resistance provided by
reinforced concrete. The shear contribution of the jacket was calculated assuming
a plastic mechanism under lateral loads that consider flexural hinging at both
ends of all flat bars along the height of the column, and from measured material
properties and dimensions. The contribution of the jacket was simply added to
the concrete and stimp contributions.
The hysteresis loops are nearly symmetrical and show considerable pinching,
especially at large drifts, and severe stifhess degradation. As expected, specimen
C-99 showed a more stable behavior than C-66 columns. Hysteresis loops of C-
66 were typical of shear-dominated columns, in which severe strength
degradation was observed after reaching its lateral strength. The lateral strength
of C-66 was about two-thirds of C-99 strength, and the drift at peak load was 2%.
C-99 and C-66 were purposely tested up to interstory drift angles of 5 and 4%,
respectively. Although it is recognized that these drifts are unrealistically high for
conventional frames, it was decided to explore if the axial load carrying
capacities could be exhausted at large displacements. It was found that both
columns were able to resist the design axial force value, although the lateral load
carrying capacity for specimen C-66 was severely diminished.
For rehabilitated columns, loops at similar drifts, exhibited better energy

dissipation capacity (wider hysteresis loops) than that of C-66. This is credited to
the participation of the steel jacket. in terms of strength, both C-66-R and C-66-S
exhibited a superior behavior as compared to C-66. The participation of the steel
jacket to the load carrying capacity of the column is therefore evident. Test C-66-
S had to be interrupted when the maximum actuator load capacity was reached
(1O00 kN, approximately).
in all cases, column behavior deteriorated when bi-directional cycles were

applied. The lateral stifiesses at the first loading peak of all columns were
comparable, thus suggesting that the conbibution to lateral stifkess of the steel
jacketing, both in a damaged and in an undamaged column, was minor and could
be disregarded for design purposes.
@Seismicisolation
@Seismicisolation
Regarding specimens’ crack patterns, C-99 exhibited hairline flexural and
inclined cracks, which were very uniformly distributed along the height; at the
end of the test (to 5% drift) minor crushing and spalling of concrete comer was
observed (Fig. 9). C-66 and C-66*, in contrast, experienced a large number of
fine inclined cracks with few horizontal cracks; failure of C-66 was characterized
by a wide inclined crack. For C-66-R, new inclined cracks were formed
suggesting the formation of inclined strut action of the concrete against the steel
skeleton. Similarly to C-66-R, specimen C-66-S exhibited inclined cracking and
spalling that indicates strut action. In Fig. 9, a proposed strut inclination, based
on the observed cracks and crushing patterns, is depicted in columns C-66-R and
C-66-S. For clarity, struts are shown for one direction of loading. The average
inclination of the struts, with respect to the vertical axis, is 35 degrees. It can be
argued that the strut action is an indication of the development of a truss
mechanism where the steel angies and straps are involved as truss chords and
ties, respectively.
For columns C-66-R and C-66-S, it is obvious that the model used for calculating
the shear strength of the rehabilitated columns, including the contribution of the
steel jacket, was quite conservative. This is mostly credited to the plastic
mechanism assumed for the jacket contribution, which only accounted for the
shear strength of the jacket when all straps hinged at their ends. Thus, the passive
confining effect that the jacket could have exerted into the column cores was not
considered. It is believed that the crack inclination suggesting the formation of a
series of concrete struts, such as in a truss analogy, is an indication of the
increase of the concrete shear strength contribution attributed to confinement.
Preliminary analysis of strain gages located on the straps indicated that at mid-
length, the straps remained elastic during the tests; yielding was recorded only at
the ends of the straps due to bending moments. The first finding indicates that the
truss action inferred from the cracking pattern may have not involved yielding of
the truss ties (i.e. straps), but rather crushing of the struts, as can be noted in Fig.
9. The fact that yielding was recorded at the strap ends is consistent with the
plastic mechanism assumed for the steel jacket. Further analyses are warranted to
better quantiS. the contribution of the possible distinct mechanisms.
FINAL REMARKS
Several remarks can be made about this ongoing investigation. Building BL has
responded elastically, with no damage under recent events. It is apparent that BL
response follows the soil fundamental ffequency. Code requirements (current and
1966) for column design in Mexico City have been experimentally assessed.
Test results from columns rehabilitated with steel jackets (comer angles and
straps) indicate that strength and energy dissipation capacities can be improved,
especially in undamaged columns.
@Seismicisolation
@Seismicisolation
ACKNOLWEDGMENTS
The interest and help from Mr. Francisco García-Jarque to get the Building BL
instrumented is gratefully acknowledged. The experimental evaluation of
rehabilitated columns is part of a joint research project between the Institute of
Engineering of UNAM, the Pacific Earthquake Engineering Research Center and
CENAPRED that has been supported through a joint grant of the Institute
Mexico - United States of the University of California and the Mexican National
Council for Science and Technology (CONACYT). CEM= Concretos,
Siderúrgica Tultitlán and SICARTSA Villacero, and MBT Mexico kindly donated
the ready-mix concrete, the steel reinforcement and the grout, respectively, for
the columns; their collaboration and support is gratefully acknowledged. The
authors wish to thank Mr.’s Carson Torres, Pablo and Carlos Olmos for their
help during the testing phase, and the Coordination of Monitoring and
Instrumentation of CENAPRED, and the Centro de Instrumentación y Registro
Sísmico (CIRES), for providing the records from Liverpool station.
REFERENCES
1. Rosenblueth, E., and Meli, R., “The 1985 Earthquake: Causes and Effects
in Mexico City,” Concrete International, American Concrete Institute,
Vol. 8, No. 5 , May, 1986, pp. 23-34.
2. Jirsa, J.O., “Repair of Damaged Buildings - Mexico City,” Proceedings of
the Pacijìc Conference on Earthquake Engineering, New Zealand, Vol. 1,
1987, pp. 25-34.
3. Departamento del Distrito Federal, Sismos de 1985 - Control de
Edijìcaciones, México, D.F. 1985-1988, Mexico D.F., 1988, 176 pp.
4. Departamento del Distrito Federal, Reglamento de Construcciones para
el Distrito Federal, Gaceta Oficial del Departamento del D.F., 1987, 100
PP.
5. Aguilar, J., et al., T h e Mexico Earthquake of September 19, 1985 -
Statistics of Damage and of Retrofitting Techniques in Reinforced
Concrete Buildings Affected by the 1985 Earthquake,” Earthquake
Spectra, EERI, Vol. 5,No. 1, February, 1989, pp. 145-151.
6. Aguilar, J., Breña S., del Valle, E., Iglesias, J., Picado, M., Jara, M., and
Jirsa, J.O., “Rehabilitation of Existing RC Buildings in Mexico City,”
Phil M. Ferguson Structural Engineering Laboratory Report 96-3,
University of Texas at Austin, August, 1996, 170 pp.
7. Alcocer, S.M., “Research Program on the Assessment of the Behavior of
Rehabilitated Buildings in Mexico City,” L4BSE Symposium Extending
the Lifespan of Structures, Vol. 73-1, San Francisco, CA, August, 1995,
pp. 127-132.
8. Migliacci, A. et al., “Repair Techniques of Reinforced Concrete Beam-
Column Joints,” Final Report, IAJ3SE Symposium on Strengthening of
@Seismicisolation
@Seismicisolation
Buildings Structures-Diagnosisand Therapy, Venice, 1983, pp. 355-362.
9. González, O., et al., “Resistencia a fuerza cortante de columnas de
concreto reforzadas con camisa de acero (2” etapa),” Revista
Construccióny Tecnología, IMCYC, enero 2001, pp. 34-42.
10. Valluvan, R., Kreger, M., and Jirsa, O., “Strengthening of Column
Splices for Seismic Retrofit of Nonductile Reinforced Concrete Frames,”
Structural Journal, ACI, Vol. 90, NO.4, 1993, pp. 432-440.
11. AC1 Committee 3 18, Building Code Requirements for Structural
Concrete (318-99) and Commentary (318R-99). American Concrete
Institute, Farmington Hills, MI, 1999.
12. Departamento del Distrito Federal, Reglamento de Construcciones para
el Distrito Federal, Mexico, D.F., 1966, 104 pp.
13. AC1 ITGíTl . l -99, Acceptance Criteria for Moment Frames Based on
Structural Testing, American Concrete Institute, Farmington Hills, MI,
1999.
TABLE 1 - EARTHQUAKES RECORDED IN BUILDING BL

Event Magnitude Depth Epicentral Maximum Acceleration(un/s’) Maximum Acceleration(cmh’)
(km) Distance Center Roof 1 Center Base South Roof 1South Base
(km) E-W N-S V E-W N-S V
Noise - - - -0.811- -0.671- 0.911- - - I
615-99 Mw7.0 70 224 -122b26.7 107125.1 12310. - - -_

3
6-21-99 Ms5.6 42 313 -2O.W-7.4 17.6/4.7 2.63/-2.0 -- - -
7-5-99 Ms 5.0 35 306 2.36/0.77 -1.65k0.5 0.46/-1.2 2.5314.7 1.6710.5 0.49/0.42
7-12-99 Ms4.7 15 395 -2.7V0.55 2.3W0.6 0.43//0.5 -2.710.5 2.3W0.6 0.44/0.36
430-99 Mw7.5 42 439 643125.6 -82.5135 7.3/-5.1 82.7125.6 -64.51-3.7 14.5113.0
10-29-01 Mc6.1 10 296 -20.415.48 -21.Z-6.95 3.4/2.5 - - .__
@Seismicisolation
@Seismicisolation
* .
Fig. 1 - Building BL
@Seismicisolation
@Seismicisolation
Transfer Function Roof Center I Base Center. Dlr E-W
25
20
-a>
.z
-
E
15
4 10
O
O 2 4 6 a 10
Frequency,Hz
O 2 4 6 8 10
Frequency, Hz
Fig. 2 - Transfer functions Roof Center / Ground Story Center of building BL
@Seismicisolation
@Seismicisolation
0.90
0.88
-
0.86
0.84
o
.s
O
n
0.82
0.80
0.78
-3 N-S 0.78
Event
Fig. 3 - Changes in the fundamental period of building BL in the EW and NS directions
@Seismicisolation
@Seismicisolation
Transfer Functions, Dir E-W
20
15
O
O 2 4 6 a 10
Frequency, Hz
20
Transfer Functions, Dir N-S
15
O
O 2 4 6 10
Frequency, Hz
Fig. 4 - Transfer functions Roof Center / Ground Story Center and Roof Center / Free
Field for the September 30, 1999 earthquake, in the EW and NS directions
@Seismicisolation
@Seismicisolation
90
r60
E,
m-
C n
30
O I I
O 1 2 3 4 5
Period,s
120 r N-S
- Base
o ! I I I I
O 1 2 3 4
Period. s
Fig. 5 - Response spectra at Ground Story Center and Free Field for the September 30,
1999 earthquake, in the EW and NS directions
@Seismicisolation
@Seismicisolation
(16-31.8 mm bars)
(9.5rnm hoops @lo0mm)
700 m m 700 mn
2000 m m 2000 m,
700 m m 700 mn
c-99 Column Series C-66 Column Series
í100x100x9.5 mm)
E70
65 mm tvp
RehabilitationScheme
(C66-R. C-66-9
Fig. 6 - Column dimensions and reinforcement details
@Seismicisolation
@Seismicisolation
4--
3--
-3--
- E-W Dü-
.........
N-S Directicm
-Bidirectional
Cycles
Fig. 7 - Test setup and loading history
@Seismicisolation
@Seismicisolation
c-99
1000
'
800
600
400 400
2 200 u- 200
m
O
J
5 O
-p -2000 -200
L -400 -400
2 a00 ao0
-800 -800
-1O00 -1000
- 5 - 4 - 3 - 2 - 1o 1 2 3 4 5 - 5 - 4 - 3 - 2 - 1o 1 2 3 4 5
Interstory Drift, % Interstory Drift, %
C-66-R
1000
800
5:
;
as o 200
-L -200
L -400
!
I-600
-800
-1000
-5 -4 -3 -2 -1 0 1 2 3 4 5 -5-4-3-2-1 o 1 2 3 4 5
Interstory Drift, YO Interstory Drift, YO
C-66-S
1O00
800
5 600
400
g 200
-
J O
2 -200
$ -400
a00
-600
-1000
-5-4-3-2-1 0 1 2 3 4 5
Interstory Drift, %
Fig. 8 - EW story shear - interstory drift hysteresis curves
@Seismicisolation
@Seismicisolation
Fig. 9 - Visible damage of the specimens at the end of the tests
@Seismicisolation
@Seismicisolation
Fig. 9 - Visible damage of the specimens at the end of the tests(cont)
@Seismicisolation
@Seismicisolation
SP 209-30
Shake Table Studies of Effects of

Foundation Flexibility on Seismic
Demand in Substandard
Bridge Piers
by M. Saiidi, B. Gopalakrishnan,
and R. Siddharthan
Svnopsis: An exploratory study of the effect of foundation rocking on the shake

table response of a quarter-scale two-column bridge bent was conducted. Several
innovative details and designs were used in the columns and the cap beam but the
footing represented typical substandard and retrofitted spread footings. The struc-
ture was tested in three modes: rocking with as-built footings, rocking with retro-
fitted footings, and fixed base. The foundation flexibility was modeled in the tests
using elastomeric bearing pads that were designed to represent the stiffness of
spread footings supported on medium soil. The rocking tests were conducted in the
elastic range, but the fixed-base model was tested to failure. The Sylmar record of
the 1994 Northridge earthquake was simulated on the shake table. The data
showed that for the same input earthquake intensity the maximum steel bar strain
near the base of the columns in an unretrofitted rocking footing was only 7 percent
of the strain in the fixed frame. When the footing was retrofitted but was still al-
lowed to rock, the maximum steel strain increased but was still only 16 percent of
the strain in the fixed frame. The retrofit design implications of the results are dis-
cussed.
Kevwords: bearing pad; bridge; earthquake; footing

flexibility; pier; rocking; shake table; steel strain
@Seismicisolation
@Seismicisolation
553
554 Saiidi et al.
M. Saiid Saiidi, FACI, is a Professor of Civil Engineering at the University of
Nevada, Reno. He is involved in analysis and shake-table studies of reinforced
concrete bridges. He is the former chair and current member of AC1 Committee
341, Earthquake-Resistant Concrete Bridges. He is also a member of AC1
Committees 342 and 352 and has served on various other AC1 committees.
Balaji Gopalakrisbnan is a staff engineer at the California Department of

Transportation in San Diego California. He conducted research on dynamic
response of bridge bents while studying at the University of Nevada, Reno.
Raj Siddharthan, is a Professor of Civil Engineering at the University of

Nevada, Reno. He is involved in research on seismic analysis and behavior of
bridge foundation systems and abutments.
INTRODUCTION
Recent major earthquakes and the failure of many highway bridges in

different parts of the world have prompted an extensive seismic retrofit program
for highway bridges. Seismic retrofit is expensive. The main components of
seismic retrofit cost include: retrofit design, construction, trafic control, and user
costs caused by traffic delays due to detours or lane closures. A major retrofit of
substructure would involve strengthening of the foundation, columns, joints, and
the cap beam. The target performance objective is typically to upgrade these
components to a level comparable to that of new bridge components. This would
mean enlarging and thickening of footings [i], retrofitting columns to form
ductile plastic hinges at the top and bottom, increasing the shear capaciîy of the
columns, increasing confinement over lap splices, strengthening the joints, and
upgrading the beams.
The benchmark to establish the degree of seismic deficiencies of existing

bridges has been primarily test results from experiments on reinforced concrete
bridge columns or piers that have been fixed at the base. In reaiity many bridge
columns are supported on spread footings that can rock during earthquakes. It is
generally known that the softening effect of rocking should be similar to that of
base isolators and should reduce earthquake forces. However, no test data are
available to demonstrate and quantifi the beneficial effects of foundation rocking.
The purpose of the investigation described in the present article was to conduct a
shake table test of a large-scale bridge bent supported on substandard footings that
were allowed to rock during earthquake simulation. Tests were also conducted on
the same model after the footings were retrofitted using current footing retrofit
methods to determine the rocking effects on a retrofitted pier. Finally the
retrofitted footings were fixed at the base and earthquakes were simulated to
@Seismicisolation
@Seismicisolation
evaluate the difference in the seismic demand on the bent elements as a result of
foundation fixity.
TEST MODELS
Structural Svstem
The high cost of maintenance and repair of aging bridges built with
conventional construction materials has necessitated an examination of
nontraditional materials for possible adaptation in bridge engineering. One of the
promising solutions for durable and reliable designs of new bridges is the use of
fiber-reinforcedplastic (FRP) fabrics with superior characteristics [2]. Therefore,
the development and application of advanced composite materials for design,
repair, and retrofit of bridges are key issues that are being considered by many
researchers. To date, the application of FRP fabrics in structural design of new
construction (as opposed to retrofit application) has been relatively rare and has
been limited to bridge deck panels. Glass and carbon fiber reinforced plastic
sections and bars have been used for specialized applications [3,4]. Their use in
new structural systems has been limited primarily due to the lack of sufficient
research data and standards. An innovative concept was explored in this study,
wherein steel reinforcement was used to provide ductility capacity and to control
the failure mode, and carbon fiber fabrics were used to provide shear
reinforcement, confinement, and longitudinal reinforcement where elastic
behavior is needed. In addition, the hinge zones were relocated [5,6] away from
member ends to limit and control the damage entirely within plastic hinge zones,
which are specially detailed, while all other portions of the structure remain
essentially elastic. To allow for the study of substandard footings and the rocking
effects the bent was built on spread footings described in subsequent sections.
Description of Specimen
The test specimen was a two-column pier with square columns. It was
constructed to % scale so that it could be accommodated on a 14 fi. x 14.5 fi. (4.3
m x 4.4 m) shake table. Unidirectional carbon fiber reinforced plastic (CFRP)
fabrics were placed externally on the frame. Figure 1 shows the bent dimensions
and also the composite layout. The lines within the elements show the direction
of carbon fibers. Figure 2 shows the beam and column cross-sections. Plastic
hinges in the columns were offset by placing the #5 bars at column ends as shown
in Fig. 2. The bent dimensions were chosen to reflect those of typical highway
two-column bents with square columns. The entire beam length and column
middle regions were designed to remain elastic. Hence FRP composites were
used for longitudinal reinforcement in these locations. Minimal #4 bars were
provided for reinforcement of the beams and columns, also #4 stirrups were used
@Seismicisolation
@Seismicisolation
556 Saiidi et al.
at 9-in. (229-mm) spacing to reinforce the fiame before the composites were
installed and to resist loads induced in the frame during its transportation. Two
horizontal and two vertical layers of CFRP fabrics were provided at each joint on
both sides of the bent for joint shear reinforcement. The joint shear design was
based on the AC1 352 provisions [7]. The concrete used in the pier was normal-
weight concrete with a maximum aggregate size of 3/8 in. (9.5 mm in.). The
target concrete compressive strength was 5 ksi (34 m a ) , while the actual
strengths were 6.26 ksi (42.8 MPa) in the columns and 5.29 ksi (36.1 MPa) in the
beam on the day of testing. The Mitsubishi CFRP product, Replark Type 30, was
used.
In the fiber direction, the measured CFRP fabric elastic modulus was
38,400 ksi (262 GPa) and its tensile strength was 380 ksi (2.59 GPa). The lamina
elastic modulus perpendicular to the fiber direction was 510 ksi (3.48 GPa) and its
tensile strength perpendicular to the fiber direction was 1.76ksi (12 MPa). The
average measured yield stress for the steel bars was 64.9 ksi (447 MPa). More
details about the installation of the fabrics are presented in [8].
Footings
The footing details and dimensions in the specimen were based on a

review of existing two-column bridge piers that were supported on spread
footings bearing on sand with an allowable soil bearing pressure of 4 ksf (0.19
MPa) under unfactored gravity loading. The ultimate soil bearing capacity was
assumed to be 7 ksf (0.34 MPa) Figure 3 shows the elevation of the footing in the
as-built model. The column reinforcement details in the figure are not fully
shown. Typical deficiencies in substandard footings are inadequate plan view
dimensions, insufficient thickness, low shear strength, and a lack of top
reinforcement [i]. The as-built footings were retrofitted to address all these
deficiencies. The performance objective in new footing construction is to
minimize damage to the footings and to force plastic hinging of the columns.
This level of performance is also expected of retrofitted footings. The moment
demand in the footing was estimated by calculating the plastic moment capacity
near the base of the column and amplifying the moment by 30 percent to account
for material over-strength in the column. Based on the combined effect of gravity
load and the amplified column plastic moment, the required plan view dimensions
of the footing were found using an ultimate soil bearing capacity of 7 ksf (0.34
MPa). An overlay was added to increase the effective depth of the footing, which
in turn led to an increase in the flexural and shear strength. The footings were
also enlarged in plan, and bottom steel bars were drilled into the bottom of the
footing to form a lap splice with the existing bars. A top mat of steel with a cross
sectional area of approximatelyone-half of the bottom steel area was placed in the
overlay [i]. The new concrete was anchored in the existing as-built footing
using dowels. Figure 4 shows the elevation of one of the retrofitted footings.
@Seismicisolation
@Seismicisolation
Beariw Pads Desim
Bearing pads were placed at the bottom and the sides of the footings to
simulate the vertical and horizontal footing flexibility. The latter pads on the
sides were placed perpendicular to the loading direction only. The pad stifhess
was based on the stifhess of a prototype footing represented by the model
footing. The analysis was done for both the as-built and retrofitted footings. The
vertical and horizontal stiffness of the prototype footing was determined assuming
that the footings are supported on dense cohesionless soil and using an iterative
algorithm developed in [9] that implemented the Schmertmann’s strain factor
method [lo]. Using this algorithm, nonlinear horizontal and vertical stress-
displacementrelationships were developed. The normal and shear stresses in the
prototype and the model were kept the same. The normal stress was based on
gravity loads and the shear stress was estimated from a dynamic analysis of the
test pier. Using these stresses and the nonlinear stress-displacement relationships,
the secant vertical stiffness for the as-built and retrofitted footings was 26.7 ksflin
(50.3 MPdm) and 34 k s f h (64.1 MPdm), respectively. The secant horizontal
stiflhess was 17.9 ksflin (33.8 MPdm) and 24.4 ksflin (45.9 MPdm)for the as-
built and retrofitted footings, respectively.
Elastomeric pads with durometer rating of 50 and the size of 10 by 10 by 2

in. (254 by 254 by 50.8 mm) were placed under the footing. Similarly pads with
the size of 10 by 10 by 1 in. (254 by 254 by 25.4 mm) were placed along each
vertical face of the footing perpendicular to the loading direction. In each of the
as-built footings nine pads were placed below and three pads were placed on each
side. The number of the pads in the retrofitted footing was 16 below the footing
and four on each side.
Testing Program
The general test setup is shown in Fig. 5. Figure 6 shows the close up of
one of the footings in the retrofitted specimen test. The footings were placed on
bearing pads on one of the University of Nevada, Reno shake tables. Steel
brackets anchored to the table with threaded bars provided vertical surfaces for
supporting the side bearing pads and prevented slippage of the footings. The bent
and the footings were extensively instrumented with strain gages, displacement
transducers (to measure displacements, footing rocking, and curvatures),
accelerometers and load cells (to measure vertical and lateral loads applied to the
specimen). For simulating a vertical load of 6.1 kipíft (88.6 kN/m) on the beam, a
number of threaded bars running between the loading frame on the top and anchor
beams (which were rigidly attached to the shake table through the steel brackets)
were stressed to 6.4 Kip (28.5 kN) each using hydraulic rams. Because the rams
remained perpendicular to the beam at all times, they did not produce a horizontal
force component and did not affect the horizontal force on the frame. A system of
accumulators was attached to the rams to minimize axial load fluctuation. To
@Seismicisolation
@Seismicisolation
558 Saiidi et al.
replicate the inertial effect of the gravity loads, a mass rig was used, which
consisted of a pinned steel fiame mechanism supporting concrete reaction blocks.
The mass rig had an effective lateral inertial load of 64 kips and it was connected
to the loading fiame by a pinned circular steel tube.
The bent was tested in three modes: (1) Rocking tests with as-built
footings (B2SA for bent with 2 square columns, --built), (2) rocking tests with
retrofitted footings (BZSR), and (3) Fixed base tests (B2SF). The testing in the
first two modes (models B2SA and B2SR) was limited to avoid yielding
anywhere in the fiame, but in the last mode the bent was tested to failure. The
failure of B2SF occurred when transverse CFRP fabrics were ruptured in the
plastic hinge zone [SI. Reference 8 also shows the correlation between the
analytical and measured results.
The earthquake history used to drive the shake table was the January 17,
1994 Northridge earthquake as measured at the Sylmar Hospital. This record was
chosen because it was representative of a typical near-fault earthquake in the
Western United States. Furthermore, analytical studies showed that this record
would lead to high ductility demand thus allowing for ultimate behavior testing.
Because the test specimen was a quarter scale model of the prototype column, the
earthquake record time axis had to be scaled by a factor of 40.25 to create a
response in the test specimen that would parallel the prototype column response.
The testing was conducted by applying scaled versions of the strong motion
acceleration record. The rocking tests consisted of six events for the as-built
model (BLSA) and four events for the model with the retrofitted footings (B2SR).
Specimen B2SF was subjected to thirteen events until visible column failure
occurred. The first test in all the models had a peak table acceleration of 0.06g
(ten percent of the peak ground acceleration in the Syimar record). Subsequent
runs were applied with target input peak acceleration increased at increments of
0.06g up to Run 6. The failure testing of BLSF is the subjected of Ref 181.
TESTRESULTS
The main purpose of the part of the research that is presented in this article
was to determine the effect of rocking on force and displacement demands.
Hence only the results specific to this aspect of the study are examined.
DisDlacement ResDonse
Figure 7 shows the cumulative measured force-displacement relationships

for B2SA (as-built specimen). The displacement shown is the horizontal
movement measured at the top of the fì-ame relative to the shake table, and the
force is the horizontal inertia force including the “p-delta” effect. The maximum
@Seismicisolation
@Seismicisolation
force on the frame was 16 kips (71.2 kN) and the maximum displacement was
0.61 in. (15.5 mm). The maximum steel bar strain was measured near the top of
the columns and was less than 1000 microstrains, far below the yield strain of
2240 microstrains. No significant cracking of the footings was observed in the
specimen. The hysteretic response seen in Fig. 7 is primarily due to rocking and
translation of the footings and the nonlinear response of the bearing pads. Figures
8 and 9 show the measured translation and rocking rotation of the right footing.
Both curves show hysteretic response reflecting the nonlinear behavior of the
bearing pads. Note the maximum horizontal displacement at the footing was 0.16
in (4.1 mm), which is 26 percent of the maximum displacement at the top of the
bent. The rocking of the footing translates into additional horizontal top
displacement that is difficult to isolate, but is estimated to be at approximately
five percent of the total displacement based on a comparison of the lateral
displacement for B2SF and B2SA.
The final earthquake event for specimen B2SR had peak amplitude of
0.24g rather than 0.36g that was used for B2SA. This was to ensure that no
yielding of the steel bars would occur in the bent and to save the bent for testing
with fixed base. Figure 10 shows the lateral load-displacement relationships
combined for all the runs and Figs. 11 and 12 show the footing response.
Comparison of Figs. 7 and 10 shows that despite the lower amplitude of the input
earthquake in B2SR by 33 percent, the lateral inertial force in B2SR was slightly
higher due to the increased stiffness in the footings, which was as a result of
increased floor plan area of the footings and the higher number of bearing pads.
The higher footing stiffness in B2SR also led to a reduction in the horizontal
translation of the footing (Fig. 11 vs. Fig. 8), but the maximum footing rotations
in the two models were comparable (Fig. 12 vs. Fig. 9).
The maximum top displacement in B2SA during the 0.36g motion was
0.51 in. (13 mm) and in B2SR during 0.24g motion was 0.55 in. (14 mm).
Assuming proportional response the extrapolated response of B2SR under 0.36g
motion would be 0.83 in. (21 mm), which is more than 60 percent higher than the
peak displacement in B2SA. The increase in displacement is attributed to the
higher stifhess of B2SR and the resulting reduction in vibration period and
increased inertial forces.
Internal Forces
To estimate the force demand in different parts of the frame, the maximum
strains in the bars in different components of the bent were evaluated. Typically
critical areas of bridge bents that govern the retrofit design are the ends of
columns. At column bases typical seismic deficiencies identified by testing of
fixed base specimens include short lap splices and small amount of transverse
steel. The aim of the tests on fixed based specimens has been to determine the
retrofit that is required to develop the yield moment in the column and to provide
sufficient rotational ductility capacity without brittle failure. At column tops,
@Seismicisolation
@Seismicisolation
560 Saiidi et al.
typical problems are insufficient lateral steel and, occasionally, inadequate
anchorage of column longitudinal bars into the joint. Again, the performance
objective is to develop a sufficientlyductile plastic hinge at the top of the column.
Figures 13 and 14 show the maximum measured longitudinal bar strains

near the base and near the top of the right column, respectively. Note that the
columns were designed to have offset plastic hinges as opposed to having plastic
hinges at the column ends. The data are for the bars at the plastic hinges and
indicate the maximum demand. From Fig. 13 it is clear that the stress buildup in
the column bars near the base of the as-built specimen was considerably less than
that of the retrofitted or the fixed-base model. For event 0.3xSyimar, the
maximum strain in the as-built case was only 7 percent of the strain for the fixed
case. In the retrofitted footing the percentage was 16. These figures clearly point
out that the demand at the base of the columns when rocking is allowed may be
far below the demand from a fixed base test specimen. The important design
implication fiom the data is that extensive retrofit of column bases may not be
necessary in real bridge columns supported on spread footings where rocking is
expected.
Figure 14 shows a similar trend for the effect of rocking on the strains near
the top of the column. However, the reduction in the maximum strains caused by
rocking is not as pronounced. Nonetheless the strains for rocking cases were
substantially lower compared to the bar strains in the fixed model. The peak
strains in the as-built and retrofitted footings were 31 and 55 percent of the strains
in the fixed model, respectively.
n i e low strains in the rocking models indicate a relatively small flexural

demand in the columns. The lower than expected flexural demand means that the
extent of yielding may be considerably less than that assumed for fixed structures
and that retrofit detailing requirements can be eased, thus leading to savings in the
retrofit costs. Additional investigations of the rocking effects are in progress by
the authors with the aim of developing specific guidelines that can aid designers
to determine the level of necessary retrofit taking into account the beneficial
effects of foundation rocking.
CONCLUSIONS
An exploratory experimental study of the effect of foundation rocking ön

the shake table performance of a quarter-scale two-column bridge bent showed
that the earthquake demand on column forces may be substantially less than what
is generally assumed or calculated using fixed-base models. The foundation
flexibility was modeled in the tests using elastomeric bearing pads that were
designed to represent the vertical and horizontal stiffness of spread footings
supported on medium soil. The &ta showed that for the same input earthquake
intensity the maximum steel bar strain near the base of the columns in an
unretrofitted rocking footing was only 7 percent of the strain in the fixed fiame.
@Seismicisolation
@Seismicisolation
When the footing was retrofitted but was still allowed to rock, the maximum steel
strain increased but was still only 16 percent of the strain in the fixed fiame.
The beneficial effect of foundation rocking became less pronounced near

the top of the columns but was still substantial. Near the top the maximum strains
in the as-built and retrofitted models were 31 and 55 percent of the strains of the
fixed models. The considerable saving that can be realized in the cost of retrofit
as a result of rocking effects justifies the inclusion of rocking in the analysis of
substandard bridge piers.
ACKNOWLEDGEMENTS
The study presented in this article was funded by the US National Science
Foundation (NSF) Grant CMS 980080 and the Nevada Department of
Transportation (NDOT). The authors are thankful to Dr. Jack Scalzi, the former
NSF program director, and Mr. Bill Crawford, Chief Bridge Engineer at NDOT
for the support.
REFERENCES
1- Saiidi, M., D. Sanders, and S . Acharya, “Seismic Retrofit of Spread

Footings Supporting Bridge Columns With Short Dowels,” Journal of
Construction and Building Materials, Vol. 15, No. 4, June 2001, pp. 177-
186.
2- Wehbe, N., Saiidi, M. and Gordaninejad, F. (1994). Behavior of

GraphiteEpoxy Concrete Composite Beams. ASCE Journal of Structural
Engineering 120:10,2958-2976.
3- Ballinger, C. A. (1991). Development of Composites for Civil

Engineering. Proc., the First Advanced Composite Mat. in Civ. Engrg.
Struct. Conf.’,ASCE, New York, N. Y., 288-300.
4- Mirmiran, A., Shahawy, M., Samaan, M. (1999). Strength and Ductility of

Hybrid FRP-Concrete Beam-Columns, ASCE J. Struct. Eng, 25:10, 1085-
1093.
5- Abdel-Fattah, B. and Wight, J. K. (1987). Study of Moving Beam Plastic

Hinging Zones for Earthquake-ResistantDesign of WC Buildings. ACZ
Structural Journal 84:S4,3 1-39.
6- Galunic, B., Bertero, V. V., Popov, E. P. (1977). An Approachfor

Improving Seismic Behavior of Reinforced Concrete Interior Joints,
Report No. UCBEERC-77/30, University of California, Berkeley.
@Seismicisolation
@Seismicisolation
562 Saiidi et al.
7- ACI-ASCE Committee 352, (2001). Recommendations For Design of
Beam-Column Connections in Monolithic Reinforced Concrete Structures,
ACI. Detroit, USA.
8- Gopalalaishan, B., M. Saiidi, and F. Gordaninejad, “Seismic Behavior Of

A Bridge Pier with Innovative Fiber Composite and Concrete Elements,”
Proceedings, International Conference on FRP Composites in Civil
Engineering, Hong Kong, December 2001.
9- El-Gamal, M., Siddharthan, R.,Maragakis, E. A. (1995).”Nonlinear

Bridge Abutment Stifhesses: Formulation Verification and Design
Curves,” Rept. No. CCEER 95-01, College of Engineering, University of
Nevada, Reno, pp12-14.
1O- Schmertmann, J. M., Hartman, J. P., Brown, P. R.(1978). “Improved

Strain Influence Factor Diagram,” ASCE GeotechnicalEngineering
Journal, vl04, pp 1134.
@Seismicisolation
@Seismicisolation
I I
Figure 1: Elevation view of the test frame (1 in.= 25.4 mm)

!
Ow Longitudinal LOFS 314?
2 Transverse Reolark
Loyers
Seclion 8-B
Column Bose
Section A-A
Beom
2 Longitudinal Replark Layers
2 Transverse Replark
Over Longitudinal Loyers
Section C-C Section 0-0

Cdumn PIasIIc Hinge Column Middle
Figure 2: Pier details and composite layout (1 in.= 25.4 mm)

@Seismicisolation
@Seismicisolation
564 Saiidi et al.
r-
7"
I .
Figure 3: As-Built footing elevation (1 in.= 25.4 mm)
mat
lap-splice
"R7- 60 d e r
to vert @ 6"
Front View (Paralld ta Frame) Inclined and vertical dowels in relrof-
Figure 4: Retrofitted footing elevation (1 in.= 25.4 mm)
@Seismicisolation
@Seismicisolation
,Labaratary Wail r r
,Mass Rig Fmme
S l
Shake Table
Figure 5: Test setup
Figure 6: Retrofitted footing on bearing pads (lughlighteá for clarity)
@Seismicisolation
@Seismicisolation
566 Saiidi et al.
Displacement (mrn)
-20 -14 -7 O 7 14 20
20 89
15
-%
3
10
59
30
z
Y
5
a Q)
e
O 0
e
-
LL 0
-2
-30 E
2 -5
3
2i -10 4
-59
-15
-20 -89
-0.8 -0.6 -0.4 -0.2 O 0.2 0.4 0.6 0.8
Displacement (in)
Figure 7: Measured force-displacement relationshipsfor B2SA
Displacement(mm)
-6.4 -4.2 -2.1 0.0 2.1 4.2 6.3
89
59
30
2
s
O @
O
LL
-30
-59
-89
-0.25 -0.15 -0.05 0.05 0.15 0.25
Displacement(in)
Figure 8: Measured force-displacementrelationships for the right footing in BZSA
@Seismicisolation
@Seismicisolation
Rotation (Deg.)
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
20 ; ' 89
15 -I As-Built
Right Footing Rotation -- 59
10 -: Clockwise Negative t-
h -- 30 h
n 5 -1 z
g Y
Y
al 0 -: --o al
2! e
O
LL -5 -: -- -30
2
-10 -:
-- -59
-15 -1
-89
-0.006 -0.004 -0.002 O 0.002 0.004 0.006
Rotation (Radians)
Figure 9: Measured force-rotation relationships for the nght footing in B2SA
Displacement (mm)
-20 -14 -7 O 7 14 20
I , ' 89
1
-20 -89
-0.8 -0.6 -0.4 -0.2 O 0.2 0.4 0.6 0.8
Displacernent (in)
Figure 10: Measured force-displacement relationships for B2SR
@Seismicisolation
@Seismicisolation
568 Saiidi et al.
Displacement (mm)
-3 -2 -1 O 1 2 3
89
59
30 h
z
?5
O t O!
IA
-30
-59
-89
- O 0.06 0.12
Figure 11: Measured force-displacementrelationships for the left footing in B2SR
Rotation (Deg.)
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
89
59
30 h
z
Y
v
0 0
eO
L
-30
-- -59
-15 -:
-201 " " I " " : " " " " : " " I " "
r -89
4.006 -0.004 -0.002 O 0.002 0.004 0.006
Rotation (Radians)
Figure 12: Measured force-rotation relationships for the left footing in BZSR
@Seismicisolation
@Seismicisolation
2000 /
1800 Retrofitted
+i+
1600
.- 1400
2
1200
.i 1000
E R. Column Bot, 13 in
(330 mm) SG 51
.G 800
2
G 600
400
200
O
O 0.1 0.2 0.3 0.4 0.5 0.6
Peak Table Acceleration ( x Sylmar)
Figure 13: Maximum tensile strains in the column bars near the base
2000
1800
1600
F
.- 1400
+2
m
1200
.g 1000
0
.E 800
2
fj 600
400
200
O
O 0.1 0.2 0.3 O .4 0.5 0.6
Peak Table Acceleration ( x Sylmar)
Figure 14: Maximum tensile strains in the column bars near the top
@Seismicisolation
@Seismicisolation
S P 209-31
Experimental Study on the Behavior

of Reinforced Concrete Buildings
with Precast Floor System
by O. López-Bátiz and H. Silva-Olivera
Svnomis: Recently in Mexico and other Latin-American Countries, reinforced

concrete (RC) buildings with precast floor systems have been widely used, al-
though such structural types are not yet included in the design codes. In order to
understand the behavior of this type of building, the response of a RC building
model to seismic loads was investigated by testing two full scale structural models.
The building models were subjected to lateral and torsional displacements. The
models simulated the first story of a four-story building, and consisted of three di-
mensional one-story moment resisting single-span frames. The model was de-
signed to assess the seismic performance of ductile frame structures following the
Mexico City Building Code.
The variable in the tests was the type of floor structural system: (a) the first
model (CR model) was cast in-situ beam-slabs structure; and, (b) the second mod-
el (PCR model) had the same structural system, but with precast elements compos-
ing the floor structure. Both models were subjected to the same loading history and
had similar instrumentation.
This paper presents the results and conclusions obtained from the compar-
ative study of the behavior of both models. One of the main conclusions is that the
presence of precast elements at the floor system was found to be an important pa-
rameter for the torsional stiffness of the whole structure.
Keywords: moment resisting frames; precast floor

system; torsional stiff ness
@Seismicisolation
@Seismicisolation
571
572 López-Bátiz and Silva-Olivera
Oscar Lopez-Batiz, Researcher at the National Center for Disaster Prevention in
Mexico and lecturer 'at the Graduate Division of the Faculty of Engineering of
UNAM. He holds B.Sc. and M.Sc. degrees from UNAM, and a Ph.D. degree
from the University of Tokyo. His research interests include the seismic behavior
of reinforced and precast concrete structures.
Hector Silva-Oiivera, was a Research Assistant during 1999-2000 at National

Center for Disaster Prevention in Mexico. He holds B.Sc. from Iberoamericana
University in 2001. His research interests include the seismic behavior of
INTRODUCTION
The use of prefabricated floor systems is more commonly used each day in Latin
American construction of reinforced concrete buildings. This system is not
clearly mentioned in the design and construction recommendations or guides for
reinforced concrete structures, which causes uncertainty in the structural design
and, in some cases, leads to selection of design parameters that are not fully
based on theoretical or experimental developments.
There are few experimental studies made on prefabricated system structures in

Latin America up to date, and only one study exists with relation to precast floor
systems (1,2). The lack of experimental information on the behavior of this type
of structures obstructs the existence of logical and rational recommendations for
analysis and design.
Design procedures of this type of structure are based on the traditional method of
reinforced concrete building design. The only difference is the consideration of a
hypothesis of less deformation capacity and rigidity of the whole structure due to
the prefabricated floor frame behavior characteristics.
In this paper, the experimental results on two full-scaled reinforced concrete

structures built and tested in the Large-scale Structures Laboratory of
CENAPRED (which will be abbreviated as LEG-CENAPRED afterward) are
shown and analyzed. The two models were subjected to lateral and torsional
reversal cyclic loading patterns. One of them was cast in situ (CR model) and the
other consists of cast in situ supporting frame and prefabricated floor system
(PCR model). The PCR model represents a structure system that is usually
considered inappropriate to be used in a high seismic zone.
TO cany out a structural analysis of buildings under lateral excitations, like
@Seismicisolation
@Seismicisolation
seismic forces, it is often necessary to consider the transmission of horizontal
inertial forces that occurs in the floor system of each level to the frames or
vertical elements of high rigidity and resistance. This transmission is generally
realized by the floor system, so that it is necessary to know the resistance and
rigidity of the elements that take part in this function. Sometimes, prefabricated
or precast elements may not have sufficient resistance against shear stresses
induced by the seismic inertial forces, principally along the joints of the different
components of the prefabricated floor system. On the other hand, the possible
less rigidity of the unit made by different components, the possibility of relative
displacement along the joints, and the limited continuity between the floor and
the supporting frame, may cause more deformations and displacements than
expected, and invalidate the analysis and design hypothesis used in
monolithically constructed buildings.
Generally, in the procedure of the structural design of structures with precast

floor systems, empirical restrictions on analysis and design are employed that
tend to be too severe for this type of structure as compared to the design
restrictions for the monolithic reinforced concrete structures. An example of
those restrictions is the force reduction factors for the seismic design spectrum:
for traditional reinforced concrete structures the reduction factors are of the order
of 2.0 or larger with a story drift ratio (which will be abbreviated as SDR
afterward) being permitted between 1.5 and 3 %, while for precast structures
usually the force reduction factors equal 1.5 (3). This is used in spite of the fact
that it is not specified in the regulations, and the limit of SDR is of the order of
1.5 %, being similar or smaller to that for waffle slab reinforced concrete frame
structures (4).
The aim of this study was to establish a direct comparison between the behavior
to lateral forces of a traditional monolithic reinforced concrete structure and a
reinforced concrete structure with a Prefabricated floor system. In the
experiment, the only variable considered was the type of floor system, while all
the other parameters and characteristics were kept constant. The comparison of
two models was made according to a pre-detmined lateral force pattern.
TEST MODELS
General dimensions
The models were constructed at full scale, and represented the first level of a
four-story building on the soft zone (type III) of Mexico City. They were
designed so that the lateral resistance of the structure will be supplied by ductile
frames. Both models had the same dimensions, one level, one span in each
direction, story height of 2 m, and the distance between the columns of 4.5 m in
both directions. The structural elements of the frame are: beams of 400x250 mm
@Seismicisolation
@Seismicisolation
and columns of 400x300 mm. The orientation of the columns was decided in the
way that the larger moment of inertia would be parallel to the direction of lateral
force. The slab had a total thickness of 120 mm in the CR model and 170 mm in
the PCR model, in which a concrete topping layer of 40 mm was included. The
selection of the slab thickness was made according to criteria of resistance, and
so that the weight per surface unit and the inertia bending moment of both slabs
might be equal. Fig.1 shows a general view of the test model and Fig. 2 shows a
detail of the characteristics of the slab of PCR model.
Desim of the models
The models were designed according to the Mexico City Building Code
(abbreviated as RCDF-NTC), so that its intended failure mechanism followed a
"strong column - week beam" type. The used strength ratio (McolMbeam) was
2.25. Also, shear and bond failures in beams and columns before reaching a
rotation ductility of 4 in beams and 2 in columns was avoided (5). The models
were designed to show ductile behavior, using the criteria established in the
RDF-NTC for structures that have proper characteristics for a seismic factor,
Q=2, which is the highest value empirically recommended for structures with
prefabricated floor system.
The slab of the CR model had a thickness of 120 mm, and was cast
monolithically with the beams. It had reinforcement for positive and negative
flexural moments with No.3 (10 mm diameter) reinforcing bars at 250 mm and
with No.3 reinforcing bars at 150 mm, respectively, that are adequate for slab
with four continuous edges. in the PCR model, the concrete topping was cast
monolithically to the beam and a welded wire mesh of 6x6/10-10 was placed.
The prefabricated small vault was made of sand and cement with a depth of 130
mm. Prefabricated prestressed joist beams, to support the prefabricated vaults,
also had a depth of 130 mm. Five centimeters of the beam edges rest inside the
concrete beam cores. The construction procedure is the common practice in
Mexico City.
Columns had longitudinal reinforcement made of 8 No.6 (19 mm diameter) bars,

uniformly placed around the perimeter, and transverse reinforcement of No.3
bars at 75 mm throughout the columns. Beams had a longitudinal reinforcement
made of 6 No.4 bars, three on each layer, with a transverse reinforcement of
No.3 bars at 175 mm. Concrete cover was 25 mm in both beams and columns.
To simulate the continuity to the next span, beams and the slabs were extended
700 mm from the axis of the columns, with the exception of that on the south
side of the model that were extended only 370 mm to adjust to the dimensions of
the loading device. Also the columns were extended 70 cm upward from the axis
of the beams to simulate the continuity to the next level.
@Seismicisolation
@Seismicisolation
The general characteristics of the materials are presented in the Table 1. The
concrete was mixed in the laboratory. The concrete for the foundation blocks and
columns had a compression strength between 36.70 and 40.80 MPa, and the
concrete for beams and concrete topping for the PCR model had a compression
strength of 28.60 MPa. The reinforcing bars used were 10 mm diameter bars of
cold rolled steel with no definitive yielding range and No.4 (13 mm diameter)
and No.6 mild steel bars with a definitive yielding range.
Prestressed beams had some shear keys in its upper part to promote good bond to
the concrete top layer. Prefabricated small vault units had the longer side of 680
mm; 6 of them were placed between 5 prestressed beams spaced at 750 mm.
Instrumentation
To grasp the main characteristics of the behavior of the two models, strains,
displacements and applied loads in various points of the models were measured
through a data acquisition device connected to a computer (6). The
instrumentation for both models was the same, so that the effect of the floor
structural systems in the general behavior of the specimens could be directly
compared. The instrumentation was classified as external and internal. The
internal instrumentation consisted of strain gauges, which were used to verify the
stress state of the reinforcing bars in the critical sections of the beams and
columns. The external instrumentationconsisted of the displacement transducers
(abbreviated as LVDT’s) and load cells.
The primary objective of the instrumentation was to determine the load -

displacement relations of the whole structure and of specific structural elements.
A general view of the external instrumentationis shown in the Fig. 3.
LoadinP amaratus
The models were subjected to lateral forces only to simulate the seismic effect.
Live loads on the slabs and the application of axial load on the columns were not
considered. Lateral displacements were applied to the models by two hydraulic
jacks of 500 kN maximum capacity. It was applied symmetricallyto the model in
points located along the axes of &ames A and B (Figs. 1 and 4), and separated
1130 mm from each other towards the center of the model.
To apply the lateral displacement, jacks were fixed to the reaction wall of LEG-
CENAPRED to push in the North direction (positive displacement) and South
direction (negative displacement) of the model. To prevent some direct tension
on the floor system during the application of negative displacement, affecting the
elements’ local behavior, high strength steel bars and steel plates were placed to
have the “pushing” effect in the negative load direction as well (see Fig. 4).
@Seismicisolation
@Seismicisolation
Loading history
The loading pattern was controlled by displacements. The displacement pattern,

shown in the Fig. 5, was determined based on the displacement (SDR) that the
model would have at the first cracking due to flexure in beams and columns,
which was estimated to be between 0.1% and O. 125 % of SDR.This was defined
as the ratio of the lateral floor displacement and the story height.
The history of application of displacement was divided in three parts. The first
part consisted of 2lcomplete cycles of displacements applied by jacks, with the
object of observing the structural behavior basically to translation in the presence
of lateral displacement before SDR reached a value of 1/50 of the floor height
(for this model, this correspond to a total displacement of 40.0 mm and 2.0 % of
SDR). In this step, the torsional effect was not considered explicitly. The second
part consisted of four cycles of torsional displacements generated by separating
the jack fiom the B axis and using only one jack of A axis to apply
displacements. In this part, the structural behavior in the presence of torsion
generated by the displacement of the mass center with respect to the rigidity
center was observed. Also, the effect of torsion on the behavior of the floor
system was studied, comparing directly the stiffness of a stnicture with a
traditional floor system and that of a structure with a prefabricated floor system.
Finally, the third part consisted of three cycles of displacements applied by the
jacks, aimed at observing the structural behavior in the presence of larger lateral
displacement to the maximum value permitted by the design regulations; that
was the SDR of 1/25 of the floor height (4.0 % of SDR).
The loading pattern does not show the cycle pattern with increasing amplitude.
After the 1 1 and ~ 19" cycle, in which a SDR of 1 and 2 % were applied
respectively, displacement reversal cycles with smaller amplitudes were applied
in order to observe the degradation of rigidity in the small range of
displacements, which is the condition for the study of the dynamic characteristic
in the structural behavior by using ambient vibration tests.
GENERAL BEHAVIOR OF THE MODELS
Hereinafter the behavior of both models observed during the test period is
described. The first flexural crack was observed at the base of the columns and at
the ends of beams in both models. It occurred at a lateral displacement of 1.O mm
(equivalent to a SDR of 1/2000 or 0.05 % of the floor height).
First yielding occurred in the longitudinal reinforcement, at first in the beams and
a cycle later in the columns. In the CR model, beam yielding of the longitudinal
reinforcement was observed at a lateral displacement of 10 mm in the peak of the
8thpositive loading cycle; and in columns at the next cycle with a displacement
@Seismicisolation
@Seismicisolation
of 10.6 mm. in the PCR model, yielding of the longitudinal reinforcement was
observed in girders and columns in the gth loading cycle and in the loth loading
cycle, respectively, when in both cases the lateral displacement reached 12 mm.
in the CR model, flexural - shear cracks were observed in columns at a lateral

displacement of 10.0 mm in the 8" loading cycle, and the first shear cracking
was observed in columns at a lateral displacement of 20.0 mm, which
corresponded with the peak of the 9" loading cycle. In the PCR model, flexural -
shear cracks were observed in columns and girders at a lateral displacement of
5.0 mm during the 6" loading cycle, and the shear crack was observed only in
columns at a lateral displacement of 10.0 mm during the lothloading cycle.
Another important variable from the point of view of defining failure for
reinforced concrete elements, is the crushing of the concrete. This phenomenon
is due to successive bending moments at the ends of the beams and in the bases
of the columns, which coincided with the designed configuration of the yielding
mechanism. in the CR model, concrete crushing in beams was observed at a
lateral displacement of 40 mm, and was observed in columns at a lateral
displacement of 55 mm. in the PCR model, this phenomenon occurred in beams
at a lateral displacement of 30 mm, and in columns at the same displacement as
in CR model. in both cases, crushing of concrete occurred during the first
loading cycle (26" loading cycle) after the four cycles of torsion (see Fig. 5),
when the first positive cycle of the maximum displacement of 60 mm was
starting.
in both models, the crack pattern varied according to the increment of the
displacement level; nevertheless after the displacement level equivalent to 2.0 %
of SDR, no additional flexural nor shear cracks were observed, and the length of
the already observed cracks showed small or no variation in girders and columns.
The strength and stiffness degradation after a displacement level equivalent to
2.0 % of SDR can be explained by the concentration of deformation in the
existing cracks, the increment of the gap of cracks, and the loss of bond between
the concrete and the reinforcement. The Fig. 6 shows the crack pattern observed
in columns and girders at SDR level of 2.0 %, representing both models that
showed almost same pattern at the final condition.
With respect to the cracks or damage observed in the slabs of the models, in both
cases the cracks occurred in upper and lower side, perpendicular to the direction
of the applied load. The orientation of the cracks showed that they were mainly
due to the contribution of the slabs to resist the bending moment of the frames in
the direction of the applied load. A flexural crack in the CR model affected the
slab on its upper side at a lateral displacement of 10.0 mm during the 7" loadihg
cycle. Ji the PCR model, the flexural crack in the compression layer (upper side
of the slab) was observed at a lateral displacement of 5.0 mm during the 5"
loading cycle.
@Seismicisolation
@Seismicisolation
Even when the models were subjected to the four cycles of torsion, diagonal
tension cracks in the slabs were not observed. At displacements larger than 40.0
mm, diagonal type cracks at the comers of monolithic slabs or precast slab units
were observed in both models. In some slab zones, the damage was severe (gap
wider than 2 mm), but there was no sign of separation of the small vault units of
the precast system during the test, including the stage of lateral displacements to
80.0 mm (equivalent to SDR of 4%). Instability, separation of the small vault
units, separation of the prestress beams from the supporting beams were not
observed in the PCR model, as can be seen from Fig. 7. In general, the two
models showed similar material spalling at high levels of SDR.
In general, the behavior of both models allowed the "weak beam - strong
columns" concept, without any significant influence of the type of the floor
system on the formation of the failure mechanism. The crack pattern of the slabs
showed that the width of the contributing slab is practically equal to half span in
the perpendicular direction to the applied load direction (see Fig. 8).
ANALYSIS OF RESULTS
The relation between the displacements of the models, the rotation demands of
the structure under torsion, and the deformations demands for the floor system
are briefly presented. The deformation characteristics of both models are directly
compared.
-
Lateral load displacement curves
To determine the lateral load - displacement curve of the models, the lateral load
was defined as the algebraic sum of the applied loads in both hydraulic jacks, and
the lateral displacement was calculated as the average of the displacement values
measured at the LVDT's located at the South side of the models on the A and B
axes. The lateral load - displacement curves measured are shown in the Fig. 9;
these relations are those computed considering only the 24 cycles for
translational loading scheme (first stage). In the same figure, the points for the
first flexural cracking (whatever structural element), the first flexural yielding
and the point related with the crushing of concrete due to the flexural demand in
beams are indicated. From the figures, is apparent that both the global
configuration and the points for cracking, yielding and concrete crushing are
almost the same for both models. Lateral strength reached by the models is
presented in Table 2. The most significant difference between the behavior of *e
models can be seen on the strength degradation during cyclic reversals. For PCR
model, strength degradation was determined to be about 3% (for a SDR less than
2%) and 5% (for values of SDR larger than 2%); this was slightly larger than the
degradation observed for the CR model. The strength degradation was defined as
@Seismicisolation
@Seismicisolation
the relation between the lateral load measured at the same displacement value on
consecutive loading cycles. However, this strength degradation level, even for
high lateral drift ratios, does not represent any indicator that the structures could
be under large lateral displacement demands (inadequate values from the Code
point of view) during non-linear inelastic seismic analysis (7).
Torsional load - displacement curves
Torsional load - displacement curves for the models were determined as follows:
the load was defined as the measured value on hydraulic jack ?A? (this because
the jack at axis ?B? was disconnected from the model), and the rotational
displacement was considered as the algebraic substraction of the displacement
values measured at the LVDT?s located on the South side of the model in the A
and B axis. Fig. 1 O shows the torsional load - displacement relations for the four
cycles of torsional demand on the models. The general tendency of the curves for
both models is similar.
The peak to peak stiffness of the models during the torsional demand were
computed; the results are shown in Fig.10. There is a slight difference between
the stiffness of the models, the peak to peak torsional stiffkess for the model CR
8% larger for the first two torsional cycles, and 5% larger for the following
torsional loading cycles.
Elongation of the floor svstem
Test, analysis and earthquake damage have indicated that the shear forces in
columns, due to yielding in beams in reinforced concrete frames, can be higher
than calculated based on an analysis that neglects the inelastic axial elongation of
the beams, which is caused by the material properties of reinforced concrete,
including crack opening. When structural systems are subjected to lateral cyclic
reversal loads, cracks appear on all the sides of the elements as a result of the
inversion of moments to which they are subjected. When the floor system,
formed by beams and slabs, is subjected to a cyclic reversal loading pattern,
cracks will appear due to several effects. Mainly due to the positive and negative
flexural moments, cracking can increase opening as the lateral displacement also
increases. With the consecutive process of crack opening and closing, these tend
to present a permanent residual width, which is in others words an elongation or
extension of the structural element on which the crack is located. In the case of
specimens tested in this work, the extension of the slab was measured by means
of the difference between the values of displacement measured on the LVDT
located on the North and South sides of the A and B axis. Fig.11 indicates the
relation obtained between the lateral displacement applied to the model (value
that is considered equal to the average of the values reported on the sensor
located on the south side of the model) and the elongation of the floor system
measured considering only the LVDT?s located on the A axis. From the graph, it
@Seismicisolation
@Seismicisolation
can be seen that the models, after being subjected to a maximum lateral
displacement of 80.0 mm, had a residual elongation of the floor system of
approximately 9.75 and 7.61 mm for models CR and PCR, respectively. This
final residual elongation of the floor system, is about 12.2% and 9.5% of the
maximum lateral displacement applied on the model, for the CR and PCR
models, respectively. The elongation pattern of the floor system under
successively increasing lateral displacements turned out to be similar for both
models, with an average rate of 1 to 6. This means that the inelastic elongation
on the floor system was about 16.6% of the value of the lateral displacement
applied. It can also be observed that the final residual elongation turned out to be
the order of 2.0mm less for the PCR model, an order of 20.5% less reminant
elongation recorded on the PCR model.
Measurement of the vertical disdacement at the middle bav of the slab
With the purpose of analyzing the deformed configuration of the floor system on
both models under bending, the vertical displacement was measured at mid-span
in the direction of the applied load. Measurement points were located on the
longitudinal axis of the jacks. The relationship between lateral force and vertical
displacement of the central points are indicated on Fig.12. From the analysis of
results in the figure, it is pointed out that on both models, the vertical
displacement in both the positive and negative direction of the floor system at the
mid-span increased. The displacement remained symmetrical in its magnitude, as
a result of subjecting the specimens to a SDR of 0.5%. When the specimens were
subjected to SDR above 1.0%, the magnitude of the positive displacement on
both models increased notably, tending the PCR model to present an
asymmetrical behavior of vertical displacements near lateral pattern
displacementspeaks.
When the structural system of idealized symmetrical frames without the

existence of slabs are subjected to cyclic reversal loading, the position of the
inflection points tend to coincide with the mid-span, therefore one can expect
null vertical displacements on these points. In the case of the models with the
presence of slabs, the stiffness symmetry of the fiame is lost, specifically the
stiffhess of the beam elements. Then when a vertical positive displacement
appears in the center of the bay on the measure points (the slab tends to go up, in
reference to it’s initial position of equilibrium) on the PCR model, it indicates
that the inflection point of the floor system of this model tends to move away
more from the center of the bay than for model CR. This can be considered as if
the contribution of the precast rigidity system in negative moment is important.
Because of this effect on the opposite side subjected to a positive moment, a
larger rotation demand can be expected on the PCR model than on the CR. Also
the positive reinforcement from the compression zone on model PCR (welded
wire mesh) will be subjected to a larger strain demand in tension due to flexure
when the structure is subjected to large relative story displacements. This last
@Seismicisolation
@Seismicisolation
aspect is corroborated with the failure of the welded wire mesh reinforcement of
the compression zone on model PCR, when is subjected to levels of SDR greater
than 3.0%.
CONCLUSIONS
For precast concrete construction in seismic areas, all joints between precast
concrete units should develop full continuity and toughness. Floor slab units
utilizing composite in-place concrete topping and jointing can develop horizontal
diaphragm action and fail-safe connections to resist lateral forces such as seismic
loads.
The most relevant conclusions fiom the experimental project comparing

structures built with traditional monolithic floor systems and a precast floor
system can be summarize as follows:
1.- The crack patterns of damage and the configuration of the yielding
mechanism did not vary significantly between the two models considered in the
study, when these are subjected mainly to translation load. This tendency was
appreciated up to levels of 3.0%of SDR. For larger values of SDR , the level of
cracking and visible damage on the precast floor system was comparably larger.
2.- The lateral load - lateral displacement behavior of the models when only the
translational displacement is considered were practically equal on both models
up to a SDR of 2.0%. For greater SDR, a small increment in the strength
degradation was observed on model PCR than that observed on model CR, the
difference between the values on the strength degradation for the models was
less than 5%. The strength to lateral load, as well as the translational stiffness,
determined by the moment resisting frame did not present significant variations
as a function of the floor system.
3.- On the basis of the two conclusions mentioned before, the design procedure
can be same as such used for typical reinforced concrete structures. However,
fiom the point of view of seismic analysis, in order to consider strength
degradation for structures with precast floor system, it must be computed a
seismic reduction factor according to the increase of lateral displacement that
should occur due to the strength degradation characteristic.
4.- When the models were subjected to representative torsion loads, the model
CR presented larger stiffness than model PCR, with a difference of 8%. Sti&ess
in the overall framing system should ensure against structural and nonstructural
building damage by minimizing drift and rotation. In seismic areas, the existence
of soft floors in the framing system can lead to severe damage; however, it is still
difficult to define the specific limit between rigid diaphragm and soft floors. The
@Seismicisolation
@Seismicisolation
acceptance of this type of precast floor system will require more research data.
For example a further study considerating the use of different orientation of the
precast prestressed joist into one floor system is necesary.
5.- Instability was not observed, nor were indications of loosed small vaults on
the PCR model. No observed slippage of the prestress joist over the main beam
occurred. In general, the phenomena of crushed and spalled material fiom the
floor systems, observed in both models at high levels of SDR, turned out to be
similar.
ACKNOWLEDGMENT
The writers express their gratitude to Mr. Edgar Ortiz, Hugo Cazamayor, Carlos
and Pablo Olmos in preparing the test specimens and during the experimental
work. The authors gratefully acknowledge the critical revision of the manuscript
by Dr. Sergio Alcocer.
REFERENCES
[ 11 Esteva L., “Investigación experimental sobre el comportamiento de

diafragmas constituidos por losas precoladas de concreto celular >’ (in
Spanish), Revista del IMCYC, Vol. 4, No. 21. July-August, 1966.
[ 2 ] Gomez-Gonzalez B., “Estudio experimental para la comparacion de la
rigidez y resistencia en su plano de sistemas de piso compuesto” (in
Spanish). Thesis submitted for Master degree at the School of Engineering,
UNAM, Mexico, 2000.
[3] Mexico City Government, “Building Code for Mexico Cio” (in Spanish),
Diario Oficial del Gobierno del Distrito Federal, 1993.
[4] Ordaz M., Miranda E. and Aviles J., “Propuesta de espectros de diseño por
sismo para el DF” (in Spanish), VI Simposio Nacional de Ingeniería
Sísmica, Santiago de Querétaro, Mexico, September, 2000.
[5] AIJ, “Design Guidelines for Earthquake Resistant Reinforced Concrete
Buildings Based on Ultimate Strength Concept”, Architecturai Institute of
Japan, Noviembre, 1988.
[6] Flores L. and Alcocer S., “Análisis y adquisición de datos (ADD), Ver.l.0,
Manual del Usuario” (in Spanish), Research Division, Structural
Engineering Area. CENAF’RED Informe IEG/01/96, Mexico. October, 1996.
[7] Lopez-Batiz O., Otani S . and Aoyama H., “Influence of Member Hysteretic
Shape on Non-linear Earthquake Response”, Proceedings of the Japan
Concrete Institute, V01.12, No.2. June, 1992.
@Seismicisolation
@Seismicisolation
TABLE 1 MECHANICAL CHARACTERISTICS OF MATERIALS
Concrete strength
Material and steel yielding Structural element
stress (MPa)
Concrete fc=40.80 Blocks and columns
Concrete fc=36.70 Beam of CR model
Concrete f c=28.60 Beam and concrete topping of PCR model
Steel bar f~428.20 No.3 steel bars for transverse reinforcement
Steel bar No.4 steel bars for longitudinal reinforcement in
W458.90 beams
Steel bar No.6 steel bars for longitudinal reinforcement in

S=458.90 columns
TABLE 2 MEASURED STRENGTH OF THE MODELS
Strength when the concrete

Maximum strength (kN)
crushing occurs in beams (kN
Model
Positive semi- 1 Negative sen!¡- Positive semi- I Negative
- semi-
cycle cycle cycle cycle
CR 662.0 623.7 741.4 690.4
PCR 671.8 624.7 719.8 679.6
@Seismicisolation
@Seismicisolation
Column section
4.50- --- 8 bars # 6
bars # 3 @ 7.5 cm
I I
1 0.30 r
Beam section
6 bars # 4
East view of the test models
I I
1 0.25 r
I I TT- - - -- ----- -3 r
- 4 7 0
I I
0.40’ 1 I II
_I_ I 12.00
I 0.30
I I
I I
Unit: meter
except those shown
in other unit
Fig.1 General view of the test model and the loading apparatus
@Seismicisolation
@Seismicisolation
Welded wire mesh
\ /Prefabricated vault F:z$!z2d

beam
(Pp) joist
View of the precasted floor system
Concrete topping
/
' 1 -
m e a m
Prefabricated vault
Transverse view of the precasted floor system
Concretetopping
I /
Pp joist beam
-I kcm
Longitudinal view of the precasted floor system
Fig.2 Details of the precast floor system
LVDT's and load cell location in the frame and floor system
Fig.3 General view of the instrumentation of the models, including the

loading apparatus
@Seismicisolation
@Seismicisolation
a) General view of the model
b) Detail of loading apparatus
Fig.4 General view of the model, instrumentation and loading apparatus
@Seismicisolation
@Seismicisolation
80, I 4
60 3
40 2
20 1
O O
-20 -1
-40 -2
-60 -3
-An
1"
I 1 -4
4 C W B Of 3 C y l S of
~ K I ofbaslation
S I tooion I baslation
I l I
Loading cycies
Fig.5 Applied loading pattern
@Seismicisolation
@Seismicisolation
CR b) Cracking in column model PCR
e) Cracking pattern at final stage (similar for both models)
Fig.6 Crack patterns in columns and girders of the CR and PCR models
at 2.0 YOof SDR and at the end of the test
@Seismicisolation
@Seismicisolation
b) PCR Model
Fig.7 View of the lower comer of slab unit of two models in the final
stage of the test
Fig.8 View of the upper face of slab of the PCR model in the final stage
of the test
@Seismicisolation
@Seismicisolation
I Lateral displacement = 61+82
Bo
t
e
-100 -75 -50 -25 D 25 50 75 100 -100 -75 -50 -25 O 25 50 75 100
Lateral displacement (mm) Lateral displacement (mm)
Fig.9 Lateral load - displacement curves for the CR and PCR models
i...., center i
Peak to peak stiffness
'......_. Kptp = (FP-FN)/(dTP-dTN)
Positive semicycle stiffness
Kp = FP/dTP
40.0
Lateral force = PI 37.5

I I Torsional displacement = 81-62
800 .
E
E 35.0
H
400 32.5
al
e
9
- O -cu> 30.0
E
L
'E
-I
-400 27.5
25.0
-800
-15 -10 -5 O 5 10 15 0.0 0.5 1 .o 1.5
Torsional displacement (mm) Story drift ratio (%)
Fig.1O Torsional load - displacement curves for the CR and PCR models
and torsional stiffness
@Seismicisolation
@Seismicisolation
82
Floor system elongation = 83-81
100
-
E
E
75
a
C
50
‘o
25 -mn
O .-
YI
U
x
-25 m
a
a
-50
-75
-100 0.00 0.03 0.06 0.09 0.12 0.15
0.0 3.0 6.0 9.0 12.0 15.0
Remanent elongation offloor system
Floor system elongation (mm) Peak displacement
Fig.11 Lateral force - floor system elongation curves of models CR and

PCR
I I Vertical displacement = 8v1 O 8v2
80
- 80
--
E
E
60
40
E
E
60
40
E 20 E 20
-om o -o 0
::
.- -20 :
.-: -20
U
-E -40
U
- -40
L
o! -60 2 -60
m
-I
-80 -80
-1 O 1 2 3 4 5 6 7 -1 O 1 2 3 4 5 6 7
Vertical displacement (mm) Vertical displacement(mm)
Fig.12 Lateral displacement- vertical displacement curves at the center

span of models CR and PCR
@Seismicisolation
@Seismicisolation
SP 209-32
Influence of the Variation of Cement

Source on Hot Weather Concreting
by G. A. Arenes-Oliva, M. Bergin, and A. M. Zayed
SvnoDsis: The effect of varying cement source on fresh and hardened concrete
properties is studied under hot weather conditions. For the seven ASTM Type II
cements studied here, the same mix proportioning was adopted at a mixing tem-
perature of 95 O F (35 OC) with a constant dosage of water reducing and air entrain-
ing admixtures. Properties of fresh concrete including slump loss over an extended
mixing period (EMP) of 90 minutes, air content, and setting times are reported.
Also, hardened properties including compressive strength development and rapid
chloride permeability test data are reported. Results indicate that the rate of slump
loss and setting times are affected by the cement compound composition, calcium
sulfate content and calcium sulfate type. The compressive strength, under hot mix-
ing conditions, is found to be dependent on composition, fineness and morphology
of cement compounds.
Keywords: admixtures; anhydrite; cement compound

composition; compressive strength; gypsum;
hemihydrate; hot mixing; setting time; slump loss
@Seismicisolation
@Seismicisolation
593
594 Arenes-Oliva et al.
AC1 Member German A. Arenes-Oliva is a graduate student at the University
of South Florida, Tampa. His research interests include hot concreting,
rheology, clinker microscopy, and pavements.
AC1 member Michael Bergin, PE is the State Structural Materials Engineer

for the Florida Department of Transportation in Gainesville, Florida. His area
of interest is durability of structural concrete and materials.
AC1 member A. M. Zayed, is Associate Professor of Materials Science and

Engineering in the Department of Civil and Environmental Engineering at the
University of South Florida.
INTRODUCTION
It is known that environmental factors, such as hot weather, influence

the properties of fresh and hardened concrete ['l. Hot weather can be defined as
any combination of environmental conditions such as high ambient
temperature, high concrete temperature, low relative humidity, etc, that affect
the rate of hydration of cement and the rate of moisture loss, often resulting in
inferior performance.
The change in the cement source of a given concrete mixture may

affect its plastic and hardened properties due to variation of the cement
compound composition. In addition, properties of cement are closely related to
the plastic and hardened characteristics of concrete under environment
conditions such as hot weather. Variation of the physical characteristics of
cement such as fineness, particle size distribution, etc. may also relate to
variation in slump loss and setting times of fresh concrete [2,31.
In addition, increase of the rate of strength development at early ages

has been attributed to chemical composition and particle size distribution of the
cement, and to the temperature of concrete [41. It has been found that high
mixing temperatures for extended mixing periods accelerate the rate of slump
loss. For extended mixing at temperatures above 90 "F (32 "C), the highest rate
of slump loss has shown to be between O to 30 minutes (30 % - 40% loss),
after which slump loss continues but at a slower rate (10 % - 20 % loss) [5*61. In
concrete with no admixtures, the rate of slump loss is affected by the wtc,
temperature, and cement properties [71.
Extended mixing period and concrete temperature are known to affect a

variety of concrete characteristics. The effectiveness of water reducers is
@Seismicisolation
@Seismicisolation
diminished with hot, and long mixing periods. Furthermore, decrease of up to
10%of compressive strength has been observed with the use of water reducers
after extended mixing periods at elevated temperatures. However, some
researchers found an increase of 10%in compressive strength after 90 minutes
of mixing ['l. A combination of variation in cement compound content and
surface area of particles also affect the amount of water consumed and the heat
liberated by the hydration process. Moreover, the chemistry of the water-
reducing chemicals used in the mixture may retard or accelerate the hydration
reactions of certain cement compounds; thus affecting the rheological
properties of the concrete.
Since cement hydration involves chemical reactions, it is expected to be

temperature sensitive. During hydration of the calcium silicates, this
temperature dependence is effective as long as the reaction is chemically
controlled (which detennines the initial set, the final set and rate of early
strength gain). Once the reaction becomes diffusion-controlled (which
determines the rate of later strength gain), it is significantly less sensitive to
temperature [41. Among the cement compounds, the C2S reactivity appears not
to be considerably affected by high mixing temperatures [*l. Variation of
cement source may be accompanied by variation of the fresh and hardened
properties of concrete. Specifically, studies made on the influence of clinker
and cement compound composition on the strength gain of concrete have
shown that variation not only of the content of the silicate compounds, but also
their morpholo y directly affects the rate of compressive strength development
of concrete i % 1 8.
MATERIALS AND PROCEDURES
Seven ASTM Type II cements labeled A, B, M, LA, LJ, X, Y were

used, along with No. 57 limestone coarse aggregate (CA) and silica sand (FA),
WRDA64 water reducing admixture, and DAREX as the air-entraining agent.
Mixture proportions are shown in Table 1. The admixtures dosages, w/c and
aggregate content were maintained constant for all concrete mixtures. A target
slump of 3 f 0.75 in. (75 & 20 111111) was adopted and the mixing temperature
was maintained at 95 f 1 "F (35 f 0.5 OC) throughout the extended mixing
period. Prior to mixing, both the coarse and fine aggregates were sieved and
stored separately by particle size. Then aggregates were recombined in the
same predetermined grading for concrete mixing.
After introducing all materials into the mixer, an initial mixing period
of three minutes was given. The mixer was stopped for two minutes followed
by additional two minutes of mixing. The first slump measurement was
recorded, and the concrete was then returned to the mixer. Next, concrete was
intermittently mixed for thirty seconds every five minutes, and three additional
@Seismicisolation
@Seismicisolation
readings of the slump were taken at 30, 60, and 90 minutes of mixing. Upon
completion of the extended mixing period, unit weight, air content and setting
times tests were performed.
Concrete cylinders 4 by 8 in. (100 by 200 mm) were cast, demolded at

24 h then and stored in saturated lime water for compressive strength testing.
RCP measurements were conducted on concrete cylinders stored in a 5%
sodium sulfate solution. A sodium sulfate solution was used in order to
examine the effect of the cement sulfate resistance on chloride ion penetration.
The oxide composition of the as-received cement and their phase analysis were
conducted; in addition, cement fineness was determined using Blaine fineness
test.
As-Received Cements
The cement compound composition was determined by Bogue formula

(using the cement chemical oxide composition) and by Quantitative X-Ray
Diffraction. These results along with those of Blaine fineness test and calcium
sulfate content are presented in Table 2. Cement compound composition, as
determined from XRD and Bogue calculations, showed variations as expected;
whereas the variations in the tricalcium silicate content were less. The
discrepancy between the two methods, was more in the case of tricalcium
aluminate and dicalcium silicate.
The total calcium sulfate content varied between 3.6%and 5.1% for the
cements studied here. In addition, the form of calcium sulfate present in the
cements was different. Predominately, calcium sulfate was present in the
hemihydrate form; however, X cement had the majority of its sulfate in the
anhydrite form while B cement had more gypsum content than hemihydrate.
Properties of Fresh Concrete
With constant admixture dosage and hot mixing conditions, the

effectiveness of the water reducer on slump loss, within the first 30 minutes,
seems to be dependent on the tricalcium silicate content of cements. This can
be seen from Figure 1 where cements with higher tricalcium silicate content
experienced lower initial slump loss. Beyond 30 minutes, it is observed that Y
cement, which has a calcium sulfate content of 3.6%,experienced the highest
slump loss among all cements studied here. Though cement Y has similar
fineness, tricalcium silicate and aluminate content to cement U,the difference
in their rate of slump loss is attributed to the difference in their calcium sulfate
content and possibly its form.
@Seismicisolation
@Seismicisolation
Air content and unit weight measurements for all concrete mixes are
presented in Figures 2 and 3, where it is observed that variations in these two
properties were not significant under the given mixing conditions. All mixtures
have a unit weight ranging between 2260 and 2290 kilograms per cubic meter,
while the air content variation was less than 1 percent.
Setting Time
Concrete setting behavior assessed by the ASTM C 403 test method is

presented in Figure 4.The results indicate a variation in the initial set time by
approximately 100 minutes between the concrete mixtures. For cements with
tricalcium silicate content of 60% to 66% and moderate tricalcium aluminate
content (5-7%), the initial set time seems to be primarily dependent on the total
calcium sulfate content. Additionally, the final set time might be affected by
the nature of calcium sulfate present. This conclusion can be drawn by
considering the behavior of cements LA, LJ, M, and Y. The first three cements
share the same calcium sulfate content and have similar initial setting time.
The final set of cement LJ shows a higher value attributed to its higher gypsum
content. Cement Y has the lowest initial and final setting times, and the lowest
calcium sulfate content. For cements A, B and X, with lower tricalcium silicate
content of 50 to 54% and 6-8 % tricalcium aluminate content, the setting
behavior seems to be affected primarily by Cas04 G A . Again, the presence
of gypsum as the predominate form of calcium sulfate in addition to the higher
sulfate to aluminate ratio, is attributable for this increase in the setting time in
the case of cement B.
Properties of Hardened Concrete
Compressive strength gain for the seven concrete mixes is shown in

Figure 5. The data indicate that for early hydration time (7days), strength gain
was primarily a function of fineness and tricalcium silicate content. This is
similar to the strength gain characteristics at ambient temperatures 111-151. At an
age of one day, LA and M mixes have the highest compressive strength due to
their high tricalcium silicate content and fineness. Cement A with the coarsest
particles showed lower early strength.
For longer hydration times, strength gain was found to be dependent

not only on the dicalcium silicate content [11-161 but also on morphology of the
clinker phases [17*181. Study of the clinker morphology using optical microscopy
revealed that the size and reactivity of the belite particles has an effect on the
strength development of concrete. For instance, cement X has a dicalcium
silicate content of 19%, compared to 26% for cement B. The compressive
strength gain for these cements between 14 and 28 days of hydration was 4.2
and 3.1 Mpa, respectively. If the strength gain at later age was solely attributed
@Seismicisolation
@Seismicisolation
to the dicalcium silicate content, then the strength gain for cement B should
have been higher. In addition both cements have similar fineness. On
microscopic examination, belite particles of clinker B were found to be coarser
than those observed for clinker X as can be seen from Figure 6 . Similar
findings were observed in the case of cements LJ and Y. Both cements IJ and
Y have similar fineness (3920 and 3850 cm2/g, respectively) and similar
dicalcium silicate content. The strength gain for LJ was 3.7 MPa between 14
and 28 days as opposed to 1.8 MPa for cement Y. Morphological examination
of cement IJ revealed that the belite particles had an average size of 10 - 25
microns while those in the cement Y had an average size between 20 to 40
microns.
In conclusion, though cement compound composition and fineness are

important when concrete strength is considered, phase morphology is also of
significance when considering the long-term concrete strength behavior.
Table 3 shows the rapid chloride penetration test data at 90 days for
different concrete mixtures. Cements X, A, and B had, in general, slightly
lower chloride penetration values than cements M, U,LA and Y. The major
distinction between these two groups is the tricalcium silicate content; this is
also depicted in Figure 7. In addition, since cylinders were cured in sodium
sulfate environment, the sulfate resistance of cements might have affected the
RCP measurements Il9’.
CONCLUSIONS
Variation of the cement source under hot-mixing conditions affected

the properties of both fresh and hardened concrete. Rate of slump loss was
found to be affected by the C3S content, CaSOdC3A, calcium sulfate content
and type. Variation in the setting time was found to be primarily influenced by
the calcium sulfate content, its nature, and CaSOdC3A. The compressive
strength was found to be dependent on cement compound composition and
morphology, in addition to cement fineness. As for rapid chloride permeability
tests, increasing the tricalcium silicate content resulted in higher permeability
values at 90 days.
REFERENCES
(1) Allred, H., Cook, H., Hot Weather Concreting, American Concrete
Institute, AC1 Journal, August 1977.
@Seismicisolation
@Seismicisolation
(2) Mahboub, K., C., Cutshaw, Q.,A., Effects of Fresh Concrete Temperature
and Mixing Time on Compressive Strength of Concrete. AC1 Materials
Journal. Title no. 98-M8. January-February2001. pp 59-62.
(3) Mindness, S. and Young, P., Concrete. Prentice Hall, New Jersey, 1981.
(4) Carino, N., Jennings, H. J., Snall, L. M., Properties of Concrete at Early
Ages. ASTM Cement, Concrete and Aggregates Journal. 1989, pp. 129-
131.
( 5 ) Klieger, P., Effect of Mixing and Concrete Temperature on Concrete

Strength. Journal of the American Concrete Institute, June 1958. pp. 1063-
1081.
(6) Whiting, D., Effect of Mixing Temperature on Slump Loss and Setting Time
of Concrete Containing High-Range Water Reducers. ASTM Cement,
Concrete, and Aggregates Journal, Vol. 2, Number 1, Summer 1980. pp.
31-38.
(7) Shilstone, J. M., Concrete Strength Loss and Slump Loss in Summer.
Concrete Construction. May 1982. pp 429-432.
(8) Bensted, J., Hydration of Portland Cement. Advances in Cement

Technology, Cement Research Institute of India.
(9) Odler, I., Strength of Cement final report). Materials and Structures,Rilem
Technical Committees. Germany, Vol. 24, pp. 143-157. 1991.
(10) Taylor, J. C., Hinczak, I., Matulis, C. E., Rietveld Full-Profile

Quantijìcation of Portland Cement Clinker: The Importance of Including a
full crystallography of the mayor phase polymorphs. Powder Diffraction.
Vol. 15, NO. 1, March 2000. pp. 7-18.
(11) Young, F., Hydraulic Cements for Concrete. American Concrete

Institute Educational Bulletin No. E3-83. 1993.
(12) Taylor, H., Cement Chemistry. Academic Press Inc. USA. 1990.
(13) Satava, V., Trettin, R., Wieker, W., Studies on Physico-Chemical

Changes in Time of Initial Setting of C3S. Hydration and Setting of
Cement. Rilem 1992.
(14) Jemmings, H., The Developing Microstructure in Portland Cement.

Advances in Cement Technology. Cement Research Institute of India.
1983.
@Seismicisolation
@Seismicisolation
(15) Bensted, J., Hydration of Portland Cement. Advances in Cement
Technology. Cement Research Institute of India. 1983.
(16) Taylor, H. F. W., The Chemistry of Cement. Department of Chemistry,

University of Aberdeen, Scotland. 1997.
(17) Hofmänner, F., Microstructure of Portland Cement Clinker.

Holderbank Management & Consultant Ltd. April 1973.
(18) Ahmed, W. U., Petrographic Methods for Analysis of Cement Clinker

and Concrete Microstructure. Petrography of Cementitious Materials,
ASTM STP 1215, American Society for Testing and Materiais,
Philadelphia, 1994.
(19) Whiting, D. and Kuhlmann, L. Curing and Chloride Permeability.

Concrete International. April 1987, pp. 18-21.
Table 1: Mix Design Proportioning
@Seismicisolation
@Seismicisolation
Table 2: Chemical and Physical Properties of Cement
1 Gypsum (%
@Seismicisolation
@Seismicisolation
Table 3: Rapid Chloride Penetration of Concrete Cylinders Cured in 5%Sodium
Sulfate Solution (ASTM 1202)
A 3955
LA Not Available
IJ
X
I 462 1
3664
I
Y
_
i 4682 1
@Seismicisolation
@Seismicisolation
-
a
a
8
e
II I,I I
a
a
rn
: I I '
a
rn
:E
i i l a
rn
rn
II
a
rn
a
: I -l I
rn
a
rn
rn
a
rn
rn
rn
rn
rn
a w
a
a
rn
EQ
U
63
i
rn
rn
a
a u
rn
a
a
a
i I II I
a
a
a
i! Il III
II -1
PJ-u
a
a
a
a
* I I . I I l I I I
I I I I I l l I
I
Fig. 1 Slump Loss of Fresh Concrete During the Extended Mixing

Period
@Seismicisolation
@Seismicisolation
2 3.0
6
V
J I
k
2 2.0
1.o
M B A LA LT X Y
Concrete Mixes
Fig. 2 Average Air Content of Concrete Mixes
-- 2280
8 2210
I 2240
M B A LA Ll X Y
Concrete Mixes
Fig. 3 Average Unit Weight of Concrete Mixes
@Seismicisolation
@Seismicisolation
Fig. 4 Concrete Setting Time
@Seismicisolation
@Seismicisolation
Fig. 5 Compressive Strength of Concrete Mixes
@Seismicisolation
@Seismicisolation
-
20 microns
Fig. 6 Morphology of Clinker B and X
U
t
rm
k
o)
3000
M
h
d
u
2000 I I I I , I
48 50 52 54 56 58 60 62 64
C3S Content (%)
Fig. 7 Rapid Chloride Penetration Test for Concrete Cylinders at an

age of 90 Days
@Seismicisolation
@Seismicisolation
SP 209-33
Evaluation of In-Place Flexural

Strength Concrete Using
Compressive and Splitting
Tensile Strengths for Cast and
Cored Specimen
by W. W. Taylor, S. D. Palmer, and R. W. Kukes
SvnoDsis: This paper presents a rational approach used for the evaluation of in-
place concrete pavement with flexural strength requirements. During the construc-
tion of a concrete paving project at McCarran International Airport in Las Vegas,
Nevada, data was developed from the testing of over 450 specimens of concrete
beams, cylinders, and cores representing samples from nearly 170 locations. Flex-
ural, compressive, and splitting tensile strength testing was performed on these
samples obtained from locations where comparison between the different types of
strength tests was possible. Relationships between this data were evaluated and a
rational approach to the evaluation of in-place concrete for compliance with flex-
ural strength requirements was developed. This approach that begins with trial
batch data and includes cast and cored specimen, could be applied to other con-
crete paving projects with similar concerns.
Kevwords: com pressive; flexura I; splitting ten si le
@Seismicisolation
@Seismicisolation
609
610 Taylor et al.
William W. Tavlor. P.E.
Mr. Taylor is the Chief Materials Engineer for Kleinfelder, Inc. He graduated
with a BS degree in Civil Engineering from Montana State University in 1974.
Mr. Taylor is registered in Montana, Nevada, Idaho, and Arizona. He has been
for responsible for projects with construction costs exceeding $2,000,000,000.
His experience entails public works, commercial, residential, and industrial
projects. He is a member of the ASCE, ASTM, ACI, and APWA.
Samuel D. Palmer. P.E., CEM

Mr. Palmer is the Division Manager for Terracon in California, Nevada, Arizona,
and New Mexico. He graduated with an MBA degree in 1992 from the
University of Nevada, Las Vegas and a BS degree in Geological Engineering in
1983 from the University of Nevada, Reno. He has 18 years of experience in the
environmental, geotechnical, and materials engineering fields. Mr. Palmer is
affiliated with the ACIA, ASCE, LVCC, ACI, and APWA.
Robert Kukes. P.E.

Mr. Kukes is presently the Area Manager at Kleinfelder, Inc in Billings,
Montana. He graduated with a BS degree in Civil Engineering from
Montana State University in 1990. He is a registered professional engineer
in the states of Montana, Wyoming, and Nevada. Mi. Kukes has vast
knowledge and experience in construction materials testing and engineering
and management. Mr. Kukes is a member of the ASCE, ACEC, and NSPE.
INTRODUCTION
This paper presents a rational approach used for the evaluation of in-place
concrete pavement with flexural strength requirements. There are a number of
established procedures for evaluating hardened concrete with compressive
strength requirements, where quality assurance testing of specimen cast from
plastic concrete indicates a potential concern. However, there is only limited
data and guidance available for the evaluation of hardened concrete with flexural
strength requirements, where testing of beams cast from plastic concrete indicate
a potential concerns.
During the construction of a concrete paving project at McCarran International

Airport in Las Vegas, Nevada data was developed from the testing of over 450
specimens of concrete beams, cylinders, and cores representing samples from
nearly 170 locations. Flexural, compressive, and splitting tensile strength testing
was performed on these samples obtained from locations where comparison
between the different types of strength tests was possible.
Relationships between this data were evaluated and a rational approach to the
evaluation of in-place concrete for compliance with flexural strength
@Seismicisolation
@Seismicisolation
requirements was developed. This approach, that begins with trial batch data and
includes cast and cored specimen, could be applied to other concrete paving
projects with similar concerns.
PROJECT
The project consisted of replacement of concrete pavement paneIs in the apron,

tarmac, and taxiway areas of McCarran International Airport. The work included
in this study included 432 mm thick concrete pavement and was performed
between October, 1999 and May, 2000.
The project owner is the Clark County Department of Aviation, who retained the
Bechtel Corporation as their program manager for the design and construction of
this project. Kleinfelder, Inc. was contracted by the Clark County Department of
Aviation to perform quality assurance and product acceptance testing. Stantec,
Inc. performed the trial batch testing for the contractor, MMC, Inc.
These specifications required testing to be performed in accordance with specific

standard ASTM procedures. These procedures were strictly followed in the field
and laboratory, including on-site, temperature-controlled initial (field) curing
structures. Equipment was precisely calibrated, including specialized calibration
of the third point loading apparatus by Construction Technologies Laboratories
to assure the uniform transfer of loads to the flexural beam specimen.
SCOPE OF STUDY
During this project, data was developed that included strengths from laboratory
batched trial mixes, field cast specimen from production materials, and concrete
cores obtained from the in-place concrete pavement. Trial batch data included
flexural strengths of cast 152 mm by 152 mm beams with a third point loading
total span of 457 mm and compressive strengths of cast 152 mm diameter by 305
mm? high cylinders. Both beams and cylinders were tested at ages of 1,3,7, 14,
21,28,35, and 90 days.
Field cast flexural strength concrete beams were also 152 mm by 152 mm in size
with third point loading total spans of 457 mm. Field cast beams were tested at
an age of 28 days. 152 mm diameter by 305 mm high cylindrical specimen were
also cast in the field from production concrete and tested for compressive and
spilt tensile strength.
Drilled cores were 100 mm diameter and approximately 432 mm length. Cores
were trimmed to the required testing length. Typically two samples were sawn
and tested for each core. Both compressive strength and splitting tensile
strengths were performed on cores, whose ages ranged from 43 to 82 days.
Flexural strength tests were performed in accordance with ASTM C 78.

Compressive strength tests of cast cylinders were performed in accordance with
ASTM C 39. Splitting tensile strength tests were performed in accordance with
ASTM C 496. Cores were obtained and tested in accordance with ASTM C 42.
@Seismicisolation
@Seismicisolation
612 Taylor et al.
Filed samples of plastic, production concrete were obtained in accordance with
ASTM C 31. The location of field samples of plastic concrete and cores were
determined by a random sampling procedure in accordance with ASTM D 3665.
The random locations of plastic concrete and core samples were determined
independently. Therefore, although the plastic concrete and core samples were
from the same sublot, they were not necessarily from the same batch (truckload)
or location. This prevented the development of relationships between the cores
and the cast beam test results. There were 3 to 4 sublots for each day of
production and typically less than 200 cubic yards of concrete per sublot.
Batches were typically 10 cubic yards in size.
The trial batches tested 3 beams and 3 cylinders for each age. The field data
produced strengths for cast beam samples at 87 sublot locations. Cores were also
obtained at 33 sublot locations.
Based upon the trial batch data, relationships were developed between flexural
strength and compressive strength. Field data of beam and cylinder specimens
cast from the same production sample in the field were used to develop
relationships between flexural, compressive, and splitting tensile strength.
These relationships were compared to published typical relationships presented

in referenced texts and the FAA Engineering Brief # 34 and Advisory Circular
No. 150/5320-6D. Using statistical procedures, a conclusion was developed
regarding the confidence of using this type of rational to evaluate in-place
concrete pavement specified using flexural strength.
DATA
Flexural, compressive, and splitting tensile strength tests were performed on this
project and are the topic for this paper. These test results are presented in the
figures that follow. Figs. 1 and 2 present the data plotted with respect to the
relationships of the data to other properties. Fig. 2 also presents this data in an
analysis form, where relationships are used to convert splitting tensile strength to
flexural strength. Figs. 3,4, and 5 present the data based upon analysis of the
relationships between the different types of strength tests.
Using the triai batch field data for cast specimens, the relationship shown in Fig.
1 was established between flexural and compressive strengths. Using the
strength data of beams and cylinders cast from plastic concrete used in the
production of the in-place concrete pavement, the relationship between flexural
and splitting tensile strengths is shown in Fig. 2.
DATA ANALYSIS
Data was analyzed using the previously described above relationships; the
relationship ranges published in the text, Composition and Properties of
Concrete, Second Edition, Troxell, Davis and Kelly, published by McGraw-Hill,
@Seismicisolation
@Seismicisolation
Inc., 1968; and also relationships presented in FAA Advisory Circular 15015320-
6D. Figs. 2 through 5 present the compressive and splitting cylinder conversions
to flexural strength using the referenced text and FAA Advisory Circular
relationships. These relationships are as follows:
ComDosition and ProDerties of Concrete--Flexural Strength equals a

constant times the square root of the compressive strength, where the constant
ranges from 8 to 10. (Results in psi, convert to MPa by multiplying times
0.006895). Flexural Strength equals the splitting tensile strength divided by a
constant, where the constant ranges from 0.5 to 0.75. (Results in psi, convert to
MPa by multiplying times 0.006895)
FAA Advisory Circular 15Ol5320-6Dand Engineering Brief # 34--

Flexural Strength equals 9.0 times the square root of the compressive strength. .
mesults in psi, convert to MPa by multiplying times 0.006895). Flexural
Strength equals 1.02 times the splitting tensile strength (psi) + 117 psi. (Results
in psi, convert to MPa by multiplying times 0.006895)
Fig. 2 shows the conversion of splitting tensile strength of cast cylindrical

concrete specimen using the relationships presented in the different sources and
developed from the field data. Fig. 3 depicts the conversion of the core data
from compressive to flexural strength using the relationships presented in the
different sources and developed during the trial batch.
The correlation coefficient for each of these relationships was used to determine
the degree to which a valid statistical relationship exists between the two ranges
of data being compared. Correlation coefficients of near 1.0 indicate a definite
statistical relationship exists between the to properties being charted. The further
away from 1.0, the less the statistical relationship that exists.
The correlation coefficient of 0.99 for the trial batch relationship of compressive
to flexural strength indicates an excellent statistical relationship exists between
these two strength properties. This was not the case for the specimen cast from
production materials in the field.
Likewise, the correlation coefficient for the relationship of splitting tensile and
flexural strength for field cast specimen was 0.0002, indicating a statistical
relationship between these two properties is nearly nonexistent. Splitting tensile
strength testing was not performed during the trial batch.
The absence of a discernible statistical relationship for the field cast specimen is
likely the result of data that is too tightly grouped to dampen the effects of testing
variables. Field operations targeted a specified strength range, where as trial
batch data was available over a broad range of strengths. For example, the trial
batch data ranged from 2.6 to 5.8 MPa (3.2 MPa) for flexural strength and 14.1
MPa to 51.4 MPa (37.3 MPa) for compressive strength. The field data ranged
only from 4.6 to 5.1 MPa (0.5 MPa) for flexural strength and 43.2 to 47.4 MPa
(4.2 MPa) for compressive strength.
@Seismicisolation
@Seismicisolation
614 Taylor et al.
The total ranges of strength data for the field cast specimen were within the
range of reproducibility of the respective tests. ASTM C 78 states: "...results of
two properly conducted tests by the same operator on beams made from the same
batch sample should not differ from each other by more than 16%." The entire
range of flexural strength values for the field data in Fig. 1 is about 11%.
Fig. 4 shows the relationship between flexural strengths calculated from

compressive strengths of cores and flexural strengths calculated from splitting
tensile strengths of cores obtained at the same location. The FAA Advisory
Circular formulas were used for both of the compressive and splitting tensile
strength calculations. The trial batch formula was used for the compressive
strength calculation. The compressive strength conversion uses the two different
formulas applied to the compressive strength of the same core. The splitting
tensile strength conversion uses a formula applied to a splitting tensile strength
of a different core from the same sample.
The calculated flexural strengths by the two different methods track each other
well, high strengths match high strengths, low strengths match low strengths.
However, the flexural strengths calculated using the splitting tensile strengths are
always higher that those calculated using the compressive strength. This
indicates that the two properties are equal in their statistical relationship
regardless of the FAA Advisory Circular formula used.
Fig. 5 contains data from 6 samples from which cylinder and beam specimen
were cast and flexural, compressive, and splitting tensile strength tests were
performed. Flexural strengths were calculated using a variety of methods and are
compared on the figure with the flexural strengths obtained directly from the
beams.
The calculated methods again track very well with each other, but not necessarily
with the beam strengths, although the range from highest to lowest value was
very small. Contrary to the dnlled core specimens, for these cast specimens, the
splitting tensile strength was always the lowest for each sample. The flexural
strength by any of the calculated methods did not very from the flexural strength
of the beams by more than 18% for any sample. Considering the uncertainty
created by the testing variability of the multiple types of tests, the correlation of
the various conversion formulas is remarkable.
CONCLUSIONS
Based upon the data from this study and the comparison of that data to published
relationships, a statistic relationship exists between flexural strength and both
compressive and splitting tensile strength. Relationships are valid for both
compressive and splitting tensile strengths from both drilled cores and cast
cylinders.
Statistically valid relationships can be developed during the trial batch testing of
a concrete mix.
@Seismicisolation
@Seismicisolation
The use of typical formulas presented in the FAA Advisory Circular can be used
to develop a reasonable approximation of flexural strength using either
compressive or splitting tensile strength conversions. Both the trial batch and
FAA Advisory Circular conversion methods fall within published ranges for the
data from this study.
Because the strength properties of concrete can vary significantly with differing
component materials, the approach of developing relationships during trial
batching is important to provide an added degree of confidence in the conversion
formulas being used. Using a multi-faceted approach of trial batch and
production data, cast and cored specimen, and flexural, compressive, and
splitting tensile tests allows the evaluator to develop a feeling of the range of
concrete strength properties of the in-place material. This will provide direction
for further investigation of potential problem areas and development of solutions
in which owner, designer, construction contract administrator, quality assurance
and control personnel, and construction contracting parties to the project can be
sufficiently confident.
As a corollary to this approach, compressive andor splitting tensile strength

testing on a regular basis during production from the same sample as the flexural
strength testing can develop substantial quantities of data to build the
relationships among the strength properties. This data and the relationships
developed from it can be used to identify sampling or testing problems with cast
specimen and to develop increased confidence in evaluation of in-place concrete
using compressive or splitting tensile strengths, should it become necessary.
References
(1) Composition and Properties of Concrete, Second Edition, Troxell, Davis,

and Kelly, McGraw-Hill, Inc., 1968
(2) FAA Advisory Circular 150/5320-6D
(3) FAA Engineering Brief # 34
@Seismicisolation
@Seismicisolation
616 Taylor et al.
i
N
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
618 Taylor et al.
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
620 Taylor et al.
1
l
I
I
I
a
I
I
I
3
I
I
I
I ,
@Seismicisolation
@Seismicisolation
SP 209-34
Monitoring Prestress losses in

High Performance Concrete Beams
with a Built-In Fiber Optic
Monitoring System
by R. L. Idriss
SvnoDsis: An optical fiber monitoring system was designed and built into a three-
span high performance concrete highway bridge. The Rio Puerco Bridge, located
15 miles west of Albuquerque, is the first bridge to be built using HPC in New
Mexico. The bridge has 3 spans with length of 29 to 30 m. It is designed to be sim-
ply supported for dead load and continuous for live load. HPC was used for the
cast-in-place concrete deck and the prestressed concrete beams. A total of 40 long-
gage (2-m long) deformation sensors, along with thermocouples were installed in
parallel pairs at the top and bottom flange of the girders. The embedded sensors
measured temperature and deformations at the supports, at quarter spans and at
mid-span. Measurements were collected during: Beam Fabrication (Casting of the
beams, Steam curing, Strand release, Storage), Bridge Construction and Service.
The data collected was analyzed to calculate the prestress losses in the girders,
compare the losses to the predicted losses using available code methods, and get a
better understanding of the properties and behavior of high performance concrete.
The project is funded by the Federal Highway Administration, the New
Mexico State Highway and Transportation Department, and the National Science
Foundation.
Keywords: bridge; fiber optic sensors; high

performance concrete; monitoring; non-destructive
evaluation (NDE); prestressed concrete; prestress
Ioss e s; te sti ng
@Seismicisolation
@Seismicisolation
621
622 Idriss
Dr Rola Idriss is a Professor of Civil Engineering at New Mexico State
University. Her area of expertise is in the development of smart structures.
She has received numerous research and teaching awards, among them the
National Science Foundation Young Investigator Award. She is an active
member of the Transportation Research Board, ASCE and ACI.
INTRODUCTION
Using HPC in prestressed concrete girders has enabled engineers to design

bridges with longer span lengths and fewer supports, shallower sections, and
increased girder spacing, which can decrease the fabrication, transportation, and
erection costs of the bridge. However, despite these benefits, there are other
factors that need to be considered. In particular, the design of a prestressed
concrete girder is dependent on the amount of prestress loss expected over a
period of time. The prestress force in a prestressed concrete girder under
service conditions can be signiiïcantly lower thanthe initial jacking force
because of prestress losses due to elastic shortening, shrinkage, creep and
relaxation. Accurate prediction of the losses is a very important step in the
design of a highly stressed HPC girder, and can a f k t the service behavior of
the girder, such as deflections, camber, and cracking of the girder. Current
methods for calculating prestress losses according to the American Association
for Highway and Transportation Officials (AASHTO) (i), and the Prestressed
Concrete Institute (2) were developed,for conventional concrete with 28-day
strength up to 41.3 MPa (6,000 psi). Actual prestress losses in HPC girders
need to be measured and compared to the predicted losses using these
equations.
To date there has been a limited number of full-scale field studies

conducted to determine the long-term behavior of prestressed HPC beams
(3,4,5,6,7,8).The research results show a need for a better, clearer
understanding of long-tem behavior of HPC. Further research is needed to
evaluate the current empirical equations and possibly modi them to provide
an accurate estimate of prestress losses for a HPC girder. Furthermore, an
understand% of the effects of high temperatures used during steam curing
needs to be developed.
The Rio Puerco Bridge is the first bridge to be built in New Mexico
with a high performance concrete mixture. The purpose of the research study
was monitoring the long-term behavior of the HPC girders. Fiber optic
deformation sensors were embedded in four of the girders to monitor their
@Seismicisolation
@Seismicisolation
behavior for one year foiiowing casting of the beams. The objectives of the
study were to:
0 Determine in-situ mechanical properties of the HPC using the built-in
monitoring system
Compare the prestress losses calculated to those estimated by empirical
equations
BRIDGE DESCRIPTION
The Rio Puerco Bridge on Route 66 is located 15 miles west of Albuquerque,

NM. The bridge is a three span, prestressed concrete bridge as seen in Figs. 1,
2. It is designed to be simply supported for dead load and continuous for live
load. High Performance Concrete (HPC) is used for the cast-in-place concrete
deck and the prestressed concrete beams.
The primary members of the bridge consist of four type BT-1600 I-

beams. The beams are prestressed with 42, Grade 270 steel tendons. The 12.7
mm (0.5 in) diameter tendons are 7-wire, low-relaxation strands. The beam
cross-sectionscan be seen in Fig. 3.
MONITOKING SYSTEM
Four beams labeled AW, AC, BW and BC were monitored in the west and
center spans as shown in Fig. 4. To measure the prestress losses, strain was
monitored in the concrete. Long-gage (2-m long) deformation sensors were
installed in the top and bottom flanges of the girders (Fig. 3) and the cables
were routed thru the top flange to the connection box installed on the web side.
Five pairs of sensors were installed in each beam, measuring deformations at
supports, at YI spans, and at mid-span for a total of 40 fiber optic sensors. To
account for temperature strain, thermocouples were installed at the supports and
mid-span of each beam at the same location as the fiber optic sensors.
The Fiber Optic Sensors
The fiber optic sensors are deformation sensors. The long gage fiber optic
sensors were embeddable, waterproof sensors offering high resolution (2pm
(79pin )), insensitivityto corrosion and vibrations and immunity to
electromagneticfields. The 2-m (6.6-fi) gage length provided an average strain
measurement in the concrete.
@Seismicisolation
@Seismicisolation
624 Idriss
Sensor Instailation
The beams were pre bricated at CSR Prestress in Albuquerque, NM. The four
beams were fabricated simultaneously on one long hbrication bed. The bed
layout is shown in Fig. 5. The cables were extended internaliy to an outlet 1-m
fiom the marked end of each beam,and routed thru the top ilange to a
connection box installed on the beam web.
MEASUREMENTS
Measurements fiom the deformation and temperature sensors are read by a 64

channel reading unit. Casting of the beams took piace on July 17,2000. The
strands were released on July 24. Following release, the beams were stored for
a month at the prestressting piant. On August 26 the beams were transported to
site. The concrete deck was placed one month later on September 26.
Measurements were collected during:
Casting of the beams
Steamcuring
Strand release
Storage of the beams,up to transport
Casting of the slab
The fuli year following construction
Samde Data
The values read &omthe data acquisition system were transferred into an Excel
program where the data could be processed and graphed. Fig. 6 shows a typical
plot of deformation versus time for the deformation sensor in beam AC.
MATERIAL PROPERTIES
Mechanical properties of the HF'C were obtained using laboratory testing and
the data measurements obtained fiom the embedded fiber optic sensor system
@Seismicisolation
@Seismicisolation
Concrete Strength
The concrete was designed for a release strength of 48.3 MPa (7,000 psi) and a
28-day strength of 68.9 MPa (10,000 psi). Cylinders were taken during the
pouring of the girders, steam cured alongsidethe girders and tested in the
laboratory. The compressive strength was found to be 50.5 MPa (7,325 psi) at
3 days (strand release), 62.6 MPa (9,076 psi) at 28 days, and 70.0 MPa (10,151
psi) at 56 days.
Coefficient of Thermal Emansion
The coefficient of thermal expansion (a)is defined as the change in unit length
per degree of temperature change. The change in length due to a change in
temperature is given by:
AL = aL(T - TO) (1)

where L = original length
T - TO= Temperature change
CL = Coefficient of thermal expansion
During storage, the daily strain fluctuationsin the beams are mainly due to the
daily temperature changes. To determine the coefficient of thermal expansion,
the change in temperature versus the strain in the concrete are measured and
plotted, and an average coefficientof thermal expansion is calculated (Fig. 7).
The coefficient of thermal expansionvalues measured at 7 days and 3 1 days
were 12.5 and 12.7 respectivelyfor an average of 12.6 microstrd°C (7
microstrain/"F). Reported values of the coefficientof thermai expansion€or
concrete range between 5.8 to 12.6 microstrain/"C (3.2 to 7 microstrain/"F),
with an average of about 10 microstrd°C (5.5 microstrain/"F)(9 ). The
coefficient of thermal expansion for saturated portland cement pastes of
varying waterkement ratios is approximately18 microstrain/"C( 10
microstrain/"F) (10). The coefficient of thermal expansion measured is
consistent with the high cement paste content in the mixture (11).
Modulus of Elasticity
The modulus of elasîicity (E) of concrete varies with strength, concrete age,
loadii type, and the characteristicsof the cement and aggregates. Using
Hooke's law:
@Seismicisolation
@Seismicisolation
626 Idriss
where O = stress in the concrete

E = strain in the concrete
E is calculated at two times: at 3 days when the prestressing force is transferred

fiom the steel tendons to the concrete, and at 60 days when the deck is poured.
At transfer, the change in stress in the concrete is that caused by the release of
the prestressed tendons, and can be determined by the following equation:
<T = -PIA f MCA (3)
where P = Jacking force

A = Transformed cross-sectional area
M = Moment caused by prestress = P * e
e = Distance fiom steel centroid to transformed section centroid
c = Distance fiom the transformed section centroid to the location of
the sensor
I = Moment of inertia of the transformed section
The strains are calculated for the bottom sensors at midspan of each beam and
compensated for temperature using the following equation:
-
E n o n - t t i d = &raw awnmete AT (4)
where craw=raw strain
amne = coefficient determhed as 12.6 x 10' microstrad"C
AT = change in temperature measured by thermocouples
The modulus of elasticity at 60 days is calculated using the stresses and strains
that occur in the concrete when the deck is poured. When the deck is poured,
the change in stress is caused by the addition of the weight of the slab:
where &lab = Moment at midspan due to weight of the slab

c = Distance fiom the transformed section centroid to the location of
the sensor
I = Moment of inertia of the transformed section
The change in strain in the concrete is determhed fiom the change in the sensor
measurementswhen the slab was poured. The strains are calculated and
compensated for temperature.
Section 8.5 of the AC1 Code (12) states that the modulus of elasticity for
no& weight concrete can be taken as:
@Seismicisolation
@Seismicisolation
E= 57,000 (f c)IR psi (6)
where E = modulus of elasticity, psi
fc = compressive strength, psi
The expression recommended for computing E for high-strength concrete by
AC1 Committee 363 (13) is:
E = 40,000 (f c)" + ( 1 ~ 1 0psi

~) (7)
As seen in Table 1, AC1 equation (6) gave a very close estimate of what was
measured in the field. Overall, ail three values obtained fiom the field and
fiom the empirical equations were found to be reasonably close.
PRESTRESS LOSSES
Four methods were used to predict the prestress losses: the Prestressed
Concrete Institute (PCI) General Method (2), the American Concrete Institute -
American Society for Civil Engineers (ACI-ASCE) Method (14), the AASHTO
LRFD Relined Method (i), and the AASHTO LRFD Lump Sum Method (I).
None of these methods were developed speciñcdy for High Performance
Concrete.
The PCI General Method is a time-step method, and therefore permits

calculation of prestress losses at speciñc time intervals. Using the PCI General
Method, the incrementalprestress losses can be calculated and compared to the
stresses measured by the sensors during that time interval. It also requires input
of materiai properties
The ACI-ASCE and the AASHTO LFWD refined methods are

component methods, where the individual components of losses due to elastic
shortening, creep and shrinkage of concrete, and relaxation of the steel are
computed based on a knowledge of material properties and summed together
for an estimate of the total loss.
The AASHTO LRFD Lump Sum meîhod provides a simple approach. It

uses a single equation to calciilate the ultimate loss. The equation to be used
depends on the beam section type and the type of prestressing.
Prestress losses measured in the field were compared to losses predicted

by these four methods. All strain measurements were compensated for
temperature. The in-situ material properties determined in the field were used
to calculate the prestress losses.
@Seismicisolation
@Seismicisolation
628 Idriss
PCI Method vs. Field Measurements
The PCI General method was used to calculate the early prestress losses that
occurred at:
0 Transfer of the prestressing force (strand release).
One month after transfer.
The cumulative losses were calculated up to one year after transfer.
As illustrated in fig. 8, it was found that:

The PCI method predicted a slightly higher prestress loss at transfer.
0 The PCI method also predicted a much higher prestress loss during the
month of storage following fabrication. The PCI method predicted large
losses due to creep during that first month. The high performance concrete
showed much less creep than factored in the PCI equation.
The PCI method also over-predicted the measured losses during the ten
months period after the slab placement.
Cornuarim Total Losses: measured vs. Dredicted
In Fig. 9, the measured prestress losses (up to one year fiom beam casting) are
compared to the total losses estimated by each method. The predictions of the
four methods are relatively similar. Overali, all methods are found to be very
conservative. The prestress losses measured up to one year are 13% vs. 22 %
predicted by the PCI general method.
EFFECTS OF STEAM CURING TEMPERATURE
Beam AC was steam cured at a lower temperature than the other three beams.
The highest temperature measured for beam AC was 70 vs. 90 "C (158 OF to
194 "F ) for the other beams. Fig. 10 shows a plot of temperature vs. time for
the beams.
It was found that beam AC exhibited much larger prestress losses then
the other three beams in the storage month following transfer as shown in fig.
1la. This pattern of much larger losses for beam AC is most pronounced in the
month following transfer, and tapers off with time, as shown in fig. 11b.
@Seismicisolation
@Seismicisolation
CONCLUSIONS
A built-in monitoring system was used successfullyto monitor a bridge during

fabricationand construction. Monitoring the beams during the fabrication
process gave an insight into the effect of the steam curing temperature.
Material properties and prestress losses were measured using the built-in sensor
system. Based on the results, the following conclusions can be drawn:
1. The current code methods over predict the prestress losses for steam
cured HPC. There is a need to modi@these methods or develop new
ones to aiiow improved prediction of prestress losses for HPC
2. Estimates by aii four methods, the complex time consuming PCI
general, the ACI-ASCE, the AASHTO LRFD rehed, and the simple
AASHTO LRFD Lump Sum, came within a close range. Overall, all
four methods over predicted the prestress losses for HPC and are found
to be very conservative
3. HPC exhibited much less creep in the month following transfer than
factored by the PCI general method.
4. The steam curing temperature can have a significant effect on the early
prestress losses in HPC. A lower curing temperature was associated
with iarger early prestress losses following transfer. This is most
pronounced in the month following transfer, and tapers off with time.
5. An embedded monitoring system can be used successfullyto measure
the in-situ material properties.
6. The coefficient of thermal expansion measured was towards the higher
values of its reported range, which is consistent with the high cement
paste content of the concrete mixture.
7. The modulus of elasticity calculated fiom beam measurements in the
field compared well with the value calcuiated using the AC1 code
equation recommended for normal weight concrete. Overall, the AC1
code normaí weight equation, the AC1 363 modified formula for high
strength concrete, and the field measurements were reasonably close.
ACKNOWLEDGMENTS
The project is sponsored by the Federal Highway Administration,the New

Mexico State Highway and Transportation Department, and the National
Science Foundation. Their support is greatly appreciated. In addition, the
authors wish to thank CSR Prestress, HDR Engineering, and Smartec CO. for
their cooperation during this project.
@Seismicisolation
@Seismicisolation
630 Idriss
REFEKENCES
1. American Association of State Highway and Transportation Officials,

1998“AASHTO LRFD Bridge Design Specifications”, 2”dEdition,
Washington,D.C.
2. PCI Committee on Prestress Losses 1975, “Recommendationfor

Estimating Prestress Losses”, PCI Journal, Vol. 20, No. 4, July-August
1975, pp. 44-75
3. Li, Degang, Chajes Michael J., and Shenton Harry W., Prestress Losses
in HPC Girders: Before and After Transfer, proceedings of the
Transportation Research Board SO* Annual Meeting, January 7-1 1,
2001. Washuigton, D.C.
4. Byle K.A., Burns N.H., and Carrasquillo R.L., “Time-Dependent

Deformation Behavior of Prestressed High Performance Concrete
Bridge B e m ” , Center for Transportation Research- Research Report
580-6, October 1997.
5. Roller, John J., Russell, Henry G., Bruce, Robert N., and Martin,
Barney T. 1993, “Performance of Prestressed High Strength Concrete
Bridge Girders”, PCI J o d , Voi. 38, No. 3, May-June 1993, pp. 34-45
6. Roller, John J., Russell, Henry G., Bruce, Robert N., and Martin,
Barney T. 1993, “Long-Term Performance of Prestressed, Pretensioned
High Strength Concrete Bridge Girders”, PCI Journal, Vol. 40, No. 6,
Nov-Dec 1995, pp. 48-59
7. Barr, P., Fekete, E., Eberhard, M., Stanton, J., Khaleghi, B., and Hsieh,
J. C.2000, “High Performance Concrete in Washington State SR 18íSR
5 16 Overcrossing: Interim Report on Girder Monitoring”, Publication
NO. FHWA-RD-00-070, April 2000
8. Albom, Theresa M., Shield, Carol K., French, Catherine W. 1996,

‘‘Behavior of Two Long-Span High Strength Concrete Prestressed
Bridge Girders”, Worldwide Advances in Structural Concrete and
Masonr~,Proceedings of the CCMS Symposium held in conjunction
with Structures Congress XIV, April 15-18 196, Published by ASCE,
pp. 141-152
9. Design and Control of Concrete Mixtures, Portland Cement

Association, 13* Edition.
@Seismicisolation
@Seismicisolation
10. Mehta, Kumar P. and Monteiro, Paul J. M. 1993, Concrete: Structure,
Proverties. and Materials, 2“6Edition, Prentice Hall Inc., Englewood
Cliffs, NJ
1 1. Neville, A. M. Properties of Concrete. Fourth Edition. Longman Group

Limited. Essex, England, 1995.
12. Building Codes for Structural Concrete (AC1 3 18-99) and Commentary
(3 18R-99)”,American Concrete Institute, Farmington Hills, Mi
13. AC1 Committee 363 1984, “State-of-the-Art Report on High Strength

Concrete”, AC1 Jourrd, Vol. 8 1,No. 4. July-August 1984, pp. 364-404
@Seismicisolation
@Seismicisolation
632 Idriss
TABLES
Table 1. Modulus of Elasticity:Empirically
Figure 1. View of the Rio Puerco Bridge
@Seismicisolation
@Seismicisolation
- 29.3 m 30.8 m ---,--- 29.3 m ,-
a. Bridge profile along centerhe
(12.63 It) (12.63 fi) (12.63 ft)
14.49 m
(47.49 ft)
b. Bridge cross-section
Figure 2. Bridge profile and cross-section
@Seismicisolation
@Seismicisolation
634 Idriss
CL CL
4; iI
It--- 11.o r
I.
21 in (0.5335m)-r
3.46 in ,'/"
(0.051 m) .,'
(0.088 m) ~ ,
/
~
j ,A'
'#3 Dummy Rebar
6 in
(0.153 m)
-L
-
53.5 in
(1 359 m)i
'
#3 Dummy Rebar 7@2in
,/' \< (0.051 m)
4
3@2 in / I ++
+I+ t
26 in(O.661 rn) 1 Il@ 2 in-c/

(0.051 m)
END VIEW MIDSPAN VIEW

Figure 3. Cross-sectional view of BT-1600 I-beam
u U
Bearing Pier#l 4 Pler #2 Bearing
Abutment #1 (I Abutment # 2
dy
Figure 4. Plan View of the bridge showing the monitored beams
@Seismicisolation
@Seismicisolation
Figure 5. Sensor installation
@Seismicisolation
@Seismicisolation
636 Idriss
BEAM AC
Sensor Reading vs. Time
18
E 16
CI
p 14
5
w1 1 2
L
g 10
C
$ 8
6
7/13/00 8/2/00 0:OO 8/22/00 9/11/00 10/1/00 10/21/00 11/10/00
0:oo 0:oo 0:oo 0:oo 0:oo 0:oo
Date + Time
[ -AClBS -AClTS AC2BS -AC2TS -ACBBC -AC3TSI
I-AC4BS -AC4TS -ACSBS ACSTS
Figure 6. Sensor reading vs. time for beam AC
-AC10 -CAClT +AUT -ICAîöB - C A U T -CAW18 -AWIT -AWJB

AWJT AWSB AW5T --%--KIB ----BClT &BG0 x BC3T -8558
-BCST -6W16 BWlT -6W3T -8MT
0.0003
0.00025
0.0002
.-e
E 0.00015
0.0001
0.00005
O
20 25 30 35 40 45 50
Temperature (C)
Figure 7. Coefficient of thermal expansion at 7 days
@Seismicisolation
@Seismicisolation
Average % Losses
Measured vs. PCI Estimate
At transfer Stress due Jan April July

to deck
Placement
1 rn Measured rn PCI Estimate I \
Figure 8. % prestress loss. Field measurements vs. the PCI tthod
Total 'YOPrestress Loss
Measured @ PCI Method ACI-ASCE LRFD LRFD Lump

lyear @lyear Method Method Sum
Figure 9. Measured losses vs. losses estimated by various methods
@Seismicisolation
@Seismicisolation
638 Idriss
85
g75
5 65
0 55
o:
$25
35
m 15
' 7/13/00
0:oo
8RI00 0:OO 8/22/00
0:oo
9/11/00
0:oo
10/1/00
0:oo
10/21/00
0:oo
11/10/00
0:oo
Date + Time
a. Temperature reading vs Time for beam AW
mte + T i
b. Temperature reading vs. time for beam AC
7/1= 8/2/aloo BIPIMOO W11Rooo lCWl/aXx) IoRlIaXx) l l / l ~

oia, o.al 0.00 Qal o.al &al &o0
Date+Tim
c Temperature mdingvs.time for beamBW
Figure 10. Temperature Reading vs. Time
@Seismicisolation
@Seismicisolation
d. Temperature readinp vs. time for beam BC
Figure 10. Temperature Reading vs. Time (cont.)
@Seismicisolation
@Seismicisolation
640 Idriss
Losses During Storage (MPa)
160
140
120
1O0
m
$ 80
60
40
20
O
AC AW BC BW
/.Measured I P C I Estimate I
a. Losses during storage
Losses from Deck Placement up to One Year
70
60
50
m 40
30
20
10
O
AC AW BC BW
I Measured H PCI Estimate 1

b. Losses fiom deck placement up to one year
Figure 1 1. Prestress losses
@Seismicisolation
@Seismicisolation
SP 209-35
Performance-Based Evaluation of
Rectangular RC Structures Using a
Flexible Diaphragm Model
by J. M. Barron and M. B. D. Hueste
SvnoDsis: Under seismic loading, floor and roof systems in reinforced concrete
(RC) buildings act as diaphragms to transfer lateral earthquake loads to the vertical
lateral force resisting system (LFRS). In current practice, horizontal diaphragms
are typically assumed to be rigid, thus neglecting the effect of the in-plane move-
ment of the diaphragms relative to the vertical LFRS. This study focuses on evalu-
ating the impact of diaphragm flexibility on the structural response of typical RC
rectangular building structures using a performance-based approach. Three-story
and five-story RC buildings with end shear walls and two aspect ratios (approxi-
mately 2: 1 and 3: 1) were developed and designed according to current code proce-
dures assuming rigid diaphragm behavior. The performance-based design criteria
outlined in the NEHRP Guidelines for Seismic Rehabilitation of Buildings (FEMA
273) (4) were used to assess the adequacy of the four case study buildings when di-
aphragm flexibility was included in the structural response. It was found that the
use of a flexible diaphragm model had the largest impact on the %story, 3: 1 aspect
ratio building and the various analysis procedures gave differing assessments of
the adequacy of this case study building. The remaining three case study buildings
generally satisfy the FEMA 273 acceptance criteria for three performance levels
based on an evaluation of critical structural elements.
Kevwords: building diaphragms; FEMA 273; nonlinear

analysis; performance evaluation; seismic analysis
@Seismicisolation
@Seismicisolation
641
642 Barron and Hueste
Mary Beth D. Hueste is an Assistant Professor in the Department of Civil
Engineering at Texas A&M University, College Station, Texas, 77843-3136.
She received her Ph.D. degree from the University of Michigan at Ann Arbor.
Her research interests include studying the behavior of concrete structures using
experimental and analytical techniques, with a focus on earthquake resistant
design of reinforced concrete structures.
Joel M. Barron is an Engineer with Thomton Tomasetti Engineers, 12750 Merit

Drive, Suite 602, LB-7, Dallas, Texas, 75251-1226. He received his M.S.
degree from Texas A&M University, College Station, Texas. His interests
include stadium and high-rise building design, along with earthquake resistant
design of reinforced concrete structures.
INTRODUCTION
Floor diaphragms in reinforced concrete (RC) buildings are typically

modeled as rigid during the design phase and so the effect of in-plane diaphragm
flexibility on the structural response is often not considered. For the rigid
diaphragm model, the diaphragm has equal in-plane displacements along its
entire length under lateral load. However, a flexible diaphragm can exhibit in-
plane bending due to lateral load, resulting in unequal horizontal displacements
along its length. This flexibility can cause lateral load transfer to vertical
elements within the structural system that were not designed to carry significant
lateral loads. If this effect is sizeable, it can lead to overloading of structural
elements and possible failure of the structure.
Damage related to diaphragm response and behavior was observed in

concrete structures following the 1994 Northridge earthquake, primarily for
precast construction. A department store in the Northridge Fashion Center
experienced damage to the roof diaphragm (concrete fill over metal deck
supported on steel beams) and to the floor diaphragms (topping slab over precast
elements) (1). The Santa Monica College precast concrete parking structure
experienced chord failure in the diaphragm (2). Precast concrete parking
structures at the Northridge Fashion Center had diaphragm movement large
enough to cause failure in some of the “gravity load columns” and the Glendale
Civic Center garage (cast-in-place, post-tensioned construction) had collector
failures (3).
The main objective of this study was to conduct a performance-based

evaluation of the seismic resistance of rectangular cast-in-place RC structures to
assess whether a rigid diaphragm assumption leads to an acceptable building
@Seismicisolation
@Seismicisolation
design. A building configuration based on a case study building developed by
AC1 Committee 374 was extended to define four buildings for this study. These
buildings were designed for the Los Angeles, California area using a rigid
diaphragm assumption, resulting in four buildings representative of current
practice. The parameters varied were aspect ratio (2:l and 3:l) and number of
stories (three and five stories). The buildings were then analyzed with a flexible
diaphragm model, and evaluated according to the NEHRP Guidelines for
Seismic Rehabilitation of Buildings (FEMA 273) (4). Three analysis procedures
were conducted for each structure: (1) linear static, (2) linear dynamic, and (3)
nonlinear dynamic. The appropriate FEMA 273 acceptance criteria were
applied to evaluate the structural response determined from these analyses for
three performance levels. The results were interpreted to conclude whether the
rigid diaphragm design assumption led to an acceptable building design for the
performance levels considered.
PAST RESEARCH
Experimental work on RC diaphragms and corresponding finite element

analysis began at Lehigh University and continued at the State University of
New York (SUNY) at Buffalo. Initially, flat plate slabs (3, beam supported
slabs (6,7), and waffle slabs (8) were investigated at Lehigh University. Chen et
al. (9) investigated all three types of slab systems. Additional research was
conducted at the State University of New York (SUNY) at Buffalo where
Panahshahi et al. (10,11,12) tested a 1:6 scale model of a one-story floor system
on a shake table and compared the measured response to predicted response
parameters from a finite element analysis. For that study, it was found that the
diaphragm flexibility was most pronounced in shorter buildings with long aspect
ratios.
Many analytical studies related to RC diaphragms examine the validity

of the rigid diaphragm assumption. This assumption has been tested for a
variety of buildings with a number of differing parameters. Most studies show
that the diaphragm, when considered flexible, can alter the natural frequency of
the building and the distribution of forces to frames located away fÌom the
primary lateral force resisting frames. Jain and Jennings (13) concluded that in-
plane diaphragm flexibility is important in long narrow buildings and in
buildings with end walls. This flexibility causes the natural frequencies and
mode shapes to change, and the resulting participation factors affect lateral load
distribution. Kunnath et al. (14) determined that a flexible diaphragm resulted in
larger middle frame displacements as the aspect ratio increases and as the
number of stories decreases. The larger aspect ratios also resulted in more base
shear distribution to the middle frames, with less base shear distributed to the
middle frames as the number of stories increases. These conclusions from
@Seismicisolation
@Seismicisolation
experimental and analytical research guided the selection of the case study
buildings for this study.
CASE STUDY BUILDINGS
The case study buildings originated from a shear wail building developed
by AC1 Committee 374. The AC1 building was elongated to arrive at
approximately 2:l and 3:l aspect ratios. The AC1 building was also shortened
from ten-stories to five and three-story buildings. These two aspect ratios and
story heights provided the overall geometry for the four case study buildings.
The plan dimensions of the 2:l (more precisely, 2.2:l) aspect ratio
buildings are 33.5 x 73.2 m (1 10 x 240 ft.), as shown in the floor plan in Fig. 1.
The 3:1 (more precisely, 3.3:l) aspect ratio buildings have a similar layout with
twelve 9 m (30 fi.) bays making the overall dimensions 33.5 x 110 m (1 10 x 360
ft.). End shear walls are provided in the central bay in the transverse direction.
For all buildings, the first story height is 6.10 m (20 fi.) and each additional
story height is 3.96 m (1 3 ft.). The column rows are numbered, starting with the
end shear wall labeled CR-1, with the middle frame labeled CR-5 for the 2:l
aspect ratio buildings and CR-7 for the 3: 1 aspect ratio buildings.
A static elastic analysis using the software SAP2000 (15) was performed
for the building design. The gravity loads consisted of dead loads of self-weight
plus a uniform 138 kPa (20 psf) partition load, and a live load of 345 kPa (50
psf) at each floor level and at the roof level. A curtain wall load of 138 kPa (20
psQ was applied over the elevation of the outer perimeter of each building.
Design earthquake loads were determined by the International Building Code
2000 (IBC 2000) (16) for the Los Angeles region. The earthquake loads were
applied to the structure according to the IBC 2000 guidelines for rigid
diaphragms, which states that the lateral load is distributed horizontally in
accordance with the relative stiffness of the vertical LFRS.
The buildings were designed according to the AC1 3 18-99 Building Code
Requirements for Structural Concrete (17) using a rigid diaphragm model. The
design of the beams and girders was controlled by gravity load demands because
the shear walls resisted the full lateral load. Therefore, the reinforcing
requirements for these members were very similar for all four buildings. The
shear wall design was controlled by the overturning moment at the base, and
was different for all four buildings. The slab was designed for gravity loads-.
only.
Member dimensions and typical reinforcement layouts are provided in

Fig. 2. Interior floor beams between column lines, 305 x 305 mm (12 x 22 in.),
were excluded from the case study building models for simplicity and because
@Seismicisolation
@Seismicisolation
they were not part of the lateral force resisting system. However, the weight
from these beams was included in the model. The shear walls are 305 mm (12
in.) thick in the three-story buildings and 61O mm (24 in.) thick in the five-story
buildings. The slab is 127 mm (5 in.) thick for the floors and was designed as a
one-way slab spanning between interior floor beams. No interior beams are
provided at the roof level; rather a post-tensioned slab with a constant thickness
of 203 mm (8 in.) is used. Normal weight concrete with a specified compressive
strength of 34.5 MPa (5000 psi) is used for all members and the reinforcing steel
specified yield strength is 414 MPa (60 ksi). Additional design details are
documented by Barron (18).
PERFORMANCE BASED EVALUATION
General
The NEHRP guidelines for evaluation of existing buildings for seismic

demands are described in FEMA 273 and were recently updated in FEMA 356
(19). The FEMA 273 guidelines include four analysis procedures with
appropriate criteria for determining the adequacy of structures. For this study,
three of the four analysis procedures were utilized: the linear static procedure
(LSP), the linear dynamic procedure (LDP), and the nonlinear dynamic
procedure (NDP). The linear analyses were conducted using SAp2000, and
IDARC2 (20) was used for the nonlinear analysis. Both the linear and nonlinear
models allowed for flexible diaphragm behavior. The full slab thickness was
used for modeling the in-plane stiffness of the floor and roof diaphragms for the
linear procedures to avoid over-estimating the effects of diaphragm deformation.
For the nonlinear analysis, the full slab thickness was also used for the initial
diaphragm stifhess. However, the IDARC2 floor slab model modifies the in-
plane stiffness of the diaphragm to account for inelastic behavior.
The case study buildings were evaluated for the 2% probability of

exceedance in 50 years event for Los Angeles, California using the FEMA 273
guidelines with the lateral load or ground motion applied parallel to the short
dimension of each building. Three performance levels were evaluated for this
event: Immediate Occupancy (IM), Life Safety (LS), and Collapse Prevention
(CP). The major results for each procedure are summarized in terms of base
shear distribution, displacements and impacts of diaphragm flexibility; the
FEMA 273 diaphragm classification is also noted. According to FEMA 273, a
diaphragm is flexible if the maximum deformation of the diaphragm is more
than twice the average inter-story drift of the story directly below the
diaphragm. The diaphragm is classified as rigid if the maximum deformation of
the diaphragm is less than half the average inter-story drift of the story directly
@Seismicisolation
@Seismicisolation
below the diaphragm. A diaphragm is classified as stiff if it is neither rigid nor
flexible.
Linear Static Procedure (LSP)
For the LSP, the design base shear varies as a function of the
performance level and the results are reported for the three performance levels.
Base Shear Distribution -- The flexible diaphragm model distributed

more of the total base shear into the middle frames as compared with the rigid
diaphragm model. However, the effect is minimal for the LSP, because the total
base shear resisted by each interior frame is still less than one percent. The two
end shear walls resist the remaining base shear.
Displacements -- Frame displacements and drifts were recorded for each

column line. The diaphragm “flex” was determined by subtracting the
displacement at the middle frame from the displacement at the end wall. The
greatest amount of flex for each case study building, as well as the highest drift
percentage, occurred at the floor level immediately below the roof level. This is
the fourth suspended floor for the five-story buildings and the second suspended
floor for the three-story buildings. Compared to the rigid diaphragm model, the
flexible diaphragm model gave a 1575% increase in the middle frame
displacement and a 9-41% increase in inter-story drift. The largest increase was
for the 3-story, 3:l aspect ratio building.
Table 1 lists the diaphragm flex and the FEMA 273 diaphragm
classification for each case study building and for each performance level. The
diaphragm flex is greatest for the 3:l aspect ratio buildings and for the three-
story buildings. However, for the LSP analysis none of the diaphragms are
classified as flexible. in addition, all story drifts for the case study buildings are
below the standard acceptable limit of two percent.
Impact of Diaphragm Flexibility -- The middle frame columns are

critical elements in terms of flexible diaphragm effects. As the diaphragms flex,,
the middle frames pick up additional lateral load. The diaphragm flexibility can
also lead to increased P-A effects in the middle frame columns. The demand-
capacity-ratio @CR) was found by dividing the maximum middle frame column
moment within each story by the corresponding column moment capacity. The
DCR’s were compared to the appropriate m values for RC columns given in
FEMA 273, where m is defined as a demand modifier to account for expected
ductility of the deformation associated with a deformation-controlled action at
the selected performance level. If a DCR is lower than the m value, then the
column is acceptable for the corresponding performance level.
@Seismicisolation
@Seismicisolation
Fig. 3 shows the maximum DCR computed for the middle frame
columns in each story for each performance level using the flexible diaphragm
model. All column DCRs for the five-story buildings and the 3-story, 2:l
aspect ratio building are below two, and are satisfactory for all three
performance levels. However, the columns for the 3-story, 3:l aspect ratio
building are not satisfactory based on the immediate occupancy (IM)
performance level.
For the IM performance level, the rigid diaphragm DCRs for the 3-
story, 3:l aspect ratio are 2.37, 1.53, and 1.16 for the 3rdstory, 2”dstory, and 1’’
story, respectively. The DCRs for the columns in the lower two stories for the
rigid diaphragm model are below the IM performance level limits, while the
flexible diaphragm model produced DCR values that exceed the IM limits for
these columns. As a result, the use of the flexible diaphragm model was critical
for the 3-story, 3:l aspect ratio building.
Summarv of LSP Results -- The case study buildings investigated with

the LSP were classified as either “rigid” or “stiff according to the FEMA 273
provisions. The base shear distribution to the interior frames was less than one
percent in all cases. Trends with aspect ratio and story height were evident,
although three of the four buildings were still acceptable for the three
performance levels for the 2% probability of exceedance in 50 years event. The
3-story, 3: 1 aspect ratio building middle frame columns exceeded the acceptance
criteria for the immediate occupancy performance level. However, this
performance objective for a 2% in 50 years event is quite conservative.
Therefore, the rigid diaphragm model gave an acceptable design for the case
study buildings, although a flexible diaphragm model appears more appropriate
for the 3-story, 3:l case study building.
Linear Dynamic Procedure íLDP)
For the linear dynamic procedure (LDP), the response parameters for the
four case study buildings were determined for a suite of ten synthetic ground
motion records corresponding to a 2% probability of exceedance in 50 years
event for Los Angeles, California. The synthetic ground motions were
developed by Woodward-Clyde (21) for the SAC project. The original suite
contained 20 records, or ten orthogonal time history pairs. The time history
having the maximum peak ground acceleration (PGA) was chosen for each pair
and applied to the case study buildings in the transverse direction. When seven
or more ground motion data sets are used, FEMA 273 allows the use of the
average value of each response parameter to determine acceptability for a
particdar performance level. Therefore, average values were used in the
evaluation of the acceptance criteria for the various performance levels.
@Seismicisolation
@Seismicisolation
Base Shear Distribution and Building Period -- For the LDP, the flexible
diaphragm model distributed more of the total base shear into the middle frames
as compared with the rigid diaphragm model, as was the case for the LSP. The
amount of base shear distributed to the interior frames is slightly larger for the
LDP than for the LSP; however, the values are still less than one percent for
each interior frame. Table 2 provides the fundamental periods of the structures
for the rigid and flexible diaphragm models, along with the ratio of the flexible
model period to the rigid model period. The most significant change in the
building period was observed for the 3-story, 3:l aspect ratio building, which
had a 20% increase when the flexible diaphragm model was used.
Displacements -- Frame displacements and drifts at each floor level were

computed for all four case study buildings. The flexible diaphragm model
generally resulted in larger average displacements for the interior frames as
compared to the rigid diaphragm model, with a 241% increase in the middle
frame displacement and a O-125% increase in inter-story drift. Again, the
largest increase was observed for the h t o r y , 3:l aspect ratio building. Table 3
lists the average and maxihum diaphragm flex and the FEMA 273 diaphragm
classification for the ground motion time history producing the greatest effect.
Again, the longer aspect ratio buildings and the buildings with fewer stories had
the largest diaphragm flex and interstory drifts. Although the LDP resulted in
more displacement and diaphragm flex than for the LSP, none of the diaphragms
are classified as flexible for the LDP.
Impact of Diaphragm Flexibility -- Maximum forces were computed for

the middle frame columns for all case study buildings for each time history
analysis using the LDP. The maximum column moment from each ground
motion was computed and the corresponding average value was determined for
each column member. The corresponding maximum DCRs were then
computed for each story in a building and compared to the appropriate m values
in the same manner as for the LSP. Fig. 4 shows the maximum column DCRs
in each story for all performance levels.
The columns in the k t o r y , 3:l aspect ratio building do not satisfj the
FEMA 273 acceptance criteria for all performance levels. One column for the
h t o r y , 3:l aspect ratio building does not meet the immediate occupancy
criteria. The other five-story building columns and the 3-sto1-y~2:l aspect ratio
building columns were found to be satisfactory for all performance levels.
The flexible diaphragm model resulted in larger DCR's compared ta

those determined for the rigid diaphragm model. When a rigid diaphragm
model was used for the 3-story, 3:l aspect ratio building, only the third story
columns did not meet all performance levels. In addition, the second story
columns did not meet the immediate occupancy criteria. The rest of the building
@Seismicisolation
@Seismicisolation
columns were found to be acceptable according to the FEMA 273 criteria when
using a rigid diaphragm model.
Summarv of LDP Results -- The analyses of the case study buildings by

the LDP with a flexible diaphragm model did not exhibit significant diaphragm
flexibility. No diaphragms were classified as flexible according to FEMA 273,
and the base shear distributed to each interior frame is less than one percent in
all cases. However, the use of a flexible diaphragm model did impact the
evaluation of the middle frame columns in the 3-story, 3:l aspect ratio building.
The LDP procedure indicates that a rigid diaphragm assumption in design is
reasonable for all case study buildings except for the 3-story, 3:1 aspect ratio
building. This is more evident for the LDP as compared to the LSP.
Nonlinear Dynamic Procedure CNDP)
The nonlinear dynamic procedure (NDP) consisted of inelastic time

history analyses of the four case study structures using the same ten synthetic
ground motion data sets used for the LDP. Because running the entire ground
motion time history is time consuming for the nonlinear analysis, only the strong
motion portion of each time history was used, as determined using the method
proposed by Trifunac and Brady (22). The reported frame reactions and
displacements correspond to the time of maximum base shear, while the
reported column plastic rotations are from maximum values that occurred during
the time history analyses.
Base Shear Distribution -- For the NDP, the flexible diaphragm model
did not result in a significant increase in base shear distributed to the interior
fiames as compared to a rigid diaphragm model. The base shear distributed to
the interior frames using a flexible diaphragm model varied between
approximately 1.7% and 2.7%, which is slightly higher than for the linear
procedures.
Disulacements -- The diaphragm flex was determined at the time of

maximum base shear for each case study building and ground motion record.
Table 4 lists the average and maximum diaphragm flex and the FEMA 273
diaphragm classification for each case study building when using the NDP. The
largest diaphragm flex for all four buildings was at the second suspended floor
level. The flex for the NDP was less than the flex for both linear procedures and
all of the diaphragms were classified as rigid. However, the story drift was'
greater for the NDP than for both linear procedures.
imuact of Diarihrapm Flexibility -- The acceptance criteria for the NDP

involves plastic rotation limits. The average maximum plastic rotation for the
ten ground motions was determined for each middle frame column for all case
@Seismicisolation
@Seismicisolation
study buildings. The maximum value within a story was then compared with the
FEMA 273 plastic rotation limits (see Fig. 5). All of the middle frame columns
for all of the case study buildings meet the IM performance level. Therefore, ail
middle frame columns are considered acceptable for the NDP. For the five-
story buildings the longer aspect ratio results in larger plastic rotations in the
middle frame columns, but that trend is not noticeable for the three-story
buildings.
Summarv of NDP Results -- Diaphragm flexibility is not a significant

issue for the case study buildings based on the NDP. The base shear distributed
to the interior frames is the same for both rigid and flexible diaphragm models,
and all diaphragms were classified as rigid. All middle frame columns were
determined to be acceptable according to the FEMA 273 plastic rotation criteria.
Four rectangular RC case study buildings were designed using a rigid

diaphragm assumption, including a 3-story, 2:l aspect ratio building; a 3-story,
3:l aspect ratio building; a h t o r y , 2:l aspect ratio building; and a htory, 3:l
aspect ratio building. Each case study building was then evaluated using the
FEMA 273 linear static procedure (LSP), linear dynamic procedure (LDP) and
nonlinear dynamic procedure (NDP) with a flexible diaphragm model. The
following conclusions were made.
1. For almost every analysis of the four case study buildings, a flexible
diaphragm model produces more frame displacement and inter-story drift
than a rigid diaphragm model. This is especially evident for frames that are
closer to the middle of the building. However, the case study building
diaphragms were not classified as flexible based on the FEMA 273 criteria.
The diaphragms for the h t o r y , 2:l aspect ratio building were classified as
rigid for the three analysis procedures. The diaphragms for the remaining
three buildings were classified as stiff for both linear procedures and as rigid
for the NDP.
2. When comparing the rigid and flexible diaphragm model analyses, the
difference in the distribution of base shear to the interior frames and the
change in the fundamental period were not significant for the case study
buildings. However, the use of a flexible diaphragm model leads to a less
favorable evaluation of the 3:l aspect ratio buildings for the FEMA 273'
linear procedures.
3. The FEMA 273 analysis procedures differ in the acceptance of structural

elements according to the criteria for the three performance levels used in
this study. For example, the middle frame columns in the 3-story, 3:l aspect
@Seismicisolation
@Seismicisolation
ratio building met all performance levels for the NDP, met only the life
safety and collapse prevention performance levels for the LSP, and did not
meet any performance levels for the LDP.
4. The use of the NDP indicates that all performance levels are met based on
the criteria evaluated in this study and the use of a rigid diaphragm model is
adequate for the design of the case study buildings. However, it is noted that
the evaluation using linear procedures did show some inadequacies for the 3-
story, 3:l aspect ratio building, especially for the LDP. Therefore, it is
recommended that the use of a flexible diaphragm model be considered in
the design and evaluation of similar low to moderate RC structures having
an aspect ratio of 3 :1 or larger.
ACKNOWLEDGMENTS
The authors wish to acknowledge the financial support provided by the

Civil Engineering Department at Texas A&M University, where this research
was conducted. The authors wish to thank Professor Sashi Kunnath for his help
in making the IDARC2 program available for this research. The opinions
expressed in this paper are those of the authors and do not necessarily reflect the
views or policies of the sponsor.
REFERENCES
1. Hamburger, R. O. (1996). “Department Store, Northridge Fashion Center.”

1994 Northridge Earthquake Buildings Case Studies Project, Seismic Safety
Commission, State of California, Proposition 122: Product 3.2, 83-100,
Sacramento, CA.
2. Phillips, R. (1996). “Santa Monica College, Precast Concrete Parking

Structure.” I994 Northridge Earthquake Buildings Case Studies Project,
Seismic Safety Commission, State of California, Proposition 122: Product
3.2, 137-142, Sacramento, CA.
3. Corley, W.G., Cluff, L., Hilmy, S., Holmes, W. and Wight, J. (1996).
“Concrete Parking Structures.” Earthquake Spectra, Supplement C to Vol.
I I , Northridge Earthquake of January 17, 1994 Reconnaissance Report,
Vol. 2, Earthquake Engineering Research Institute, Oakland, CA.
4. Applied Technology Council (ATC). (1997a). NEHRP Guidelinesfor the

Seismic Rehabilitation of Buildings. (FEMA Publication 2731, Building
Seismic Safety Council, Washington, D.C.
@Seismicisolation
@Seismicisolation
5. Karadogan, H.F., Huang, T, Lu, L.W. and Nakashima, M. (1980). “Behavior
of Flat Plate Floor Systems Under In-Plane Seismic Loading.” Proceedings,
Seventh World Conference of Earthquake Engineering, Vol. V, 9-16.
6. Nakashima, M., Huang, T. and Lu, L.W. (1982). “Experimental Study of

Beam-Supported Slabs Under In-Plane Loading.” AC1 Journal, Proceedings,
79(1), 59-65.
7. Nakashima, M., Huang, T. and Lu, L.W. (1984). “Effect of Diaphragm

Flexibility on Seismic Response of Building Structures.” Proceedings,
Eighth World Conference on Earthquake Engineering, Vol. lV,735-742.
8. Ji, X., Chen, S.J., Huang, T. and Lu, L.W. (1985). “Deflections of Waffle
I
Slabs Under Gravity and In-Plane Loads.” Deflections of Concrete

Structures, AC1 SP-86, Detroit, 283-294.
9. Chen, S.J., Huang, T and Lu, L.W. (1989). “Diaphragm Behavior of RC

Floor Slabs.” Proceedings, Ninth World Conference on Earthquake
Engineering, Vol. IV, 9WCEE Organizing Committee, Japan Assn. For
Earthquake Disaster Prevention, Tokyo, Paper 6-5-20, 565-570.
10. Panahshahi, N., Kunnath, S.K., Reinhom, A.M., Lu, L.W. and Huang, T.
(1990). “Inelastic Modeling of RC Buildings with Flexible Floors.”
Proceedings, Fourth US. National Conference on Earthquake Engineering,
Vol. II, 369-378.
11. Panahshahi, N., Reinhom, A.M., Kunnath, S.K., Lu, L.W., Huang, T. and
Yu, K. (1991). “Seismic Response of a 1:6 Scale Reinforced Concrete Scale-
Model Structure with Flexible Floor Diaphragms.” AC1 Structural Journal,
88(3), 3 15-324.
12. Panahshahi, N., Reinhorn, A.M. and Kunnath, S.K. (1994). “Earthquake
Simulation Study of a One-Sixth Scale-Model RC Building with Flexible
Floor Diaphragms.” Proceedings, F$h US. National Conference on
Earthquake Engineering, Vol. II, 641-650.
13. Jain, S.K. and Jennings, P.C. (1985). “Analytical Models For Low-Rise
Buildings With Flexible Floor Diaphragms.” Earthquake Engineering and
Structural Dynamics, 13(2), 225-241.
14. Kunnath, S.K., Panahshahi, N. and Reinhom, A.M. (1991). “Seismic

Response of RC Buildings with Inelastic Floor Diaphragms.” Journal of
Structural Engineering, ASCE, 117, 1218-1237, April.
@Seismicisolation
@Seismicisolation
15. Computers and Structures, Inc. (1 999). Integrated Finite Element Analysis
and Design of Structures: Getting Started (SAP2000). Computers and
Structures, Inc. Berkeley, CA, USA.
16. Intemational Code Council (1998). International Building Code 2000; Final
Draft July 1998. Intemational Code Council, Falls Church, VA.
17. AC1 Committee 3 18 (1999). Building Code Requirements for Structural

Concrete (ACI 318-99) and Commentary (ACI 318R-99). American
Concrete Institute, Detroit, MI.
18. Barron, Joel Martin (2001). “Performance Based Evaluation of the Seismic
Resistance of Structures with Concrete Diaphragms”, MS Thesis,
Department of Civil Engineering, Texas A&M University, College Station,
TX, 176 pages.
19. American Society of Civil Engineers (ASCE) (2000). “Prestandard and

Commentary for the Seismic Rehabilitation of Buildings” (FEMA
Publication 356), Federai Emergency Management Agency, Washington,
D.C.
20. Reinhom, A.M., Kunnath, S.K. and Panahshahi, N. (1988). “Modeling of

RC Building Structures with Flexible Floor Diaphragms (IDARC2).”
Technical Report NCEER-88-0035, Department o f Civil Engineering, State
University of New York at Buffalo, Buffalo, NY, Sept. 7, 1988.
21. Woodward-Clyde (1997). “Development Suites of Time Histories, Draft

Report.” Subcontractor: Woodward-Clyde Federal Services Pasadena, CA.
htt~://ciuiver.eerc.berkelev.edu:8080/studies/svstem/~ound
motions.htm1.
March 21,1997.
22. Trifunac, M.D. and Brady, A.G. (1975). “A Study of the Duration of Strong
Earthquake Ground Motion.” Bulletin of the Seismological Society of
America, 65, EI Cerrito, CA.
@Seismicisolation
@Seismicisolation
~
Table 2. Fundamental Period of Case Study Buildings.

1 Description 1 1 1
Rigid Diaphragm Flexible Diaphragm Flexible/Rigid 1
(sec.) (sec.)
s-story, 2:1 0.51 0.53 1.O4
5-Story, 3:l 0.61 0.66 1.os
3-Story, 2:1 0.34 0.37 1.o9
3-StOV, 3:l 0.40 0.48 1.20
@Seismicisolation
@Seismicisolation
Flex (mm) FEMA 273

Description
Classification
5-Story, 2: 1, 2ndFloor Rigid
5-StOrv. 3:l. 2ndFloor 61 77 Rieid
3-Story, 2:1, 2ndFloor 12 28 Rigid
3-Storv, 3:1, 2ndFloor 41 61 Rigid
@Seismicisolation
@Seismicisolation
33.5 m
(11 0 - 0 )
CR-1__.
CR-2-
CR-3-
CR-4 -
CR-5- 8 Q 9.14 rn (30'-O)

= 73.15 I (240'-o"j
CR-6-
CR-7 -
CR-8-
79.14 m
(30'-O)
CR-9-
Fig. 1. Floor Plan of 2:1 Aspect Ratio Case Study Building.
@Seismicisolation
@Seismicisolation
127 mm 127 mm
(5") (5") 1
+ T p : m m
(22")
305 rnrn i 21 9 mrnJ

(12") (48")
Floor Beam Cross Section Floor Girder Cross Section 203 mm
(8")
559 mm
(22")Fmrn (22")
Column Cross Section

icû4! -+ 1219 mm
(48")
Roof Girder Cross Section

rnm
(16)
Shear Wall Cross Section
Fig. 2. Typical Reinforcing Bar Layouts for Case Study Buildings.
@Seismicisolation
@Seismicisolation
-a
r;
5 1
I h
-
-+3 Story 3:l
-
3 story 2:l
+- 5 Story 3:l
5 story 2:l
o1 I
O 1 2 3
m- Factor
(a) Immediate Occupancy (DM)Level
-
-+- 3 Story 3:l
-
3 story 2:l
+- 5 Story 3:1
5 story 2:l
-u
o1 I
O 1 2 3 4
m- Factor
(b) Life Safety (LS) Level
-
>
4
-
-+ 3 Story 3:l
-
9 3 3 story 2:l
b -A- 5 Story 3:1
9 2 5 story 2:l
v1
1
o! I I I
O 1 2 3 4 5
m- Factor
(c) Collapse Prevention (CP) Level
Fig. 3. Maximum Middle Frame Column Moment DCR’s, U P .
@Seismicisolation
@Seismicisolation
5 yc
3 3
i?
4
-
-A-
3 story 2 1
5 Story 3:l
8 2 -)c 5 Story 2:l
CA
1 -IM Level
o1 I
O 1 2 3 4
m - Factor
(a) Immediate Occupancy (IM)
Level
-- 3 Story 3:1
3 story 2:l
t-5 story 3:1
5 story 2:l
-i-+
-Ls
o ! I
O 1 2 3 4 5
m - Factor
(b) Life Safety (LS) Level
-
t3
-A-
Story 3:l
3 story 2:l
5 Story 3:l
+5 story 2:1
-CP
o1 I l
O 1 2 3 4 5 6
m - Factor
(c) Collapse Prevention (CP) Level
Fig. 4. Maximum Average Middle Frame Column Moment DCR’s, LDP.
@Seismicisolation
@Seismicisolation
-3Story2:l
+5 Story 3: 1
-t5 Story 2 1
IM Level
. ......
- 0
0.000 0.005 0.010 0.015 0.020 0.025
Plastic Rotation (rad.)
Fig. 5. Middle Frame Column Rotation, NDP, for All Performance Levels.
@Seismicisolation
@Seismicisolation
SP 209-36
Behavior of RCS Beam-Column-Slab

Subassemblies Under
Earthquake-Type loading
by X. Liang, G. J. Parra-Montesinos, and J. K. Wight
Svnor>sis:Test results of four composite RCS beam-column-slab subassemblies

under cyclic loading are presented in this paper. The specimens were designed for
beam plastic hinging and controlled joint deformations through the use of a defor-
mation-based design model for RCS connections. Two types of simple connection
details were used in the specimens, one with overlapping U-shaped stirrups pass-
ing through the steel beam web, and the other with steel band plates wrapping
around the column just above and below the steel beam flanges. The performance
of the test specimens was evaluated in terms of load-displacement response, beam
rotations, joint deformations and energy dissipation capacity. The accuracy of the
deformation-based joint model was also evaluated. Excellent response was ob-
served for all specimens, with large beam rotations, minor to moderate joint dam-
age, and significant energy dissipation capacity, indicating that RCS frame con-
struction is adequate for use in zones of high seismicity. Good agreement was also
found between predicted and experimental shear force vs. shear deformation re-
sponse in RCS connections.
Kewvords: beam-column connections; composite

structures; earthquake-resistant structures; hybrid
structures; joint she ar d ef o rmat¡ o n; perf o rma n c e- based
design; RCS frames
@Seismicisolation
@Seismicisolation
661
662 Liang et al.
Xuemei Liang is a Ph.D. candidate at the Dept. of Civil and Environmental

Engineering Universiq of Michigan, Ann Arbor. Her research interests include
earthquake resistant design of composite structures.
Gustavo J. Parra-Montesinos is an assistant professor of Civil Engineering at the

Universityof Michigan, Ann Arbor, where he received his Ph.D. in 2000. He is a
member of ACI Committees 335 and 352. His research interests include seismic
behavior and rehabilitation of reinforced concrete and hybrid steel-concrete structures.
James K. Wight, FACI, is aprofessor of Civil Engineering at the University of

Michigan, Ann Arbor. He is the Chairman of ACI Committee 318 and member of
Committees 352 and 445. His research interests include earthquake resistant design
and rehabilitation of reinforced concrete and composite structures.
INTRODUCTION
Hybrid RCS frame systems that consist of Reinforced Concrete (RC)

columns and Steel (S) beams have gained popularity in the past twenty years.
These systems combine the advantages of both RC and steel frame structures.
While steel floor systems have the advantage of lightweight and easy
construction, RC columns have been shown to be significantly more cost
effective in terms of axial strength and stiffness compared to steel columns (1).
The efficiency of RCS fiame systems in terms of construction and

economy has led to increasing attention from researchers and practitioners.
Extensive research programs on RCS systems have been conducted during the
past twenty years in the United States and Japan, most of them focusing on
understanding the force transfer mechanism within the composite connection. In
the U.S., studies at the University of Texas at Austin (2,3) and Cornell
University (4) led to the development of design guidelines for RCS joints in low
to moderate seismic regions (5). More recently, a research program was
undertaken at the University of Michigan to study the behavior of exterior RCS
connections under seismic loading (6). Test results from this research program
showed that RCS joints designed following a strength-based approach may
experience excessive deformations, leading to severe joint damage and large
story drifts. Thus, a deformation-based model was developed for estimating the
shear force vs. shear distortion response in order to control joint deformations
and damage in RCS joints (7). To evaluate the seismic behavior of RCS beam-
@Seismicisolation
@Seismicisolation
column-slab subassemblies designed for controlled joint deformations and
following a strong c o l d w e a k beam philosophy, a new research program was
undertaken at the University of Michigan that consisted of the testing of four
RCS beam-column-slab subassemblies under reversed cyclic loading.
In this paper, results from the experimental program are presented in

terms of load-displacement response, joint deformations, beam rotations, energy
dissipation capacity, and stifiñess degradation. In addition, the ability of the
deformation-based model to limit joint distortions, and thus joint damage, is
evaluated.
RESEARCH SIGNIFICANCE
The research described in this paper provides information on the seismic

performance of RCS beam-column-slab subassemblies with connections
designed for controlled deformations. This research contributes to a more
rational design of RCS connections and provides new data that support the use
of RCS frame construction in regions of high seismicity.
DEFORMATION AND RESISTANCE MECHANISMS IN RCS

CONNECTIONS
Two types of joint deformations have been identified in RCS

connections (3), Joint Shear Deformation and beam Rigid Body Rotation. Joint
Shear Deformation (JSD) is similar to that of typical RC and steel connections,
and Rigid Body Rotation (RBR) of the steel beam within the column is caused
by high bearing stresses in the column regions just above and below the beam
flanges. The summation of JSD and RBR is the Total Joint Distortion (TJD) in
an RCS connection. Large joint deformations in RCS frames can lead to severe
joint damage as well as excessive story drift, and therefore should be avoided in
seismic design.
The shear strength of RCS connections is provided by the summation of

shear resistance of steel beam web, inner concrete strut and outer concrete panel
in the connection region (3,7), as shown in Fig. i(a), (b), and (c), respectively.
The inner concrete strut is surrounded by the beam flanges and face bearing
plates (FBPs) at the column faces, while the outer concrete panel is mobilized
outside the steel beam flanges through the use of shear keys, such as embedded
steel columns and/or steel band plates. Design equations for these mechanisms
were proposed in the ASCE design guidelines for RCS connections (5).
However, these equations were developed primarily based on strength equations
for structural steel and reinforced concrete beam-column connections without
@Seismicisolation
@Seismicisolation
664 Liang et al.
explicitly considering joint deformations. From a research program conducted at
the University of Michigan during the late 1990s (6), it was shown that exterior
RCS connections designed following the ASCE guidelines could experience
excessive deformations under cyclic loading, and have lower strength than
predicted. Therefore, a new deformation-based joint model was developed for
both interior and exterior RCS connections subjected to large load reversals.
This analytical model can be used to estimate the envelope of cyclic shear force
vs. shear deformation response in RCS connections. In addition, the strains in
the concrete panel and steel transverse reinforcement can also be estimated.
The deformation-based model for RCS connections developed by Parra

and Wight (7) is based on the state of strain in the connection loading plane by
defining a ratio, k,c, between the principal tensile and compressive strains. The
ktc ratio is influenced by connection details and types (interior and exterior).
Once the state of the strain in the connection is determined for a given shear
deformation, the shear resistance from the inner and outer concrete struts, as
well as steel web panel, can be estimated. The total shear force vs. shear
deformation response in the composite connections is predicted for joint shear
deformations up to 1.2% (0.012 rad), a deformation at which ultimate joint
strength is assumed to be developed. Detailed information about this model can
be found elsewhere (7).
Based on the deformation-based model for RCS connections, design

equations were developed for RCS connections for moderate damage and
ultimate strength level, which approximately correspond to 0.5% and 1.2% joint
shear deformation,respectively (8).
As mentioned earlier, the total joint shear strength is the summation of

shear resistance from the steel beam web, and inner and outer concrete struts.
The horizontal shear strength of the steel beam web, Vwh, can be calculated as
follows:
fY
VWh= k, - t,h,
J?
where kw = 0.9 and 0.8 for interior and exterior joints, respectively, f y and tw
are the yield strength and thickness of the steel web panel, respectively, and h,
is the column depth. The shear strength of the inner and outer concrete struts, K h
and Voh,respectively, can be obtained from the following equations:
yh=kif,'(-0.0048J;,'
+1.13)h,(bf - t,)
y,,, =k,f,'(-O.O048J;.'+1.13)hcb,, (3)
@Seismicisolation
@Seismicisolation
where k, and ko are strength factors, and bf and bo are the width of the beam
flanges and outer concrete panel, respectively. Procedures for estimating the
outer panel width, bo, can be found elsewhere (7). In Eqs. (2) and (3), fc must
be expressed in MPa. Values of ki and ko for 0.5% and 1.2% shear deformation
levels have been previously developed (8), and are listed in Table 1.
The total joint shear strength for the design joint shear deformation must
be larger than the maximum shear force transferred to the joint by the adjoining
members. Once the joint shear strength is estimated, the bearing strength of the
composite connection can be checked according to the ASCE design guidelines.
Because joint shear failures are more stable than bearing failures in RCS frames
(4), the joint bearing strength must be larger than the joint shear strength.
EXPERIMENTALPROGRAM
The experimental program consisted of the lateral load testing of two

interior and two exterior RCS beam-column-slab subassemblies at the
University of Michigan. RC columns, with a length of 2240 mm and a 380 mm
square cross section, had similar dimensions and reinforcement layout for all
specimens. The RC columns, which were designed according to the AC1
Building Code (9), had longitudinal reinforcement of 12#6 (19 mm) bars,
corresponding to approximately 2.4% of the column gross area. All four
specimens had W12x16 (305 mm deep) steel beams, supporting a 90 mm RC
slab. Shear studs were welded on the top flange of the steel beams so that full
composite behavior (10) could be achieved between the steel beam and the
concrete deck. Minimum reinforcement as required by the AC1 Building Code
(9) was used in the RC slab. The W12x16 steel beams had width-thicknessratios
of 7.5 for the flange and 49.4 for the web. The flange width-thickness ratio
slightly exceeded the limiting ratio of 7.4 given in the AISC seismic provisions
(1 1). W6x16 (150 mm deep) steel sections, welded to the beam flanges, were
embedded in the RC column. These light steel columns are commonly used in
RCS construction for erection purposes and shear transfer within the connection
region. A description of the four test specimens is given in Table 2.
Connection Details
Two types of simple connection details, which performed well in prior

experimental investigations, were used in this test program. The connection
detail shown in Fig. 2(a) consists of overlapping U-shaped stirrups passing
through holes in the steel web panel, and several layers of closely spaced
stirrups in the column region just above and below the steel beams for
@Seismicisolation
@Seismicisolation
666 Liang et al.
confinement and shear transfer purposes. It is assumed that transverse beams do
not frame into the connection for this particular joint detail. The joint detail
shown in Fig. 2(b) is recommended for cases when transverse beams frame into
the connection. For this case, steel band plates, which wrap around the column
regions just above and below the steel beams, are used to provide confinement
to the connection region. The excellent confinement provided by the steel band
plates enables the total elimination of joint transverse reinforcement, as well as
the stirrups in the column just above and below the joint region (6). However,
attention needs to be paid to potential buckling of the column longitudinal bars
over their unsupported length through the joint. Therefore, layers of small ties,
anchored in the joint core without penetrating the steel beam webs, are provided
within the beam flanges as shown in Fig. 2(b). In this experimental program,
each connection detail was used in one interior and one exterior subassembly
(Table 2). Both joint details shown in Figs. 2(a) and (b) include small embedded
steel columns welded to the beam flanges, and 13 mm thick Face Bearing Plates
(FBPs) welded to the beam web and flanges at the column faces to mobilize the
concrete region within the beam flanges.
The connections of the two interior specimens were designed for a target
shear deformation of approximately 0.5%. For the two exterior specimens,
because the same composite beam was used as for the interior specimens, lower
joint deformations were expected as will be explained later.
Test setup, Loading History and Instrumentation
The test setup used for this experimental program is shown in Fig. 3. All
columns and beams were pin-connected at their ends to simulate inflection
points at midspan of the member. Lateral cyclic loads were applied at the top of
the column through a 450 kN hydraulic actuator. A small axial compression load
equal to approximately 4.0% of the column axial capacity was applied to the
columns through two hydraulic jacks, as shown in Fig. 3. It should be mentioned
that column axial load has been shown to have little effect on the seismic
performance of RCS connections (4). A total of twenty lateral displacement
cycles were applied to the specimens, with displacement amplitudes ranging
from 0.5% up to 5.0% story drift (ratio between lateral displacement and story
height). Each cycle to a new drift level was repeated once to study the specimen
stiffness and strength loss at that drift level. A load cell and an LVDT were
used to monitor the applied lateral load and displacement. Linear potentiometers
and inclinometers were used to measure joint deformations and beam rotations.
Strains in the steel beams, and column and slab reinforcement were monitored
through linear and rosette strain gages.
@Seismicisolation
@Seismicisolation
Material Properties
The concrete compressive strength obtained at test dates for the columns
in all specimens ranged fiom 39.2 to 49.7 MPa, and 31.5 to 44.8 MPa for the
slabs. Dual Grade A36/A572-Grade 50 steel with a yield strength of
approximately 320 MPa was used for the steel beams, and A36 steel was used
for FBPs and steel band plates. Grade 60 (420 MPa) deformed steel bars were
used for all the reinforcement in the columns and slabs.
Cracking Pattern and Lateral Load vs. Story Drift Response
Typical load vs. story drift responses for interior and exterior beam-
column-slab subassemblies are shown in Figs. 5(a) and (b), respectively, with
total drift corrected to account for deformations in the test setup. For the exterior
specimens, positive displacement corresponds to positive bending (slab in
compression) in the beam. The joint cracking pattern and beam deformation at
5.0% story drift for one interior and one exterior subassembly are shown in Figs.
4(a) and (b).
For the two interior subassemblies (Specimens 1 and 2), first diagonal
cracking occurred in the joint region at approximately 1.0% story drift. For
Specimen 2, which had beams in the transverse direction, diagonal cracks
originated from the tips of the bottom flange of the transverse beams. Flexural
cracks across the width of the concrete slabs were also observed at 1.0% drift.
Up to this displacement level, Specimens 1 and 2 behaved in the cracked-elastic
range. For larger drift levels, beam yielding started to occur, and at 3.0% drift,
after significant beam yielding had taken place, local buckling was observed in
the beam flanges and web (Fig. 4(a)). Both Specimens 1 and 2 reached the
maximum lateral load during the first cycle at 4.0% drift. During the second
cycle to 4.0% drift, local flange buckling became severe, leading to a drop in the
lateral strength of the subassemblies. New diagonal cracks continued to form in
the joint region up to 4.0% drift when maximum lateral loads were achieved, but
most of the joint cracking occurred before 3.0% story drift. The joint region in
Specimens 1 and 2 sustained only moderate damage at the end of the tests (Fig.
4(a)). Crushing of slab concrete also occurred during the 4.0% and 5.0% drift
cycles due to large beam rotations.
In terms of load vs. displacement response for the interior specimens

(Fig. 5(a)), slight pinching can be noticed in the hysteresis loops for the cycles
@Seismicisolation
@Seismicisolation
668 Liang et al.
below 3.0% drift because of joint diagonal cracking and beam rigid body
rotations within the joints. At larger displacement cycles, during which large
beam flexural deformations occurred, full hysteresis loops were observed. Both
Specimens 1 and 2 exhibited stable responses, retaining more than 75% of their
peak strength at the end of the tests.
For the two exterior beam-column subassemblies, Specimens 3 and 4,

little joint damage occurred due to lower shear forces transferred into the
connection compared to the interior subassemblies. For these two specimens,
hairline diagonal cracks started to form at approximately 1.O% drift in the joint
region, and several new diagonal cracks formed at about 1.5% and 2.0% drift.
No significant cracking occurred in the joint region for larger drift levels as the
beam started to yield, limiting the shear force transferred into the joint. In
Specimens 3 and 4, beam yielding was observed at 1.5% drift, and beam local
buckling began to take place at 3.0% drift. The cracking pattern and beam local
buckling of Specimen 4 are shown in Fig. 4(b). The load vs. story drift response
for the exterior subassemblies was not symmetrical, as shown in Fig. 5(b), due
to the presence of the concrete slab. The specimens were stronger for the
positive loading direction, for which the concrete slab was engaged in
compression. Because of the minor damage in the joint region, the overall
response of Specimens 3 and 4 was dominated by beam rotations, which led to
full hysteresis loops and excellent energy dissipation capaciîy. While a post-
yield strength gain can be observed in-the positive loading direction, drops in the
lateral strength for the negative direction were observed due to local buckling of
the steel beams during the 4-5% drift cycles.
Beam Moment vs. Rotation Response
In the test specimens, beam rotations were measured over a length equal
to 75% of the composite beam depth from the column face using linear
displacement transducers. A typical beam moment vs. rotation response for the
test specimens is shown in Fig. 6 . As shown in the figure, first yielding in the
composite beam occurred at approximately 1.0% drift. During the cycles
performed at 3.0% and 5.0% drift, beam rotations in excess of 0.01 rad and 0.03
rad were measured in the test specimens, respectively. The severe beam local
buckling at 4.0% drift caused a drop in the beam moment strength, and thus the
applied lateral load. The large beam rotations combined with the wide hysteresis
loops shown in Fig. 6 indicate that beam rotations contributed to a large
percentage of the energy dissipated in the test specimens.
@Seismicisolation
@Seismicisolation
Deformation of the Composite Connections
The joint shear deformation (JSD), beam rigid body rotation (RBR) and
total joint deformation (TJD) of the test specimens were measured using
inclinometers. Response envelopes for these joint deformations in Specimen 2
(with steel band plates) are shown in Fig. 7. Both JSD and RBR kept increasing
until the maximum lateral load of the specimen was achieved, remaining
approximately constant after the lateral load dropped due to beam flange local
buckling. For Specimen 2, maximum TJD was about 1.1%, with maximum JSD
being effectively limited to 0.5% as targeted, and 0.6% for RBR. At the end of
the test, the joint retained more than 30% of the stiffness at 1.0% drift. For
Specimen 1, which had standard joint details, the JSD was remained below 0.5%
up to 3.0% drift. The maximum TJD achieved for this specimen was about
1.4%, with 0.6% of JSD and 0.8% of RBR. For Specimens 3 and 4, TJD at the
end of the test was less than 1.0%, with JSD and RBR being approximately
0.3% and 0.6%, respectively.
Evaluation of the Shear Force vs. Shear Deformation Joint Model
The predicted lateral force vs. shear. deformation responses for the
interior and exterior RCS connections are shown in Figs. 8 (a) and (b),
respectively. For the two interior subassemblies, the total moment capacity of
the composite beam, considering material overstrength and a 10% increase due
to strain hardening, would lead to a maximum joint shear force of approximately
1100 kN (actuator lateral force of 160 kN). As can be seen from Fig. 8 (a), the
predicted joint shear deformations at this force level were 0.7% and 0.4% for
interior connections with standard details (Specimen 1) and steel band plates
(Specimen 2), respectively. For the two exterior connections, a maximum
transferred joint shear force of approximately 700 kN (lateral load of 100 kN)
was estimated. Predicted joint shear deformations were approximately 0.3% and
0.2% for exterior connections with standard details (Specimen 3) and steel band
plates (Specimen 4), respectively, as shown in Fig. 8 (b). Comparing these
predicted JSD with the maximum values obtained during the tests as explained
in the previous section, it is shown that the shear force vs. shear deformation
model developed by Parra and Wight predicts joint shear deformation with
reasonable accuracy.
A comparison between the predicted and experimental joint shear force

vs. shear deformation response for Specimen 1 is shown in Fig. 9. At 0.5% joint
shear deformation, the difference between predicted and measured lateral force
was less than 5%. For Specimen 2, this difference was approximately 10% at
0.5% JSD. For the exterior connections, similar results were obtained.
@Seismicisolation
@Seismicisolation
670 Liang et al.
A typical joint strength contribution from the steel beam web and inner
and outer concrete struts is shown in Fig. 10. Contribution from the steel beam
web was about 40% of the total shear strength of the composite connection at
0.2% JSD, and decreased to approximately 30% at ultimate strength level (1.2%
JSD). The contributions from inner and outer concrete struts were about 30% of
total shear strength each at 0.2% JSD, and they increased to 35% each at
ultimate strength level.
Energy Dissipation Capacity
The energy dissipation capacity of the test specimens was studied by

evaluating the ratio between the energy dissipated during each displacement
cycle i, ED¡, and the energy dissipated by a perfect elasto-plastic system, b p i
(using initial stiffness of the specimen and strength corresponding to cycle i).
Fig. 11 shows the energy dissipation ratios for Specimens 2 and 4, which are
representative of the interior and exterior subassemblies, respectively. The
results show that RCS beam-column-slab subassemblies possess excellent
energy dissipation capacity, with energy dissipation ratios for the first cycle
performed between 2.0% and 5.0% drift ranging from 40% to 55% for interior
subassemblies, and from 55% to 65% for exterior subassemblies.For the second
cycle performed at drifts equal to or larger thag 2.0%, energy ratios ranged from
27% to 45%, and from 46% to 56% for interior and exterior subassemblies,
respectively. As most of the energy was dissipated through plastic deformations
in the steel beams, the energy dissipation ratio for RCS subassemblies is similar
to that of steel beam-column subassemblies,and significantly higher than that of
RC subassemblies. Because the steel beam in RCS subassemblies runs
continuously through the connection, the risk of joint weld fracture, which is
often of concern in steel connections, is eliminated.
Stiffness Degradation
To evaluate the stiffness degradation of RCS beam-column-slab

subassemblies under cyclic loading, the effective stiffness for the test specimens
was calculated for each displacement cycle by dividing the lateral force at
maximum drift in each cycle over the peak lateral displacement measured at that
particular cycle. For exterior subassemblies, the effective stiffness was
significantly different for the positive and negative loading directions for which
the RC slab was in compression and tension, respectively. Typical results for
one interior and one exterior subassembly are shown in Fig. 12. At 5.0% drift,
the stiffness of the interior subassembly (Specimen 1) was approximately 25%
of the stiffness at 1.0% drift, when the specimen was behaving in the cracked-
@Seismicisolation
@Seismicisolation
elastic range. For the exterior subassembly shown in Fig. 12 (Specimen 4), both
positive and negative stiffness at the end of the test were approximately 20% of
the measured 1.O% drift stiffness. However, most of the degradation of effective
stiffness was a result of large beam plastic rotations and local buckling, rather
than joint deterioration.
Story Drift Components
Four major mechanisms contribute to the total story drift' in RCS

subassemblies: 1) beam rotations, including elastic and plastic rotations, 2)
column rotations, 3) joint shear deformations, and 4) beam rigid body rotations.
The drift contribution from these four components was calculated for ail four
specimens. Figs. 13(a) and (b) show typical percentage contributions from these
four mechanisms for an interior and exterior subassembly, respectively. The
calculated drift values were all within 8% of actual story drift, as can be
observed in the figures.
As shown in Figs. 13(a) and (b), beam rotations contributed the most to
total story drift. For interior subassemblies, it represented about 50-60% of total
drift up to 3.0% drift. At 5.0% story drift, the beam contribution increased to
approximately 75% of the total -drift due to significant beam plastic
deformations (Fig. 13(a)). At this lateral displacement level, the RBR and JSD
accounted for approximately 15% and 7% of the total drift, respectively, column
rotations contributing less than 5% of the total drift. For the exterior
subassemblies, the response was different for the positive and negative loading
directions because of the asymmetry in the composite beams. In the positive
direction, for which the slab was in compression, the component contribution
was similar to that of interior joints. In the negative direction, however, beam
rotations contributed to up to 95% of the total story drift for the 5.0% drift cycle,
as shown in Fig. 13(b).
This paper reports results from the testing of four RCS beam-column-
slab subassemblies under large load reversals. The specimens were designed for
beam plastic yielding and controlled joint damage through the use of'a
deformation-basedjoint model. Two simple joint details were used. Joints with
overlapping U-shaped stirrups passing through the steel beam web were used
when transverse beams were assumed to be framed outside the connection
region. Steel band plates wrapping around the column regions just above and
below the steel beam flanges were used when transverse beams framed into the
@Seismicisolation
@Seismicisolation
Liang et al.
connection. All specimens showed stable load vs. displacement response with
inelastic activity dominated by beam yielding. Beam rotations larger than 0.03
rad were measured during all tests, leading to substantial energy dissipation
capacity in the specimens. The connections sustained only minor to moderate
damage during the tests, with joint shear deformations approximately equal to or
less than 0.6% (0.006 rad), and total joint deformations less than 1.4%. In
addition, the deformation-based design model accurately predicted the joint
shear force vs. shear deformation response for the test specimens. Experimental
results indicate that RCS frame structures with simple joint details exhibit
excellent seismic response and are suitable for use in regions of high seismicity.
ACKNOWLEDGEMENT
This research was sponsored by the National Science Foundation under

the grant MS-9710204 and the University of Michigan. The opinions expressed
in this paper are those of the writers and do not necessarily reflect the views of
the sponsors.
REFERENCES
1. Griffis, L., (1986). “Some Design Considerations For Composite-Frame

Structures,” AZSC Engrg., 23(2), 59-64.
2. Sheikh, T. M., Yura, J. A., and Jirsa, J. O., (1989). “Beam-Column Moment
Connections for Composite Frames: Part 1,”JStruc. Engrg., ASCE i 15(1 i), 2858-
2816.
3. Deierlein, G.G., Yura, J. A., and Jirsa, J. O., (1989). “Beam-Column Moment
Connections for Composite Frames: Part 2,” J. Struct. Engrg., ASCE 1 1 5(1 i),
2811-2896.
4. Kanno, R., (1993). “Strength, Deformation, and Seismic Resistance of Joints

Between Steel Beams and Reinforced Concrete Columns, ” Ph.D. thesis, Cornel1
University, Ithaca, NY.
5. ASCE Task Committee on Design Criteria for Composite Structures in Steel and
Concrete, (1994). “Guidelines for Design of Joints Between Steel Beams and
Reinforced Concrete Columns,” J. Struct. Engrg., ASCE 120(8), 2330-2357.
6. Parra-Montesinos, G., and Wight, J. K., (2000). “Seismic Response of Exterior RC

Column-To-Steel Beam Connections,”J. Struc. Engrg., ASCE 126( IO), 1 1 13-1 121.
@Seismicisolation
@Seismicisolation
7. Parra-Montesinos, G., and Wight, J. K., (2001a). “Modeling Shear Behavior of
Hybrid RCS Beam- Column Connections,” J. Struc. Engrg., ASCE 127( i), 3- 11.
8. Parra-Montesinos, G., Liang, X., and Wight, J. K., (2001b). “Towards

Deformation-based Design of RCS Beam-Column Connections,” Us - Japan
Seminar on Advanced Stability and Seismicity Concept for Performance-based
Design of Steel and Composite Structures, Kyoto, Japan.
9. AC1 Committee 3 18, (1999). “Building Code Requirements for Structural

Concrete, ” Report No. AC1 318-99, American Concrete Institute, Farmington Hills,
Michigan.
10. AISC, (1997). “Loud & Resistance Factor Design,” American Institute of Steel
Construction, Inc., Chicago, Illinois.
11. AISC, (1997). “Seismic Provisions for Structural Steel Buildings, ” American
Institute of Steel Construction, Inc., Chicago, Illinois.
@Seismicisolation
@Seismicisolation
674 Liang et al.
Table 1. k Factors Recommended for RCS Connections
Connection Detail
Target shear
k factor Standard Steel Band Plates
deformation
~
ki 0.25 0.17 0.32 0.24

0.5%
k0 0.14 0.09 0.17 0.12
ki 0.32 0.21 0.40 0.29
1.2%
k0 0.17 0.1 1 0.22 0.15
Table 2. Specimen Description
Interior subassembly, overlapping U-shaped stirrups, no

Specimen 1
transverse beams.
Interior subassembly, steel band plates, no joint transverse
Specimen 2
reinforcement, transverse beams.
Specimen 3 Exterior subassembly, same details as in Specimen 1.
Specimen 4 Exterior subassembly, same details as in Specimen 2.
@Seismicisolation
@Seismicisolation
(a) Steel web panel
I
f
dbeam
(b) Inner concrete strut
dbeaml
(c) Outer concrete strut
Fig. 1 RCS joint shear resistance mechanisms

(Not all external forces are shown)
@Seismicisolation
@Seismicisolation
676 Liang et al.
(a) RCS joint with U-shaped stirrups

Embedded
Steel Column
Small
(b) RCS joint with steel band plates
Fig. 2. Proposed RCS joint details

450 kN Actuator
Fig. 3. Test setup
@Seismicisolation
@Seismicisolation
(a) Specimen 1 (b) Specimen 4

Fig. 4. Cracking pattern and beam deformation at 5% drift
izo , , , , , , . , , , , , . , , , ,
=QvRiff m
(b) Exterior Subassembly (Spec. 4)
Fig. 5. Lateral load vs. displacement response
@Seismicisolation
@Seismicisolation
678 Liang et al.
Fig. 6. Beam moment vs. rotation Fig. 7. Joint deformation envelopes

Response (Specimen 1) (Specimen 2)
290, , , I , . , , , , , 7
m- Spec. 4
o am am aca aca am aoiz am4
(a) Interior connections (b) Exterior connections
Fig. 8 Predicted lateral load vs. joint shear deformation responses
@Seismicisolation
@Seismicisolation
.. .. .. ..
.
..................................................
. ..... e
. < <
l
.-cO
2 0.8
2
0 0.6
5
0.4
e!
a2
0.002 0.004 0.008 0.008 0.01 0.012

JSD imd)
Fig. 9. Predicted vs. experimental Fig. 10. Predicted percentage load

lateral load vs. joint shear deform. contribution from shear strength
response for Specimen 1 mechanisms for Specimen 1
a7 I ' ! ' ! . ! 5 . , . ! . , , $ . I .
' -
................. j ........... .-
o3 _
02 ..........................
................
i ...............*Spec4l*cyde."
~
2
1 i ...........
...........................
.........
aiL I
I 5 1 I
o ~ " I ' I ' I ' ' '
2 3 4 5 O 1 2 3 4 5 6
saoiy Driff (W stay Riff (W
Fig. 11. Energy dissipation ratio Fig. 12. Stiffness degradation
@Seismicisolation
@Seismicisolation
680 Liang et al.
12 - - !!!= L I 1.2 L -
1 ........ 1
I
C
O
‘a
5 a8
‘a = û8
P
a 5
a
5 a6
C
8 û6
o
O
*s a4 F
n
su
a2 a2
O O 4
1 1.5 25 5 25 5
a) Interior subassembly (Spec. 2) b) Exterior subassembly (Spec. 4)

(Negative Loading)
Fig. 13. Drift components for RCS subassemblies
@Seismicisolation
@Seismicisolation
SP 209-37
Performance of Sacrificial Exterior

Shear Keys Under Simulated
Seismic Loading
by P. F. Silva, S. Megally, and F. Seible
Svnopsis: Performance evaluation of sacrificial exterior shear keys under simulat-

ed seismic loading is presented in this paper. The performance evaluation was con-
ducted in terms of damage levels observed during testing. The experimental pro-
gram consisted of six experiments, and the variables investigated during testing
were; (1) inclusion and exclusion of back and wing walls to determine influence of
these walls in the seismic response of the shear keys, (2) adoption of different de-
sign details such as the use of sacrificial flexural keys, and (3) prestressing of the
abutments with the main objective of decreasing damage propagation into the
abutment walls. These experiments provided results that were used to make realis-
tic assumptions of the load-deformation response of sacrificial exterior shear keys
as well as their post-peak performance under cyclic loads. Experimental and ana-
lytical results were also used to develop a hysteretic model for exterior sacrificial
shear keys at the abutments. The development of this hysteretic model is also dis-
cussed in the paper.
Kewvords: abutments; capacity protection; sacrificial

shear keys; seismic performance; shear friction; shear
sliding
@Seismicisolation
@Seismicisolation
681
682 Silva et al.
ACZ Member Pedro Silva is an Assistant Professor of Civil Engineering at the
University of Missouri, Rolla. He received his Ph.D. from the University of
California, San Diego. His research interests include the development of
innovative performance-based procedures for the design and use of fiber
reinforcement polymers for the structural rehabilitation of civil structures.
ACZMember Sami Megally is an Assistant Project Scientist in the Department of

Strucîural Engineering at the University of California, San Diego. He received his
Ph.D. from the University of Calgary, Canada. He is a member of AC1 Committee
421.His research interests include the seismic performance of bridges, long-term
concrete stresses and deformations and punching shear of slabs.
ACZ Member Frieder Seible is a Professor of Structural Engineering in the

Department of Structural Engineering and Executive Associate Dean at the
University of California, San Diego. He received his Ph.D. from the University of
California, Berkeley. His research interests include seismic design and
rehabilitation,use of fiber reinforcementpolymers in civil engineering structures.
INTRODUCTION
Current design procedures stipulate that damage to abutments under a major
seismic event is admissible provided that any damage in the abutments does not
'.
result in collapse of the bridge or unseating of the superstructure In addition,
seismic loads should not control the number of piles required at the abutments . '
Thus, in order to control damage to the abutments and their supporting piles, the
transverse seismic input force should be limited by construction of sacrificial
shear keys at the abutments. Transfer of seismic loads to the abutments is
controlled by designing sacrificial shear keys such that the ultimate capacity of
the shear keys does not exceed 75 percent of the total shear capacity of the piles
plus one of the wing-walls '.
A critical design aspect of sacificial shear keys is the determination of their peak
capacity; therefore, it is of utmost importance to discern that assumptions
regarded as conservative for the design of sacrificial shear keys, are actually non-
conservative. Another design aspect of shear keys is their residual capacities and
behavior under cyclic loading. In addition, protection of the abutment walls
against severe damage following an earthquake must also be considered in the
detailing of these keys in order to avoid full reconstruction of the bridge
abutments after an earthquake. These design aspects are discussed in this paper
based on experimental results of six sacrificial exterior shear keys.
@Seismicisolation
@Seismicisolation
TESTSPECIMENS
The experimental program consisted of six exterior shear keys designed at a 2 6
scale and divided in three test series (see Figs. 1 to 3). Test series I consisted of
II
two test specimens that were built
on the same foundation support
and the details employed in the
construction of these test units un,,lB
conformed to existing as-built
details.
As shown in Fig. 1, specimen 1A
was built without abutment walls,
and specimen 1B was constructed
A b u m m i , ~
with the abutment back and Wing Stemwail Found,vv..
walls. SUPPOfl
Test series II specimen 2A was a -

Fig. 1 Test Specimens Test Series I
redesign of specimen 1A with the
difference that in specimen 2A
the shear key was cast against the
hardened and smooth concrete
surface of the abutment stem
wall. As an alternative detail in
the construction of shear keys,
specimen 2B was designed to
display a more predominant
flexure-shear response. Fig. 2
depicts series II test units and the
position of load application
during testing.
Test series III, specimens 3A and -
Fig. 2 Test Specimens Test Series II
Construction Construction
3B were a redesign of specimen
1A with the difference that in
specimens 3A and 3B the
abutment stem wall was post-
tensioned and a smooth
construction joint was imposed at
the interface between the shear
key and the stem wall. Fig. 3
depicts series III test units and the
position of the lateral load. A
similar test setup was used for ail
the test series and the typical test
setup is shown in Fig. 4, which
depicts test series I. -
Fig. 3 Test Specimens Test Series III
@Seismicisolation
@Seismicisolation
684 Silva et al.
TESTSETUP
As shown in Figs. 4 and 5,
typical features of each test
series included: (1) two test
units that were cast on the
same foundation support, (2)
all the test units had the same
dimensions to the exception of
internal reinforcement details
and construction phases, (3)
the same test setup was used
for all the test series, (4) the
lateral load was applied by two
servo-controlled hydraulic
actuators connected to a
loading fiame, and ( 5 ) the
loading fiame was restrained
against twisting and rotation
by two frames one at each end
of the loading fiame.
(Depicted is Test Series I)

REINFORCEMENT
DETAILS
As shown in Figs. 5 and 6, the foundation block was post-tensioned to the strong
floor by using ten tie-downs placed on the sides of the shear keys and one central
tie-down to the top of the stem wall. In a typical Caltrans bridge, the
superstructure is constructed over square steel bearing pads spaced along the
abutment stem wall. The internal tie-down bar (see Fig. 5) was utilized in order
to generate a downward force on the abutment stem wall corresponding to the
bridge superstructure tributary load to the bearing pads adjacent to the exterior
keys.
As previously described, series I test units conformed to existing Caltrans

abutment as-built bridge design details. in series I, the U-shaped shear
reinforcement consisted of 4 rows of 6- #3 bars (see Fig. 6). in addition, the
horizontal reinforcement at the top level of the stem wall, which is often defined
as the tension tie 2, consisted of a single layer of 5-#3 bars, conforming to the
minimum requirements specified by Caltrans. The horizontal and vertical
reinforcement on the sides of the shear keys consisted of #3 bars at 4.7511.
(121mm) on centers. For specimen lB, the shear key and the abutment walls were
coupled together internally with #3 bars at 4.75in. (121mm) on centers matching
an integral as-built abutment wall.
@Seismicisolation
@Seismicisolation
for h i d e of W-Scction
-
@) Typical Test Setup Side View
Fig. 5 Typical Test Setup (Depicted Test Series I)
Fig. 6 Reinforcement Layout - Test Series L
@Seismicisolation
@Seismicisolation
686 Silva et al.
Series I tests have shown that extensive damage is likely to occur in the abutment
walls after an earthquake. In order to avoid extensive damage levels into the
abutment walls, new design alternatives were suggested for the construction of
series II units. In specimen 2A the shear key was cast against the hardened and
smooth concrete surface. Before casting of the shear keys the hardened stem wall
concrete top surface was lubricated with hydraulic oil to act as a bond breaker. in
order to reduce the amount of reinforcement crossing the horizontal plane
between the shear key and the abutment stem wall, only the U-shaped 24- #3 bars
(see Fig. 7) crossed this plane. In addition, the tension tie steel was changed to a
single layer of 6-#3 bars, which was designed to ensure that a sliding-shear failure
would develop; however, test results show that significantly higher amounts of
steel are required.
8? .-.
Fig. 7 Reinforcement Layout - Test Series II
Thick
Tie
Fig. 8 Reinforcement Layout - Test Series III
@Seismicisolation
@Seismicisolation
Specimen 2B was designed to display a more predominant flexure-shear response
by placing the applied load at the top of the key and reducing the width of the
shear key to 8in. (203mm). As shown in Fig. 7(b), the key was reinforced with
18-#5 longitudinal bars, which were embedded into the foundation support. Near
the stem wall-shear key interface the horizontal reinforcement consisted of #3 ties
at 1.7511. (44.5mm)on centers. in specimen 2B the tension tie steel was changed
to two layers of 6-#3 bars for a total of 12 bars.
Redesign of specimen 2A was not adequate in preventing excessive damage into
the abutment walls. As a result, a few modifications were implemented in series
III test units (see Fig. 8). These modifications were: (1) the abutment stem wall
was post-tensioned horizontally with 309kips (1,375kN) using two un-bonded
1.75in. (44.5mm) high-strength bars running inside two ducts with an inside
diameter of 2in. (50.8mm), (2) in specimen 3B the construction joint was
positioned 3in. (76.2mm) above the shear key-stem wall interface, (3) the shear
key reinforcement was modified to straight 8-#5 headed bars embedded into the
foundation support, (4) in specimen 3B the shear key reinforcement was coupled
at the shear key-stem wall interface with mechanical splices (couplers), and ( 5 )
the tension tie consisted of 4-#5 headed bars and 4-#5 straight bars. Other
reinforcement details are shown in Fig. 8.
MATERIALPROPERTIES
The concrete used in the construction of the specimens was specified to have a
minimum 28-day compressive strength of 3,250psi (22.4MPa) with an aggregate
size of 0.50in. (13mm). The cylinder day-of-test strength results for all the test
specimens were obtained Com three cylinders and are presented in Table 1.
Tensile tests were performed on three, 36in. (914mm) bars to determine their
properties. Reinforcement tensile test results are presented in Table 2.
Table 1 Concrete Table 2 Reinforcement Material Pro erties
Material Properties
Series
I Specimen 1 Day-of-Test ¡ Designation
ksi MPa
Series I #3 65(448) 98(676)
#3 84(579) 124(855)
Series II
#5 65 14481 105í7241
#3 I 79(545) I 119(821)
#5 I 72í497) I 99í683)
DESCRIPTION OF DAMAGE LEVELS
Observations recorded during testing were used to characterize the load

deformation response, and establish damage levels for exterior shear keys
subjected to seismic loads. Description of five damage levels is presented in Table
3 and these levels are described next.
@Seismicisolation
@Seismicisolation
688 Silva et al.
Table 3 Damage Levels Defmition
LEVEL I I
DamageEvahation Performance Evaluation I
I I NOREPAIR I CRACKING I
II I POSSIBLEREPAIR I YIELDING I
iII I MINIMUM REPAIR I INITIATION OF LOCAL MECHANISM
~
I
FULL DEVELOPMENT OF LOCAL
N REPAIR MECHANISM
V I REPLACEMENT I STRENGTH DEGRADATION I
LEVEL I: This damage level was characterized by onset of cracking, which are
barely visible and close after load removal. LEVEL II: This damage was defined
when the reinforcement has yielded, and was quantified visually by cracks that
were clearly visible but most likely would not require repair. LEVEL III: This
damage level was characterized by the onset of concrete spalling or large open
cracks, which are typically good indicators of local damage, and would require
minimum repair. LEVEL N: Qualitatively, this level was described when cracks
and spalling extended over the full region of the shear keys. Depending on the
prevalent failure mechanism, damage at this level would require some form of
major repair. LEVEL V: This damage level was defined when the load carrying
capacity of the shear key is compromised due primarily to fiacture of the
reinforcement, and would require replacement or major repair of the bridge
abutments.
ANALYTICAL MODELS
STRUT-AND-TIE MODEL
ANALOGOUS
According to this model the shear key capacity was calculated according to a
strut-and-tie mechanism. The reinforcement is assumed to act as a tension
member, or tie, with the concrete acting as compressive struts *.Using this model
the capaciîy of the key was computed as:
Where: VCis the concrete contribution to the shear strength of the key, as a result
of aggregate interlock and dowel action along the inclined crack, and is expressed
as:
2.04 fc b h (psi)
0 . 1 7 G b h (MPa)
Where b is the width of the shear key or width of the abutment stem wall, and h is
the height of the abutment stem wall. in order to compute VN and VS three
different failure models were considered. in the first model, which is depicted in
Fig. 9, it was assumed that a diagonal crack initiated at point B, and propagated
@Seismicisolation
@Seismicisolation
into the abutment stem wall towards point A. Using this mechanism the steel
contribution to the shear strength of the key was expressed as:
T , h + T , d + n h ~h2
d21
, , ~ + n , T , , v --
2s h + a
(3)
Where: T I = A , l f , l is the force

developed in the tension tie,
Tt=A,,2 f y . 2 is the force
developed in the first row of
steel crossing the shear key
interface, nh and n, are the
number of side faces with
horizontal and vertical side
reinforcement, and Ti,h=A,,hfy.h
and T,,=As,~ is the
horizontal and vertical side
faces reinforcement that
crosses the inclined crack as
shown in Fig. 9. in Eq. (3) it is
important to recognize that
only the first line of shear
reinforcement ( i.e. T2 ) is
included in this expression
because the remaining legs do
not cross the inclined crack. In
general As,h= A , , = A , , and Jh Fig. 9 Failure Model I
=fs.v=fs.s, thus; Specimens 1A and 1B
Where: A , / is the total area of steel along the tension tie TI, As,2 is the total area of
steel along Tz, and A,,, is the cross sectional area of the side reinforcement. In the
model depicted in Fig. 10, it is assumed that the horizontal crack BC begins to
incline towards A only after it crosses the first line of shear reinforcement.
Utilizing this mechanism, the steel contribution to the shear strength of the key
Was:
d
As,lfy,ïh + 's,2fy,2 -+ nh A s , s f y , s (5)
2
Because a construction joint was provided between the shear key and the
abutment stem wall, the vertical side reinforcement is not continuous in Fig. 10,
and Vs includes only the horizontal side reinforcement. In addition, the distance
term in the shear key reinforcement should be reduced to dl2 because it is now
assumed that all the shear key reinforcement contribute to A.2, due to the shear
reinforcement extending further into the stem wall.
@Seismicisolation
@Seismicisolation
690 Silva et al.
Finally, the third model
depicted in Fig. 11, assumes
that the diagonal crack never
developed, because the post-
tensioning force, F, is large
enough to prevent inclination
of the horizontal crack into
the abutment stem wall. This
results in a drastic change in
the failure mechanism.
Utilizing this mechanism the
total capacity of the shear key
was computed as:
Fig. 10 Failure Model II Specimen 2A
FIG. 11 FAILURE
MODEL111
SHEARFRICTION
SLIDING MODEL
The shear keys of test series III displayed a similar response to the mechanism
described in Fig. 11; however, a horizontal crack separated the shear key from the
abutment wall and the force in the diagonal compression strut, depicted as CCJin
Fig. 11, could not be transferred into the stern wall resulting in a sliding shear
ñiction mechanism. The nominal capacity, Vn,was
@Seismicisolation
@Seismicisolation
V" = P 4fY 0
Where A, is the area of shear key reinforcement, fy is the yield strength of the
steel, p is the coefficient of fiiction taken as 1.4R for concrete cast monoliîhically
and less than 1.02 for sections across construction joints, and i1 is taken as 1.0 for
normal weight concrete.
LOADING
PROTOCOL
To the exception of specimen 2B, all the specimens were displaced according to
the loading protocol depicted in Fig. 12(a) and loading directions described in
Figs. 1 and 3. During testing of specimen lA, three cycles were supposed to be
performed at 1.4in. (36mm), but without going to zero load the actuators were
inadvertently displaced to 2.4in (61mm). For this test unit, testing proceeded then
at this displacement level.
For specimen lB, which 8.0 203.2

included the abutment +
c 6.0
walls, a direct comparison .G 152.4
E
between the lateral Q>0 Y
*
responses of the two keys 8 !2
4.0 101.6 E
was desired. Therefore, the o
0
3
H
same load I displacement
history was used for n
2.0 50.8 9
n
specimen 1B.
0.0 0.0
Specimen 2B was O 4 8 12 16
subjected to full reversed Number of Cycles
cycles up to failure with (a) Series I and III, and Specimen 2A
the loading direction
depicted in Fig. 2 and the 4.0 101.6
loading protocol depicted
in Fig. 12(b). Initially, the
test unit was subjected to
c
c
v
*
2.0
E
50.8 z-
c
EE
single cycles under force E 0.0 0.0
0
m
o
control, and
displacement control, the
in
a
,a -2.0 -50.8 e.-3
test unit was displaced
n
-4.0 -101.6
with three cycles to the
ductility levels of p l , pl.3, O 5 IO 15 20 25
Number of Cycles
p.2, p3, p4, p6,and @) Specimen 2B
Fig. 12 Loading Protocol
@Seismicisolation
@Seismicisolation
692 Silva et al.
TEST OBSERVATIONS
Development of crack patterns for specimen lA, 1B and 2A were very similar, as
shown in Fig. 13.
(a) Specimen 1A
(c) Specimen 2A (d) Specimen 2B
(f) Specimen 3B
Fig. 13 Test Units Response at Level V
@Seismicisolation
@Seismicisolation
in specimen lA, 1B and 2A, the applied horizontal load was transferred to the
footing through a diagonal compressive strut, as evidenced from the diagonal
cracks that formed in the abutments. Because in specimen 2A the shear key was
cast against the hardened and relatively smooth surface of the abutment stem wall,
a crack propagated horizontally kom the base of the shear key until it intersected
the first row of shear reinforcement, followed by the development of diagonal
cracks into the abutment stem wall. Final failure of specimen lA, 1B and 2A was
in the diagonal compression strut and kacture of the steel, which require
extensive repair or reconstruction of the abutment walls after an earthquake.
Although the constructionjoint between the shear key and the abutment stem wall
did not change drastically the failure mode of specimen 2A, the damage to the
concrete abutment wall was localized below the shear key rather than spreading
along the abutment stem wall, as in specimens 1A and 1B (see Fig. 13).
As anticipated, specimen 2B showed a predominant flexure-shear response. Fig.

13(d) shows specimen 2B at its ultimate displacement (level V), which shows that
a flexural plastic hinge was formed near the key interface to the stem wall,
requiring only localized repair after an earthquake.
The failure mode of specimen 3A and 3B was very similar, as shown in Fig. 13.
in these specimens, a horizontal crack developed above the construction joint. As
this crack propagated completely through the key, the concrete of the shear key
was effectively separated from that of the abutment stem wall. This indicates that
the applied horizontal load was transferred to the abutment wall through a shear
friction mechanism, with the shear key itself behaving as a rigid body connected
to the abutment by the reinforcement. Comparing to specimens lA, lB, and 2A
the failure modes depicted in Fig. 13(e) and (I> indicate that the combined action
of the construction joint with bond breaker and post-tensioning of the abutment
stem wall was effective in preventing significant spreading of damage into the
abutment wall.
RESPONSE
LOAD-DISPLACEMENT
Fig. 14 shows the lateral load versus displacement of test series I, II, and III. Fig.
14(a) to (c) also show the computed values for VC and VS,which were previously
described in terms of the analytical models presented in Figs. 9 and 10. Specimen
1A and 1B values were computed in terms of Eqs. (2) and (4) and specimen 2A
capacity was computed in terms of Eqs. (2) and (5).
Comparison of the curves for specimen 1A and 1B indicates that the abutment
back and wing walls contributed significantly to the capacity of the shear keys.
The amount of vertical reinforcement crossing the shear key-abutment wall
interface was reduced in specimen 2A compared to specimen 1A. As a result, the
capacity of specimen 2A was less than that of specimen lA, and the capacity of
specimen 2A was relatively maintained with increased displacements.
On the other hand, a significant drop in the capacity of specimen 1A and 1B was
observed following the peak load. This indicates that reduction of the shear key
@Seismicisolation
@Seismicisolation
694 Silva et al.
vertical reinforcement and presence of the joint between the shear key and the
abutment walls tends to enhance the displacement ductility capacity of shear keys.
Lateral Displacement (mm) Laterai Disdacement ímmì
0.00 25.40 50.80 76.20 101.60 127.00
1250
1Ooo 5
h
750 3cl
500 7ü
&
250
O
0.00 1.00 2.00 3.00 4.00 5.00 0.00 1.00 2.00 3.00 4.00 5.00
Lateral Displacement (in.) Lateral Displacement (in.)
(a) Specimen 1A (b) Specimen 1B
Lateral Displacement (mm)
-101.6 -50.8 0.0 50.8 101.6
0.00 25.40 50.80 76.20 101.60 127.00
100
0.00
0.00
1.00 2.00
(c) Specimen 2A
50.80 101.60
3.00
Lateral Displacement (in.)

4.00
152.40
5.00
203.20
li4.0
0.00
LEVELIV
-2.0
V
0.0
Lateral Displacement (in.)
(d) Specimen 2B
Lateral Diviacement (mm)
50.80 IOI.60
2.0
152.40
4.0
500
250
203.20
.e>
g
cl
%3
cl
300
2,50 O LEVELII ----

v1 O LEVELIII
.-P A LEVELIV
&ZOO
X LEVELV
-2
3 150
';d
c
6
2 50
O
0.00 2.00 4.00 6.00 8.00 0.00 2.00 4.00 6.00 8.00
Lateral Displacement (in.) Lateral Displacement (in.)
(e) Specimen 3A ( f ) Specimen 3B
Fig. 14 Load vs. Displacement Response
@Seismicisolation
@Seismicisolation
Specimen 2B hysteretic response displayed stable hysteretic loops with significant
amounts of energy dissipation capacity, as it is expected in a flexural dominated
response. This indicates that the design of specimen 2B was adequate in
developing a flexural dominated response. The theoretical first yield and ideal
capacity of specimen 2B were computed in terms of a moment-curvature analysis?
In series IIZ test units, two significant drops were observed in the horizontal load.
Post-test observations indicate that the first drop was caused by fracture of four of
the #5 bars crossing the sliding plane. These four bars were located in-between
the post-tensioning ducts, which restricted significantly the deformation of these
bars under sliding. The second drop was caused by fracture of the remaining four
#5 bars, which were located in the vicinity of the cover concrete. Thus, in future
tests all the bars should be positioned outside of the post-tensioning ducts in order
to improve the performance of the shear keys. Capacity of these test units was
computed using Eq. (7) and replacing the reinforcement yield strength by the
ultimate strength.
PERFORMANCE ASSESSMENT FOR EXTERIOR S H E A R KEYS

Fig. 15 shows an idealized load versus displacement envelope for sacrificial
exterior shear keys that display a performance within the failure models described
in Figs. 9 and 10. For the other test units, a different hysteretic model was
developed, and will not be described in this paper. The idealized hysteretic model
described in this section was expressed in terms of two models, which correspond
to the concrete and steel components.
in Fig. 15 A,, Ao,Am,Aw, and A, are the keys top lateral displacements
specified at the five different Damage Levels, which were previously defined. In
addition, VI, VII,VIII,VW, and VV are the capacity of the keys at the respective
Damage Levels. Finally, k o , ~ k, o,~l,k o , ~ ~ kl , ,,,IV, and k o , are
~ the loading branches
stiffhess between Damage Levels.
The concrete component model was defined as an idealized elasto-plastic model

as shown in Fig. 16(a), but with a strength loss model as depicted in Fig. 16@).
The strength loss model was implemented in order to describe the reduction in the
concrete component due to loss of the aggregate interlock beyond the peak load.
The steel component model was also defined as an idealized elasto-plastic model
as shown in Fig. 17(a), and with a strength loss model depicted in Fig. 17 (b). As
before, the steel strength loss model was implemented in order to describe the
reduction of the steel component due to hcture of the reinforcement.
Observation of Fig. 15 suggests five distinctive loading branches. The first branch
described the response of the test units up to damage level II. The next is the
ascending branch up to the peak load, which is defined in terms of damage level
HI. The third branch can be characterized as the descending curve between
@Seismicisolation
@Seismicisolation
696 Silva et al.
damage Levels IIï and IV. The next branch is depicted by a flat curve behveen
Levels IV and V, and finally, the fifth branch is the descending branch beyond
Level V. Quantificationof these damage levels is described next.
Lateral Load
-
---
Hysteresis Rule
Concrete Component
......-..Steel Component
' AI AI1 A"1 'N

Lateral Displacement
AV AF
Fig. 15 Hysteresis Rule Exterior Shear Keys
(a) Concrete Component
@) Concrete Strength Reduction Variation
Fig. 16 Concrete Component Model
@Seismicisolation
@Seismicisolation
v, .............................................................................
I
‘11 (a) Steel component
$ A
All Alii Av A,
(b) Steel Strength Reduction Variation
Fig. 17 Steel Component Model
LEVEL i: This damage level was characterized by onset of cracking. Lateral load
at this level was computed based on principal tensile stresses and developed based
on a FEM analysis of the shear keys. As shown in Fig. 18(a), the shear stresses
vary linearly fiom zero at the outside edge of the stem wall to the maximum shear
stress at the inside edge of the stem wall. Similarly, as shown in Fig. 18@),the
normal stresses vary linearly at the level of the stem wall according to the flexural
theory.
(a) Shear Stresses (b) Normal Stresses

Fig. 18 FEM Results
Based on the FEM results, at the level of the stem wail maximum shear stresses
were computed as:
2v
v, =-
bd
Based on the FEM results, at the level of the stem wall maximum normal stresses
were computed as:
6 Vu
f, =-bd2 (9)
@Seismicisolation
@Seismicisolation
698 Silva et al.
Where a is the distance fiom the top of the stem wall to center of application of
the lateral load, V (see Fig. 10) and it was assumed that a=kd. Expressingf; as
the cracking strength of concrete and according to the principal stresses the shear
force required to cause cracking was:
VI = Abd
(10)
3k+J9k2+4
LEVEL II: This level corresponds to onset of yielding of the shear key
reinforcement and corresponds to point B in Fig. 15. Lateral load at this damage
level was computed according to the strut-and-tie analogous models previously
described. The lateral load at yielding was given either by Eqs. (4) or (9,and an
additionalterm to account for the concrete component as:
Where V, is the concrete component to the shear resisting mechanism and is given
by Eq. (2), and AI1 and AlIl are the shear keys top displacement at damage levels
II and III. The shear key top displacement at this level was:
Where Ld is the Leq= 2 L, +La

reinforcement development
length, and L,is the cracked
region as shown in Fig. 19.
Based on test observations I
the L, value is
approximately the width of
the stem wall, b.
III ES. (12) the t e m fi is
implemented because the
reinforcement crosses the
crack at approximately 45'.
The geometric
4K-F
ratio (A + d ) / in
Eq. (12) was derived based
on the assumption that the
shear keys deform as a rigid (a) Shear Key Deformed Shape
body with the center of Fig. 19 Shear Keys Response Mechanism
rotation at the base of the
stem wall near the
compression zone identified
as point A in Fig. 19.
@Seismicisolation
@Seismicisolation
LEVEL ZIZ: This level corresponds to the peak load and corresponds to point C in
Fig. 15. Immediately after this level increase in the crack width opening leads to
reduction of the concrete contribution to the shear resisting mechanism. The
lateral load at yielding was given by Eq. (l), and the shear key top displacement
was computed by assuming that all the bars that cross the cracked zone have
reached yielding. Thus:
Where s is the reinforcement spacing in the stem wall. The variable s in the
denominator of Eq.(12) is different fiom the denominator in Eq. (13) ( Jh2+d2)
because the crack width is now taken at the reinforcement located near the base of
the stem wall (i.e. point A in Fig. 19).
LEVEL ZV: At this level full reduction in the concrete contribution is expected
and the capacity was computed only in terms of the steel component, VS. This
damage level corresponds to point D in Fig. 15. Because of lack of research data
to justify the steel strains necessary to produce full reduction in the concrete
compression mechanism, steel strains were back calculated h m the research data
by assuming rigid body deformationsand computed as:
Correlation of this equation to research data indicates that full degradation of the
concrete contribution to the shear resisting mechanism is likely to occur at a steel
strain of approximately O.O05in/in. In the same way as in level III, the top
deflection at Level IV was:
LEVEL V: At this level fiacture of the reinforcement is initiated and corresponds

to point E in Fig. 15. As in level IV, steel strains at this level were back calculated
h m research data using Eq.(14). At this level the correspondent steel strain was
0.007idin. It is important to recognize that these strains do not correspond to the
strains in the reinforcement at the top of the stem wall but instead to those near
the base of the stem wall. In the same way as in levels III and N, the top
deflection at Level V was:
And the ultimate displacement AF, depicted in Fig. 15 was:

A, = A v + vc
Av
@Seismicisolation
@Seismicisolation
700 Silva et al.
CONCLUSIONS
The following are some conclusions dram fiom this research program:
1. Reducing the amount of shear reinforcement crossing the failure plane and
providing a construction joint between the shear key and the abutment stem wall
enhances the ductility capacity of the shear key and localizes the damage under
the shear key.
2. Flexural shear keys have significantly higher ductility and energy

absorption capabilities than sacrificial shear keys. In addition, test results indicate
that in specimen 2B damage was localized over a small region.
3. Post-tensioning of the abutment stem wall changes the failure mode of the
shear keys fiom a strut-and-tie to a sliding shear friction mechanism, and localizes
damage at the interface of the key to the abutment wall, resulting in minimum
repair after an earthquake.
4. The capacity of shear keys as in specimens lA, 1B and 2A was accurately

calculated by summation of the concrete, V, and the steel, V,, components. In
specimen 2B a simple flexural
REFERENCES
1. Caltrans,Bridge Design Specifications, 1993.
2. Schlaich, J., and Schafer, K.,“Design and detailing of Structural Concrete
Using Strut and Tie Models”, ïñe StructuralEngineer,V.69, No. 6, 1991, 13
PP.
3. AC1 Committee 3 18, Building Code Requirementsfor Reinforced Concrete
(ACZ 318-99),American Concrete Institute, Farmington Hills, MI, 1999.
4. King, D.J., Priestley, M.J.N., Park, R.,“Computer Programfor Concrete
Column Design, Department of Civil Engineering, University of Canterbury
”
Christchurch, New Zealand, Report No. 86-12, May 1986.
ACKNOWLEDGMENTS
Caltrans is acknowledged for funding of this research under contract No. 59
A005 1.
@Seismicisolation
@Seismicisolation
SP 209-38
Thermal Effects on Chloride

Diffusion in Concrete
by B. D. Miller, M. A. Miltenberger,
and E. K. Attiogbe
SvnoDsis: Corrosion of reinforcing steel in concrete usually initiates from chloride

ingress. The temperature of the service environment and cementitious material
blends influence chloride ingress. This paper presents data from a multi-year ex-
periment designed to illustrate the effects of temperature and supplementary ce-
menting materials on chloride diffusion in concrete. This experiment consists of
six concrete mixtures, five of which contained different supplementary cementing
materials. Test specimens were submerged in 6% sodium chloride solution, and
stored at a temperature of 10, 23, and 32 "C for approximately three and one-half
years. The chloride ingress data are presented, and the impact of temperature on
chloride diffusion coefficients along with its implications for corrosion service life
modeling are discussed. Presently, most service life prediction models adjust for
the service temperature of a structure by using the Arrhenius Equation. The time-
to-corrosion data obtained in this study suggests that the net result of temperature
related effects does not match the predictions based on the Arrhenius Equation.
Keywords: chloride diffusion; corrosion; thermal effects
@Seismicisolation
@Seismicisolation
701
702 Miller et al.
Biographical Sketches
ACI member Brian D. Miller is a Project Engineer with Master Builders, Inc.,
Cleveland, Ohio, U.S.A. He has a B.S. in Civil Engineering and has been with
Master Builders since 1995. His research interests include chloride transport
in concrete, corrosion of reinforcing steel, and fiber-reinforced concrete.
ACI member Matthew A. Miltenberger is a Senior Project Engineer with

Master Builders, Inc., Cleveland, Ohio, U.S.A. He has a M.S. in Structural
Engineering and has been with Master Builders since 1993. His research
interests include corrosion of reinforcing steel, fiber reinforced concrete,
chloride transport in concrete, and modeling. He is a member of ACI
committees 355, 365 and 222. Matt was a co-recipient of the ACI Leonard C.
Wason Award for Most Meritorious Paper in 2000.
ACI member Emmanuel K. Attiogbe is the Director of Technical Services,

Master Builders, Inc., Cleveland, Ohio, U.S.A. He holds a Ph.D. in civil
engineering from the University of Kansas, specializing in structural
engineering and concrete materials. He was awarded ACIS Wason Medal for
Materials Research in 1995. He is a member of ACI Committees 231,
Properties of Concrete at Early Ages; 236, Materials Science; CRC, Concrete
Research Council; and TTTC, TAC Technology Transfer Committee.
To predict corrosion service life, a diffusion coefficient is normally

determined for the particular concrete. There are several factors that affect
chloride diffusion in concrete. Determining the effect of these factors and how
to adjust for them in modeling is necessary for a reliable prediction. The
effects of temperature and supplementary cementing materials on chloride
diffusion in concrete are some of the factors that influence service life
predictions. Previous studies including temperature effects on chloride
diffusion in cement pastes have shown that an increase in temperature leads to
an increase in chloride diffusion (1,2). The Arrhenius equation (2) is
commonly used to adjust for temperature based on these studies. This paper
presents data on the effect of temperature on chloride diffusion in concrete
containing various types of supplementary cementing materials and its relation
to corrosion service life prediction.
@Seismicisolation
@Seismicisolation
INTRODUCTION
Corrosion of reinforcing steel in concrete is a primary cause of

premature deterioration of concrete bridges, parking and marine structures (3).
Corrosion is usually initiated from chlorides, which are present in deicing salts
or seawater. Chlorides enter into concrete by several mechanisms and build up
at the reinforcing steel. Build up continues until the chloride threshold for the
particular concrete has been reached or exceeded and then corrosion may
initiate. In concretes that are in a saturated condition or constantly exposed to
moisture, diffusion is the primary mechanism of chloride ingress (43). By
definition, diffusion is the ingress of ions under a concentration gradient and
for concrete, occurs in the pore solution. Under constant exposure to moisture,
the chloride concentration at the surface is a function of the solution
concentration and does not change appreciably over time. In environments
where structures are exposed to cyclic wetting and drying conditions, diffusion
still contributes as a mechanism for chloride ingress, although usually not as
the primary mechanism. As the build-up of chloride at the surface occurs over
time, ingress begins with capillary absorption pulling the chlorides into the
concrete. With each cycle of wetting and drying, more chlorides are pulled
into the concrete and the existing deposits are driven further inward. This
process continues until the chloride build-up reaches a depth where the interior
of the concrete is essentially saturated. At this point, chloride transport in the
interior occurs primarily through diffusion (4).
Models based on Fick's second law of diffusion ( 6 ) are commonly used

to predict the service life of concrete structures with respect to corrosion. The
service life of a structure is comprised of two components, the time-to-
corrosion initiation, and the time for corrosion propagation. This paper focuses
on the time-to-corrosion initiation. The chloride diffusion coefficient for a
particular concrete is one input parameter which the models use to estimate the
time for corrosion initiation. This evaluation was performed to determine the
effect of service temperature on chloride diffusion in concrete containing
various types of supplementary cementing materials and to verify its
implications in service life modeling.
The experimental program consisted of five concrete mixtures

containing supplementary cementing materials and a normal portland cement
concrete as a reference. The supplementary cementing materials were dosed
by weight of cement replacement and are as follows:
1. Fly ash (Class F) at 20% replacement
@Seismicisolation
@Seismicisolation
704 Miller et al.
2. Ground granulated b l a s t h a c e slag (GGBS) at 50% replacement
3. Manufactured calcined clay at 25% replacement
4. Metakaolin at 8% replacement
5. Silica fume at 8% replacement
Each concrete mixture was tested for compressive strength, rapid

chloride permeability, and chloride diffusion. Two methods, Chloride
Migration and Bulk Diffusion, were used to determine the chloride diffusion
coefficient. B u k diffusion was used to evaluate the effects of temperature on
the apparent chloride diffusion coefficient. The chloride exposure process for
Bulk Diffusion testing is reasonably simple and is very efficient for long-term
temperature studies. The Chloride Migration method is very efficient for
repetitive testing and was performed at room temperature to evaluate changes
in the effective diffusion coefficient over time. This data is important in
determining the effect of supplementary cementing materials on the diffusion
coefficient over time.
Materials and Mixture Proportions
The materials used in this evaluation conformed to the applicable

ASTM specifications. All concrete mixtures had nominal cementitious
contents of 356 kg/m3with a nominal maximum aggregate size of 1-inch and a
water-cementitious materials ratio (w/cm) of 0.40. Concrete mixing followed
a 3-minute mix, 3-minute rest, and 2-minute remix sequence. The air-contents
were approximately 6% f l % and the slump ranged between 190 and 235 mm.
Plastic property tests conducted and specimens cast were in accordance with
ASTM C 192.
SDecimens and Curine Procedures
Chloride diffusion and rapid chloride permeability specimens were

cored from 305 x 305-mm slabs at the time of testing. These slabs were broom
finished and cured in their forms under plastic and wet burlap for 24 hours.
The slabs were then placed outside with their forms intact and covered with
wet burlap and plastic for seven days. The forms were then removed and the
slabs kept outside until testing. The cylinders for compressive strength tests
were moist cured until the time of testing.
@Seismicisolation
@Seismicisolation
Test Methods and Procedures
Compressive strength tests were performed in accordance with ASTM

C 39. Three 100 x 200-mm cylinders were tested for compressive strength at
28 days, 365 days and approximately 3.5 years. Cylinders were moist cured at
23 OC until test age.
Rapid chloride permeability tests were performed in accordance with

ASTM C 1202. Two 95-mm diameter specimens were cored from slab
specimens for each test. Rapid chloride permeability tests were performed at
28 days, 365 days and approximately3.5 years.
Bulk diffusion testing was performed as follows. For each concrete

mixture, six 95-mm diameter cores were taken from a slab that was cured for
28 days. The outer circumference and bottom of the cores were painted with
three coats of epoxy paint. The cores were saturated with water prior to being
submerged in a 6% sodium chloride (NaCl) solution. The container was sealed
to minimize evaporation. To determine the effects of temperature on the bulk
diffusion rate, the cores from each mixture were stored in three
environmentally controlled rooms during the test. These rooms consisted of a
room at a constant temperature of io O C , a room at a constant temperature of
23 "C and a room at a constant temperature of 32 OC. At approximately three
and one-half years, cores from each concrete mixture were taken out from the
temperature-controlledrooms. The outer 10-mm of each core was removed to
minimize edge effects. To create a chloride profile, powder samples were then
collected from the core in increments of 2 mm using a precision vertical
milling machine. The powder samples were titrated for acid-soluble chloride
content by means of potentiometric titration with silver nitrate. Titrations were
performed in accordance with ASTM C 1152. The chloride profile was used
to determine an "apparent" diffusion coefficient, Da. Da is essentially the mean
value for chloride ingress during the test period that incorporates both bound
and ñ-ee chlorides. The apparent diffusion coefficient was determined by
fitting the chloride profile data to Fick's second law of diffusion and applying
nonlinear regression analysis (7).
@Seismicisolation
@Seismicisolation
706 Miller et al.
The Chloride Migration procedure described below is an accelerated
test method for determining the effective chloride diffusion coefficients (De)of
concrete (8). De is essentially a measure of fi-ee chloride and the conductivity
of the pore structure of the concrete.
Chloride migration testing was performed at 28,365 days and 3.5 years
as follows. Test specimens, 95-mm diameter x 50-mm thickness, were cored
fi-om the slabs at each test age. The outer circumference of each specimen was
coated with two applications of epoxy paint. Afier the epoxy cured, the
specimens were submerged in tap water until testing, typically at least 48
hours. This insured that the specimens were saturated with water prior to
chloride exposure. A diagram of the test cell is presented in Fig. 1. This
procedure uses an electrical potential of 12 volts to drive chloride ions fi-om a
2.5M NaCl solution into the specimen until a constant chloride concentration is
obtained in the exit-side solution (catholyte). This concentration of chloride is
verified through weekly monitoring of the catholyte with chloride ion sensitive
indicator strips. Once the concrete pore solution is saturated with chloride,
both reservoirs are emptied and refilled with a 2.5M NaCl solution. At this
point, the conductivity of the cell is measured and the effective chloride
diffusion coefficient of the concrete is calculated from the diffusivity ratio (8).
It should be noted that during application of the electrical potential, anions in
the pore solution, such as hydroxide, are driven out of the specimen as chloride
ions migrate inward (9). Therefore, the conductivity of the cell is due to
saturation of the concrete with chloride ions with minimal impact fiom other
ions.
TEST RESULTS
Compressive Strength
The compressive strength results are presented in Fig. 2. All concrete

mixtures had compressive strengths greater than 43 MPa after 28 days. The
strength gain between the ages of one and three and one-half years was low,
with the strength exceeding 55 MPa at three and one-half years.
@Seismicisolation
@Seismicisolation
Rapid Chloride Permeability
Fig. 3 shows the results from rapid chloride permeability testing

(ASTM C 1202). The reference concrete had a rapid chloride permeability
rating of low (i.e., less than 2000 coulombs) and the pozzolan-treated concretes
had rapid chloride permeability ratings of very low @e., less than 1000
coulombs) at three and one-half years. The resistance to chloride ion
penetration increased with time for all the concrete mixtures.
Chloride Diffusion
The effective chloride diffusion coefficients from chloride migration

are presented in Table 1. Table 2 presents the apparent diffusion coefficients
&om bulk diffusion. The diffusion coefIicients are presented in units of
mm2/yr (lmm2/yr = 3.171 x m2/s). These units were chosen to simpli@
the presentation of results, and can be directly used in corrosion service life
calculations because the unit of measure for concrete cover is mm and the unit
for corrosion service life is years. Fig. 4 illustrates the change in diffusion
coefficient with time.
DISCUSSION OF RESULTS
All the concrete mixtures had compressive strengths exceeding 43 MPa

at 28 days. Also, all mixtures, except the reference and the fly ash-treated, had
very high resistance to chloride ion penetration at 28 days (Figs. 3,4). The
compressive strengths and the resistance to chloride ion permeability increased
with time. The high compressive strengths and resistance to chloride ion
penetration indicate concrete mixtures of high quality. The diffusion
coefficients determined from chloride migration testing for the concrete treated
with supplementary cementing materials, with the exception of the fly ash-
treated concrete, are relatively low at 28 days as shown in Table 1. Typically,
diffusion coefficients determined from chloride migration testing for high
quality concrete have 28-day values of 30 mm2/yr or less. The results suggest
that supplementarycementing materials, with the exception of fly ash, enhawe
the durability of concrete at early ages. The higher chloride diffusion
coefficients at early ages for the concrete containing fly ash is due to the slow
rate of hydration associated with Class F fly ash. In addition, the effects of all
the supplementary cementing materials on the long-term chloride diffusion
coefficients are similar as can be inferred from Fig. 4.
@Seismicisolation
@Seismicisolation
708 Miller et al.
The exposure temperature of a concrete structure affects both the
mobility of the chloride ion and the hydration rate of the cementitious paste.
With availability of moisture, the process of hydration continuously changes
the pore structure of the concrete over time. As temperature increases the rate
of hydration increases. This increased hydration rate essentially begins to
restrict the pore structure and therefore, would reduce the rate of chloride
ingress over time. This is especially true with pozzolans such as fly ash.
However, as the temperature increases, the mobility of chloride ions also
increases. This increased mobility increases the chloride ingress rate. These
effects, along with thermal expansions and contractions and other interactions
combine to have a net effect on the chloride diffusion coefficient. From
previous studies performed in paste, the temperature effect has been modeled
using the Arrhenius equation. Fig. 5 illustrates the theoretical effect of
temperature on the diffusion coefficient based on this equation. The activation
energies used for this chart ranged from O to 50 kJ/mol(2). Fig. 6 shows the
actual diffusion coefficients determined from the test data. The results do not
follow the prediction based on the Arrhenius equation. The data suggest that
the combination of factors related to temperature change have a net effect of
reducing the diffusion coefficient at higher service temperatures.
To illustrate the Arrhenius equation's effect on corrosion service life,

Fick's second law of diffusion was solved for time-to-corrosioninitiation. The
Arrhenius equation in combination with an average activation energy of 35
kJ/mol was used to adjust the apparent diffusion coefficient results from 23 OC
to 10 and 32 O C , and the time-to-corrosion was determined at each temperature.
The results are compared against the time-to-corrosion obtained from the
actual test data. Fig. 7 shows the results for the time-to-corrosion predictions
for the fly ash-treated concrete. Both the Arrhenius-adjusted and the test data-
based predictions decrease with increasing temperature. However, the test data
indicate a departure from the theoretical behavior above 23 OC. This was the
case for all of the concretes containing supplementary cementing materials.
However, it should be noted that the test data at 32 OC represents 3.5 years of
accelerated curing. Under hot weather field conditions, the normal variations
in ambient temperature could result in lower average curing temperatures.
Therefore, the Arrhenius-adjusted time-to-corrosion prediction may provide a
conservative estimate for most hot climate situations. However, in cold
climates (represented by 10 OC in Fig. 7), the Arrhenius equation adjustment
appears to overestimate the time-to-corrosion prediction when compared with
the test data.
In summary, the data from this study suggest that the net effect of
temperature on the chloride diffusion coefficient is not explained solely by the
Arrhenius equation. Further research is needed to quantify the net effect, on
chloride diffusion, of factors related to temperature change.
@Seismicisolation
@Seismicisolation
CONCLUSIONS
1. The data suggest that the net effect of temperature on the chloride diffusion
coefficient is not explained solely by the Arrhenius equation.
2. The Arrhenius-adjusted time-to-corrosion prediction, using an activation
energy of 35 kJ/mol, appears to be conservative for concrete structures in
hot climates, but may not be conservative at temperatures typical of cold
climates.
3. The addition of the various supplementary cementing materials, with the
exception of Class F fly ash, enhance the early-age durability of the
concrete mixtures.
4. The long-term diffusion coefficients for concrete containing various
supplementary cementing materials are similar.
@Seismicisolation
@Seismicisolation
710 Miller et al.
REFERENCES
1. Collepardi, M., Marcialis, A., and Turriziani R.,"Penetration of Chloride

Ions into Cement Pastes and Concretes," Journal of The American Ceramic
Society-Discussionsand Notes, Vol 55, NO. 10, 1972, pp 534-535.
2 . Page, C. L., Short, N. R., and El Tmas, A., "Diffusion of Chloride Ions in
Cement Pastes," Cement and Concrete Research, Vol 11, 1981, pp. 395-
406.
3. Chaker, V., ed., "Corrosion Forms & Control for Infrastructure," STP
1137, Amencan Society of Testing and Materials, 1992.
4. Hong, K., Hooton, R. D., "Effects of Cyclic Chloride Exposure on the

Penetration of Concrete Cover," Cement and Concrete Research, Vol 29,,
1999, pp 1379-1386.
5 . Miller, B. D., and Miltenberger, M. A., "The Effects of Corrosion-

Inhibiting Admixtures on Chloride Transport in Concrete, '' international
Conference on Ion and Mass Transport in Cement-Based Materials,
Proceedings, Special Volume, Toronto, Canada, 1999, pp 367-376.
6 . Bamforth, P. B., and Price, W. F., "Factors Influencing Chloride Ingress

into Marine Structures," Concrete 2000: Economic and durable
construction through excellence, E & FN Spon, 1993, pp 1105-1 1 18.
7. Nortest Method NT Build 443, "Concrete, Hardened: Accelerated Chloride

Penetration", 1995.
8. Miltenberger, M. A., Luciano, J. J. and Miller, B. D., "Comparison of

Chloride Diffusion Coefficient Tests for Concrete," #' International
Conference on Durability of Building Materials and Components,
Vancouver, Canada, 1999.
9. Streicher, P. E., Alexander, M. G., "A Chloride Conduction Test for

Concrete," Cement and Concrete Research, Vol 25, No. 6, 1995, pp. 1284-
1294.
@Seismicisolation
@Seismicisolation
Table 1. - Average Effective Chloride Diffusion Coefficients from Chloride
Migration Tests at 23'C
Specimen I.D. De(mm2/yr) De (-'/Y) De (m'/~r)

@ 28 days @ 1 year @ 3.5 year
Reference 43 36 30
Fly Ash 51 17 14
GGBS 17 12 8
Calcined Clay 23 13 9
Metakaoln 13 12 16
Silica Fume 19 17 17
Note: lmm2/yr= 3.171 xlO-I4m2/s
Table 2. - Average Apparent Chloride Diffusion Coefficients from Bulk

Diffusion Tests
I.D. @ 10 OC @, 23 OC @, 32 OC
Reference 45 46 52
Fly Ash 28 47 38
GGBS 23 22 11
Calcined 31 28 23
Metakaoln 19 26 26
Silica 38 50 34
Note: immz/yr= 3.171 x1O-l4mZ/s
@Seismicisolation
@Seismicisolation
712 Miller et al.
, 50mm
7 Test Specimen
1. Schematic of Chloride Migration cell.
2. Average compressive strength results.
@Seismicisolation
@Seismicisolation
04
O 200 400 600 800 1000 1200 1400 1600
Time (days)
-t Reference 4-Fly Ash -A- GGBS +Calcined Clay -Et Metakaolin +Silica Fume
3. Variation of Rapid Chloride Permeability results with time.
5
E
-
E
cl'
O I
O 200 400 600 800 IO00 1200 1400 1600
Xm(daYs)
-+Reference 4-Fly Ach -A- GGBS ++Calcined Clay 8 Metakaolin -0- Silica Fume
4. Variation of effective chloride diffusion coefficients with time from

Chloride Migration tests.
@Seismicisolation
@Seismicisolation
O 10 20 30 40 50 60
Temperature (%)
-+U = O Jim1 -m- U 20000 Jlmol -A- U = 35000 Jim01 +e U = 50000 Jlmol
5. Effect of temperature on diffusion coefficients using the Arrhenius
Equation.
tReference +-Fly Ash -A- GGBS Jt CalcinedClay 8Metakaolin 4 Silica Fume
6. Apparent diffusion coefficients vs. temperature from Bulk Diffusion.
@Seismicisolation
@Seismicisolation
25
ci1
*o .. .- - ...- . - ....
E
.-O
.-m U
-E 15.
CI
_.._._..........
.-O
I:
t 10. .____._..__..._.
4 O
5 ------
O 5 10 15 20 25 30 35
Temperature (OC)
-8-Arrhenius-Adjusted Prediction -W- Test DataBased Prediction
7. Temperature effects on time-to-corrosion for Class F Fly Ash Concrete.
@Seismicisolation
@Seismicisolation
SP 209-39
Spectral Response of Mortar

Surfaces During Curing
by 1. Kindervater, R. luna, and A. Belarbi
Svnopsis: Imaging and non-imaging sensors that collect spectral data of surface
materials are rapidly becoming available to engineers due to advances in electro-
optics and sensor technology. Applications of remote sensing for the identification
of surface materials and determination of some of their characteristics have been
developed in the geological sciences. Transportation research systems are moving
aggressively towards using these types of technologies for materials such as soil
subgrades, concrete, asphalt, and, to a lesser extent, steel. A series of experiments
were identified to analyze the spectral response of laboratory prepared surfaces,
primarily of materials with a mineralogical origin, including soil, aggregate, and
concrete. This paper presents the experimental procedure and results of a series of
tests performed on a mortar mixture. Temperature, strength, and spectral re-
flectance were measured for a period of time during curing of the mortar. Results
revealed apparent correlations between temperature, water content (curing rate),
and spectral response.
Keywords: mineralogy; mortar curing; spectral

response; spec t rora diomete r
@Seismicisolation
717
@Seismicisolation
718 Kindervater et al.
BiomaDhical Sketches:
Lori Kindervater, Graduate Student of Civil Engineering at the University of Missouri-Rolla.

Her research interests include stnictural engineering, construction materiais, and earthquake
engineering. She is a student member of American Society of Civil Engineers, Earthquake
Engineering Research Institute, and National Society of Professional Engineers. She received
her Bachelor’s degree in 2001 and is currently pursuing a Master’s degree.
Ronaldo Luna, Associate Professor of Civil Engineering at the University of Missouri-Rolla.

His research interests include geotechnical engineering, material characterization, GIS
applications, and remote sensing. He is a member of the American Society of Civil Engineers,
Geo-Institute and American Society of Engineering Education. He serves in the technical
committees on Soil and Rock Properties (ACLOI) and Earth Materiais Exploration and
Classification (ACLO2) at the Transportation Research Board.
Abdeldjelil Belarbi, FACI, Associate Professor of Civil Engineering at the University of

Missouri-Rolla. His research interests include constitutive modeling of reinforced and
prestressed concrete as well as use of advanced composite materials and smart sensors in civil
engineering infrastructures. He is a member of several AC1 technical and educational
committees, and he is the current Chair of AC1 Committee E801, Student Activities; and
subcommittee 445-5, Torsion of Structurai Concrete. He is past President of AC1 Missouri-
Chapter.
INTRODUCTION
Remote sensing is a simple concept that involves gathering information about

an object without being in direct contact with it. There are many different
kinds of remote sensing technologies and applications. Common examples
include the use of X-ray imaging to gather detailed information about the inner
human body and the use of dogs trained to “sniff out” illegal drugs and
imported food items. The usefulness of remote sensing stems from its ability
to decrease the time and effort required to identify specific object or material
traits.
The form of remote sensing used in this project is spectral data collection. The
spectrum referred to herein is the electromagnetic (EM) spectrum (see Fig. 1).
The EM spectrum relates the amount of radiant energy present to its
Wavelength or frequency. As a result, energy reflectance can be considered as
a function of wavelength. Essentially, when radiant energy from ihe
surroundings hits an object, some of the incident energy will be reflected, and
some will be absorbed. The characteristics of the object being observed
determine the amount of incident energy reflected at a particular wavelength.
A plot of the percent reflectance versus wavelength presents the spectral
signature of the object (1). This information is useful because the spectral
@Seismicisolation
@Seismicisolation
signature of an object has particular traits, just as the object itself has particular
characteristics. The purpose of this study was to determine whether or not the
spectral signature of mortar changes as the properties of the mortar change
during hydration and curing.
Once these fundamental relationships are established and calibrated, this

research will lead to techniques for rapid assessment of the curing state of
concrete surfaces. These techniques can provide new methods of quality
control that do not require destructive tests. Since this process makes relative
measurements of the surface at different stages of curing, further research
could establish whether or not the water content of concrete can be accurately
determined during curing. If this is possible, then the availability of portable
imaging hyperspectral devices could allow creation of a map of the curing
stages of concrete sections. This technology would be useful for research and
quality control purposes.
Background - Hvperspectral Imaeing
Hyperspectral imaging (HSI) is a relatively new tool, and its applications are
still experimental. However, there are several advantages to using HSI instead
of conventional imaging techniques. Imaging spectrometers produce an image
similar to a photograph, where complete spectral information (beyond the
visible range) for each pixel in the image is recorded. This produces a spectral
data set containing spatial relationships that can be directly used for mapping.
As a result, HSI can be used to create maps of minerals, plants, soils, pollution,
manmade materiais, water, and other surface features (2).
Because of the increasing availability of portable and airborne sensors that can
produce laboratory quality spectra from the field, many disciplines are
interested in the possible applications of hyperspectral imaging. One well-
researched area is identification and composition of materials (mainly minerai-
based). At high resolutions, minerals can be identified, and it may also be
possible to determine their elemental composition (3). If these techniques cag
be well developed, spectral imaging could be very useful for mineral
identification because it is quick and does not require the traditional extensive
laboratory testing. They could also be used in the field since
spectroradiometers are currently available as portable handheld or backpack
units.
@Seismicisolation
@Seismicisolation
Similarly, it is possible to use HSI technology to perform quality control
checks in some manufacturing processes. A series of tests with concrete
samples has been performed in an attempt to determine if a spectroradiometer
could provide information necessary to monitor the manufacture of cement (4).
The results of these tests showed that a spectroradiometer could indeed provide
qualitative and quantitative information for detecting the different
characteristics of materials used in the manufacturing process (4). The
changes of the spectral response curves of mortar during curing are less
understood and are the engineering focus of this paper. This work determines
some apparent relationships between spectral response and mortar curing.
Engineering Significance of Curing
Hydration is the chemical reaction between cement and water that results in the
setting and hardening of concrete. It is during the chemical and physical
processes of hydration that concrete gains the properties of compressive
strength, durability, watertightness, wear resistance, and volume stability that
are important in hardened concrete (5). Hydration can also cause undesirable
effects such as concrete shrinkage. Because of the desirable properties
developed during hydration, civil engineers are very interested in achieving
good hydration. This is done through a process known as curing. “The aim of
curing is to ensure as much hydration as possible at reasonable cost” (6). Good
curing provides the necessary temperature, moisture, and time needed for
hydration, and in return, “all of the desirable properties of concrete are
improved” (5). The required level of temperature, moisture, and time
experienced by concrete during curing depends on the particular concrete
mixture and the conditions under which the concrete is placed (7). There are
many methods that can be used to provide satisfactory curing, and several
studies have been done to determine the appropriate curing method for
different types of concrete. For instance, it is more important to moist cure fly-
ash concrete than normal concrete (8), and water curing concrete for 6 days
after demolding reduces drying shrinkage in hot and arid climates (9). A good
reference for proper curing requirements by different methods and for various
construction types is AC1 308.1-98, Standard Specification for Curing
Concrete, found in the AC1 Manual of Concrete Practice.
@Seismicisolation
@Seismicisolation
Preparation of Specimens
A total of eight (8) specimens, formed into 50 x 50 x 50 mm (2 x 2 x 2 in.)

cubes, were prepared for this experimental study. Seven of the cubes were
made of mortar with sand as the aggregate, and the eighth cube was made only
of paste (cement and water). The number of cubes required was determined
prior to the experiment based on the following criteria. One mortar cube was
designated as the control cube, and was monitored for spectral response
throughout the testing program. Three other cubes were monitored for changes
in internal temperature of the mortar using a temperature sensor placed
approximately 25 mm (1 in.) from the surface. Three more mortar cubes were
needed for strength tests after seven, fourteen, and twenty-eight days of curing.
The last cube was made only of paste and was also monitored for internai
temperature change.
Materials and Mortar Mixture
The aggregate used for the mortar mixture was an oven-dried, poorly-graded
Missouri river sand composed mostly of chert. Other grain size analysis
results include a Dl0 of O.lmm (0.004 in.), a coefficient of curvature of 1.15,
and a coefficient of uniformity of 2.38. The cement was a Type I portland
cement. This mixture had only fine aggregate (sand) with a sanucement ratio
of 3.5 by weight. Tap water was used, and the final mortar mixture had a w/c
of 0.52. No admixtures or curing agents were included in the mortar mixture.
Testinn Eauiument
A non-imaging spectroradiometer, the Fieldspec Pro FR (lo), was used to

obtain the spectral signatures. This particular instrument has a range of 350 to
2500 nanometers (the visible to short wave infrared part of the EM spectrum)
with a spectral resolution of 10 nanometers. Also, a System 4101 Concrete
Maturity Meter (model H-2680) was used to monitor the temperature of four of
the curing cubes throughout the testing program. Strength testing of the cubes
was performed with a Tinius Olsen Testing Machine. The eight prepared
cubes and the associated maturity meter are shown in Fig. 2. The spectral
testing setup (including the spectroradiometer, portable computer, light
sources, and control cube) is shown in Fig. 3.
@Seismicisolation
@Seismicisolation
722 Kindenrater et al.
Suectral Signatures and Schedule
A schedule to collect the spectral signature of each cube during curing was
developed. Because the hydration process is most rapid during the first three
to seven days (3,spectral signatures were collected at more frequent time
intervals during the first seven days of curing than during the remainder of the
experiment. Both the control cube and the paste cube were scheduled for
spectral readings at each time indicated. Spectral readings of the seven mortar
cubes were scheduled such that at least one cube was monitored each day.
These readings were intended to check the progress of the cubes in comparison
with the control cube. The comparisons showed no apparent difference in
temperature or spectral response.
RESULTS AND INTERPRETATION
The results of this experiment show several characteristics of the spectral

signature of curing mortar. One of the first noticeable characteristics is that,
for this particular mortar mixture, the signature of the mortar is definitely
controlled by the cement component of the mixture. Fig. 4 shows just how
closely the signatures of the mortar and the unhydrated cement resemble each
other. At an age of 114 days, the two are indistinguishable. The shape of the
signature of the sand used in the mortar mixture does not seem to contribute
significantly to the shape of the mortar cube’s spectral signature.
A second characteristic of the mortar’s spectral response is its increased

intensity of reflection as the mortar cures. This behavior was expected since it
is well known that water absorbs energy evenly across the EM spectmm. Fig.
5 shows the spectral signature of the surface of the control cube during the first
28 days of curing. Not all collected signatures are shown, but it can be seen
from this selection that the general trend is for the intensity of the curve to
increase as the mortar cures. Notice the large increase in intensity between the
4-day (96 hrs) and 6-day (144 hrs) signatures. The cubes were kept wet during
the first 96 hours of curing using wet towels and water mist before being
allowed to slowly air dry. Over the next few days, the mortar continued drying
as the hydration process progressed without added-moisture, and the intensity
of the spectral signature increased considerably.
The two characteristic dips in the spectra at 1450 nm and 1950 nm should also
be noticed. These two areas are the water absorption bands and can be found
in almost all mineral compounds (2). Both dips changed shape as the mortar
@Seismicisolation
@Seismicisolation
cured, but the band around 1450 nm seems to change significantly at a later
age. The depth of an absorption band is parameter used to characterize its
shape and can be measured by taking the difference between the intensity at
the top of the dip and the intensity at the bottom of the dip. Not only does the
depth of the 1450 nm band change from 0.04 to 0.01, but the width of the band
also gets much smaller. However, the width does not exhibit a drastic change
until after the mortar has dried out (after 6 days). Both bands do seem to have
a minimum depth that can be reached, as can be seen from the relatively
constant depth after 12.5 days (shown in Fig. 6). The top and bottom of the
1450 band were taken at 1380 nm and 1416 nm, respectively. For the 1950
band, the top and bottom considered were at 1850 nm and 1950 nm. For
example, the intensities at 1850 nm and 1950 nm after 316 hours were 0.5213
and 0.3916. The difference in these two values gives a depth of 0.1297.
Fig. 7 shows another view of how the intensity in the mortar signature changed
with time. Here, four particular wavelengths are shown. Two of the
wavelengths are in the water absorption bands (1450 and 1950 nm) and can be
expected to behave similarly. For the other two wavelengths, one short
wavelength (612 nm, in the visible range) and one long wavelength (2180 nm,
in the infrared range) were chosen. Comparison of the intensity trends among
these four wavelengths shows that each specific wavelength in the signature
appears to have changed at the same rate and in the same manner during
curing.
Fig. 7 also shows a peculiar dip in the middle of the intensity increase between
100 and 200 hours. Only two data points out of the 50 points in the data set
make up this dip, so it is not reasonable to consider this a definite feature for
this experiment.
The relationship between the internal temperature and the spectral signature of
the mortar was also investigated, because the amount and rate of heat evolution
from the mortar is an indication of the reaction occurring during the curing
processes (6). Fig. 8 illustrates this relationship.
The reflectance intensity line in Fig. 8 represents the intensity of the mortar
signature at 612 nm. As mentioned previously, all the wavelengths seemed to
change in the same way, so this wavelength can be considered to be
representative of the entire signature. A technique to define the trend of a
dataset called the "moving average trend" was used for the raw temperature
data. Since the temperature was recorded in 1-hour increments with a n
accuracy of one degree Celsius (1.8" Fahrenheit), it produced a very blocky
graph with a stair-step pattern that is difficult to interpret. The moving average
curve represents the raw temperature data well and is much easier to compare
to other data, such as the measured reflectance intensity. The temperature line
in Fig. 8 is the moving average trend (with a period of 30) of the internal
@Seismicisolation
@Seismicisolation
Kindenrater et al.
temperature of the mortar cube. This type of trend uses previous data in the
complete data set to determine a trend and predict the next point in the data set
using the following equation:
Ft = (A, + At., + ... + Ai-"+,)I n (1)
In Equation 1, At is the actual data value at time t, n is the number of values in

the period, and F, is the predicted value at time t. These predicted points,
based on the actual data, are what make up the moving average trend line.
The shape of the temperature curve is consistent with the shape of the typical
rate of heat evolution during cement hydration (see Fig. 9) except for the final
increase in temperature beginning around 550 hours when the hydration
reaction should have been proceeding very slowly. The two bumps in the
temperature curve from the experiment could be characteristic of the hydration
of tricalcium silicate (C3S) and tricalcium aluminate (C3A) when the cement
hydration is chemically controlled (6). Individually, these processes occur
around 10 and 40 hours, respectively, but this can vary depending on the
mortar mixture and the quality of the components. However, it does not seem
likely that the hydration reactions would be so far into the life of the mortar as
shown in the results of this experiment. Because the recorded data are actually
internal mortar temperatures, it may be better not to look for a correlation here.
The internal mortar temperature may not accurately reflect the amount of heat
given off by the mortar during curing, which is the heat evolved. Actual heat
evolution data would be better for a direct comparison.
It should also be noted that the increase in internal mortar temperature between
100 and 200 hours, as shown in Fig. 8, is accompanied by an increase in the
intensity of the spectral signature of the mortar surface. However, there is no
accompanying intensity decrease to go along with the decreasing temperature
trends or dips at the 350 and 550 nm. Nor is there a second intensity increase
with the second (or third) temperature increase.
CONCLUSIONS
One of the expected findings of this study was the overwhelming effect of the
portland cement on the signature of the mortar mixture. Even though the
amount of fine aggregate was more than three times the amount of cement, the
cement controlled the signature of the mixture. This can most easily be
explained by the effective coating of the fine aggregate by the cement. The
spectral signature of the mortar used in this experiment was distinctly related
to the signature of its constituent cement, a Type I portland cement. More
studies should be undertaken to see if this is a trait of all mortars and concretes
@Seismicisolation
@Seismicisolation
or a particular condition of the mixture used herein. Once a generalization
about all mixtures can be made, it could be a good method of determining the
type of cement in existing concrete mixtures, without the need for further
material characterization.
Results revealed that there is a relationship between the stage of curing of

mortar and its spectral signature. The most easily recognized characteristic of
this relationship is the way the spectral signature reacts to the free water
content of the mortar. The intensity of the signature increases as hydration
progresses until a certain point where the signature becomes relatively constant
and is controlled by the unhydrated cement. The water absorption bands also
experience changes in their shape until the signature becomes constant with
time, indicating that the material possesses a specific ability to absorb a certain
amount of light through its water content. The point in time at which this
occurs may be related to the water content of the mortar.
It also appears that there may be a relationship between the internal

temperature of curing mortar and its spectral signature. These studies are
preliminary in nature and could be the basis for more extensive and rigorous
research, including studies of the effects of larger aggregate on the spectral
properties of concrete. It was the intent of the authors to present these early
findings to the AC1 community.
ACKNOWLEDGMENTS
The work reported herein was sponsored by a grant from the Office of
Academic Affairs and by the Department of Civil Engineering, University of
Missouri-Rolla. The authors are thankful for valuable assistance from their
fellow students and faculty members in their Department.
CITED REFERENCES
1. Remote Sensing Tutorial. Hp. Dec. 1999. Online. National Aeronautics

and Space Administration. Available: http://www.rst.nasa.gov. 5 May
2000.
2. Kruse, F. A. “Introduction to Hyperspectral Remote Sensing Concepts and

Sensors.” Analytical Imaging and Geophysics. Hyperspectral Data
Analysis and Image Processing Workshop. Boulder, Colorado. July 6-9,
1999.
@Seismicisolation
@Seismicisolation
3. Clark, R. N., Trude, V.V. King, M.K., Swayze, G.A., and Vergo, N.,
“High Spectral Resolution Reflectance Spectroscopy of Minerals”.
Journal of Geophysical Research. V. 95, No. B8, Aug. 10, 1990, ~12653-
12680.
4. “NIR Analysis of Concrete Samples”. Application Note 109.01E. Online.

Analytical Spectral Devices, Inc. Available: ftp://ftp.sni.net/asd. 5 March
2001.
5 . White, G.R. Concrete Technologv. Van Nostrand Reinhold Company.

New York: 1977, pp. 86-88.
6. Mindness, S. and Young, J.F., Concrete. Prentice-Hall, Inc. New Jersey:

Englewood Cliffs, 1981, pages 76-86,302-307.
7. Marotta, T.W. and Herubin, C.A., Basic Construction Materials. 5”

edition. Prentice Hall, Inc. New Jersey: Upper Saddler River, 1997, page
203.
8. Gopalan, M.K. and Hague, M.N., “Effect of Curing Regime on the

Properties of Fly-Ash Concrete”. ACI Materials Journal. V. 84, No. 1,
Jan. 1987, p14-19.
9. Fattuhi, H.I. and Al-Khaiat, H., “Shrinkage of Concrete Exposed to Hot

and Arid Climate”. Journal of Materials in Civil Engineering. V. 11, No.
1, Feb. 1999, p66-75.
10. Analytical Spectral Devices, Inc. Online. Available: http://www.asdi.com

29 April 2002.
@Seismicisolation
@Seismicisolation
0.01-0.4~m
Wavelength: < 0.00
f...)
Near Infrared
0.4-0.7~m 0.7-1.5pm
Fig. 1. The Electromagne pectrum (note: lmm = 1000 nm)
Fig. 2. Mortar specimens with maturity meter.
@Seismicisolation
@Seismicisolation
~~
Fig. 3. Equipment setup.
0.7 I i
0.6
0 0.5
o
E
(D
U
$ 0.4
=
E
œ 0.3
E
0
e
g 0.2
0.1
O !
350 850 1350 1850 2350
Wavelength (nm)
sand O unhydrated cement + mortar at 6 days A mortar at 114 days

Fig. 4. Spectral signatures of sand, unhydrated cement, and control cube.
@Seismicisolation
@Seismicisolation
0.6
0.5
h
0.4
y
C
m
c
0.3
u
Q
E:
E 0.2
o.1
O
350 850 1350 1850 2350
Wavelength (nm)
0intial 0 2 days Q 4 days A 6 days rn 8 days + 16 days 0 24 days A -28 days
Fig. 5. Spectral signatures of the control cube.
0.3
I
0.25
h
al
o
S
;0.2
al
r
2!
0.15
v
5
a
p 0.1
U
S
o
0.05
-1450 band ------.I950

band
Fig. 6. Depth of absorption bands.

@Seismicisolation
@Seismicisolation
0.6
0.5
h
o
o
5
c
0.4
o
0.3
8
v
.-h
U
2 0.2
-d
S
o. 1
o l t I
O 100 200 300 400 500 600 700

firne (hours)
Fig. 7. Variation in reflectance intensity at different wavelengths,
25 I 0.5
24
I I
-- 0.45
23 -- 0.4
A
c) 22 al
Y -- 0.35
5
c
21
-- 0.3
Q
p
c
m
5 20 c
-- 0.25 å!
E 19 8
8 -- 0.2
18
17 -- 0.15
16 o. 1
O 100 200 300 400 500 600 700
Time (hours)
-_..._.
reflectance intensity at 612nm -temperature
Fig. 8. Variation in temperature and reflectance during curing.

@Seismicisolation
@Seismicisolation
Stage I
/ +-Stage II- Stager 111and IV c-Stage V
C,S hydration
I
-,I I
Time (hours)
-
Fig. 9. Typical heat evolution curve for the hydration of portland cement (6).
@Seismicisolation
@Seismicisolation
SP 209-40
The Performance of Fiber

Reinforcement in Concretes
Exposed at High Temperatures
by G. Giaccio and R. Zerbino
SvnoDsis: The benefits derived from the ability of the fibers to control crack prop-
agation have been recognized for many years. In addition, the development of
high-performance concretes has enhanced this situation as the increases in strength
lead to a more brittle behavior of the material. The introduction of steel-fiber rein-
forcement in these concretes is probably the best way to improve the performance
of concrete when higher tenacity is required. This paper shows the contribution of
fiber reinforcement in both conventional and high-strength concretes exposed to
temperatures up to 500°C. Concretes with different types and content of fibers are
analyzed, mainly regarding the failure mechanism and tenacity. The post-peak be-
havior under conpressive and flexural loads is studied using a close loop system.
NDT was also used to evaluate the damage. The residual mechanical properties of
fiber-reinforced concretes are affected in a similar way than those corresponding to
plain concrete. Nevertheless, it can be seen that the residual parameters tend to in-
crease as the strength increases when high carbon-steel fibers instead of low car-
bon-steel fibers are used, and when fiber reinforcement is introduced.
Kevwords: com pression; flexure; high perfor mance

concrete; high temperature; steel fibers; toughness
@Seismicisolation
@Seismicisolation
733
734 Giaccio and Zerbino
Graciela Giaccio: Professor at the Civil Eng. Dept. of La Plata Univ. (UNLP).
Researcher of the Commission of Scientific Research, Bs As Prov. CIC-LEMIT.
Argentina. Her main research interests are additions, failure mechanisms, HPC.
Raúl Zerbino: Professor at the Civil Eng. Dept of La Plata Univ. (UNLP).
Researcher of the Nat. Council of Scientific and Tech. Research, CONICET.
Argentina. His main research interests are deformability, fracture and FRC.
INTRODUCTION
The use of fiber-reinforced concrete (FRC) has increased progressively

over the years. It has appeared as an excellent alternative to improve the
performance of concrete structures where it is required to control the formation
and propagation of cracks or when they are going to be exposed to dynamic
actions. The development of high-performance concretes has enhanced this
situation as the increases in concrete strength lead to a more brittle behavior of
the material. The introduction of steel-fiber reinforcement is probably the best
way to improve the performance of concrete when higher tenacity is required.
Concrete structures may be exposed to high temperatures, by accidental

causes or by the characteristics of the structurai application. As a consequence,
concrete undergoes changes that may result, in many cases, in extensive
cracking. In this sense, it is interesting to study the contribution of fibers to
control crack formation and propagation.
Many researchers have verified the degradation of the microstructure of

concrete due to the exposure to high temperatures and its effects on the behavior
of concrete, the changes being a function of the heating and cooling cycle (1 -8).
Most of the studies refer to the compression behavior and report reductions in
the ultrasonic pulse velocity, the dynamic modulus of elasticity, strength, etc.
The evaluation of the tensile strength shows a great variability, and it is verified
that it is more sensitive than the compression strength to the changes produced
by the exposure to high temperatures. The type of aggregate used plays a critical
role (9,lO). A study on the residual properties in flexure and direct tension of
high-strength concretes exposed to high temperature veri@ the decrease in
tensile strength and toughness, the increase in characteristic length, and only
minor effects on the fiacture energy ’.
’ Felicetti, R. and Gambarova, P. G. “On the residual tensile properties of high
performance siliceous concrete exposed to high temperature”, in the Proceedings of the
International Workshop in honor of Z.P. Bazant‘s Anniversary (Prague, 1998), pp 1-20.
@Seismicisolation
@Seismicisolation
Recently the authors (1 1) have studied the residual mechanical
properties of plain concrete in tension. It was found that the degradation of the
structure notably affects crack propagation and that this is reflected in the shape
(ascending and softening) of the load-deflectioncurves. While damage produces
a drop in strength, the reduction of the fracture energy is less important, due to
the greater branching in crack propagation. The residual tensile strength appears
to be more affected by the cracking than the compression strength.
This paper presents some results on the residual mechanical properties of

fiber-reinforced concretes exposed to high temperatures. Flexural and
compressive tests were performed including the post-peak behavior.
EXPERIMENTAL
Materials
Fiber-reinforced concretes (FRC), with compressive strengths over 50

MPa, were prepared by adding the desired dosage of each fiber type into a base
concrete. Two types (low and high carbon) of hooked end collated steel fibers,
with aspect ratio 80 and different lengths were used. Granitic crushed stone,
siliceous river sand, silica fume (BET 22800 m2/kg), two cements and a
naphthalene superplasticizer were used. Table 1 shows the main characteristics
of the fibers and the cements used.
Concretes A-50 and B-50 (cement a) were prepared with the same type
and dosage of fiber reinforcement (50 kg/m3 of fiber L) and different strength
level. The effect of fiber content is analyzed comparing two FRC (C-40 and C-
80) prepared with the same mixture proportions and 40 or 80 kg/m3 of H fiber
respectively. A plain concrete (C) was also made. Concrete mixture proportions
are shown in Table 2.
Experimental Details
To attain uniform curing conditions and avoid the superposition of other

factors (different hydration rates, moisture conditions, etc.) on the effect of high
temperatures, all specimens were cured in a fog room (23 OC, 95 % RH) during
60 days and then they were stored in laboratory ambient. At the ages of 7
months (A-50 and B-50) and 4 months (C, C-40 and C-SO), specimens were
heated in an electric oven, and kept at maximum temperature, 150 or 500 OC, for
24 or 1 hour respectively. Heating rate was 100 "C/h. After heat treatment the
specimens were kept 30 days in the laboratory and tested. All tests were
@Seismicisolation
@Seismicisolation
performed at the ages of 8 months for concretes A-50 and B-50, and 5 months
for C, C-40 and C-80.
The damage produced by the exposure to 500 OC can be considered

representative of a degradation due to fire. On the other hand, the exposure to
150 OC could be associated with a condition of structural application, as the
shrinkage measured during this treatment (24 hr at 150 OC) was in the order of
350 microstrains.
To estimate the alteration produced during the treatment, loss of weight,

and variations in ultrasonic pulse velocity and dynamic modulus of elasticity
were measured (9).
Flexure tests
Several standards have been proposed to evaluate the behavior of FRC.

Some of them as ASTM C 1018 or AENOR UNE 83-510 (12,13) use prismatic
specimens with four point load and evaluate the relative toughness comparing
the areas below the load-deflection curve.
ASTM C 1018 defines several non-dimensional indices (In) that are

ratios of areas up to certain displacements divided by the area up to first-crack,
where n indicates the value obtained for In in the case of an elasto-plastic load-
deflection curve, According to ASTM, first-crack is “the point on the load-
deflection curve at which the form of the curve first becomes non linear” or “the
point at which the curvature first increases sharply and the slope of the curve
exhibits a definite change”. An alternative definition given by the NBN B 15-
238 (14), is based on the maximum load achieved before the intersection of the
load-deflection curve with a line drawn parallel to its initial linear part and
displaced 0.05 mm from the origin. Since these indices are based on the first-
crack, any difference in the identification of this load leads to significant
changes in the values of the indices (15-17).
More recently it has been proposed to use notched specimens to evaluate

the increase of the width of the fissure (CMOD), with the advantages of using
this variable as the parameter for test control (18,19). The advantages of central-
point loading, avoiding the asymmetries in the measures of the deflections that
are produced after the formation of the first-crack when third-point loading is
used, were also considered.
RILEM TC 162-TDF recommendation (20) adopted a center-point

loading arrangement on notched beams measuring deflections and, as an option,
crack mouth opening displacement (CMOD). Based on the load-displacement
curves, the limit of proportionality and two equivalent flexure tensile strengths
@Seismicisolation
@Seismicisolation
which identify the behavior of the material up to a selected deflection are
calculated. An offset of 0.05 on beams 150 high and 450 mm span is proposed.
The equivalent tensile strength (feq) represents an average post-peak

strength over a range of CMOD (w). The values of feqcan be used in the drop-
constant models of the tensile constitutive stress-crack opening, where the o-w
curve is modeled by a vertical drop from the tensile strength and a constant
residual strength (21).
In these experiences, the flexural behavior was analyzed following the

general guidelines of RILEM TC 162-TDFrecommendation. Prisms of 75 x 105
x 430 mm were tested using center-point loading configuration over a span of
350 mm. Each beam was turned 90" from the casting surface and then sawn at
mid-span. The depth of the notch was 20 mm. In all cases the notch was made
two days before the tests were performed. As smaller beams were used, an
offset of 0.035 mm was adopted (instead of 0.05), identifying as FU the
deflection at the limit of proportionality Fu (maximum load in the interval). The
equivalent tensile strengths feq,2and feq,3, calculated from the mean load
estimated from the area below the curve until deflections 82 (calculated as 82 =
FU + 0.455mm) and 83 (calculated as 8 3 = 8 ~ +" 1.855mm)were obtained.
The tests were controlled by the average of the central deflection using
an initiai rate of 0.01 &min which was increased after the peak-load. The
central deflection was measured by LVDT mounted at mid-span on two
independent rigid rods that were fixed to each side of the beam, at mid-height,
above its supports (with pins that permit rotation perpendicular to the beam
axis). The sensors measure the displacement of the bottom (tensile face) of the
beam with respect to the rod. The rollers at the supports and loading points, as
well as the rest of the loading setup, were designed according to ASTM C 1018.
This system eliminates torsional effects due to imperfections of the molds and
non-symmetric deformations during loading (Fig. I). The CMOD was also
measured with a clip gage. Load - deflection and load - CMOD curves of sound
and damaged fiber-reinforcedconcretes were compared.
Compressive tests
Compressive tests were performed in a 2MN hydraulic machine using

cylinders of 150 x 300 mm, testing a minimum of 3 specimens. The load was
measured with a load cell mounted between the actuator and the loading points;
axial and lateral deformations were simultaneously recorded by LVDT mounted
on the system usually employed for the determination of the elastic and Poisson
coefficients (ASTM C 469). Three loading cycles up to 40 % maximum stress
were applied during the test before achieving maximum load. Strength, elastic
static modulus, and initiation (finit) and critical stresses (fcrit)were calculated.
@Seismicisolation
@Seismicisolation
Initiation stress is defined as the stress at which there is a clear increase in
lateral / axial strain ratio, and it is associated with the beginning of cracking
through the matrix. Critical stress is defined by the minimum in the volumetric
curve and it can be associated in plain concrete with the unstable propagation of
cracks through the matrix.
Cylinders of 100 x 200 mm were also used in compressive tests of

concretes Cy C-40 and C-80. These tests were performed in a servohydraulic
INSTRON 1O00 kN system under closed-loop control including post-peak
behavior with the purpose of comparing the toughness of the material. Axial and
lateral deformations were simultaneously recorded by a clipgage and a LVDT
mounted on the device employed for the determination of the elastic and
Poisson coefficients (ASTM C 469). The signal of the axial strain was used to
control the test (deformation rate was 0.5 mm/m.s). Again, three loading cycles
up to 40 % maximum stress were applied during the test before maximum load
was achieved. Strength and elastic static modulus were measured. Post-peak
behavior was recorded and the area under the stress-strain curve up to peak-
load, and 80 % and 60 % of the maximum load of the descending branch were
calculated and compared as a measure of toughness.
TEST RESULTS AND ANALYSIS
The results obtained on reference concretes, identified as Control, and

concretes exposed to high temperatures are analyzed in this section. The residual
properties of concretes exposed to 150 and 500 "C are identified with a
superscript and superscript 'O0 respectively.
Non destructive tests
Non destructive tests are an useful tool that indicate the grade of damage
produced by high temperature in concrete, and they allow the verifcation of the
degradation in specimens of different shape and size. Previous work on plain
concretes affected by high temperature (9,10,11) has shown that the resonant
frequency method is more sensitive than the ultrasonic pulse velocity to detect
damage in elements exposed to temperatures up to 700 OC.
Tables 3 to 5 include values of ultrasonic pulse velocity (UV), dyn&ic

modulus of elasticity (Ed) and weight loss after heating, measured before
flexure and compressive tests.
There are large reductions in pulse velocity in concretes exposed to 500

OC. Reductions of velocity ranged between 25 and 40 % in beams and between
@Seismicisolation
@Seismicisolation
33 and 67 % in cylinders of 150 mm diameter. The alteration decreased in
cylinders of 100 mm diameter (between 22 and 31 %). In concretes exposed to
150 "C only minor differences (<lo %) can be detected. The dynamic modulus
of elasticity changed in a similar way as ultrasonic velocity, the residual values
being lower than 80 % at 150 OC and 50 % at 500 OC, both in beams and in 150
mm cylinders.
NDT tests indicate a greater alteration in the internal structure of FRC,

the same as loss of weight. In this sense, it could be assumed the presence of
defects at the fiber-matrix interfaces. Nevertheless, as it will be discussed, the
residual mechanical performance was better in FRC.
Flexure tests
Modulus of rupture (MR) calculated using the maximum load, and the
other parameters that characterize FRC behavior, the first crack strength (ffct,fl)
and the equivalent tensile strength (feq,2and fq3) are given in Table 3. They
were measured on notched beams following the general guidelines of RILEM
recommendation.
It must be noted that concrete C-80, prepared with high-carbon steel

fibers, achieved a MR of 14.5 MPa and equivalent strengths-(feq,land feq,3)over
11 MPa.
The residual mechanical properties of FRC are affected in a similar way

than those corresponding to plain concrete. The residual parameters in FRC tend
to increase as strength increases, as it appears when comparing A-50 and B-50.
In a similar way, the incorporation of fibers leads to an increase in the residual
strength (C vs C-40 or C-SO). Finally, it can be noted that concretes prepared
with high-carbon steel fibers (H) tend to show rather higher residual values.
The residual flexural strength of concrete C was 3.5 MPa (53 % of the
sound concrete) while in C-40 and C-80 it increases up to 7 and 9.2 MPa (70
and 63 % of the sound concrete). These FRC present values of residual
equivalent strengths over 5 and 7 MPa respectively for the most severe exposure
condition. As it was expected, the matrix degradation leads to a reduction of the
residual mechanical properties of concretes.
Figure 2 presents the load-displacement curves corresponding to sound

and damaged fiber-reinforced concretes (A-50, B-50 and C-80). Both load -
deflection and load - CMOD curves are represented. It also includes the flexure
behavior of concrete C.
@Seismicisolation
@Seismicisolation
In FRC exposed to 150 OC the shape of the curves is similar to the
undamaged concrete. The differences are specially notable at first-crack level
indicating that the fibers are particularly effective after matrix cracks. After first
crack, there is an increase in inelastic deformation, then the action of fibers
strongly increase the concrete loading capacity. The reductions in residual
flexural strength were lower in FRC than in plain concrete.
For the most severe condition (500 OC) the degradation of the material is
reflected by an increased non-linearity, nevertheless the FRC still exhibit a
strengthening type behavior and they keep an almost constant load capacity
during the post-peak, even exceeding CMOD values of 3 mm. In most cases the
maximum load is achieved at deflections as high as 1 mm, that is 10 times
higher than plain concrete.
Comuressive tests
Limited work exists on the effect of fibers on the compressive behavior

of concrete because the increases in strength are small and the effects of
variables such as fiber aspect ratio or volume are negligible. Nevertheless, it is
evident that the addition of steel fibers changes the mode of failure with respect
to plain concrete (22,23).
The development of the failure process under compressive loads can be

followed by means of the analysis of the stress - strain curves. Based on the
transversal, axial and volumetric strains, the initiation and critical stresses can
be obtained, representing the starting of matrix crack growth and the unstable
propagation of cracks in concrete. When fibers are incorporated in concrete,
these stresses tend to increase. The fibers delay the starting of the growth of
cracks at the matrix and extend the period of crack propagation, leading to a
higher failure stress (24).
The effect of fiber reinforcement on the compressive behavior of High-

Performance Concrete was similar to that observed on normal concrete,
confirming that the fibers enhance crack control both differing crack formation
and extending the period of crack propagation.
Previous work on the behavior of plain concrete affected by high

temperature (9,ll) has shown that the exposure to 500 "C caused significant
damage to the internal structure.
The results of compressive tests performed on 150 x 300 m m cylinders

are given in Table 4. Fig. 3 presents the initiation stress (finit), the critical stress
(fcrit), and the compressive strength (f'c) of each concrete for the different
exposure temperatures.
@Seismicisolation
@Seismicisolation
It can be seen that the residual strengths of concretes exposed to 150 OC
are similar to those ones corresponding to the reference concrete (always at 20
OC). The difference is lower than 10 %. When concretes were exposed to 500 OC
the residual strengths significantly decreased ranging between 60 and 78 %.
Residual compressive strengths of fiber-reinforced concretes are very similar or
slightly higher than the residual strength of plain concrete, as it can be seen
comparing C and C-40 or C-80.
The alteration of the concrete structure as it affects the failure

mechanism, is reflected in the values of initiation stress and critical stress. When
concretes are exposed to 500 OC, finit and fcritdecrease, being the decrease in fcrit
more significant. This means that, due to the cracking generated by high
temperature, the growth and propagation of matrix cracks start earlier. The
period of stable crack propagation in the mortar is reduced (given by the
difference between fcrit- finit) and, although the period of unstable crack growth
is extended (given by the difference between fc - fCr& the capability of
controlling crack propagation decreases leading to premature failure (9). This
behavior is verified in concrete C (Fig. 3). When the exposure temperature was
150 OC, the residual stresses show a small decrease following the same
tendencies.
In the case of fiber-reinforced concretes exposed to high temperature,

they present similar behavior as plain concretes. Comparing concretes C, C-40
and C-80, it can be seen that the presence of fibers leads to values of finit , fcrit
and f'c slightly higher, and they tend to extend in relative terms the periods of
crack propagation, specially when high contents of fibers are used.
The same as occurred in plain concrete, the reductions in the modulus of

elasticity are higher than in compressive strength. The results summarized in
Tables 3 and 4 show that the residual compressive strength tends to be higher
than the residual modulus of rupture. This is in accordance with the results
obtained in other experiences on the behavior of plain concrete affected by high
temperature (1 1). It must be mentioned that this effect was more important when
exposure temperature was 150 OC.
Table 5 shows the results of compressive tests performed on cylinders of

100 x 200 mm of concretes C, C-40 and C-80. To study the post-peak behavior
a system under closed-loop control was used. The compressive strength (fc),
the static modulus of elasticity and three values of toughness (Tpk, Tso and T60),
calculated as the areas below the axial strain-stress curve up to peak load, up to
80 % and 60 % of maximum load on the descending branch are shown.
As it can be seen, the changes in f'c and E follow the same tendencies
observed in Table 4 (cylinders 150 x 300 mm), being rather lower the residual
values of concretes exposed to 150 "C. (It must be noted that the modulus of
elasticity was measured using the same device (gage length 125 mm) in both
@Seismicisolation
@Seismicisolation
size cylinders, then there appear some differences between the values
corresponding to small and big specimens).
Comparing the behavior of control and damaged concrete C, the values

presented in Table 5 show that the residual compressive strength was 77 and 81
% for temperature exposition equal to 500 and 150 O C , and the residual modulus
of elasticity was 50 and 73 %, respectively This is consistent with the results
obtained in previous studies. Regarding the effects of fibers in damaged
concrete, in FRC (C-40 and C-80) the residual values tend to be higher at 150
OC and they remain equal at 500°C. In all cases the higher residual values
correspond to C-80, this means the highest fiber content.
Figure 4 shows the stress vs. axial and lateral strain curves corresponding
to plain and FRC with the higher content of fibers (concretes C and C-80
respectively). The stress - strain curves correspond to the specimen which best
fits the average value. They include the post-peak branch and allow the
discussion of some characteristics of concrete structure and the failure process.
Plain and fiber-reinforced concretes behaved in a different way. After the peak-
load was achieved, in sound concretes, the descending branch in fiber-reinforced
concrete was more gradual.
The differences mentioned are clearly reflected in the calculated

toughness (Table 5). At peak-load, the values of Tpk corresponding to C, C-40
and C-80 are similar. On the contrary, comparing the toughness Tgo, it increases
as fibers are incorporated, having greater increments as the content of fibers
increase. When T60 is considered, these differences are even higher.
Figure 4 shows that there were significant changes in the shape of the
curves of plain concrete C as a result of cracking. After the peak-load is
achieved, the descending branch in Control specimens C is less gradual than in
damaged concretes, especially concrete exposed at 500 "C. The values of Tpk are
lower in damaged concrete, but as unloading continues, the toughness increased.
Again, in FRC a different stress-strain behavior occurs when exposed at a high
temperature -500 OC- (Fig. 4). C-80 shows a higher deformation at peak-load
and a more extended and gradual descending branch. As a result, the toughness
increases even at peak-load, being the residual Tpk in the order of 175 %. Plain
and fiber-reinforced concretes exposed to 150 OC had lower Tpk, Tgo or T60 than
the reference concretes.
CONCLUSIONS
The residual mechanical behavior under flexure and compressive loading

of FRC with compressive strengths over 50 MPa, exposed to high temperatures
was studied. From these tests the following conclusions can be drawn:
@Seismicisolation
@Seismicisolation
- The flexural residual parameters in FRC tend to increase as fibers are
incorporated and when concrete strength increases. Concretes with high carbon
steel fibers tend to show rather higher residual values than those prepared with
low carbon steel fibers.
- In FRC exposed to 150 OC, the shape of the curves is similar to the undamaged
concrete. After first-crack there is an increase in inelastic deformation. The
reductions in residual flexural strength were lower in FRC than in plain
concrete.
- For the most severe condition of exposure (500 OC), the residual flexural
strength was near 60 % and the residual equivalent tensile strength ranged
between 60 to 70 % of the sound concrete. The degradation of the material is
reflected by an increased non-linearity, nevertheless the FRC still exhibited a
strengthening type behavior and kept an almost constant load capacity during
the post-peak, even exceeding CMOD values of 3 mm.
- Fiber-reinforced concretes exposed to high temperature present similar
compressive behavior as the plain concretes. The presence of the fibers leads to
values of finit , c i t and f c slightly higher, and they tend to extend, in relative
terms, the periods of crack propagation, specially when high contents of fibers
are used.
- In FRC exposed to high temperature, the same as occurs in plain concrete, the
reductions in the modulus of elasticity are higher than in the compressive
strengths. In a similar way, the residual compressive strength tend to be higher
than the residual modulus of rupture. This effect was more important when
exposure temperature was 150 “C.
- NDT tests indicated a greater alteration in the internal structure of FRC, the
same as loss of weight. In this sense it could be assumed the presence of defects
at the fiber - matrix interfaces. Nevertheless the residual mechanical
performance were equal or better in FRC.
REFERENCES
1. Bazant, Z. and Kaplan, M. “Concrete at High Temperatures’’ (Longman,

First edition, Essex,. UK, 1996).
2. Abrams, M.S. “Compressivestrength of concrete at temperatures to 1600
OF”, in “Temperature and Concrete”, AC1 SP-25, 1971, pp 33-58.
3. Zoldners, N.G., Malhotra, V.M. and Wilson, H.C. “Effect of high

temperature on concretes incorporating different aggregates, Proc. ASTM
Jour. 60,1960, pp. 1087-1108.
4. “Concrete for Nuclear Reactors”, AC1 SP-34 (American Concrete Institute,
Detroit, 1972).
@Seismicisolation
@Seismicisolation
5 . Baker, G. “The effect of exposure to elevated temperatures on the fracture
energy of plain concrete” Materials & Structures, V. 29, N”190, RILEM,
París, 1996, pp. 383-388.
6. Papayianni, M. and Valiasis, T. “Residual mechanical properties of heated
concrete incorporating different pozzolanic materials” Materials &
Structures, V. 24, N0140, RILEM, Park, 1991, pp. 115-121.
7. Zhang, B. Bicanic, N., Pearce, C.J., and Balabanic, G. “Residual fracture
properties of normal- and high-strength concrete subjected to elevated
temperatures” Magazine of Concrete Research, V. 52, N02,2000, pp. 123-
136.
8. Mohamedbhai, G.T.G. “Effect of exposure time and rates of heating and
cooling on residual strength of heated concrete” Magazine of Concrete
Research, Vol 38, N”136, 1986, pp. 151-158.
9. Barragán, B., Di Maio, A., Giaccio, G., Traversa, L. and Zerbino, R.
“Effects of high temperature on residual mechanical and transport properties
of concrete”, Proc. Fifth CANMET/ AC1 Int. Conference on Durability of
Concrete, AC1 SP-192, Barcelona, 2000, pp.983-1000.
10. Di Maio, A., Giaccio, G. y Zerbino, R. “Relationship between the static and
dynamic modulus of elasticity of concretes exposed to high temperatures”
Proc. V Iber. Cong. of Const. Pathologies and VI1 Congress of Quality
Control (Montevideo, 1999), Ed. Ana I. de la Fuente ISBN 9474-39-191-1,
V.l, pp. 331-338. (in Spanish).
11. Barragán, B., Giaccio, G. and Zerbino, R. “Fracture and failure of thermally
damaged concrete under tensile loading” Materials & Structures, V.34,
RILEM, París, 2001, pp.312-319.
12. ASTM C 1018, “Standard Test Method for Flexural Toughness and First-
Crack Strength of Fiber-Reinforced Concrete (Using Beam With Third-Point
Loading), Annual Book of ASTM Standards, V. 04.02, American Soc. for
Testing and Materials, Philadelphia, USA, 1992.
13. AENOR UNE 83-510, “Concretes with Steel and/or Polypropilene Fibers:
Determination of the index of tenacity and strength at the first crack”,
Madrid, (in Spanish)
14. NBN B 15-238, 1992, “Tests on Fibre Reinforced Concrete - Bending Test
on Prismatic Samples”, Norme Belge, Institut Belge de Normalisation
(IBN), Brussels, Belgium, 1989.
15. Gopalaratnam, V.S., Shah, S.P., Batson, G.B., Criswell, M.E.,
Ramakrishnan, V. and Wecharatana, M. “Fracture Toughness of Fiber
Reinforced concrete”, AC1 Mater. J., V. 88, N04, 1991, pp. 339-353.
16. Mindess, S . , Chen, L., and Morgan, D.R. “Determination of the First-Crack
Strength and Flexural Toughness of Steel Fiber-Reinforced concrete”, Adv.
Cem. Based Materials, V. 1, 1994, pp. 201-208.
17. Banthia, N., and Trottier, J.-F. “Test Methods for Flexural Toughness
Characterization of Fiber Reinforced Concrete: Some concerns and a
proposition”, AC1 Mater. J., V. 92, Nol, 1995, pp. 48-57.
@Seismicisolation
@Seismicisolation
18. Gopalaratnam, V.S., and Gettu, R. “On the Characterization of Flexural
Toughness in Fiber Reinforced Concretes”, Cem. Concr. Composites, V. 17,
1995, pp. 239-254.
19. Jamet, D.,Gettu, R., Gopaiaratnam, V.S., and Aguado, A. “Toughness of
Fiber-Reinforced High-Strength Concrete From Notched Beam Tests”,
Testing of Fiber Reinforced Concrete, AC1 SP-ISS, ed. D.J.Stevens et al.,
ACI, Detroit, USA, 1995, pp. 23-39.
20. RILEM TC 162-TDF: Test and design methods for steel fibre reinforced
concrete. Recommendations. Materials & Structures, V.33, RILEM, París,
Jan-Feb. 2000, pp.3-5.
21. Olesen J.F. and Stang H. ‘Designing FRC slabs on grade for temperature
and shrinkage induced cracks’, in ‘Fibre-Reinforced Concretes’ (P. Rossi
and G. Chanvillard, eds., RILEM, Cachan, 2000), pp.337-346.
22. Traina, L. A. and Mansour, S. A. “Biaxial strength and deformational
behavior of plain and steel fiber concrete”, AC1 Mater. J., V. 88, N”4, 1991,
pp. 354-362.
23. Mangat, P.S. and Motamedi Azari, M. “Influence of steel fibre
reinforcement on the fracture behaviour of concrete in compression”, Int.
Jour. of Cement Composites and Lightweight Concrete, V.6, N”2, nov 1984,
pp 219-232.
24. Giaccio, G. and Zerbino, R. “Fiber Reinforced High Strength Concrete:
Evaluation of Failure Mechanism’’ III International Conference on High
Performance Concrete and Performance and Quality of Concrete Structures,
Recife, 2002. (to be published).
@Seismicisolation
@Seismicisolation
Table 1. Characteristics of Fibers and Cements.
Collated hooked-end Fibers
Fiber tensile strength max. elongation aspect ratio length
L low carbon >1100MPa >0.8 % 80 60 nun
H highcarbon >2300MPa >1% 80 35 nun
Physical tests
Fineness
75 pm ( passing %)
Surface area, Blaine (m2/kg)
Setting time initial
a
5.6
Cements
b
5.8 c3s
355 298 c2s
155 165 C3A
+
Composition (%)
67
12
4.6
a
65
10
4.0
l b
(min) final 240 250 C4FA 9 9

Autoclave Expansion (YO) -0.01 -0.01
Compressive 1 day 14.3 12.1 so3 2.71 2.77
Strength (MPa) 2 days 26.0 27.0 Loss on ignition 1.45 0.70
28days 49.3 52.6 I
Table 2. Concretes.
Concrete A-50 B-50 C (2-40 C-80
Fibers Type L L None H H
&g/m3) 50 50 O 40 80
Water (kg/m3) 158 151 168 169 170
Cement “a” (kg/m3) 360 450 -
Cement “b” (kg/m3> - 480 480 480
Silica fume (kg/m3) 25 48 48 48
Fine aggregate (kg/m3) 920 785 760 750 740
Coarse aggregate (kg/m3) 900 900 900 890 880
Maximum size (nun> 19 19 16 16 16
Superplastizicer (~/m~) 3 7 8 8 8
SlUmD ímmì 30 50I I
170 140 130 I I
@Seismicisolation
@Seismicisolation
Table 3. Results of flexure tests.
Relative values
Property Concrete A-50 B-50 C C-40 C-80
A-501B-501 C IC-401C-80
t
I NDT tests - -
Control 4.61 4.67 4.45 4.38 4.49 100 1O0 1O0
*150 4.18 4.29 4.13 4.03 4.06 91 92 90
*500 3.11 2.85 3.33 3.05 2.94 67
- 70 65
*150 3.4 3.8
*500 6.4 5.7 5.8
Control 47.2 51.6 41.2 38.2 41.3 100 1O0 1O0
*150 38.5 41.6 35.1 31.2 35.7 82 82 86
*500 18.3 16.5 21.7 17.0 13.0 39 -45 31
-
I Flexure tests _. - _.
6.6 1O0 100 100 1O0 100

4.5 71 84 83
3.5
- 56 70 63
+
7.0 8.4 1O0 1O0 1O0
6.8 7.4 74 97 88
4.7 5.2 - 56 -67 62
-
7.0 11.2 1O0 1O0 1O0
5: I
8.6
;:i 67
52
13.1 1O0 1O0
94
79
1O0
82
63
100
7.3 10.8 67 85 82
6.3 8.4 54 63 -73 -64
@Seismicisolation
@Seismicisolation
Table 4. Results of compressive tests (cylinders of 150*300 mm).
Relative values
'roperty Concrete A-50 B-50 C C-40 (2-80
A-501B-501 C IC-401C-8û
NDT tests
Control 4.47
UV
*I50 4.07
*500 - 2.99
Weight *150
loss % *500 ~
Control 42.3
Ed
*150 32.2
(GPa) *500 17.1
Compression tests - -
Control 52.1 71.6 59.3 59.4 75.4 1O0 1O0 1O0
"50 51.1 77.0 55.3 57.3 70.0 98 1O8 93
*500 40.4- 53.6
- 35.8
- 36.3
- 47.7
- 78 -
- 75 60 _.
Control 34.7 38.4 33.3 30.9 33.8 1O0 1O0 1O0

*150 34.2 34.6 27.4 26.5 29.3 99 90 82
*500 18.2 22.6 13.3 11.0 12.0 52 59 40
Control 47.5 65.3 53.9 51.8 65.1 1O0 1O0 1O0 100 100
*150 41.4 64.7 47.6 48.1 56.0 87 99 88
*500 - 27.7 37.5
- 22.0
- 20.0
- 24.6
- -58 -57 41
-
Contro1 91 91 91 87 86
*I50 81 84 86 84 80
*500 69 70 62 55 51 -
Control 38.9 57.2 43 .O 43.0 49.3 1O0 1O0 1O0 100 100
*150 32.4 55.2 39.7 40.8 41.5 83 97 92
*500 - 26.1 36.3 20.0
- 19.5
- 22.4
- 67
- 63 -
- 47
Control 75 80 73 72 65
*150 63 72 72 71 59
*500 - 65 -68 -56 -54 47
@Seismicisolation
@Seismicisolation
Table 5. Results of compressive tests (cylinders of 100*200 mm).
relative values
Properties Concrete C (2-40 C-80
C IC-40 C-80 I
NDT tests
UV
( k w
4.33
4.01 1 1 I
100
95
100
93
100
93
Weight loss *150

% *500 I 4.9 I 5.9 I 5.9
Compressior ests -
Control 66.6 1O0
f C
*150 54.0 93
(MP4 *500 51.2 77
Control 39.7 39.5 41.5 100 100 1O0
*150 29.1 34.0 36.4 73 86 88
*500 20.0 18.6 20.6 50 47 - 50
Control 0.09 0.08 0.09 100 100 1O0
*150 0.05 0.07 0.08 55 88 89
*500. 0.08 0.09 0.16 89 178
113 -
Control 0.10 0.12 0.16 100 100 1O0
Tso 0.08 0.10 0.13 80 84 81
*150
(MPa*dm)
*500 0.11
0.14 - 175
-
Control 0.14 0.16 1O0
T60
*150 0.15 0.14 84
(MPa*d m )
*500 0.22 0.14 I 0.32 I 157 I 88 133
@Seismicisolation
@Seismicisolation
Conversion factors SI (METRIC) TO INCH-POUND
To convert from To Multiply by

"C "F VF)= 1.8 Vc) + 32
MPa Psi 145
kg/m3 lb/yd3 1.685
mm Inch 0.039
kN pound-force 224.8
~/rn~ in3/yd3 46.6
@Seismicisolation
@Seismicisolation
........ .... .
._
i
-
..................
-,
I I
L=3h
Fig. 1. Deflection measurement device used in flexure tests.
@Seismicisolation
@Seismicisolation
1400 I I 700 r I
1200 600
-g 1 O00
800 3 400
500
U
9
U
600 3
-I
300
400 200
200 1O0
Ol I l I I I
I O
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1 .o I .5
Displacement (mm) Displacement (mm)
1400 I 1 1400
1200 1200
1O00 1O00
g 800 3 800
U U
3
-I
600 3
J
600
400 400
200 200
0 1 I I I I I
I Ol I I I , ,
1
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Displacement (mm) Displacement (mm)
-
Fig. 2. Load - deflection and load CMOD curves corresponding to sound
and damaged concretes (A-50, B-50, C and C-SO).
@Seismicisolation
@Seismicisolation
70
-2 60
50
1 40
fcrit
v) 30
20
10
O
.o .o .o .o .o .o .o *o .o .o .o .o .o .o -0
@$&9@$&9@$&9@$$?@$&9
Maximum exposition temperature
Fig. 3 Initiation stress (finit), critical stress (f& and compressive strength (fc) of
concretes expossed to different temperatures.
@Seismicisolation
@Seismicisolation
80 , I
l I
60
i
-2 7050 -
%40 -
3 30 -
20 -
10 -
0 -
-9000 ô000 -3000 O 3000 6000 9000
Strain ( p m h )
-9000 -6000 -3000 O 3000 6000 9000

Strain (prnirn)
Fig. 4 Stress vs. axial and lateral strain curves corresponding to concretes
C and C-80.
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
SP 209-41
Reinforced Concrete Beam-Column

Design: An Artificial Neural
Network Approach
by M. E. Haque
Svnopsis: The basic problem in beam-column design is to establish the propor-

tions of a reinforced concrete cross-section whose design strength is just adequate
enough to support the factored axial load and moments. Since the stress distribu-
tion due to the axial load and moment depends on the cross-section’s proportions,
which are initially unknown, column design cannot be carried out directly. Instead,
the proportions of a cross-section must be estimated and then investigated to deter-
mine whether its design capacity is adequate for the factored loads and moments.
The dimensions of a beam-column cross-section and the area of reinforcing steel
required to support a specific combination of axial load and moment can be estab-
lished by using the column design interaction curves, where an interaction curve
represents all possible combinations of axial load and moment that produce failure
of the cross-section. The bending resistance of an axially loaded column about a
particular skewed axis due to biaxial moments can be determined through itera-
tions and lengthy calculations. These extensive calculations are multiplied when
optimization of the reinforcing steel or column cross-section is required. This pa-
per investigated the suitability of an Artificial Neural Network (ANN) for model-
ing a preliminary design of reinforced concrete beam-column. An ANN back-
propagation model has been developed to design a beam-column which predicts
column cross-section and reinforcing steel requirements for a given set of inputs
which are concrete compressive strength, reinforcing steel strength, factored axial
load and moment. The trained ANN back-propagation model has been tested with
several actual design data, and a comparative evaluation between the ANN model
predictions and the actual design has been presented.
Keywords: artificial neural network; axial load;

beam-column; interaction curve; moment; reinforced
concrete
@Seismicisolation
757
@Seismicisolation
758 Haque
Biographical Sketch:
MOHAMMED E. HAQUE,PkD., P.E.

Mohammed E. Haque is an Associate Professor of the Department of
Construction Science at Texas A M University at College Station, Texas. He
has about twenty years of professional experience in analysis, design, and
investigation of building, bridges and tunnel structural projects of various city
and state governments and private sectors. Dr. Haque is a regrstered
Professional Engineer in the states of New York, Pennsylvania and Michigan,
and members of ASEE, ASCE, and AC1 Dr. Haque received a BSCE fuom
Bangladesh University of Engineering and Technology, a MSCE and a Ph.D. in
CiviWStructural Engmeeringfrom New Jersey Institute of Technology, Newark,
New Jersey. His research interests include fracture mechanics of engineering
materials, composite materials and advanced construction materials, computer
applications in structural analysis and design, artijîcial neural network
applications, knowledge based expert system developments, application based
software developments, and buildings/ infrastructure/ bridgedtunnels inspection
and database management systems.
INTRODUCTION
Artificial Neural Networks (ANNs) are revolutionary computing paradigms that

try to mimie the biological brain. These ANNs are modeling techniques that are
especially useful to address problems where solutions are not clearly formulated
[I] or where the relationships between inputs and outputs are not sufficiently
known. ANNs have the ability to learn by example. Patterns in a series of input
and output values of example cases are recognized. This acquired “knowledge”
can then be used by the ANN to predict unknown output values for a given set
of input values. Alternatively, ANNs can also be used for classification. In this
case, the Artificial Neural Networks’ output is a discrete category to which the
item described by the input values belongs. This paper describes an ANN back-
propagation Multi-layer Perceptron (MLP) model to design a beam-column,
which predicts column cross-section for a given set of inputs: concrete
compressive strength, column types (square tied and circular spiral), reinforcing
steel ratio (0.01 - 0.08), factored axial load, Pu,and moment, W. The trained
ANN back-propagation model has been tested with several actual design data,
and a comparative evaluation between the ANN model predictions and the
actual design is presented.
ARTIFICIAL NEURAL NETWORKS - CONCEPTS A N D DEFINITIONS
ANNs are composed of simple interconnected elements called processing

elements PES) or artificial neurons that act as microprocessors. Each PE has an
input and an output side. The connections are on the input side correspond to the
dendrites of the biological origin and provide the input from other PES while the
@Seismicisolation
@Seismicisolation
connections on the output side correspond to the axon and transmit the output.
Synapses are mimicked by providing connection weights between the various
PES and transfer functions or thresholds within the PES. Fig. 1 illustrates a
simple processing element of an ANN with three arbitrary numbers of inputs
and outputs [2]. The activation of the PE results from the sum of the weighted
inputs and can be negative, zero, or positive. This is due to the synaptic weights,
which represent excitatory synapses when positive (wi>O) or inhibitory ones
when negative (wi<O). The PES output is computed by applying the transfer
function to the activation, which as a result of the synaptic weights, can be
negative, zero, or positive. The type of transfer fùnction to be used depends on
the type of ANN to be designed. Currently, back-propagation is the most
popular, effective and easy to learn model for complex networks [2,3].
For the last few years, the author has been using various ANN back-propagation
Multi-layer Perceptron (MLP) modeling techniques in materials science and
engineering fields including ANN prediction models for fatigue crack growth in
DP steel [2], corrosion fatigue behavior of DP steel [3], mechanical behavior of
powder metaliurgy steel [4], dry sliding wear in Fe-2%Ni based PM alloy [5],
and the effect of heat treatment on mechanical properties in MIM alloy [ 6 ] . To
develop a back-propagation neural network, a developer inputs known
information, assigns weight to the connections within the network architecture,
and runs in the networks repeatedly until the output is satisfactorily accurate.
The weighted matrix of interconnections allows the neural networks to learn and
remember [7].In essence, back propagation training adapts a gradient-descent
approach of adjusting the ANN weights. During training, an ANN is presented
with the data thousands of times (called cycles). After each cycle, the error
between the ANN outputs and the actual outputs are propagated backward to
adjust the weights in a manner that is mathematically guaranteed to converge
Pl.
ANN BACKPROPAGATION MODEL
The neural network used for the proposed model was developed with
NeuroShell2 software by Ward Systems Group, Inc., using a back-propagation
architecture with multi- layers jump connection, where every layer (slab) is
linked to every previous layer (Fig. 2). Two ANN models were developed. The
model-1 network was trained for column size. The inputs were type of column
(Type = 1 for square tied column, and type = 2 for circular spiral column),
factored load, Pu, Moment, W, and steel ratio, pg. In this model concrete
ultimate compressive strength, Z' was 4,000 PSI (28 MPa), and reinforcing steel
yield strength, fywas 60,000 PSI (4 14 m a ) . The model-:! network was trained
for determining the values of Po, P,, and Pny, which determined the axial load
capacity of a biaxial bending column using Bresler equation [9]:
UP,, = I / P , + iP,,
-, , ; for P,,>= O. lP,,
l/P,
where P,, = nominal axial load capacity of a column under biaxial bending
P , = nominal axial load capacity of a column when the load is placed at an
eccentricity e,.
@Seismicisolation
@Seismicisolation
760 Haque
P, = nominal axial load capacity of a column when the load is placed at an
eccentricity ey.
Po = nominal axial load capacity of a column when the load is placed with a zero
eccentricity.
The input for model-2 was h, pp, eh,and y. The value of y was estimated using
the equation:
y = I -(5h), where the column size, h was in inches.
The number of hidden neurons, for which the logistic activation function,
f(x)=I/{I + exp(-x)) was used, was determined according to the following
formula [IO]:
Number of hidden neurons = O.S(lnputs + Outputs) + .\/@?umber of training
patterns)
Given the properties of the training data used: 4 inputs, 1 output, and 340
training patterns - the number of processing elements was determined to be 21.
The training data was obtained from a series of column design Interaction
Diagrams. A set of data was randomly separated from the training set, and was
not used in the training model. These data were used to evaluate the trained
model.
ANN MODEL TRAINING
Network training is an act of continuously adjusting their connection weights

until they reach unique values that allow the network to produce outputs that are
close enough to the desired outputs. This can be compared with the human
brain, which basically learns from experience. The strength of connection
between the neurons is stored as a weight-value for the specific connection. The
system learns new knowledge by adjusting these connection weights. The
learning ability of a neural network is determined by its architecture and by the
algorithmic method chosen for training. The training method usually consists of
one of three schemes:
(1) Unsupervised learning where no sample outputs are provided to the network
against which it can measure its predictive performance for a given set of inputs.
The hidden neurons must find a way to organize themselves without help from
the outside.
(2) Reinforcement learning where the connections among the neurons in the
hidden layer are randomly arranged, then reshuffled as the network is told how
close it is to solving the problem. Reinforcement learning is also called
supervised learning, because it requires a teacher. The teacher may be a training
set of data OT an observer who grades the performance of the network results.
Both unsupervised and reinforcement suffers from relative slowness and
inefficiency relying on a random shuffling to find the proper connection
weights.
(3) Back propagation method is proven highly successful in training of multi-
layered neural nets. The network is not just given reinforcement for how it is
doing on a task. Information about errors is also filtered back through the system
@Seismicisolation
@Seismicisolation
and is used to adjust the connection weights between the layers, thus improving
performance.
The accuracy of the developed model, therefore, depends on these weights.
Once optimum weights are reached, the weights and biased values encode the
network’s state of knowledge. Thereafter, using the network on new cases is
merely a matter of simple mathematical manipulation of these values.
in the present research, several different ANN back-propagation trial models
with different layers/slabs connections, weights and activation functions
(including linear, Sine, Symmetric Logistic, Gaussian, Gaussian Complement,
etc.) were trained. In addition, pattern selections including “Rotation” and
“Random” were used with weight updates using Vanilla, Momentum and
TurboProp. The presented ANN back-propagation Multi-layer Perceptron
(MLP) model with logistic activation function, “Rotation”for pattern selection,
and “TurboProp” for weight updates was the best one among all other trials,
which converges very rapidly to reach the excellent statistical performance (as
illustrated in System Performance). Figs. 3 - 6 demonstrate the graphical
comparisons between the actual column sizes and the network predicted column
sizes during training and evaluation phases of the developed ANN model. They
clearly demonstrate very good agreement between the actual and predicted
performance.
ANN TRAINED MODEL PERFORMANCES
The neural network used for the model presented herein demonstrated an
excellent statistical performance as indicated by the R2 and r values. During
network training, R2 was obtained as 0.9980 and 0.9984 during network
evaluation, which were very close to 1.0 indicating a very good fit between the
actual and the network prediction. R2is a statistical indicator usually applied to
multiple regression analysis, and can be calculated using the following formulae
[lo]:
R2 = 1 - (SSE/SSw)
Where SSE = Z(u - y)2,SS, = Z(u - y)2,y is the actual value, y is the predicted
value ofy, andy is the mean of they values.
The correlation coefficient r, is a statistical measure of the strength of the
relationship between the actual vs. predicted outputs. During network training, I
values were obtained as 0.9991, and 0.9993 during network evaluation, which
were very close to +1.0 indicating an excellent fit between the actual and the
network prediction. The formula for r:
r = S S , / ~ / ~ S SSSJ
~
Where
SS, = zxy - (lin) { -
(zx) ( ~ y ) ;) SS, = zx2 - (í/n)(Z x)’; SS, = z ~ Z (l/n)(Z
Y12
where n equals the number of patterns, x refers to the set of actual outputs, and y
refers to the predicted outputs.
EXAMPLES USING THE ANN MODELS
@Seismicisolation
@Seismicisolation
762 Haque
Several reinforced concrete short columns with uniaxial and biaxial bending
moments were designed using the ANN models; a few of the results are shown
in Tables 1 and 2. For predicting the required column sizes, the inputs for the
ANN model-1 were type of column (Type = 1 for square tied column, and type
= 2 for circular spiral column), factored load Pu,Moment M,,, and steel ratio pg
(selected within 0.01 to 0.08). For a circular spiral column with biaxial
moments, the resultant moment, Mu was calculated by taking the square root of
the sum of the squares of the moments from two perpendicular directions, Mu,
and For a square tied column with biaxial moments, Mu for the ANN
model-1 was calculated using the equation [ii]:
qMm'= Mm + M , {(I- p) /ß], where ßvaries fiom 0.55 to 0.65
In the example problems, the load-contour curvature with the parameter, ß =
0.65 was used. Once the size of the squire tied column with biaxial bending was
predicted through the ANN model-1, a reasonable size was selected, and the
capacity of the column, @Pnwas calculated using the Bresler equation as stated
earlier fiom the values of P,, P,,,,, and Po.The values of Pm,PnY,and Powere
obtained fiom the ANN model-2. In general, all the cases, the ANN models
produced reasonably good results in determining column sizes with varying steel
ratio.
CONCLUSIONS
It was demonstrated in this paper that the Artificial Neural Network (ANN)
based back-propagation model, particularly, the Multi-layer Perceptron (MLP)
networks can be applied to design reinforced concrete beam-columns. The
neural network for the presented models demonstrated an excellent statistical
performance in the network training as well as in the evaluation of the trained
networks. Several reinforced concrete short columns with uniaxial and biaxial
bending moments were studied and it was found that the ANN models produced
good results in determining column sizes with varying steel ratios. The
application of an ANN model certainly minimized the extensive calculations,
especially required for column with biaxial bending during the optimization of
the reinforcing steel or the column cross-section.
REFERENCES
1. Chester, M. (1993) Neural Networks - A Tutorial, Prentice Hall: Englewood

Cliffs, NJ, USA
2. Haque, M.E., Sudhakar, K.V. (2001) ANN based Prediction Model for
Fatigue Crack Growth in DP Steel. International Journal of Fatigue &
Fracture of Engrneering Materials and Structures, Vol. 24, Issue 1, pp. 63-
68.
3. Haque, M.E., Sudhakar, K.V. (2001) Prediction of Corrosion-Fatigue
behavior of DP Steel through Artificial Neural Network. International
Journal ofFatigue, Vol. 23, Issue 1, pp. 1-4.
@Seismicisolation
@Seismicisolation
4. Sudhakar, K.V., and Haque, M.E. (2001) "Mechanical behavior of Powder
Metallurgy steel - Experimental Investigation and Artificial Neural Network-
Based Prediction Model, ''Journal of Materials Engineering and
Performance, 1O( i), pp. 3 1-36.
5 . Haque, M.E., and Sudhakar, K.V. (2000) "An Artificial Neural Network
Prediction for Dry Sliding Wear in Fe-2%Ni based PM Alloy," Brazilian
Journal OfMaterials Science andEngineering, Vol. 3 , N.l, pp. 37-44.
6. Sudhakar, K.V., and Haque, M.E. (2000) "Effect of Heat Treatment on
Mechanical Properties and Neural Network Simulation in MIM Alloy,"
Brazilian Journal of Materials Science and Engineering, Vol. 3 , N. 1, pp. 5-
14.
7. Obermeier, K., and Barron, J. (1989) Time to Get Fried Up, BYTE,14 (8) pp.
227-233
8 . Rumelhart, D., Hinton, G., and Williams, R. (1986) "Parallel distributed
processing," MIT Press, Cambridge, MA., USA
9. Bresler, B (1960) "Design Criteria for Reinforced Concrete Column Under
Axial and Biaxial Dending," AC1 Proc., Vol. 57, pp. 481-490.
10.NeuroShell 2 User's Manual (1996) Ward Systems Group, Inc., Frederick,
MD, USA
11. Design Handbook, vol. 2, Columns,ACZ Publ. SP-l7a(78), 1978.
@Seismicisolation
@Seismicisolation
764 Haque
Table 1Examples on column size prediction using ANN Model-1
640 330 O 0.03 19.55 20

(2847) (447) (0.497) (0.500)
600 140 O 0.03 15.60 16

(2669) (190) (0.396) (0.406)
408 O 0.015 18.36 20

(1815) (0.466) (0.500)
424 O 0.041 15.64 16

(1886) (0.397) (0.406)
378 158 0.02 19.04 22

(1681) (214) (0.484) (0.559)
432 144 0.03 16.23 18

(1922) (195) (0.412) (0.457)
378 158 0.02 21.09 22

(1681) (214) (O. 536) (O. 5 59)
432 144 0.03 17.61 18

(1922) (195) (0.447) (0.457)
Note:
(a)@) See Table 2 for Column Load Capacity by Bresler Equation using A" Model-2
@Seismicisolation
@Seismicisolation
Table 2 Column Load Capacity by Bresler Equation using ANN Model-2
PO PIK p, P, (I) Check cpPn Check

Kips Kips Kips Kips P, >=O. lP, Kips cpP, >= Pu
(W (kN) (kN) (W (W
'(")1817.25 824.12 945.75 581.23 OK 406.86 OK

(8083.13) (3665.69) (4206.70) (2585.31) (1 809.71)
'@)1633.52 1068.44 934.89 717.66 OK 502.36 OK

(7265.90) (4752.42) 4158.39) (3192.15) (2234.50)
~~
Note:
'(a), '@) See Table 1 for the tied column size
@Seismicisolation
@Seismicisolation
766 Haque
- Output signal to other

PES (corresponds to
the axon in biological
neurons)
Transfer function
Activation function
Inputs coming from

-4 other PES (corresponds
XI xz x3 to the dendrites of a
biological neuron)
Fig. 1 Processing element of an ANN model with three arbitrary numbers of inputs and outputs
@Seismicisolation
@Seismicisolation
Fig. 2 Five layers jump connection
@Seismicisolation
@Seismicisolation
768 Haque
50 b - -
- - - ANN PREDICTED COLUMN SIZE
LL
O
10
O
- o ) 2 % 2 G S % Z C Z % r bn - =- ~ m ~ ~ ~ z
PATTERNS PROCESSED
Fig. 3 Square Tied Column Sizes for the ANN Training Patterns Processed
60
50
-z -ACTUAL COLUMN DIAMETER
- - - ANN PREDICTED COLUMN DIAMETER

2 40
w
I-
w
5
5 30
O
z
I
3 20
8
10
PATTERNS PROCESSED
Fig. 4 Circular Spiral Column Diameters for the ANN Training Patterns Processed
@Seismicisolation
@Seismicisolation
60 7- l,._....l. __...__
~l___..____..____,_..
~ I...__-_-.~l_,.L__
_l~.____.._,.
~ .._____._~_l____.._....~..
~ ~
50 -
-ACTUAL SIZE
- - - ANN PREDICTED SIZE
540-
2I
3
o 30-
o
U.
O
gu> 2 0 -
10 -
o-r 8 I I I I I I I I , I I I I I I I I I , , , , I
60 ..................................................................................................................
-ACTUAL COLUNM DIAMETER

50 -
-5 I ANN PREDICTED COLUMN DIAMETER
p! 40 -
W
LI
3 30 -
P
2
i55 2 0 - .
O ..
o
10- - - - " . I
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
PAlTERNS PROCESSED
Fig. 6 Circular Spiral Column Diameters for the Patterns Processed during
ANN Model Evaluation
@Seismicisolation
@Seismicisolation
SP 209-42
Assessment of Structural Systems

Evaluation Methods Based on local
Seismic Demands
by H. S. Lew and S. K. Kunnath
Svnopsis: This paper examines the effectiveness of nonlinear static procedures for
seismic response analysis of buildings. Nonlinear static procedures are recom-
mended in FEMA 273 (Federal Emergency Management Agency-Guidelines for
Seismic Rehabilitation of Buildings) for assessing the seismic performance of
buildings for a given earthquake hazard representation. Three nonlinear static pro-
cedures specified in FEMA 273 are evaluated for their ability to predict deforma-
tion demands in terms on inter-story drifts and potential failure mechanisms. Two
steel and two reinforced concrete buildings were used to evaluate the procedures.
Strong-motion data recorded during the Northridge earthquake are available for
these buildings. The study shows that nonlinear static procedures are not effective
in predicting inter-story drift demands compared to nonlinear dynamic procedures.
Nonlinear static procedures were not able to capture yielding of columns in the up-
per levels of one of the selected buildings. This inability can be a significant source
of concern in identifying local upper story failure mechanisms.
Keywords: dynamic a na lysis; earthquake eng inee ring;

nonIine a r static procedures; perf o rma nc e - b a se d
design; story drift
@Seismicisolation
771
@Seismicisolation
772 Lew and Kunnath
AC1 Fellow H.S. Lew is a senior research engineer at the Building and Fire
Research Laboratory, National Institute of Standards and Technology,
Gaithersburg, Maryland, USA. He is a former director of AC1 and currently
serves on AC1 3 18, Building Code and several other AC1 technical committees.
AC1 Member Sashi K Kunnath is an associate professor, Civil & Environmental

Engineering, University of California, Davis, California,USA. He serves on
several technical committees including AC1 374, Performance-Based Seismic
Design of Concrete Buildings.
INTRODUCTION
The American Society of Civil Engineers (ASCE) is in the process of producing

an U.S. standard for seismic rehabilitation of existing buildings. It is based on
Guidelines for Seismic Rehabilitation of Buildings [i] which was published in
1997 by the U.S. Federal Emergency Management Agency (FEMA). The
evaluation guidelines recommended in FEMA 273 consists of three basic parts:
(a) definition of performance objectives; (b) demand prediction using four
alternative analysis procedures; and (c) acceptance criteria using force andor
deformation limits which are meant to satisfy the desired performance objective.
FEMA 273 recommends four different analytical methods to estimate seismic

demands: (i) linear static procedure (LSP); (ii) linear dynamic procedure (LDP);
(iii) nonlinear static procedure (NSP); and (iv) nonlinear dynamic procedure
(NDP). Given the limitations of linear methods and the complexity of nonlinear
time-history analyses, engineers favor NSP as the preferred method of analysis.
Following the analysis of a building, the safety and integrity of the structural
system are assessed using acceptance criteria. For linear procedures acceptance
criteria are based on demand-to-capacity ratios, and for nonlinear procedures, they
are based on deformation demands.
This paper examines the ability of the FEMA 273 nonlinear static procedures to
predict deformation demands in terms of inter-story drift and potential failure
mechanisms in the system.
Nonlinear Static Procedures for Seismic Demand Estimation
There are several procedures that can be adopted for conducting a nonlinear static
analysis. While the basic approach to analysis is essentially the same, different
procedures use alternate ways to characterize the lateral force distribution that is
applied to the structural model. FEMA-273 recommends the following three
forms of lateral load distribution:
@Seismicisolation
@Seismicisolation
i) Inverted Triangular Pattern (FEMA-1)
A lateral load pattern represented by the following FEMA-273 equation:
i=l
where: F, = lateral load at floor level x

wx, i = mass at floor level x, i
h ,i = height fiom base to floor level x,i
k = 1.0 for T < 0.5 s
k = 2.0 for T > 2.5 s, with linear interpolation for intermediate values
V = total lateral load (base shear) to be applied to the building
This load pattern results in an inverted triangular distribution across the height
of the building when k = I . O and a parabolic distribution when k = 2.O, and is
normally valid when more than 75 % of the mass participates in the
fundamental mode of vibration.
ii) Uniform Load Pattern (FEMA-2)

A uniform load pattern is based on lateral forces that are proportional to the
total mass at each floor level.
This pattern is expected to simulate story shear forces and tends to amplify
demands at the lower levels of the building.
iii) Modal Load Pattern (FEMA-3)

A lateral load pattern proportional to the story inertia forces consistent with
the story shear distribution calculated by a combination of modal responses.
Each of the above lateral load distributions will be evaluated using four sample
structures: 6-story steel frame building; 13-story steel fiame building, 7-story
concrete building, and 20-story concrete building. These buildings have strong-
motion records obtained from instrumented data from the 1994 Northridge
earthquake. The strong-motion responses at the roof level were used to calibrate
the building models.
Benchmark Response
@Seismicisolation
@Seismicisolation
774 Lew and Kunnath
Time-history analyses were performed on the four buildings, and the results were
used as benchmark values to which the results of the nonlinear static analyses
were compared. The target displacements were based on either the actual
measured building response or a displacement consistent with the guidelines of
FEMA-273. The following aspects of the building response behavior were
examined:
i) The ability of each nonlinear static procedure to predict the inter-story

drift demands.
2) The ability of each nonlinear static procedure to identifj the distribution of
yielding (plastic hinge formation) in the structure leading to a local or
global failure mechanism.
MODELING AND CALIBRATION OF BUILDINGS
ó-Stow Steel Building
This building, located in the Los Angeles area, was designed in 1976 based on the
1973 Uniform Building Code (UBC) [2]. The primary lateral load resisting
system is moment fiames around the perimeter of the building. Since the
structural system is symmetrical, only one of the perimeter frames was considered
in the evaluation. The interior frames were designed as gravity frames and consist
of simple shear connections only. In constructing the building model, the base
was assumed to be fixed. The plan view of the building and the elevation of a
typical frame used in the analysis are shown in Fig. 1.
The building was instrumented with a total of 13 strong-motion sensors at the

ground, 2nd, 3rd and roof levels.
The building performed well in the Northridge earthquake with no visible signs of
damage. Recorded data indicate that the building response was essentially elastic.
Minimal calibration was required to obtain a good match of the computed
response with the observed roof response (Fig. 2).
13-Stow Steel Building
A 13-story steel moment-resisting frame building is located about 5 km southwest

of the epicenter. It was designed based on the 1973 UBC, and built in 1975. The
layout of the perimeter frames and a typical elevation of one of these perimeter
frames are shown in Fig. 3. The overall building plan dimensions are 48.7 m x
48.7 m (160 ft x 160 A). The typical floor system consists of about 64 mm (2.5
in) of concrete fill over 76 mm (3 in) steel decking. In modeling the columns are
fixed at the basement level.
@Seismicisolation
@Seismicisolation
The acceleration recorders were placed at the basement and on the sixth and
twelfth floors. Recorded accelerations at the basement indicate that the building
experienced a peak ground acceleration (PGA) of 0.41g in the N-S direction and
0.32g in the E-W direction. Approximately 12 % of the connections of the west
perimeter ( N - S direction) frame íkactured during the earthquake. The number of
connections that fractured on the remaining three sides was less than 6 %.
Since the building was symmetrical in both directions, only one frame was
analyzed. Gross section properties were used to compute the initial stiffness
values. Each perimeter frame was assumed to carry 50 % of the building mass in
each direction.
The building response in the east-west direction is shown in Fig. 4. This indicates
that the building stiffness and mass modeling are reasonably accurate.
7-Story RC Building
A seven-story nonductile reinforced concrete building was designed and

constructed during the mid-1960s in accordance with the 1964 Los Angeles City
building code. The building sustained damage during the 1971 San Fernando and
the 1994 Northridge earthquakes. The building suffered relatively minor damage
in the 1971 event but was heavily damaged during the 1994 event.
This building has a rectangular plan (Fig. 5) with overall dimensions of

approximately 19 m (63 ft) in the north-south (transverse) direction and 46 m
(150 ft) in the east-west (longitudinal) direction. The total height of the building is
20 m (66 ft). The floor system consists of reinforced concrete flat slabs.
Perimeter spandrel beam-column fiames resist primarily the lateral load in each
direction. The interior slab-column frames are also expected to carry a significant
portion of the lateral load. The north side of the building has four bays of brick
masonry infill walls between the ground and the second floor. Nominal
expansion joints separate the walls from the columns and the spandrel beams.
These walls were not designed as the lateral load resisting system.
The building was repaired after the 1971 event. During the 1994 event, damage
was primarily confined to the longitudinal perimeter frames with the most severe
damage between the fourth and fifth floors of the south perimeter frame in the
form of shear failure of columns immediately below the fifth floor spandrel beam.
Many beam-column joints suffered minor to moderate shear cracks below the
fifth floor level. Concrete spalling and hairline flexural cracks were observed in
several spandrel beams. In the transverse direction, the damage was limited to
minor flexural cracks in the beams in the end bay.
The two interior frames and the two exterior frames were considered to be
identical for purposes of the modeling. The building was assumed to be fixed at
the base. The response of the calibrated building model to the input gound
@Seismicisolation
@Seismicisolation
776 Lew and Kunnath
motion is shown in Fig. 6. The comparison indicates that the resulting model is
adequate for the analytical studies.
20-Stow RC Buildine
The building was designed in 1967 and constructed in 1968 in accordance with
the Los Angeles City building code. As shown in Fig. 7, the building has a
rectangular plan with dimensions of 56 m x 18 m (184 ft x 58 fi) in the
longitudinal (E-W) and transverse (N-S) directions, respectively. There are
twenty stories above ground and a basement. The typical story height is 2.7 m
(8.75 fi). The bottom two and top two levels have different heights as indicated in
the figure. This building experienced a PGA of 0.33g at the basement level
during the 1994 Northridge earthquake.
The gravity load resisting system of the building consists of concrete slab
supported on concrete beams and columns. The primary lateral force resisting
system consists of moment-resisting frames with strong shear walls in the
basement only. Beams are provided between all column lines except between
lines 2 and 3 for intermediate frames (also shown in the figure). Post-earthquake
survey of the building indicated that the building suffered significant non-
structural and content damage.
Two typical frames were considered in the modeling of the building. Eight
frames with three continuous beams across all three spans and seven frames with
discontinuous beams were modeled separately. The intenor slab between the
discontinuous beams was assumed to remain elastic. A total of 19 column types
and 19 beam types were used to construct the building model. A fictitious stiff
column was added at the basement level to simulate the effect of the rigid walls at
this level. Since the recorded ground motion was at the basement level, it was
considered essential to model the basement level. The columns are considered
fixed at the basement level.
Using gross section inertia and uncracked concrete stiffness, a fundamental

building period of 2.46 s was obtained. Since this value is in good agreement to
the recorded data, no additional tuning of the model was required. Figure 8 shows
a plot of the time-history response of the roof, which indicates that the building
model is adequate.
EVALUATION OF NONLINEAR STATIC PROCEDURES FOR

SEISMIC DEMAND ESTIMATION
Analysis of 6-Story Steel Building;
@Seismicisolation
@Seismicisolation
As indicated previously, this building did not sustain damage during the
Northridge earthquake. Instrument records and the calibrated response indicate
that the structural response was essentially elastic. To examine the effectiveness
of nonlinear static procedures to estimate seismic demands, the recorded
acceleration record at the base of the building was scaled up uniformly to achieve
a roof displacement in the yielded state of the system (the “target” displacement
per FEMA 273). This approach allowed a better comparison of the different
methods of analyses.
The target roof displacement was determined according to FEMA-273. The

computed target roof displacement was 450 mm (18 in). The scale factor for the
recorded base motion was established through trial-and-error until the computed
roof displacement matched the target displacement of 450 mm (18 in).
The displacement profile at maximum roof displacement is shown in Fig. 9. It is

observed that all nonlinear static procedures predict a similar profile. In
comparison with nonlinear dynamic procedure, the displacements are over-
predicted.
The driît profiles over the height of the building are shown in Fig. 10. The dnft at
the first level is over-estimated by all nonlinear static procedures but under-
estimated at all other levels except the second level. It should be noted that the
maximum inter-story drift in the case of the time-history response could occur at
different times.
A comparison of the hinge patterns at the target displacement is shown in Fig. 11.
The plots show the distribution of beam and column yielding for each of the
analyses using a different lateral load distribution along the height of the building.
Analysis of 13-Stow Steel Building
This building experienced widespread connection damage during the Northndge

earthquake. Since this implies inelastic action, it was reasonable to use the
recorded roof displacement as the target displacement for the comparative study.
Figure 12 compares the displacement profile of the building at the target

displacement for the different nonlinear static procedures. Figure 13 compares
the maximum inter-story drift along the building height.
While the displacement profiles for all methods look very similar, the predicted
maximum inter-story drifts vary considerably for the nonlinear static procedures
when compared to the dynamic response indicating the presence of higher modes
in the response that cannot be captured by the static methods.
@Seismicisolation
@Seismicisolation
778 Lew and Kunnath
The locations of plastic hinges predicted by the different analytical methods are
shown in Fig. 14. Since FEMA-1, FEMA-2 and FEMA-3 resulted in almost the
same hinge distribution, only one of three patterns is shown.
Analysis of 7-Story Concrete Building
This building experienced inelastic behavior and significant structural damage

during the Northridge earthquake. Therefore, it was not necessary to establish a
projected target displacement for the response analyses. The actual displacement
obtained from the nonlinear time-history analysis was used as the target
displacement.
The displacement profiles using the different shapes of lateral load distributions
are shown in Fig. 15. The corresponding drift values are shown in Fig. 16. The
overall trends in both displacement and drift are similar. The nonlinear static
procedures tend to underestimate drift in the upper levels. As seen in Fig. 17, the
plastic hinge formation in the upper levels is not estimated adequately by the
nonlinear static procedures.
Analysis of 20-Story Concrete Building
This building did not experience damage during the Northridge earthquake.
hsîmment records and the cahbTated response indicate that the structural
response was essentially elastic. As was the case in the 6-story steel building
(Section 3.1), the acceleration record at the base of the building was scaled up
uniformly to achieve a roof displacement in the yielded state of the system. The
displacement profiles at the target roof displacement are shown in Fig. 18.
The drift profiles predicted by different procedures along the height of the
building are compared in Fig. 19. It is seen that all nonlinear static procedures
underestimate the drift at almost all story levels above the third floor. Figure 20
presents a summary of the plastic hinge patterns resulting from each of the
different lateral load distributions. The locations of plastic hinges are consistent
with the inter-story drift profiles.
CONCLUSIONS
This paper examined the effectiveness of nonlinear static procedures in evaluating

the inelastic response of buildings. Specifically, the FEMA 273 methodology is
evaluated to see whether nonlinear static procedures can predict deformation
demands in terms of inter-story drift and potential failure mechanisms in the
structure.
The evaluation was carried out using four buildings for which instrumented data
were available:
@Seismicisolation
@Seismicisolation
1) Six-Story Steel Moment-Frame Building,
2) Thirteen-Story Steel Moment-Resisting Frame Building,
3) Seven-Story Reinforced Concrete Moment Frame Building, and
4) Twenty-Story Concrete Moment Frame Building.
A frame model of each of the above buildings was first calibrated against
observed instrument data. Then, each of the building models was analyzed using
a detailed nonlinear time-history analysis followed by a series of nonlinear static
(pushover) procedures:
1) A lateral load pattern represented by an inverted triangular load (FEMA-1).

2) A uniform load pattern based on lateral forces that are proportional to the total
mass at each floor level (FEMA-2).
3) A lateral load pattern proportional to the story inertia forces consistent with
the story shear distribution calculated by a combination of modal responses
using a response spectrum analysis (FEMA-3)
The following conclusions are drawn:
1. Nonlinear static procedures are generally not effective in predicting inter-story

drift demands compared to nonlinear dynamic procedures. Drifts are
generally under-estimated at upper levels and sometimes over-estimated at
lower levels.
2. The peak displacement profiles predicted by both nonlinear static and

nonlinear dynamic procedures are generally in agreement. This suggests that
the estimation of the displacement profile at the peak roof displacement by
nonlinear static procedures is reasonable so long as inter-story drifts at the
lower levels are reasonably estimated. Primarily the drift at the lower story
levels affects the displacement profile.
3. Nonlinear static methods did not capture yielding of columns at the upper
levels. This inability can be a significant source of concern in identifying local
upper story mechanisms. Consequently, nonlinear static methods were also
ineffective in predicting locations of potential plastic hinges in upper stones.
REFERENCES
1. Federal Emergency Management Agency (FEMA), NEHRP Guidelines
for the Seismic Rehabilitation of Buildings, Washington, D.C. 1997
2. International Conference of Building Officials, 1973 Uniform Building

Code, Whittier, CA
@Seismicisolation
@Seismicisolation
780 Lew and Kunnath
6 BAYS @ 6.1 m /
9.1 9.1 91 9.1 1'
9.1
6 bays
dimensions in meters
Figure 1 Plan and Elevation of 6-Story Steel Building
@Seismicisolation
@Seismicisolation
-o
I
50
-+z 30
10
5
0 -10
4 -30
Q
2
P
-50
O 5 10 15 20
TIME (SEC)
Figure 2 Recorded vs. Computed Response of 6-Story Steel Building
W14
3 - G X246
--o W14
Y287
: -42 l2
W14
X314
Pm
:-+a W14
X398
'--o 9 3rn
2
WWY7
aaxiI W14
X426
0 W14
4 4m X500
-o E'ev - Groun L L 1
Figure 3 Plan and Elevation of Perimeter Frame of 13-Story Steel Buiiding
@Seismicisolation
@Seismicisolation
782 Lew and Kunnath
- - - . OBSERVED
_ 1
5
h
30
40 - SIM ULATED
;20
10
35 -10
0
ci -20
2
p -30
40
-50
O 5 10 15 20
TIME (S)
Figure 4 Recorded vs. Computed Response of 13 Story Steel

Building
lm
N
35m
lm
0 bays @ 5.72m
Figure 5
@Seismicisolation
Typical Floor Plan and Elevation of 7-StOry RC Building
@Seismicisolation
25 , 1
20
5 15
c, 10
4
p. -5 -
g -10 -
C; -15 -
2 -20 -
-Recorded
-25 -
-30 I I
O 5 10 15 20
TIME (c)
Figure 6 Recorded vs. Computed Response of 7-Story RC Building
I
'6m 5.4m 6m
PLAN Section A-
A
Figure 7 Plan and Elevation of 20-Story RC

Building
@Seismicisolation
@Seismicisolation
784 Lew and Kunnath
................
O 5 IO 15 20
TIME (SEC)
Figure 8 Recorded vs. Computed Response of 20-Story RC Building
$ 4
J
iOZ 3
$ 2
O
O 10 20 30 40 50 O 2 4 6 8
DISPLACEMENT (CM) STORY DRIFT (Yo)
Figure 9 Displacement Profiles of 6 Figure 10 Inter-story Drift Profiles of

Story Steel Building 6 Story Steel Building
@Seismicisolation
@Seismicisolation
ia) NDP (b) FEMA-
I C ) FEMA-2 Id) FEMA-
Figure 11 Hinging Patterns of 6-Story Steel Building

at Target Displacement
14 14
12 12
10 10
-!
W
> 8 a
Y
eo
' 6 6
4 4
2 2
0 "I
...
O
O 10 20 30 40 50 O 10 20 30 40 -50
DISPLACEMENT (CM) DISPLACEMENT (CM)
Figure 12 Displacement Profiles of Different NSPs vs.NDP of 13-

Story Steel Building
@Seismicisolation
@Seismicisolation
_____
786 Lew and Kunnath

14
12
10
-1
w
' 8
Y
e
'
v)
6
O 1 O
O 0.5 1 1.5 2 O 0.5 1 1.5 2
STORY DRIFT (%) STORY DRIFT (Yo)
Figure 13 Inter-story Drift Profiles of 13-Story Steel Building
(a) NDP
!FEEH3 (b) NSP
Fiaure 14 Hinaina Patterns of 13-Stow Steel Buildina
@Seismicisolation
@Seismicisolation
6- +FEMA-I
+FEMA-P
5-
1
w
w 4-
J
>
$ 3-
u3
2-
O
O 10 20 30
DISPLACEMENT (CM)
Figure 15 Displacement Profiles of 7- Figure 16 Inter-story Drift Profiles of

Story Concrete Building 7-StOry Concrete Building
(a)NDP
(b) FEMA-1and -3 (c) FEMA-2
Figure 17 Hinging Patterns of 7-Story Concrete

Building
@Seismicisolation
@Seismicisolation
788 Lew and Kunnath
21 I,
18
15
-I
W
2
J
12
& 9
P
v)
O 30 60 90 120 150 O 30 60 90 120 150

DISPLACEMENT (CM) DISPLACEMENT (CM)
Figure 18 Displacement Profiles of 20-Story Concrete Building
21 ,
18 -
15 -
-1
W
zj 1 2 -
-.-FE MA-1
>
cc
c
v)
9 -
6 -
3 -
0.00 2.00 4.00 6.00 8.00

0.00 2.00 4.00 6.00 8.00
STORY DRIFT (%) S T O R Y D R I F T (Yo)
Figure 19 Inter-story Drift Profiles of 20-Story Concrete Building

@Seismicisolation
@Seismicisolation
l
Frame 1 Frame 2
(a) Nonlinear Dynamic Procedure
Figure 20 Hinging Patterns of 20-Story Concrete Building
@Seismicisolation
@Seismicisolation
790 Lew and Kunnath
Frame 1 Frame 2
(B) Nonlinear Static Procedure FEMA 1,

FEMA 2, FEMA 3
Figure 20 Hinging Patterns of 20-Story Concrete Building(cont.)
@Seismicisolation
@Seismicisolation
SP 209-43
Hybrid Fibers and the Development

of Cracking in Concrete
by J. S. Lawler, D. Zampini, and S. P. Shah
SvnoDsis: Cracking in concrete is fundamentally altered by the addition of rein-

forcing fibers. A combination of microfibers (less than 22 pm [0.0009 in.] in diam-
eter) and macrofibers (500 pm [0.0197 in.] in diameter) that contribute in comple-
mentary ways to performance, is presented as a means for controlling cracking and
improving the lifecycle behavior of concrete. In previous work, a hybrid blend of
these fibers in a mortar matrix demonstrated better mechanical performance and
lower cracked permeability than was seen with a single fiber type. The research
presented in this paper attempts to realize the potential of such blends in concrete.
A mixture proportioning method that achieves good workability and cohesion in
concretes containing microfibers was used to produce a cast concrete. The me-
chanical performance and shrinkage cracking resistance of this material were eval-
uated. In the hybrid reinforced concrete, the microfibers delayed the development
of macrocracks and so the composite demonstrated greater strength and cracking
resistance than a similar matrix reinforced with macrofibers only. However, this in-
fluence was less pronounced than was observed with a mortar matrix and was con-
fined to smaller crack openings.
Keywords: concrete; fiber-reinforcement; fracture

process; hybrid; microfibers; shrinkage; workability
@Seismicisolation
791
@Seismicisolation
792 Lawler et al.
AC1 member, John S. Lawler is currently employed with Wiss, Janney, Elstner
Associates, Inc. in Northbrook, Illinois. This work was performed while he was
a Research Assistant in Civil Engineering at Northwestern University in
Evanston, Illinois, from where he earned his Ph.D. in 2001. His interests include
fiber hybridization for concrete reinforcement.
Davide Zampini, is the Research Group Leader at MBT (Schweiz) AG, R & D
Switzerland in Zürich. Areas of research include fiber-reinforced cement-based
materials, rheology, cement hydration and durability.
AC1 Fellow, Surendra P. Shah, is a Walter P. Murphy Professor of Civil
Engineering at Northwestern University and the Director of the Center for
ACBM. His research interests include high performance cement-based
materials, fiber-reinforced composites, non-destructive testing, constitutive laws
and fracture mechanics of cement-based systems. He has authored numerous
technical papers, edited twelve and Co-authored three books.
INTRODUCTION
Discontinuous fibers are included in concrete to increase both strength

and toughness under tensile loading conditions. Fibers retard crack development
and carry load across cracks that have opened, modifying the failure mechanism
of the composite material. The effectiveness of a particular fiber type at
improving a specific concrete property, such as strength, toughness or resistance
to shrinkage cracking, depends on the material properties and geometry of the
fibers. Hybrid blending of fibers, i.e. combining two or more fiber types that
improve different categories of behavior in an additive and complementary
manner, is a method for maximizing the effectiveness of fiber-reinforcement in
concrete.
Some fundamental approaches (1-5) in the use of hybrid systems have

been presented in the literature. in the research presented in this paper,
microfibers (less than 22 pm [0.0009 in.] in diameter) were blended with
macrofibers (500 pm [0.0197in.] in diameter) since, as is discussed below, the
size of the fibers is pivotal in determining how they interact with the fracture
process.
The role of fibers in the fracture process and the resulting influence on
mechanical performance in mortar reinforced with micro- and macrofibers was
examined in a previous study using Subregion-Scanning Computer Vision
(SSCV), which is a high-resolution, full-field method for measuring cracking
(6). The results from that work, summarized briefly below, showed that the
blending of fibers in this manner produced better performance than the use of a
single fiber type. Based on these findings, such a blend was included in a
concrete matrix in an attempt to achieve similar improvements in performance.
in the work described in this paper, a method for designing concrete mixtures
@Seismicisolation
@Seismicisolation
that accounts for the high paste volume demand resulting from the use of
microfibers was employed and the mechanical performance and shrinkage
cracking resistance of the resulting concrete was characterized.
Hybrid Fiber Blends and the Fracture Process in Mortar
To best utilize fibers for reinforcement of concrete, the mechanisms by which

they influence the failure process must be considered. Through the use of
SSCV, applied to specimens in uniaxial tension, it was observed that in both
reinforced and unreinforced materials the failure process occurred in three stages
(Fig. 1) (6). First, fine, discontinuous ‘microcracks’develop as soon as stress is
applied to concrete. During the pre-peak regime of the stress-strain relationship,
these discontinuous microcracks link to form longer but still narrow coalesced
microcracks. Eventually, a crack evolves that is long enough, e.g. crossing the
full specimen width in a uniaxial tensile specimen, to govern the stress in the
surrounding material. This crack begins to widen, becoming a larger crack
called a macrocrack. In concrete containing conventional reinforcement,
subsequent specimen deformation is concentrated at this macrocrack, and
localization is said to have occurred. The rate at which the process occurs is
significant because the development of a macrocrack and the localization of
deformation at that crack determine the maximum load that the composite can
canya
Microfibers primarily interact with fracture evolution by delaying the

coalescence of microcracks and then hampering the widening of coalesced
cracks. Since both crack coalescence and widening are part of the process by
which the microcracks form macrocracks, this delay results in improved pre-
peak performance and strength. Macrofibers, due to their greater size, are more
efficient at carrying load across the wide macrocrack as it opens and so
contribute to the post-peak ductility of the material. Multiple macrocracks will
develop if enough reinforcement exists such that the amount of energy required
to open the existing cracks is greater than that required to open a new one (7).
SSCV utilizes digital image correlation (DIC), a technique based on

comparisons of subsequent digital images captured of a deforming specimen, for
measurement of full-field displacement at high resolution. This method was
used to observe the fracture process of mortar (1:0.45:1, cement : water: sand by
weight) containing no fibers, 0.5% steel macrofibers only, 0.5% steel or
polyvinyl alcohol (PVA) microfibers only and finally, blends of 0.5% steel or
PVA microfiber and 0.5% steel macrofibers (6). The properties of these fibers
are given in Table 1. It should be noted that no specimens were tested that
contained 1% by volume of a single type of fiber. Mixtures with 1% microfiber
content would have been unworkable, whereas the workability of the hybrid
fiber mixture was acceptable.
@Seismicisolation
@Seismicisolation
794 Lawler et al.
While cracks in macrofiber-reinforced mortar evolved in a similar way to
the unreinforced material, the use of microfiber reinforcement resulted in
multiple, narrow coalesced cracks before the peak load. Because of the smaller
inter-fiber spacing of the microfibers, the likelihood that the crack path would
intersect with the fibers was high. As a result, the microfibers were able to
bridge the coalesced cracks while still narrow and maintain tensile stresses
throughout the material. Multiple cracking increased the deformation capacity
of the material and the corresponding delay in the development of a critical
macrocrack resulted in increased strength. The blend of PVA microfiber and
steel macrofiber produced the best composite mechanical performance of all the
specimens examined and induced multiple macrocrack development. The
mechanical performance was evaluated using three-point bending tests of
notched-beams and the responses of representative specimens are pictured in
Fig. 2.
While strength and toughness are of primary concern when evaluating a

material for use in a structural application, the durability of the material must
also be considered if the full viability of a structure is to be evaluated. The
durability of concrete is intimately related to its permeability because concrete is
susceptible to degradation through mechanisms that involve the ingress of water.
Cracks effectively determine the permeability of dense materials such as high
performance concrete and an alteration in the crack topology resulting from
fiber-reinforcement has a significant effect on permeability.
Under the assumption of laminar flow, the rate of flow through a parallel
sided slot, such as a crack, is directly proportional to the crack width cubed.
Therefore, for similar overall displacement, a multiply-cracked material will
demonstrate a lower permeability than one exhibiting a single crack.
Permeability tests that measured the flow of water through the specimen versus
tensile deformation were conducted on mortar specimens with the same matrix
and fiber content described above (6). A significant reduction in flow rate for a
given displacement was observed for composites that included microfibers. The
best performance resulted from a hybrid blend of steel macrofibers and PVA
microfibers. This is attributed to the multiple cracking, as observed with the
SSCV, that characterized the failure of this composite.
Concrete
Hybrid fiber-reinforcement improved mechanical response and reduced

cracked permeability in mortar in the studies discussed above. While such a
material may be useful in repair applications, achieving similar performance
improvements in a concrete matrix is clearly desirable. However, effective use
@Seismicisolation
@Seismicisolation
of microfibers in concrete is difficult because the presence of aggregate
influences the role of these fibers in the fracture process.
In comparison to mortar, in concrete there is both a greater amount of

interfacial zone and a lower paste volume fraction, defined as VJVfot= volume
of paste / total volume. The interfacial zone, where the density of the cement
matrix is reduced as a result of its proximity to the aggregate, may dominate the
fracture process and essentially eliminate the influence of the fibers. This
mechanism of failure, typical in unreinforced normal strength concrete, was
circumvented in the work discussed here because a strong interfacial bond
between matrix and aggregate was achieved through the use of concrete with a
low water-to-cement ratio. More significant, in this case, is the large reduction
in paste volume fiaction, which negatively impacts workability and so may
influence the dispersion of the microfibers and certainly influences the overall
quality of the placed material.
A previous study concentrating on the effects of paste volume on

workability and fiber dispersion developed a mixture design process intended to
overcome the rheological limitations of using microfibers (6). It was found that
a sufficiently large volume of paste was required to coat the high surface area of
the microfibers before attractive flow and acceptable cohesion properties could
be achieved. Based on the relationship between flow and paste volume, an
optimum paste volume fraction could be determined that efficiently maximized
the flow achievable for a given fiber blend while minimizing the cement content.
In the current study, the procedure for proportioning mixtures was

applied to concretes containing hybrid blends of reinforcement consisting of
steel or PVA microfibers with a steel macrofiber. The mechanical response and
shrinkage crack resistance of a PVA hybrid-reinforced concrete was evaluated,
while for comparison, the mechanical performance of a steel microfiber hybrid
was also assessed. Flexural testing, which determines the strength and
toughness, or capacity to absorb energy, was used to quantify the mechanical
response of the material. Shrinkage cracking, which develops as result of drying
shrinkage in concrete structural elements whose movement is restrained by
environmental conditions, is known to be significantly reduced by fibers (8). A
restrained ring configuration was used to measure the relative effectiveness of
the reinforcements considered here.
This study was executed to verify that the procedure for designing
mixtures that was developed could be used to produce high performance hybrid
fiber-reinforced concrete that includes PVA microfibers. The impact of this
fiber blend on the behavior in such a matrix was then quantified. This study
demonstrates that hybrid blending of micro- and macrofiber can be used
effectively in a concrete matrix and quantifies this improvement.
@Seismicisolation
@Seismicisolation
796 Lawler et al.
Mixture desim
Mixtures containing a blend of steel macrofiber and steel or PVA

microfiber were developed according to the method discussed above, whereby
the optimum paste volume, i.e. the minimum paste content that produced stable,
maximized flow, was determined. This was done by evaluating the percent
flow of îrial batches using a drop table according to ASTM C230-98.
Superplasticizer dosage was maximized in an incremental process to give the
best workability possible while avoiding bleeding, which results from an
overdose. Cohesion was also considered, and mixtures in which bleeding and
non-uniformity, or separation of the constituents, was observed were rejected.
Fly ash replacement improved flow and was helpful for achieving a cohesive
mixture. A low water-to-binder ratio (wh) was used to reduce susceptibility to
bleeding and to maximize the strength of the hardened concrete. During this
evaluation, macrofiber-to-concrete volume was held constant, as was microfiber
to paste volume. The relationship between flow and paste volume obtained for
the mixtures examined here is plotted in Fig. 3. The mixtures chosen for casting,
demarcated by a box in the figure, are at the optimum paste volume fraction.
In addition to the hybrid reinforced concrete, control mixtures of

unreinforced concrete and macrofiber reinforcement only, were cast (Table 2).
Also appearing in this table is the superplasticizer dosage and the percent flow
measured from the cast batches. The superplasticizer used was a
polycarboxylate type. The wfb (by weight) was 0.322, and 0.336, for the
mixtures without and with fly ash respectively, since waterhinder (by volume)
was kept constant. Type I portland cement was used in this study. Class F fly
ash was included in the hybrid mixtures, at 15% replacement by volume, which
was found to be most effective for increasing workability (6). Coarse aggregate
(4-8 mm [0.157 - 0.315 in.]) and fine aggregate (< 4 mm [0.157 in.]) made up
42% and 58% of the total aggregate by weight, respectively.
While most normal concrete has a maximum aggregate size of 25 mm (1

in.), the concrete examined in this study had a maximum aggregate size of 8 mm
(3/8 in.) The failure process of the concrete may be modified to a small degree
by this difference but the influence of coarse aggregates on the effectiveness-of
the microfiber hybrid-reinforcement, namely the decrease in paste volume and
the increase in interfacial zone proximity, is comparable for coarse aggregate
with either maximum aggregate size.
Mixing, done in a 40-liter (2441 in.3) planetary mixer at medium speed,

was performed in the following order: cementitious material (including fly ash),
fine and coarse aggregate were mixed together with about a quarter of the water
@Seismicisolation
@Seismicisolation
for one minute, then the remaining water was added, along with a third of the
superplasticizer for the hybrid mixtures or with all of the superplasticizer for the
non-hybrids. At this time the macro-, then microfibers were added gradually,
after which the remaining superplasticizer was added, before final mixing was
performed. Final mixing time, which was five minutes for the mixtures without
microfiber and 16 minutes for those containing the microfiber, was governed by
the significant difference in superplasticizer dosage. Very workable and
cohesive mixtures, normally difficult to achieve, were produced through this
process.
The specimens were cast in three lifts and both vibrated and rodded to
aid compaction. Each lift was vibrated for a minimum of 60 seconds and rodded
25 and 75 times for the flexural beams, and shrinkage rings, respectively. The
beam specimens were covered and left to cure for 24 hours before they were
demolded and placed in a 100% RH environmentuntil testing.
Flexural Testing
Flexural mechanical properties of the mixtures were measured by three-

point bending tests of the beams with a notch cut to li3 of the specimen height at
the mid-span, based on the procedure recommended in (9). The tests were
performed with a closed-loop servo-hydraulic testing machine with 89 kN (20
kip) capacity and an 89 kN (20 kip) load cell. The beams were 50 x 100 x 460
mm (1.97 x 3.94 x 18.11 in.) in size and crack mouth opening displacement
(CMOD) measured at the notch was used to control the test and was advanced at
a rate of 0.15 &min. (0.006 idmin.) The displacement of the specimen
parallel to the loading direction was measured using an average of two 5.08 mm
(0.2 in.) linear variable differential transformers (LVDTs) supported by a
“Japanese-style” yoke. Three beams were tested for each mixture 28 days after
casting. Stress is calculated using the total specimen depth.
Restrained Ring Shrinkage Cracking
The effect of the fibers on shrinkage cracking was evaluated in a

restrained ring configuration, based on AASHTO Designation PP34-98. The
concrete is cast around a steel annulus and relies on the axisymmetric nature of
the specimen to simulate an infinitely long slab. This test was canied out using
a 50 mm (1.97 in.) thick concrete specimen, 150 mm (5.91 in.) in height, cast
around a steel ring of 25 mm (0.98 in.) thickness and 300 mm (1 1.8 1 in.) outside
diameter. After 24 hours of sealed curing, the outer mold was removed and the
top surface of the ring was sealed with silicone caulking, resulting in a
arrangement that permitted drying from the outer surface only. The specimen
@Seismicisolation
@Seismicisolation
798 Lawler et al.
and ring were transported to an environmental chamber where conditions were
maintained at 22 "C and 50% RH. Daily visual inspection was conducted and
for each specimen the age at which cracks were first observed was recorded. In
addition, the crack width at 44 days was measured, using an optical microscope
at three heights, approximately 38, 75 and 113 mm (1.50, 2.95 and 4.45 in.) from
the base, for all cracks observed. These values were averaged to determine the
width of the individual cracks, which were summed to determine the total crack
width for that specimen. Fig. 4 depicts the ring and microscope setup. Four
specimens, cast two at a time, were tested for each mixture.
RESULTS
Flexure
Both the flexural strength and toughness of the composites can be

determined from the flexural stress-displacement relations pictured in Fig. 5 for
representative specimens. (FA ("h)in the figure refers to the percentage of fly
ash included in the mixture.) The stress at the development of the first
macrocrack, i.e. where the curve makes a marked break from the pre-peak
linearity, also known as the bend over point, is significantly increased by the
addition of the microfibers. After an initial drop in load that occurs at the bend
over point, the PVA hybrid exhibits strain-hardening behavior, while the steel
hybrid shows a drop in performance with greater deformation that continued
through the test. The macrofiber-reinforced concrete displays much more
prolonged strain-hardening and carries a greater stress than any of the hybrids
after approximately 0.8 mm (0.031 in.) of displacement.
To give a better picture of the performance for the mixtures than is

visible in one specimen alone, average properties for all beams are plotted in bar
chart form below. The line in the plot is not an error bar, but rather gives the
maximum and minimum performance seen in that mixture. The work of fracture
(Wf), Le. the area under the curve, is calculated up to deformations of W1000 =
0.4 mm (0.016 in.) and U200 = 2 mm (0.079 in.), where L = span used for
testing (Fig. 6). The PVA hybrid blend outperformed the macrofiber
reinforcement at small displacements. However, when the full displacement
pictured here is considered, the ability of macrofibers to bridge openmg
macrocracks is apparent and superior in toughness is observed.
When comparing the ultimate nominal strength of the specimens, the

maximum flexural stress carried by the steel hybrids and macrofiber concretes
was similar (Fig. 7), but neither was as high as the PVA hybrid. As expected, all
fiber-reinforced mixtures are stronger than the unreinforced concrete. The first
crack stress, determined as the stress where the strecs-displacement relation
@Seismicisolation
@Seismicisolation
differs significantly from a line fit to the curve between 10 and 40% of the peak
stress, is consistently stronger in the hybrid reinforced mixtures. In the hybrid
fiber-reinforcedmortar, the microfibers participated most actively in the fracture
process up to the formation of the macrocrack and it is apparent that they are
performing similarly in concrete.
The modulus of elasticity, E, was determined according to the method

given by RILEM (9) which is based on the initial compliance calculated from
the load vs. CMOD relationship between 10 and 40 % of the peak load. The
results obtained are pictured in Fig. 8. The trends in this relationship are similar
to those seen in the first crack stress, with the hybrid fibers significantly
increasing the modulus over the unreinforced mortar, which in turn
outperformed the macro fiber-reinforced material.
Shrinkage Cracking
Cracking was first observed to occur after approximately nine days in

these mixtures. All of the rings containing fibers experienced the development
of two cracks, while those containing no fibers saw the development of only one
crack (Fig. 9). Only one sample, a specimen containing 0.82% macrofiber,
exhibited a third crack. Multiple cracking was not seen in the unreinforced
concrete because the opening of the crack relieved the stress on the composite.
Fibers however, carry load across the crack and as additional shrinkage occurs,
additional cracks initiated in the reinforced material. The specimens containing
the PVA microfiber hybrid had the latest first crack development and then took
longer to show evidence of second cracks. The average total crack width, Le. the
s u m of the width of all cracks in the specimen, for all four specimens measured
at 44 days, is also graphed in Fig. 9 for these specimens. The unreinforced
concrete saw the largest crack opening by a significant margin. The total crack
width was similar for both types of fiber-reinforced concretes examined.
DISCUSSION
Processing heavily influenced the workability of the microfiber-

reinforced mixtures. The trial batches used for the mixture design process were
produced in a smaller mixer and required five minutes of mixing time to achieve
a uniform appearance. However, when batches of sufficient size to cast the
beam and ring specimens were prepared, a much longer mixing time was
necessary for the superplasticizer to become fully absorbed by the paste and a
reduced superplasticizer capacity was observed. Shorter mixing times resulted
in bleeding but the procedure detailed here produced a workable and cohesive
mixture. While the steel microfiber hybrid mixture demonstrated lower
@Seismicisolation
@Seismicisolation
800 Lawler et al.
workability than the macrofiber mixture, the PVA microfiber hybrid mixture
exhibited comparable workability.
Since similar experimental procedures were pursued in measuring the

performance of hybrid-reinforced concrete reported here and of hybrid-
reinforced mortar mixtures reported previously (6), comparisons can be made to
see what differences in the performance reveal about the interaction between the
fibers and the matrices. Given the fundamental differences between concrete and
mortar and the difference in w/b for the mixtures that were examined, the
increase in strength that was observed in the concrete was expected. While the
initial trend resulting from the inclusion of the microfiber was similar,
specifically an increased load carrying capacity at small displacements, it was
less pronounced in concrete than in mortar. The increase in load carrying
capacity at large displacements resulting from the microfibers throughout the
post-peak in mortar did not occur in the concrete. Instead, at displacements
greater than 0.5 mm in the flexural test, a significant drop in stress began that
continued throughout the range tested.
It was observed during testing that the step-like drops in the post-peak
region of the stress-displacement relationship pictured in Fig. 5 resulted from
macrofiber fracture. This type of behavior was not seen in the mortar. However,
these occur with greater frequency in the concrete reinforced with the hybrid
blend than with macrofiber only, suggesting that, since the matrix is nearly
identical, the macrofibers were more likely to break in the presence of the
microfiber. The microfibers, may locally reinforce the matrix around the
macrofibers, increasing the pull-out resistance and consequently the stress on the
macrofiber. Such a modification in failure mode is detrimental to the
mechanical response since an individual fiber consumes more energy (and
contributes more effectively to the composite performance) if it pulls out rather
than breaking.
An increase in modulus was related to microfiber dispersion in Ref. (6),

where it was hypothesized that the initial compliance, based on the early portion
of the load-CMOD relationship, is influenced by the effect that microfibers have
on the fracture process. It has been shown through scanning electron microcopy
that cracking begins as soon as load is applied to the specimen (10) leading to
non-linearity in the specimen response. The higher modulus of elasticity
measured here suggests that the microfibers are already influencing crack
development even at this stage of the process by reducing some of this non-
linearity. The steel hybrid showed the greatest increase in modulus, making this
the lone property where it showed the best performance. The significantly
higher modulus of the steel fibers themselves, which is nearly six times higher
than the PVA, may play a role in this. This increase in the modulus of elasticity
that resulted from microfibers observed in the concretes did not occur in the
mortars. This suggests that the microfibers may influence crack evolution in the
concrete earlier than in the mortar or that the impact of the microfibers on
@Seismicisolation
@Seismicisolation
workability in the mortar results in greater porosity, which offsets increases in
modulus resulting from modifications of the fracture process.
The trend for restrained ring shrinkage results is similar to that seen in
the flexural test results with the microfibers having the greatest influence on
mechanical performance at small deformations. The hybrid blend, containing
the microfibers, was able to influence cracking early in the development process.
The initiation of cracking was impeded by the microfibers and the first crack
occurred at a greater age. After the cracks are well-established, however, the
width to which they opened was nearly the same for both mixtures. It is likely
that the use of fly ash in the concrete also had some influence on the age at
which cracking became visible, since because the hydration process is somewhat
slower in fly ash, a greater amount of creep may have occurred relieving some of
the stress on the composite.
CONCLUSIONS
The objective of this study was to demonstrate that a hybrid blend of

microfibers and macrofibers that previously exhibited attractive reinforcement
properties in mortar could be effectively utilized in concrete. The first step in
the process was the production of workable and cohesive concretes containing
these fibers, which was achieved based on the relationship between flow, as
measured with a drop table, and the paste volume fraction. The flow,
mechanical performance and resistance to restrained ring shrinkage cracking
were evaluated for selected mixtures.
In terms of ultimate strength, the steel hybrid performance was

comparable to the macrofiber, while the PVA hybrid exceeded it by a significant
margin. The toughness of the PVA hybrid was greater than that of the
macrofiber mixture in flexural displacements less than 0.5 mm (0.020 in.) but
not superior at larger deformations. Beneficial effects of the microfiber in the
flexural testing were seen up to and shortly after the peak. In this region of the
materials’ mechanical response, the microfiber retards the process by which
microcracks grow and coalesce to form macrocracks, the development of which
signals a change in damage mechanism from crack development to crack
opening and determines the strength of the composite. The modulus of elasticity
was higher in the hybrids, particularly with the steel microfiber, suggesting that
the microfibers influence the fracture mechanism at stress levels even below
40% of the ultimate strength.
The mixture containing the PVA microfiber hybrid showed retarded

development of restrained ring shrinkage cracking as the microfibers again
delayed the coalescence of cracks. However, at 44 days of age, the total crack
@Seismicisolation
@Seismicisolation
802 Lawler et al.
width was approximately the same as that seen for the macrofiber specimen and
much below that observed in the unreinforced concrete.
Contrary to the trends observed in mortar, the PYA microfiber hybrid-

reinforcement was not as effective at improving toughness when compared to
macrofiber only reinforcement in concrete. While initial strength was improved
for the hybrid, this improvement was not maintained much past the peak load as
it was in the mortar matrix. This may be explained by the difference in the
failure mechanism of the fibers in the concrete.
In this matrix, the microfibers showed some promise in terms of

mechanical response but did not match the performance achieved with a mortar
matrix. A careful examination of the fracture surface may give insight into the
discrepancy in the effectiveness of the hybrid in mortar and concrete. However,
the similarity in the microfiber influence at small displacements in the two
materials suggests that microfibers may be used to reduce cracked permeability
in concrete as well as mortar. Clearly, given the potential for improved
performance through blending micro- and macrofibers, additional work is
warranted in this field.
@Seismicisolation
@Seismicisolation
REFERENCES
1. Bentur, A. and Mindess, S., (1993) Fiber Reinforced Cementitious

Composites, Elsevier Applied Science, U.K.
2. Banthia, N., and Sheng, J., (1990) “Micro-reinforced cementitious materials,”

in: S . Mindess, and J. Skalny (eds.), Fiber Reinforced Cementitious Materials,
MRS Symposium Proc. Vol. 21 1, Boston, pp. 25-32.
3. Betterman, L.R., Ouyang, C. and Shah, S.P., (1991) “Fiber-Matrix Interaction

in Microfiber-Reinforced Mortar,” Advanced Cement Based Materials vol. 2,
pp. 53-61.
4. Mobasher, B. and Li, C.Y., (1996) “Mechanical Properties of Hybrid Cement-

Based Composites,” AC1 Materials Journal, vol. 93, no. 3, pp. 284-92.
5 . Qian, C.Y. and Stroeven, P,, (2000) “Development of hybrid polypropylene-

steel fiber-reinforced concrete,” Cement and Concrete Research, no. 30, pp. 63-
69.
6. Lawler, J.S., (2001) Hybrid fiber-reinforcement in mortar and concrete, Ph.D.

thesis, Northwestern University, Evanston, IL.
7. Tjiptobroto, P. and Hansen, W. (1993). “Tensile Strain Hardening and

Multiple Cracking in High-Performance Cement-Based Composites Containing
Discontinuous Fibers,” AC1 Materials Journal, vol. 90, no. 1, pp. 16-25.
8. Gryzbowski, M. and Shah, S.P., (1990) “Shrinkage Cracking of Fiber

Reinforced Concrete,” AC1 Materials Joumal, vol. 87, no. 2, pp. 138-48.
9. RILEM, (1991) “Determination of fracture parameters (KI: and CTOD,) of

plain concrete using three-point bend tests, Draft recommendation, TC 89-FMT
Fracture Mechanics of Concrete-Test Methods,” Material and Structures, vol.
23, pp. 457-60.
10. Attiogbe, K. and Darwin, D., (1988) “Strain due to submicrocracking in

cement paste and mortar,” ACT Materials Journal, vol. 85, no. 1, pp. 3-1 1 .
@Seismicisolation
@Seismicisolation
804 Lawler et al.
Tables
Tahlc 1- Rihcr tvncn.
hooked 500 30 1115 210

straight 22 6
PVA monofilament 14 12 1500 36
Table 2. Mixtures examined.
@Seismicisolation
@Seismicisolation
Figures
Pre-peak At peak Post-peak
Fig 1. Three stages of the fracture process.
Fig. 2. Flexural Stress vs. midpoint displacement.
@Seismicisolation
@Seismicisolation
806 Lawler et al.
0.310 0.330 0.350 0.370 0.390 0.410 0.430 0.450

Paste Vol. Fraction
Fig. 3. Flow versus paste volume fraction for hybrid mixtures.
, -
Steel -
-
400mm
Fig. 4. Restrained ring shrinkage cracking setup.
@Seismicisolation
@Seismicisolation
\.--'-/
i L.
Lr--
I
I
Fiber : FA (%): VON,,
- : O : 0.43
.a2 Oh Macro : O : 0.43
.5% Macro + .29% PVA : 15 : 0.39
.5% Macro + .32% Steel : 15 : 0.43
I I
I I
1 I I
1
O 0.4 0.8 1.2 1.6 2
Displacement (mm)
Fig. 5. Stress vs. midpoint displacement for concrete mixtures.
u.u
5.0
4.0
4 3.0
9 2.0
1.0
0.0
MIX Mix
Fig. 6. Toughness determined for small (0.4 mm) and large (2 mm) deformation.
@Seismicisolation
@Seismicisolation
808 Lawler et al.
ii 45
40
-2 40
35
r
c 35 e30
‘y 3 0
5 25 gu 2 5
20
E 20
E 15 15
g 10 : 10
2 05 2 05
g O0
Microfiber h/pe
M~woM~c,o(%)~
Fly Ash (5)

0; I 1 “”I,””’
0.BO:O
Mix
Steel PVA
10.5,; 1 I 1 O0
0;O û.8;:O
Steel
10.5:0.32
Mi
15
PVA
0.5 : 0.29
15
Fig. 7. Ultimate strength and first crack stress.
-2 38
37
Line s h m peiiormance ange
e 36
35
e 34
33
32
-9 31
McmMum(%)~
Fly Ash (%)
28
Microfiber iype
0;O 1 1 I 10.80:O
MIX
Steel
0.5,5OJz
PVA
0.515””
Fig. 8. Modulus of Elasticity.
@Seismicisolation
@Seismicisolation
Plain Conc. 0.82% Macro. 0.5% Macro. ' Plain Conc. 0.82% Macro. 0.5% Macro.
+ 0.29% PVA + 0.29KPVA
Fig. 9. Average age at which cracks were first observed (lefi) and average total
crack width at 44 days (right).
@Seismicisolation
@Seismicisolation
SP 209-44
Ferro-Silicate Slag from ISF Zinc

Production as a Sand Replacement:
A Review
by R. Hooper, C. McGrath, C. Morrison,
and K. Lardner
SvnoDsis: Recent European and UK policies, for example the Landfill Directive
and the UK landfill tax, have set an agenda to promote reuse and recycling by con-
trolling and minimising the landfilling of secondary materials. The UK Govern-
ment and it?s agencies are actively pursuing policies of sustainable and environ-
mentally responsible construction. The changing attitute towards secondary
materials has encouraged investigation into the use of ferro-silicate slag from the
Imperial Smelting Furnace (ISF) production of zinc in construction processes.
The UK ?Ten Year Transport Pian?, including the development of the high-
way infrastructure, offers opportunities to successfully demonstrate the consump-
tion of small volume streams of secondary materials, including ISF slag, within the
local area. Pavement construction offers several opportunities for consumption, the
most credible of these being the replacement of the sand fractions by the slag in
bound mixtures, cement and bituminous. This paper focuses upon cementitious
mixtures alone.
The presence of zinc and lead ions in the ISF slag are proven to have an
impact on the setting characteristics of concrete mixtures, although there is little
difference in the compressive strengths after 28 days. The leaching characteristics
of the slag suggest that the retardation is not linearly related to the quantities of
zinc or lead leached. Additionally, leaching tests in combination with pulverised
fuel ash (fly ash) and ground granulated blastfurnace slag indicate that it may be
possible to minimise retardation of set in by including these materials in the con-
crete mixture.
Kevwords: c em entitio us mate ria Is; f erro-si li c ate sl a y;

Imperial Smelting Furnace
@Seismicisolation
@Seismicisolation
811
812 Hooper et al.
Biographical Sketches:
Dr Rebecca Hooper. BRE. Watford. UK. Rebecca is a Chartered Chemist and

principal consultant within BRE's Centre for Concrete Construction. She has a
background in glass and ceramic materials and works on a v h e t y of projects
including: secondary materials as alternative aggregates in construction;
specification to minimise damaging ASR and assessment of ceramic flooring.
Ms Clodagh McGrath. McGrath Environmental Consultants. Cork. Eire.

Clodagh has worked with the UK Environment Agency, industry, regulators,
waste management companies and Government on a variety projects to
provide innovative solutions and independent advice for waste energy and the
environment.
Miss Caroline Morrison, B E . Watford, UK. Caroline is a Ph.D. student with
University of Sheffield Centre for Waste Immobilization working at BRE. She
is a graduate of Natural Sciences from the University of Bath specialising in
materials science and environmental issues.
Mr Kevin Lardner. BRE, Watford. UK. Kevin is a consultant in BREs

Materials Laboratory Service, specialising in the production, testing and long
term monitoring of novel and innovative concrete systems.
INTRODUCTION
Sustainability is becoming an increasingly important issue within

today's construction industry and the changing attitude towards secondary
materials has encouraged investigation into the use of ferro-silicate slag from
the Imperial Smelting Furnace (ISF) production of zinc in construction
processes. This paper reviews the current UK policies on sustainable
construction, particularly with reference to cementitious mixtures in pavement
construction, examines the secondary materials currently accepted for use and
the implications for use of ISF slag. Finally, the paper disseminates the results
of recent technical investigations into the use of the ISF slag in cementitious
mixtures.
@Seismicisolation
@Seismicisolation
REVIEW OF ISF SLAG
Availability
In 2001, Europe consumed 31% of the zinc metal produced world-wide

and was responsible for 31% of the worlds zinc metal production [i], through
plants in Belgium, Bulgaria, Finland; France; Germany; Italy; Macedonia;
Netherlands; Norway; Poland; Romania; the Russian Federation; Spain and
the UK [2]. Of the European producers with the largest annual capacity
(Belgium, France, Finland, Germany, Italy, Netherlands, Norway, Spain, and
the UK), four manufacturers use ISF production, others use electrolytic
plants [3]. The UK produces 105,000 tonnes of metal per annum [3] and
approximately 80,000 tonnes of ISF slag with a stockpile of 2.5 million
tonnes [4]. An ISF production facility in Italy produces 75,000 tonnes of zinc
metal every year [3] and 60,000 tonnes of slag [SI. There are equivalent zinc
production plants in France and Germany [2] each producing 100,000tonnes
of metal per annum [3], with an estimated slag production of 80,000 tonnes.
The USA and Canada produce 1,142,000 tonnes of zinc per annum and
consume 1,580,000tonnes [2]. Canada is the world's second largest metal
producer and the USA the largest consumer of zinc metal [2]. OveraI1, 13.2 %
of the world's primary refined zinc is produced using the ISF process [2],
suggesting slag production of approximately 975,000 tonnes.
Chemical ComDosition and Phvsical ProDerties
Table 1 contains details of several ISF slag compositions and compares

these to other metal industry by-products used in the construction industry [6-
91. A proportion of the calcium oxide in steel slag will be present as "free
lime" which will undergo volumetric expansion when combined with water.
For use as an aggregate, blastfurnace slag is assessed according to BS
1047:1992 [IO] including assessment of:
o Iron unsoundness: cracking, disintegration, shaling or flaking after 14 days
soaking in water.
0 Dicalcium silicate unsoundness:
if (1.2 Sioz+0.4 Alzo,+ 1.75 S ) 2 (Ca0+0.8 MgO) (mass percentages)
or (0.9 SO,+ 0.6 A1,0,+1.75 S ) 2 Ca0 (mass percentages).
The comparison of chemical compositions of the blastfurnace and steel

slags show similar compositional levels of SiO, and Cao, demonstrating that
the ratio of these two quantities cannot be used to judge dicalcium silicate
unsoundness or the presence of free calcium oxide. However, the lower levels
of these materials present in the ferro-silicate slags suggest that this will not be
@Seismicisolation
@Seismicisolation
814 Hooper et al.
an issue. Of more concern are the high quantities of iron, potentially present as
iron sulfide, which could result in iron unsoundness. However, the apparent
stability to weathering suggests that long-tem stability of ISF slag as an
aggregate is not a concern [4]. The slag is glassy and granular in nature and
has a similar particle size range to sand [4, 111, but it has a higher specific
gravity [4, 8-91, and lower wafer absorption than river sand (Table 2).
CURRENT EUROPEAN AND UK POLICIES ON WASTE

MANAGEMENT AND SUSTAINABLE CONSTRUCTION
The Landfill Directive
There are increasing restrictions on the disposal of waste in Europe that

could increase the amount of pre-treatment needed before disposal. The recent
Landfill Directive [ 121outlines the necessary measures to tackle the increasing
waste generation in Europe. The aim of the Directive is to provide measures,
procedures and guidance to prevent or reduce negative effects to the global
environment and all its cycles from landfilling of waste during the whole
lifecycle of the landfill site. The Directive encourages reuse and recycling of
components and materials through various means, including growing disposal
costs.
Subsidies to assist this change are few and far between, however a
positive response has been witnessed by European Member States
implementing waste management plans. In some Member States, these plans
have served to set targets and increase levels of reuse and recycling. In others,
there is a growing wealth of infomation available to help them improve
performance. Funding from the EU and European Member States helps
support research and development in new techniques and technologies, and
more recently dissemination of established knowledge (for example, Thematic
Network on reusing foundry wastes and Thematic Network on recycled
aggregates).
UK Fiscal Policies
The Landfill Tax was introduced on 1st of October 1996 and it applies
to waste that is disposed of in licensed landfills. The tax seeks, as far as is
practicable, to ensure that the price of landfill fully reflects the impact that it
has upon the environment. It provides an incentive to reduce the waste sent to
landfill sites and to increase the proportion of waste that is managed at higher
levels of the waste hierarchy.
@Seismicisolation
@Seismicisolation
The Landfill Tax Credit Scheme was established to permit up to 20%
of the taxes collected by landfill operators to be used for the purpose of
implementing social and environmental projects complying with specific
"approved objects" in the regulations. The revenue is used to encourage the
use of more sustainable waste management practices and technologies, or to
establish partnerships between landfill operators and local communities. Thus
the Scheme benefits both the community and the waste industry by providing
opportunities to enhance social and environmental conditions and services at
the local level. The Landfill Tax Credit Scheme is managed by "Entrust" and
individual projects and funds that are approved by Entrust have to be managed
by an authorised environmental body.
In April 2000 the Treasury agreed to introduce an tax on primary

aggregates (that is virgin material) from April 2002. Secondary and tertiary
aggregates will not be subject to the tax, which should encourage a greater use
of recycled aggregates in low to high-grade applications. A large proportion of
the tax will be used for a "Sustainability Fund' that will be used to support
various initiatives that are currently out for consultation. This will include
development, improvement and research and development within the industry,
its facilities and its impacts to the local populations. Similar to the Landfill
Tax, it is envisaged that a competent authority will oversee the management
and approval of the fund and projects.
Sustainable Construction Strategy
The need to reduce waste at all stages of construction was central to the
message of "Rethinking Construction" [13], the 1998 report of the
Construction Task Force on the scope for improving the quality and efficiency
of UK construction. Improving the efficiency of the construction industry is a
key objective for the UK Government, as set out in its strategy for more
sustainable construction "Building a Better Quality of Life" [141. The Strategy
identifies priority areas for action, and suggests indicators and targets to
measure progress. It sets out action that the Government has already taken,
further initiatives that are planned, and highlights what others can do.
The Sustainable Construction Strategy emphasises the importance of

reducing waste at all stages of construction by focusing on the need to
consider long-term impacts of design, construction and disposal decisions so
that materials and other resource use is optimised. The Strategy encourages the
industry (including clients) to consider refurbishment or renovation as an
alternative to new buildings and structures. It highlights the need to avoid
over-specification in materials and the scope for standardisation of
components.
@Seismicisolation
@Seismicisolation
816 Hooper et al.
Main Barriers to Consumution of Small Volume Secondarv Materials
In recent years the quantities of aggregates produced for construction is

approximately 205 million tonnes per annum in the UK [15]. It is interesting to
note this is a reduction of approximately one third compared to ten years ago
[15]. This is potentially a result of increased consumption of secondary
materials in construction. Despite the substantial work that has been conducted
in the UK relating to the use of certain secondary materials in construction,
this research have tended to focus upon the larger volume materials, such as
ferrous slags and power station ashes. More recently there has been
considerable research examining the use of construction and demolition
wastes, including the reuse of highway-pavement materials. The non-ferrous
industry has been given little regard, however there are potential uses for such
secondary materials that could provide cost benefits to both the producer and
the user.
These secondary materials are often generated only in small quantities

at many, widespread locations and there are consequently process issues (such
as handling, storage, transport and specification as well as overall lack of
awareness of the opportunity) that currently prevent their utilisation as
construction materials. The requirement to deal with these materials
consequently results in increased costs for the responsible party, consumes
valuable landfill sites and, in the case of the manufacturing industry, lowers
the economic viability of the organisation.
The ISF slag is an example of such a material, along with foundry

sands, glass cullet collected through recycling schemes, minestone spoil,
quarry residues, spent oil shale and cement kiln dust. The quantities of
material produced in each instance may be considered small in relation to the
volumes of aggregates consumed annually by the UK construction industry.
However, the comparatively small volumes could be readily consumed in the
locality of production on an annual basis. Such use would provide
considerable environmental benefits without significantly affecting local
economies reliant on production of primary aggregates. Pavement construction
and maintenance is a major concern of local authorities. The synergy between
the use of these secondary materials in local construction projects, and the
support for the local manufacturing industry and recycling projects, and hence
the economic viability of the area will be important in establishing the local
consumption of these materials.
@Seismicisolation
@Seismicisolation
THE USE OF SECONDARY MATERIALS IN PAVEMENT
CONSTRUCTION
In July 2000 the UJS Government announced Transport 2010, the ten-
year strategy to improve the transport system. This includes E121 billion
capital investment through public and private sector partnership [16]. The
strategy includes major works to the road and rail systems as well as the
support of other schemes, for example "super-trams" and "park and ride"
schemes. With the anticipated investment into the highway network, there will
be many construction projects local to the secondary material production
facilities, including UK ISF zinc production [17].
UK Highways APency Policv on Use of Secondarv Materials
As part of its policy the Highways Agency [18] aims to "improve its
environmental performance by:
reducing waste through re-use and recycling and by using refurbished and
recycled products and materials where such alternatives are available;
encouraging manufacturers, suppliers and contractors through
specijìcations to develop environmentally preferable goods and services at
competitive prices;
working with contractors to improve environmental performance where
this is relevant to the contract and to the achievement of value for money".
The degree of interest in recycling and using secondary materials is

evidenced by the research resources committed to these issues by the
Highways Agency (E2,265,000 between 1993 and 2001) [19].
The Design Manual for Road and Bridges, one of the Highways
Agency's specifying documents, contains specific references to the use of
secondary materials in construction [20]. "Zt is government policy to
encourage use of reclaimed and marginal materials wherever possible to
obtain environmental benefits and reduce the pressure on sources of natural
aggregates. Suitable materials may be those reclaimed from roads during
reconstruction, from residues of industrial processes including mining and
from the demolition of other construction projects. '' Equally the Specifications
for Highways Works [21] specifies the use of several secondary materials, as
shown in Table 3. Although ISF slag is not specifically mentioned in this
Table there is a clear indication that appropriate drivers are in place to support
the use of secondary materials in highway construction. The Highways
Agency recognises the need for flexibility in specification to enable the use of
secondary materials: "It is expected the quantities of materials recycled will
increase as the range of industrial by-products permitted in the specification is
extende$' [20].
@Seismicisolation
@Seismicisolation
818 Hooper et al.
Overview of Hirrhwav
- Construction
A typical diagram of pavement construction is given in Fig. 1 [22].

The foundation compromises the underlying subgrade, capping and sub-
base. It is the “theplatform upon which the more expensive and
structurally signijhnt layers are placed‘ 1221. The subgrade is the soil
underlying a pavement and may include a fill material. Capping is an
optional subgrade improvement layer protecting the subgrade from
damage. It is made up of coarse or fine graded granular material or
material stabilised with cement or lime [23].The sub-base is the platform
layer upon which the main structure of a pavement may be laid [22].
The surfacing is the upper layers of the pavement designed to carry traffic
directly. The roadbase is the main structural layer of the pavement. The
surface dressing is a single/double/triple layer of aggregate combined with
one or more layers of binders to form a running surface. The roadbase and
surfacing are combined in concrete pavements.
Assessment of Leaching:
- from Secondm Materials
The increased use of secondary materials and increased concern over

the protection of the environment and human health has led to a more detailed
scrutiny or regulation of unfamiliar processes and materials usage. In the UK,
regulation is the responsibility of the Environment Agency, and local
inspectors make decisions according to a number of guidelines and precedents.
Established practice in the UK has normally been built around the usage of
natural materials plus a small number of waste materials that are exempted
from general inspection.
For other secondary materials, leaching tests need to be carried out. If

the leachate conforms to quality standards for drinking water, there should be
no concerns regarding the use of the material. This is appropriate where the
material will be in direct contact with aquifers, or that water coming into
contact with it could pose a significant risk to aquifers. Where the leachate
does not conform to quality standards for drinking water, the extent to which a
material can be used is open to interpretation. It is possible for the material to
be encapsulated (wholly or partially) andor for the material to be used in a
non-sensitive situation, but this will be at the discretion of the inspector.
A variety of leaching test methods have been used, but it seems likely
that these will narrow down to those being developed as European Standards.
The European Standardisation Technical Committee CEN/TC 292
“Characterisation of Waste” has produced a series of drafts (prEN12457 Parts
1-41241) for finelgranular material (mostly crushed to below 4 mm) for general
@Seismicisolation
@Seismicisolation
compliance purposes, and two others for monolithic material. The relevant
committee for construction aggregates, CEN/TC 154 "Aggregates", has
produced its own extraction method prEN1744-3 [25] based on a German
method. A recent European Union Project, "Alternative materials in road
construction", undertook a review of leach testing [26]. The objective of this
study, and the resulting programme outcomes, was to assist the specification
of harmonised tests through CEN/TC 292.
POTENTIAL AREAS FOR USE OF ISF SLAG IN PAVEMENT

CONSTRUCTION
The methods of pavement construction suggest the following areas of

potential use of the ISF slag:
1) As fill materials for subgrade soils,
2) As aggregates in unbound base layers in highway construction,
3) As aggregates in bituminous and cement bound layers in highway
construction,
4) As aggregates in pavement quality concrete,
5 ) As feedstock for cement production.
This paper focuses particularly on the use of the ISF slag in

cementitious construction, although the other areas of construction are
mentioned here. The use as an aggregate in bituminous bound applications is
also a credible application [27].
Fill Material for Subgrade Fills and Capping and Unbound Aggregates
Consideration of the use of ISF slag in both subgrade and unbound

basecourses of highway construction has several advantages over other types
of use: proving the technical feasibility is reasonably readily achieved (testing
to comply with Standards); minimal treatment required by the supplier.
Leaching is likely to be a significant concern when proposing the use of these
materials and the potential for use in a full market is very limited. Already
significant quantities of construction and demolition waste are consumed in
this way and it will be difficult to replace these materials with other untried
products. Given these difficult market conditions, it would appear more
appropriate to focus on the technical proving of the ISF slag in bound
applications.
@Seismicisolation
@Seismicisolation
820 Hooper et al.
&repates Bound in Bitumen for Roadbase Construction
Although there are fewer secondary materials specified for use in

bituminous bound courses (Table 3) it is worth noting that both blastfurnace
slag and steel slag have specific provisions for use. Hughes [28] reported that
zinc smelter waste could be substituted for sand in sand-asphalt mixtures
without detriment to the mix. Stripping of the binder from the aggregate may
need to be addressed when investigating the use of glassy ISF slag in
bituminous systems. The use of ISF slag in bituminous bound road courses
appears to be worthy of technical evaluation as bound applications are likely to
cause less concern with regard to hazardous leachates than unbound
applications.
&regates Bound in Cement for Roadbase Construction
For pavement construction the aggregates for these cement bound

layers are similar to those for concrete. The types of cement bound materials
(CBM) are sub-divided into five classifications, CBM 1 being the lowest
compressive strength, suitable for sub-base, CBM 5 being the highest
compressive strength, suitable for roadbase [21]. A change is also reflected in
the particle size distribution of the aggregates across these classifications.The
use of the secondary materials given in Table 3 [21] is specified as follows:
CBM 1 All the secondary materials listed can be used if they comply
with the strength requirements and do not swell or crack
following immersion in water for 7 days.
CBM 2 All of the secondary materials listed can comply with the
requirements, but the 10%fines values must be greater than
50 kN (indicating mechanical properties).
CBM 3 - 5 Crushed concrete, crushed air-cooled blastfurnace slag and
pulverised fuel ash are all specified. China clay sand is
potentially suitable. In these categories both crushed rock and
slag must be tested for their coefficient of thermal expansion.
The ISF slag may be appropriate for use as an aggregate in these

layers. The available literature indicates that slags from incineration of "urban
solid waste" can be successfully accommodated in cement bound layers [29].
However, it is known that the inclusion of available lead and zinc in cement
can retard the hydration and strength development of cement pastes [30-341.
There is no indication that this inclusion will have detrimental effects upon the
final strength development [9]. However, retardation could prevent
compliance, as the required strength at 7 days may not be achieved. Depending
on the properties of these zinc and lead containing materials as aggregates, for
@Seismicisolation
@Seismicisolation
example durability, crushing resistance and so on, there may be advantages to
the Highways Agency in the specification of these materials where available.
Sand Replacement in Concrete Construction
For the most part the Highways Agency specifications are in

accordance with BS 5328, "Concrete" [35-381.The British Standard for
concrete is flexible enough to allow the use of alternative materials, once
compliance is proven [35]: "Where non-standard materials are used, there
should be satisfactory data on their suitability and assurance of quality
control." All highway structures using cement, that is including sub- and
roadbases, must take account of the potential for alkali aggregate reaction, an
expansive reaction between the siliceous component of aggregates and alkalis
from the cement matrix.
Several authors have examined the use of secondary materials from the
non-ferrous metals industry as aggregates in concrete construction, including
materials from zinc production, materials from the production of alumina from
bauxite (black sand) and ferro-nickel slags [5,9,28,39-431. As stated
previously, the presence of zinc and lead in cement mixtures is associated with
retardation of set [30-341. However the retardation of set could imply that the
ions are becoming bound to the cement paste matrix, thus being unavailable,
or less available, for leaching.
Cement Production
ComDosite Cements--Both pulverised fuel ash (pfa) [44] and ground

granulated blastfurnace slag (ggbs) [45] are consumed as cementing materials
within the UK construction industry in composite cements (either blended
during cement production or as additions to cement after production).
Pulverised fuel ash is a pozzolan, a material containing silica in a reactive
form, ggbs is a latently hydraulic cement, requiring the presence of alkali
(provided by the portland cement) to activate its hydraulic behaviour. For use
with portland cement, the blastfurnace slag component is required to (in
addition to other requirements) [45]:
Contain at least two-thirds by mass of glassy slag,
Consist of at least two thirds by mass of the sum of Cao, MgO and SiO,,
The ratio by mass (Ca0 + MgO)/ (SiO,) will be more than 1.0 and (Cao)/
(SiO,) less than 1.4.
There has been one report of the use of waste slags from lead
production in composite cements, and although the time to set was retarded,
the strength development over 28 days was similar [46]. The data in Table 4
@Seismicisolation
@Seismicisolation
822 Hooper et al.
assesses compliance of the ISF slags to some of the clauses in BS 6699 [45]
and a comparison with pfa composition is also made. Pfa for use with portland
cement should have no more than 10%by mass calcium oxide, as the chemical
composition in the Table 4 suggests. This Table indicates the ISF slag fails to
meet these selected requirements of the British Standards for ggbs and pfa,
suggesting its use in composite cements is not appropriate.
The viability of a composite cement blending, as described for lead

smelting slag [46], would depend upon the consistency of the slag products, in
terms of the set-retarding components, and the availability of the materials. As
mentioned previously, the retardation of set implies the lead and zinc
components of the slags are bound in the cement matrix, perhaps making this
use of the slag more attractive. However, there will need to be a careful study
of the health and safety implications of grinding such lead bearing materials to
fine powders. Equally, if these materials can be consumed as aggregates this
would be preferable as the processing required is significantly less.
Cement Kiln Feedstock-- In essence the production of cement involves

the high temperature combination of three materials with the components,
calcium oxide, aluminium oxide and silicon dioxide. Natural materials
consumed in this production are often limestone or chalk, clay and shale.
Small additions of iron oxide are made in some situations to meet final
production requirements [47]. After passing through a production kiln the
resulting "clinker" is then ground to produce portland cement.
Iron oxide is added to cement kilns particularly when sulfate-resisting

portland cement is in production. The high iron contents of the ISF slag would
be a possible replacement for this addition. This potential use is reliant upon
the ability of local production facilities to make these additions and the levels
of zinc and lead in the slag, their implications for processing, in addition to the
possible retardation of setting [47]. Given the current consumption of
materials by the cement industry, and the comparatively low volumes of ISF
slag generated, it is unlikely that further investigation would be worthwhile.
The expense involved in testing such materials for the cement companies, and
the emission regulations they are subjected too, make such propositions
unattractive to cement companies. However, this situation should be
reconsidered regularly in light of changing EU and UK legislation.
ISSUES RELATED TO THE USE OF ISF SLAG IN CEMENTITIOUS

MIXTURES
Hughes and Haliburton [28] suggest that the ISF slag should not be
used in concrete due to the retardation of set, low strengths and the high alkali
reactivity of the material. Atzeni et al [ 5 ] , however, found that the partial
@Seismicisolation
@Seismicisolation
replacement of the sand for ISF slag in concrete was not significantly
detrimental to the compressive strength of the concrete after one month. These
results are borne out by those of Sugita et al [9], who demonstrated increases
in compressive strengths of concretes after two years when 30% of the sand
was replaced by slag. Additionally, the replacement at 30%, 50% or 100% of
the sand was accompanied by a reduction in the depth of carbonation after two
years. However, these concretes also under went expansion, up to three times
the expansion of the concrete containing sand alone, over the two years. Sugita
et al suggest this could be due to the expansive oxidation of the iron
component in the slag, although the possibility of ASR should not be
dismissed.
The harmful effects of heavy metals in our environment have, in turn,

generated a variety of methods to remove them from wastewater and industrial
effluent. It is anticipated that some of this understanding can be applied to fix
any soluble zinc and lead ions from ISF slag into cementitious mixture. This
fixation is required to ensure the concrete sets within a reasonable time-scale,
although significant experimentation would be required to successfully
transfer this technology from the treatment of effluents to the highly alkaline
environment of a cementitious mixture.
A review of waste processing literature [48-581 highlighted various

cement additions that may be able to counteract this, including fly ash,
blastfurnace slag and sodium silicate. As mentioned previously, pulverised
fuel ash (pfa) and ground granulated blastfurnace slag (ggbs) are common and
inexpensive UK concrete construction materials. Furthermore, particular types
of pfa and ggbs are specified in the UK to minimise ASR [59].
STUDIES INTO ISF SLAG AS A CONCRETE AGGREGATE
Technical studies at BRE, to test the extent of retardation of concrete

set when ISF slag is included in concrete mixtures and to assess the ability of
pfa and ggbs to minimise the leaching of lead and zinc ions in highly alkaline
environments, are reported here. Experimentation was carried out on two
different ISF slags that reflect variations in the production process ( S 1 and
S2). Table 5 provides an indication of the major components of the two slags
determined by X-ray Fluorescence (XRF) analysis.
Experiments Assessing the Retardation of Concrete Setting
Temperature monitoring of setting concrete was carried out on

mixtures containing the ISF slags to assess the impact on concrete set and
@Seismicisolation
@Seismicisolation
824 Hooper et al.
cement hydration. An increase in temperature of the mix shows
reactionhydration of the cement in the concrete. Table 6 details the control
and test mixtures containing the ISF slags and Fig. 2 shows the temperature
changes during the setting of these mixtures. The delay in set ranges from 1
day to 6 days, depending on the slag composition and the quantity of slag
present.
All mixtures reach approximately the same maximum temperature

(37°C +. 2"C), however the mixtures containing the ISF slags show a delay in
achieving this maximum temperature although the temperature change profiles
are very similar. This demonstrates that the hydration reaction occurs to the
same extent in each mixture, but that the ISF slags retard the onset of
hydration. This conclusion concurs with 28-day compressive strength
measurements made on mortar specimens, which show little variation in
average strength of sand, s a n d slag and slag alone mixtures, despite
retardation in set evidenced by the time to demould (Table 7).
ExDeriments Assessing Leaching in Highlv Alkaline Environments
Leaching experiments were carried out to determine the levels of zinc

and lead released from the slags in various test solutions. Of each ISF slag,
25 g were added to 100 ml of the test solution and then shaken for 24 hours.
The samples were then filtered and the solutions tested for the presence of zinc
and lead ions by atomic absorption spectroscopy. The final solution pH was
also measured using a pH electrode. Figs 3 and 4 show the experimental
results for the leaching of zinc and lead ions into the test solutions and the
solution pH.
The lead ions appear to be more accessible for leaching than the zinc
ions, although a higher quantity of zinc is present in both the slag samples
(Table 5). Greater concentrations of lead and zinc ions were leached from S 1
than S2, which might be anticipated from the compositional analysis (Table 5).
There is a connection between the pH of the leaching solution and the
quantities of ions leached. Leaching is greater at the high pH anticipated in
cementitious mixtures (12-13) than at the lower pH associated with water (7-
8). However, other chemical reactions clearly play a part in the levels of free
ions in solution. For example, less zinc and lead ions are leached from S1 into
the pfa solution, at pH 10.3, than the saturated calcium oxide solution in
combination with pfa, at pH 9.8. Very low levels of ions were leached from
the slags into solutions of pfa and ggbs, although higher levels were leached
from S1 into the solution containing sodium silicate, at similar pH. These
results indicate competition between increased ion solubility at high pHs and
chemical reaction which removes the ions from solution.
@Seismicisolation
@Seismicisolation
Discussion of Results
Thomas et al. [31] suggest that, within cementitious mixtures, metal

ions such as zinc and lead form a gelatinous layer around the cement grains
that prevents cement hydration from occurring. Water diffusing through this
gelatinous layer finally reacts to form hydrated cement. Such a diffusion
controlled reaction mechanism implies that the rate of hydration of the cement
would be slowed in the presence of lead and zinc. However, Fig. 2 shows that
hydration is prevented for a time, but that when it occurs the reaction
temperature profiles are almost identical to mixtures without these ions
present. These results show that, under these circumstances, the cement
hydration reaction is not diffusion controlled but that once water has
penetrated to the cement grain, the gelatinous layer is completely disrupted
and hydration can take place unhindered.
The longest delay in cement hydration is seen with the complete

replacement of the sand with the slag S2, although there are generally less lead
and zinc ions leached from this slag than from S 1. Thus, the retardation period
is not linearly connected to the concentration of ions present but to the ratio in
which these ions are present. This ratio may affect the chemical composition
of the gelatinous layer formed, which in turn affects the layers’ permeability to
water and hence the delay in the onset of cement hydration.
Retardation of set by the formation of a permeable gelatinous layer,

implies that the ability of pfa and ggbs to “mop-up” ions from the mix water,
demonstrated in Figs 3 and 4, may prevent the formation of a gelatinous layer
and hence eliminate the retardation of set. Reduction in ion availability will
not only be beneficial in reducing the potential for delayed set but could also
minimise the probability of detrimental in-service leaching of ions into the
surrounding environment. Additionally, the inclusion of pfa and ggbs,
specified measures in the control of alkali-silica reaction (ASR) [59], could
eliminate any concerns about the glassy nature of the aggregate and its
susceptibility to ASR [28].
CONCLUSIONS
Implications of Current UK Policy for Use of ISF Slap,
Of the potential uses of ISF slag in pavement construction, this review

suggests that investigations into the use of this material as cement bound
aggregate would be most beneficial, based upon technical feasibility, costs,
market requirements and likelihood of success. Concrete construction can be
regarded as a continuous manufacturing process throughout the UK and it is
@Seismicisolation
@Seismicisolation
826 Hooper et al.
likely that the ISF slag could be exploited locally, without incumng high
transportation costs. This strategic technical work could provide a consistent
use for these materials in the medium and long term. Drivers for sustainable
construction, specifically the tax on primary aggregates, may well support and
encourage this use. Using these aggregates in a bound medium will reduce the
risks of hazardous leachates and such a form of encapsulation is likely to be
more attractive to the users and regulatory authorities. However, given the
nature of these materials, processing, handling and end of life issues would all
need to be addressed in any technical programme.
Conclusion from Technical Studies
Technical studies on cementitious systems have confirmed the

retardation in set due to the presence of zinc and lead ions in the ISF slag.
Concrete set is delayed for a period and then proceeds as it would in a control
sample without slag. This delay may be caused by the formation of gelatinous
layer around the cement grains, the water permeability of this layer controlled
by the ratio of lead and zinc ions in the mix water. The use of pfa and ggbs
appear to have the potential to control the release of ions within the mixture
and mitigate the retardation of set, in addition to minimising the risk of
deleterious ASR.
The experimental results demonstrate it is technically feasible to use

ISF slag as an aggregate in cementitious mixes. The partial replacement (50%
or less) of the sand fraction with the slag causes less technical difficulties, as
the delay in set is less, though the technical barriers to the use of high
replacement volumes (50 to 100% replacement) are surmountable.
ACKNOWLEDGEMENTS
BRE gratefully acknowledge the support of the UK Department of

Trade and Industry and Rowan House Limited for their support in this work,
along with Britannia Zinc Limited.
REFERENCES
[i] P. Dawson, Zinc Statistics, International Lead and Zinc Study Group,
www.ilzsg.org, April, 2002.
@Seismicisolation
@Seismicisolation
[2] Intemational Zinc Association, Zinc production and consumption,
www.iza.com, October, 2001.
131 European IPPC Bureau, Draft Reference Document on Best Available

Techniques in Non Ferrous Metals Industries, World Trade Centre,
Seville, 2000.
[4] A. Nash, Britannia Zinc Ltd., Personal Communication with R. Hooper,

BRE, Watford, UK, 1999.
[5] C. Atzeni, A. Massida, U. Sanna, Use of granulated slag from lead and
zinc processing in concrete technology. Cem. Concr. Res., 26 (9) (1996)
1381-1388.
[6] W. Gutt, P. Nixon, Use of waste materials in the construction industry.

Materials and Structures, 12 (70) (1979) 255-306.
[7] H. Ito, T. Nakamura, M. Kawahara, K. Morinaga, Phase transformations

and leaching characteristics of lead and zinc slag after heat treatment, in:
T. Azakami (Ed.), Zinc and Lead '95, Proceedings of the International
Symposium on the Extraction and Applications of Zinc and Lead,
Tohuku University, Sendai, 1995, pp 502-51 1.
[8] M. Kawahara, K. Morinaga, H. Ito, T. Nakamura, The potential of zinc

slag as part of brick material source, in: T. Azakami (Ed.), Zinc and
Lead '95, Proceedings of the International Symposium on the Extraction
and Applications of Zinc and Lead, Tohuku University, Sendai, 1995, pp
516-519.
[9] S . Sugita, M. Shoya, T. Sugawara, Studies on the properties of concrete

with zinc slag as fine aggregate. CAJ Review (1988) 92-95.
[lo] British Standards Institution, BS 1047:1983, Air-cooled blastfurnace

slag aggregates for use in construction, BSI, London, 1983.
[I 11 British Standards Institution, BS 882: 1992, Specification for aggregates

from natural sources for concrete, BSI, London, 1992.
[12] European Commission, Council Directive 99/31/EC of 26'" April 1999

on the landfill of waste, Brussels, 1999.
[I31 Her Majesty's Stationary Office, Rethinking construction: the report of

the Construction Task Force [Chairman: Sir John Egan] - July 1998 -
Sponsored by Construction Industry Sponsorship Division, Her
Majesty's Stationary Office, London, 1998.
@Seismicisolation
@Seismicisolation
828 Hooper et al.
[ 141 Department of the Environment, Transport and the Regions, Building a
Better Quality of Life: A Strategy for More Sustainable Construction
April 2000, Copies of this document are available from: DETR Free
Literature PO Box 236 Wethererby LS23 7Nl3, Product code 99CD1065,
London, 2000.
[15] B. Q . Blackwell, Movement for Innovation, Department of the

Environment, Transport and the Regions, Personal communication with
Clodagh McGrath, BRE, Watford, March 2001.
[ 161 Department of the Environment, Transport and the Regions, Transport

2010, The 10 year plan, DETR, London, July 2000.
[17] Department of the Environment, Transport and the Regions, Transport

2010, The 10 year plan, Regional Factsheets, South West (Cornwall
(including the Isles of Scilly), Devon, Dorset, Gloucestershire, Somerset,
South Gloucestershire, Wiltshire and unitary authorities of Bristol, Bath
and north East Somerset, North Somerset, South Gloucester, Swindon,
Torbay, Poole and Bournemouth), DETR, London, July 2000.
[ i 81 Highways Agency, Greening operations, Green policy statement and

action plan, Her Majesty's Stationary Office, London, 1999.
[i91 Highways Agency, Compendium of Highways Agency Research 1994-

1999, Her Majesty's Stationary Office, London, 1999.
[20] Highways Agency, Design manual for roads and bridges, Volume 7:
Pavement design and maintenance, Section 1-Preamble, Part 2-HD
35/95 Technical Information, Chapter 2 - Conservation and the use of
reclaimed materials in road construction and maintenance, Her Majesty's
Stationary Office, London, 1995.
[21] Highways Agency, Manual of contract documents for highway works,

Volume 1, Specification for highway works, Her Majesty's Stationary
Office, London, 1998.
Pavement design and maintenance, Section 1-Preamble, Part 1-HD
23/99 General Information, Her Majesty's Stationary Office, London,
1999.
Geotechnics and Drainage, Section 1-Earthworks,Part 6-HD 74/00
Treatment of Fill and Capping Material Using Either Lime or Cement or
Both, Her Majesty's Stationary Office, London, 2000.
@Seismicisolation
@Seismicisolation
[24] CENRC 292, prEN12457 Parts 1-4, Characterisation o f Waste,
European Committee for Standardisation, CEN Central Secretariat,
Brussels, 2000.
[25] CEN/TC 154, prEN1744-3, Aggregates, European Committee for

Standardisation, CEN Central Secretariat, Brussels, 2000.
[26] J. M. Reid, Deliverable D3, literature review, Volume 1, Transport

Research Laboratory, Crowthome, 1999.
[27] R. Hooper; C., McGrath, R. Collins, The use on non-ferrous metals

industry wastes in construction, BRE 423, CRC Ltd, London, 2001.
[28] M. 1. Hughes, T. A. Haliburton, Use of zinc smelter waste as a highway

construction material, A Highway Research Record, 1973 (430), pp 16-
25.
[29] T. Pizzocchero, M. Pasetto, The use of urban solid waste slags in soil
cements and cement bound granular materials, Institute of Concrete
Technology Conference, Concrete and the Environment, April 1993.
[30] F.M. Lea, The chemistry of cement and concrete, Third Edition, Edward
Amold Ltd., London, 1970.
[3i ] N.L. Thomas, D.A. Jameson, D.D. Double, The effect of lead nitrate on
the early hydration of portland cement. Cem. Concr. Res., 11 (1981)
143-153.
[32] I.N. Stepanova, L.G. Lukina, L.B. Svatovskaya, M.M. Sychev,

Hardening of cement pastes in presence of chlorides of 3d elements. J.
Appl. Chem. c/c Zh. Prokl. Khim., 54 (1981) 885-888.
[33] C. Tashiro, H. Takahasi, M. Kanaya, I. Hirakida, R. Yoshida, Hardening

property of cement mortar adding heavy metal compound and solubility
of heavy metal from hardened mortar. Cem. Concr. Res. 7 (1977)
283-290.
[34] Bhatty, J. I., The role of minor elements in cement manufacture and use,
Research and Development Bulletin RD 109T, Portland Cement
Association, Skokie, Illinois, 1995.
[35] British Standards Institution, BS 5328-1:1997, Concrete. Part 1: Guide

to specifying concrete, BSI, London, 1997.
[36] British Standards Institution, BS 5328-2: 1997, Concrete. Part 2:

Methods for specifying concrete mixes, BSI, London, 1997.
@Seismicisolation
@Seismicisolation
830 Hooper et al.
[37] British Standards Institution, BS 5328-3:1990, Concrete. Part 3:
Specifications for the procedures to be use in producing and transporting
concrete, BSI, London, 1990.
[38] British Standards Institution, BS 5328-4:1990, Concrete. Part 4:

Specifications for the procedures to be used in sampling testing and
assessing compliance of concrete, BSI, London, 1990.
[39] S. Sugita, M. Shoya, T. Sugawara, Study on the effective use of water

granulated slag as fine aggregates of concrete, CAJ Review, 1986, pp
138 -141.
[40] S . Sugita, M. Shoya, T. Sugawara, Frost resistance of concrete with non-

ferrous metal slags as aggregates, Proceedings of the Fourth
International Durability of Building Materials and Components, Volume
2, Singapore, 1987, pp 991-996.
[41] Y. Hatakana, The use of black sand for portland cement concrete,
Transactions of the Japan Concrete Institute, Volume 5, 1983, pp 105-
110.
[42] M. Shoya, S. Sugita, Y. Tsukinaga, Freeze-thaw resistance of concrete

incorporating ferro-nickel slag as fine aggregates, J. Soc. Mat. Sci.
Japan, Vol. 43, No. 491, 1994, pp 976-982.
[43] T. Katayama, Petrographic Studies on the potential alkali-silica

reactivity of ferro-nickel slags for concrete aggregates, Proceedings of
the 9"Intemational Conference on Alkali-Aggregate Reaction in
Concrete, London, 1992, pp 497-509.
[44] British Standards Institution, BS 3892-1:1997 Pulverized-fuel ash.

Specification for pulverized-fuel ash for use with Portland cement, BSI,
London, 1997.
[45] British Standards Institution, BS 6699:1992, Ground granulated

blastfumace slag for use with Portland cement, BSI, London, 1992.
[46] S. L. Sarkar, D. Bonen, Non-ferrous lead smelting slag waste as a

potential constituent in the production of blended cement, Proceeding of
the 2"' International Symposium on Blended Cements, Malaysia, 1994,
pp 108-113.
[47] K., McQuillan, G. Smith, Personal communication with Rebecca

Hooper, BRE, Watford, 2001.
@Seismicisolation
@Seismicisolation
[48] R.K. Sharma, S. Kumar, A.K. De, P.K. Ray, Use of fly ash as an ion
exchanger in water filtration studies for the removal of heavy metals. J.
Environ. Sci. Health, A25 (6) (1990) 637-651.
[49] M-H. Leu, J-E. Chang, I-P. Chen, Evaluation of the detoxification
process for treating biosolids from an integrated industrial wastewater
treatment plant. Proceedings 13" International Conference Solid Waste
Technology Management, 4A/6, 1997, 1-7.
[50] J.V. Rouse, Removal of heavy metals from industrial effluents. J.

Environ. Eng. Div., 1976,929-936.
[51] G. Schünard, Comparison of the effect of lime and a cation exchanger on

the fixation of heavy metals in soil. Landwertech. Forschung, 32 (4)
(1979) 395-404.
[52] J. García-Miragaya, M. Dávalos, Sorption and desorption of Zn on Ca-

Kaolinite. Water, Air and Soil Pollution, 27( 1986) 217-224.
[53] V. Albino, R. Cioffi, M. Pansini, C. Colella, Disposal of lead-containing

zeolite sludges in cement matrix. Environmental Technology, 16 (1995)
147-156.
[54] C.S. Poon, C.J. Peters, R. Perry, Mechanisms of metal stabiliztion by

cement based fixation processes. The Science of the Total Environment,
41 (1985) 55-71.
[55] C.S. Poon, A.I. Clark, C.J. Peters, R. Perry, Mechanisms of metal
fixation and leaching by cement based fixation processes. Waste
Management and Research, 3 (1985) 127-142.
[56] R.K. Salas, Disposal of liquid waste by chemical fixation/solidification

the Chemfix@process. Toxic and Hazardous Waste Disposal in: R.
Pojasek (Ed.), Vol 1, Processes For Stabilization/Solidification, Ann
Arbor Science Publishers, MI, 1979, pp 321-348.
[57] J.R. Conner, method of making wastes non-polluting and disposable.

United States Patent #3837872, September 24 1974.
[58] J.R. Conner, R.J. Polosky, Method of improving the quality of leachate
from sanitary landfills. United States Patent #3841102, October 15 1974.
[59] BRE Digest 330, Alkali-silica reaction in concrete, CRC publishing,

London, 1999.
@Seismicisolation
@Seismicisolation
832 Hooper et al.
TABLES
Table 1-Compositions of ISF slags

Oxide or ISF slags Blastfurnace Steel
metal slag slag
Wt.% present
As 0.05 0.02 0.05
5.5 8.25 7.77 8.25 7.66 13.5 3.8
Ba0
Ca0 3.3 13.66 14.80 13.66 15.66 40.9 40.3
CaO/SiO, 1.2 1.85
Cu 0.6 0.42 0.68 0.26 0.85
52.9 30.66 38.17 39.66 38.28 0.5 16.3
%O 0.8
MgO 5.5
Mn203
1.13 0.8 4
Na20 0.7
Pb 1.4 0.61 0.76 0.61 1.o2
S 3.55 2.92 3.55 2.84 0.5
SiO, 11.3 18.13 19.69 18.13 16.12 34 21.7
Tio, 0.8
9.8 9.48 9.01 7.62 6.93
Table 2-Particle Size Distribution of ISF Slag

% Mass passing through sieve
BS 882: 1992 [ l i ]
Sieve Overall Coarse Medium Fine ISF Slag [4]
Size Limits Sand Sand Sand
lmm
5.00 89 to 100 - 1O0 99
2.36 60to 100 60-100 65-100 80-100 99 96
1.18 30to 100 30-90 45-100 70-100 88 89
0.60 15 to 100 15-54 25-80 55-100 54 60
0.30 5 to70 5-40’ 5-48 5-70 17 35
Sand ISF Slag
Approximate specific gravity
- 2.59 3.76 3.68
Ref. [91 [81 [91
Approximate % water absorption 2.7 1.84
@Seismicisolation
@Seismicisolation
Table 3 - Specifications for Secondary Materials in UK Highway Construction
Appli- Embank Cap- Un- Bitu- Cement Cement Pave-
cation -ment ping bound men bound bound ment
and fill sub- bound sub- road- quality
Material base layer base base concrete
Crushed B A C A A A
A
Concrete
Bitu-
minous B B C A B C C
Planings
Demo-
lition B B C C B C C
Wastes
Blast-
furnace A B A A A A A
Slag
Steel
C C A A B C C
Slag
Burnt
Colliery A B A C B C C
Spoil
Unburnt
Colliery B C C C B C C
Spoil
Spent
Shale B B A C B C C
Oil
Pulver-
ised Fuel A A C C B A A
Ash
Furnace
Bottom B B C C B C C
Ash
China
Clay B B B B B B B
Waste
Slate
B B B C B B B
Waste
A Specific Provision
B General Provision
C Not permitted
@Seismicisolation
@Seismicisolation
834 Hooper et al.
Table &Assessment of Compliance to BS 147:1996and BS 3892-1:1997
Compl- Blast- Pfa ISFslag Lead
% by mass iance furnace smelting
Criteria slag slag waste
ALO, 13.5 26 5 to 8 2.7
Ca0 40.9 10 3 to 16 14.1
Fe203 0.5 4 37 to52 37.8
MgO 5.5 2 9.4
Pb - - 0.5 to 1.4 -
SiO, 34 48 11 to34 23.4
Zn
~~~
- - 7 t o 10 -
Compliance to BS 147:1996
Ca0 + M a 0 + SiO, >0.66 0.8 0.4 0.1 to0.5 47
(Ca0 + MgO)/ (SiO,) >1.O 1.3 0.3 0.3 to 0.5 1.o
(Ca0 + MgO)/ (SiO,) 4 . 4 1.2 0.21 0.3 to0.5 0.6
Compliance to BS 3892-1:1997
% Ca0 <10 40.9 10 3 to 16 14.1
Table 5-XRF Analysis of ISF slag

Oxide S1% S2%
Silicon dioxide SiO, 16.30 19.83
Iron (III)oxide Fe,O, 33.33 37.17
Calcium oxide Ca0 11.98 14.49
Lead(II)oxide PbO 7.39 1.46
Zinc oxide ZnO 11.37 14.23
Sulphur trioxide SO, 10.26 4.42
Table 6-Composition of test concrete mixtures

Test Cement Free ISF Slag c5 mm 10-5 mm 20-10 mm
Mixture (kg/m3) Water (kg/m3) Crushed Crushed Crushed
Reference (kg/m3) Rock Rock Rock
(kg/m3) (kg/m’) (kg/m3)
Control 270 200 O 995 332 663
50% S1 270 200 654 497.5 332 663
1OO%S1 270 200 1307 O 332 663
50% S2 270 200 1307 O 332 663
1OO%S2 270 200 654 497.5 332 663
@Seismicisolation
@Seismicisolation
Table 7-Compressive Strength of Test Mortars after 28 Days
Mortar Composition (kg/m3) Compressive Average Standard Time to
specimen strength strength deviation demould
reference OMpa) (ma) (days)
Control Cement = 510 40.5 40.2 0.3 1
<5 mm gravel = 1530 40.0
Water = 270 40.0
100% S1 Cement = 510 33.0 33.3 1.o 3
ISF slag = 2264 34.5
Water = 273 32.5
100% S2 Cement = 510 34.0 34.3 1.5 6
ISF slag = 2263 36.0
Water 278 33.0
50% S1 Cement = 510 42.5 40.7 1.6 1
<5 mm gravel = 765 40.0
ISF slag = 1132 39.5
Water i295
50% S2 Cement = 510 38.5 37.3 1.o 1
<5 mm gravel = 765 37.0
ISF slag = 1130 36.5
Water = 297
@Seismicisolation
@Seismicisolation
836 Hooper et al.
Flexible, Flexible Rigid
Composite and Rigid
Com pos¡te
Figure 1-Schematic Diagram of Typical Pavement Design
40
A
35
a
L
a
5 30
a
P
i 25
I-
20
O 2 4 6 €3 10
Time (days)
Figure 2 -Temperature Profiles of Setting Concrete
@Seismicisolation
@Seismicisolation
~~
Contmi (de- pH 13 buffer Saturated Saturated pfa with de- ggbs with de- sodium
ionised water) solution Ca0 solubon Ca0 in ionisedwater ionised water stlicale with
combinabon de-ionised
with pfa water
Figure 3-Leaching of Zinc into Test Solutions
_____ ___..
.
Control (de- pH 13 buffer Saturated Saturated pia with de- ggbs with de- sodium
ionised solubon Ca0 SoIubon Ca0 in ionised water ionised water silicate with
water) combination de-ionsed
with pia water
Experimental solution
Figure 4-Leaching of Lead into Test Solutions
@Seismicisolation
@Seismicisolation
SP 209-45
Effect of Reinforcing Bar

Contamination on Steel-Concrete
Bond During Concrete Construction
by L. H. Taber, A. Belarbi, and D. N. Richardson
SvnoDsis: During concrete construction, form oil, bond breaker, concrete splatter
and other types of contaminants often contaminate reinforcement. Current specifi-
cations and quality control measures require the removal and clean up of these
contaminants before the placement of concrete due to a concern of a reduction in
bonding capacity. This is costly, labor intensive, and may not be necessary.
Currently, there is limited research on the effect of reinforcing bar contami-
nation on the bond between the deformed steel reinforcing bar and concrete. Be-
cause of this lack of data, specifications are conservative and require the removal of
the contaminant. Inspectors often cite AC1 301-96, Standard SpeciJications for
Structural Concrete, which states, “When concrete is placed, all reinforcement shall
be free of materials deleterious to bond.” If it could be conclusively proven that this
level of care is unnecessary, the construction industry would benefit greatly.
To address the effects of contaminants on bond characteristics of deformed
steel reinforcing bars, a preliminary study was completed at the University of Mis-
souri-Rolla. The research program focused on three contaminants often seen dur-
ing construction: form oil, bond breaker and concrete splatter. Other variables in-
cluded size of reinforcing bar, strength of concrete and epoxy versus uncoated
reinforcing bar. This paper will provide the experimental program and test proce-
dures as well as the test results and observations. The results reveal that in the ma-
jority of situations tested, the ultimate bond stress was not significantly affected by
the three contaminants tested. In some cases, the bond breaker and form oil affect-
ed the smaller epoxy coated bars, while the effect of concrete splatter was insignif-
icant.
Kevwords: bond strength; construction specifications;

contaminants; reinforced concrete; reinforcing bars;
slippage
839
@Seismicisolation
@Seismicisolation
840 Taber et al.
AC1 member Lawrence H. Taber is a Structural Engineer with Black and
Veatch Corp. in Kansas City, Missouri and a former Graduate Research
Assistant in Civil Engineering at the Universis of Missouri-Rolla. He earned
both his BSCE and MSCE from the University of Missouri-Rolla.
Abdeldjelil Belarbi, FACI, Associate Professor of Civil Engineering at the

University of Missouri-Rolla. His research interests include constitutive
modeling of reinforced and prestressed concrete as well as use of advanced
composite materials and smart sensors in civil engineering infiastructure. He
is a member of several AC1 technical and educational committees and the
current Chair of AC1 Committee E801, Student Activities; and ACI-ASCE
subcommittee 445-5, Torsion of Structural Concrete.
AC1 member David N. Richardson is Associate Professor of Civil Engineering

at the University of Missouri-Rolla. His teaching and research activities focus
on properties, production and use of concrete and related materials. Current
research involves aggregate gradation optimization, specimen testing
variables and GGBFS. He is currently a director of the Missouri Chapter of
AC1 and a member of the certification committee.
INTRODUCTION
A good bond between steel deformed reinforcing bar and concrete in

concrete structures is crucial for structural and serviceability performance. If
this bond is inadequate, behavior and failure characteristics will be altered.
The bond mechanism allows the forces to be transferred between the concrete
and steel.
In concrete construction, many types of contaminants are present on the

site, such as form oil for coating the forms and bond breaker used in tilt-up
construction. The reinforcement could be contaminated during construction if
care is not taken. If contaminated, there is concern regarding the bond strength
and specifications are in place to guide the action to be taken.
When constructing reinforced concrete (RC) structures, inspectors have

the duty to enforce several specifications dealing with concrete construction
(1,2). One of the goals of the specifications is to maintain a clean and safe
construction environment. Often, reinforcing bar is subjected to various
construction contaminants, such as form oil or mud during concrete
construction, and the specifications require the reinforcing bar to be cleaned
prior to placing the concrete. The American Concrete Institute (ACI) provides
several of these specifications. AC1 301-96 (1) states, “When concrete is
@Seismicisolation
@Seismicisolation
placed, all reinforcement shall be free of materials deleterious to bond.” This
specification also addresses form oil specifically when it states, “Do not allow
formwork release agents to contact reinforcing steel or hardened concrete
against which fresh concrete is to be placed.” When dealing with oil and
concrete splatter or mortar, ACI’s Manual of Concrete Inspection ( 2 ) says,
“reinforcement should be clean, and oil or non-adherent mortar which has been
spilled on it should be cleaned off.” Again, the primary concern of these
specifications is assuring a good bond, but do the contaminants reduce the
bond strength enough to warrant these specifications? Without detailed
research into this issue, most construction specifications are conservative and
require the removal of these contaminants from the reinforcing bar. This is
time-consuming, costly, and may be unnecessary (3).
Previously, several studies have been performed in regard to the

variables that affect bond strength, including the effect of the amount of
concrete cover, casting position, slump and consolidation on epoxy-coated
reinforcing bar (4), and the effect of rust and scale (5,6). It was found that up
to a certain degree of rust, the bond was not significantly reduced and the
specifications were relaxed. These specifications now recognize that slightly
rusted steel reinforcing bar does not cause a significant reduction in the bond
stress and allows contractors to use this reinforcing bar without having to first
remove the rust. This saves the contractor considerable time and money
without significantly affecting the quality and strength of the RC structure.
Princiule of Bond Between Reinforcing Bar and Concrete
Bond is defined as a force transfer between two materials (7). Bond

behavior involves the bond stresses, transfer mechanisms and ultimately the
complete failure mode. Bond stress is defined as the shear stress at the steel-
concrete interface, which by transferring the load between the reinforcing bar
and surrounding concrete, modifies the steel stresses (8). The average bond
stress is typically represented as fiuavg
and its equation is:
where: f i v B= average bond stress, dfs= change of steel stress over unit length,
db = diameter of reinforcing bar, and lb = embedment length.
There are several mechanisms that transfer the load between the
concrete and steel. The three primary mechanisms are chemical adhesion,
mechanical interlock and frictional resistance. Each method contributes to the
@Seismicisolation
@Seismicisolation
842 Taber et al.
bar and conditions under which the concrete is placed. For deformed steel
reinforcing bar, the greatest contribution comes from the mechanical interlock,
with the frictional and chemical bonds both helping to a lesser extent. The
bearing of concrete on the steel ribs causes the mechanical interlock. As the
forces are transferred, the concrete is placed under a shearing stress; thus
eventually causing bond failure. With plain (smooth) reinforcing bar, the
chemical and frictional bonds would be the primary mechanisms, with the
mechanical interlock almost non-existent.
Bond Test Methods
Several test methods are commonly used to determine steel-concrete

bond strength. One method is the pullout test specified by RILEM (9). The
advantages of this test are the easy setup and simple specimens, yet a concern
is the additional confinement provided by the compression induced into the
specimen around the anchorage area. Furthermore, it is not representative of a
beam because reinforcing bars are in tension and concrete is in compression.
A flexural test avoids the additional confinement and more realistically
simulates the embedment in the tension zone in concrete beams, where both
reinforcing bar and concrete are in tension, in addition to the presence of
shearing forces and dowel action. Two flexural tests are common, one
specified by RILEM (10) and the other specified by ASTM (1 1). The RILEM
beam test is a simply supported beam with a gap in the middle, allowing for
easy calculations of the tension force and ultimately the bond stress. The
ASTM A944 Beam-End test is a rectangular specimen placed into flexure
through a cantilevering action. Both of these test methods are widely used in
testing laboratories.
The test method selected for this research program was the RILEM
Bond Test for Reinforcing Steel-Beam Test, Specification Number 7-II-28 D
(10). This test method was chosen over other methods for several reasons. A
flexural test was preferred over a direct pullout test because it more
realistically simulates the actual conditions in a beam or slab. The RILEM
beam test was chosen over the Beam-End method for its simplicity. In the
authors? opinion, both the RILEM test and Beam-End test provide similar
results and similar testing characteristics.
There is a need for research into the specific problem of the effect of
reinforcing bar contamination on the steel-concrete bond in reinforced concrete
structures. Structural and construction engineers are often conservative and
@Seismicisolation
@Seismicisolation
require the reinforcing bar to be cleaned before placing the concrete. The
engineers do not know the extent of the effect of contaminants on bond stress.
This study will provide a basis for more research into this topic, with the goal
of eventually having sufficient data to confirm or deem unnecessary the need
for the current requirements. If it can be shown that some of the most common
construction contaminants do not have a detrimental effect on the steel-
concrete bond, the specifications could then be revised to possibly allow a
specific amount of reinforcing bar contamination, resulting in considerable
savings to the concrete construction industry.
Several factors influence the bond strength of the steel-concrete

composite. These factors are related to both the concrete (compressive
strength, amount of cover, etc.) and reinforcing bar (size, surface coating,
embedment length, rib face angle, etc.). Additionally, the type of contaminant
(form oil, mud, bond breaker, concrete splatter, rust, etc.) may produce
different bond characteristics. For this preliminary study, four variables were
included. They were the surface coating condition (epoxy coated or uncoated),
size of reinforcing bar, concrete compressive strength, and type of
contaminant. Of the many types of Contaminants, the study focused on three:
water-based form-release agent, a water-based bond breaker, and concrete
splatter. These contaminants are commonly seen during construction
operations. The effects of rust have been already studied and the specifications
modified accordingly.
The test program consisted of 32 tests, with three replicate specimens

per test for a total of 96 specimens. The surface coating was varied between
epoxy coated and uncoated. The two sizes of reinforcing bar chosen were #4
(13 mm) and #7 (22 mm). Nominal concrete compressive strengths at 28 days
of 4,000 psi (28 MPa) and 7,000 psi (48 MPa) were selected. Instead of the
7,000-psi (48 MPa), a higher strength would have provided a clearer
understanding of the effect of concrete strength, yet to attain this and maintain
workability, admixtures most likely would have been used. Use of admixtures
was not desirable in order to avoid any possible unforeseen chemical
interaction between the contaminants and admixtures. The contaminants were
applied along the entire embedment length, simulating the worst-case situation.
The specimens were batched in groups of 12 beams, with the type of
contaminant being the only variable per batch. The test program is
summarized in Table 1.
The specimen identification code consisted of four characters. The first

letter is the surface coating condition, either epoxy coated (C) or uncoated (U).
The second number in bold is the reinforcing bar size, either #4 (13 mm) or #7
@Seismicisolation
@Seismicisolation
844 Taber et al.
(22 mm). The third number is the 28-day compressive strength, either a 4 for
4,000 psi (28 MPa) or a 7 for 7,000 psi (48 MPa). The fourth letter is the type
of contaminant applied to the reinforcing bar, C for control (clean), S for
concrete splatter, and B for bond breaker and F for form oil. The number in
parenthesis is the number of replicates tested.
Test Specimen
The test specimens used in this study were fabricated according to

RILEM Standards (10). Figure 1 illustrates the configuration of the test
specimen. The specimen was a flexural beam separated in the middle with the
reinforcing bar crossing the gap. The two rectangular prisms were joined
together with a steel hinge in the compression zone and the reinforcing bar to
be tested was located in the tension zone. The reinforcing bar was embedded
into the center of each prism with a length of 10 times its diameter (lodb), as
specified by RILEM. PVC tubing was provided outside the 104 length to
prevent the concrete from bonding to the steel along the non-embedded portion
of the reinforcing bar. The dimensions of the beam, casting position and
additional reinforcement were as specified by RILEM. The beams were cured
per RILEM specification for 28 days at approximately 60% relative humidity
and 20 f 2OC.
Materials
The concrete mixture proportions used in this study are given in Table
2. The mixture used Type I portland cement, 9/16 in. (14.3 mm) nominal
maximum size, clean, crushed limestone from the Gasconade Formation in
Missouri, and fine aggregate sand from the Little Piney River in central
Missouri. The water was tap water from Rolla Municipal Utilities in Rolla,
Missouri.
The reinforcing steel used was ASTM A615-96A Grade 60 (12). The
epoxy coating was applied per ASTM A775-96 (13). The tested average yield
stress was determined at 443 MPa (64.2 ksi) and the average tensile strength
was 713 MPa (103.5 ksi). The uncoated reinforcing bar was free from any rust
or other contaminants and the epoxy coated reinforcing bar had no chips or
scratches in the epoxy.
Three contaminants were used in this study: namely bond breaker, form
oil and concrete splatter. The bond breaker was a water-based lifting agent
that is used in tilt-up construction. The form oil was a surface consolidating
water-based form release agent that is used in precast and poured concrete
@Seismicisolation
@Seismicisolation
applications. The concrete splatter had the same mixture proportions as the
mortar portion of the concrete in the specimen, simulating a multiple lift pour
where this contamination often occurs.
The method of applying these contaminants was an attempt to simulate

field conditions. The form oil and bond breaker were applied 30 minutes prior
to placing the concrete using a brush to get complete coverage. The concrete
splatter was applied 18 hours before placement of concrete. From a series of
trial batches with varying setting times, 18 hours was chosen as the worst-case
situation because at this point the concrete has hardened, yet not achieved a
good bond to the steel. The splatter was dropped (or splattered) onto the bar
while slowly rotating the bar to ensure complete coverage. Figs. 2 and 3 show
the form oil and concrete splatter contaminated reinforcing bar, respectively,
prior to concrete placement.
Test SetuD and Procedure
The beam tests were conducted using a 1IO-kip (500 kN) MTS servo-
valve controlled machine. All loads were applied using ram stroke control to
better observe the failure and post-failure behavior. The free-end slip was
measured using Linear Variable Differential Transformers (LVDT’s) placed on
each end of the reinforcing bar. The beams were transported from the curing
location to the testing machine using an overhead crane and lifting system
designed to avoid relative movement between the two halves of the beams.
Placing a wooden block in the hinge location and bolting two boards to either
side of the beam restrained the two halves from moving relative to each other.
The beams were tested under four-point loading in accordance with the
RILEM specification with the load being applied symmetrically to both halves
of the beams (Fig. 4). The LVDT’s were attached securely to the beam,
assuring that only relative slip between the reinforcing bar and concrete was
measured. As one side of the beam failed, a restraint device was placed on that
end of the reinforcing bar to prevent further slip and to allow the other half to
reach failure.
RESULTS
Concrete Cornmessive Strength
The compressive strength of the concrete was determined according to

ASTM specification C39. However, four cylinders instead of three were tested
@Seismicisolation
@Seismicisolation
846 Taber et al.
per batch of concrete. The design strengths selected for this study were 4,000
psi (28 MPa) and 7,000 psi (48 MPa); however, the actual strengths were 4,780
psi (33 MPa) and 6,710 psi (46 MPa), respectively.
The actual strength for the 7,000-psi (48 MPa) mixture is slightly low
and the mixture proportions could have been adjusted or admixtures added to
get higher strengths. Because of possible unforeseen chemical interactions
with the contaminants, admixtures were not used. Workability requirements
prevented the wlc from being lowered. The workability was a concern with
the RILEM test because of the additional reinforcement required in the beam.
Bond Stress Calculations
The bond stress on the bonded portion of the bars was determined in
relation to the applied load and geometry of the test setup. The relationship
between the applied load and the bond stress can be shown to be:
Pa
T, =--
T --
2jd
zdlb rdl,
Where P = Applied Load, a = distance between support and load, jd = lever

arm,d = reinforcing bar diameter, and ìb = embedment length.
Substituting the following values; lb = Iod, u = 450 mm (17.7 in) and jd = 160
mm (6.3 in) then
P
T, =-
7.11rdz
where P is in kN and d is in mm
For a 13 mm (#4) bar the average bond stress is calculated:
For a 22 mm (#7) bar the average bond stress is calculated:
@Seismicisolation
@Seismicisolation
Bond Test Results
The results of the RILEM beam tests are summarized in Fig. 6 . The
ultimate bond stresses of the various tests did vary somewhat. The statistical
significance of the amount of variance is presented later.
From Fig. 6 , it is evident that #4 (13 mm) bars show a higher bond
strength compared to that of #7 (22 mm) bars. All #7 (22 mm) bars exhibited
bond-splitting failure, where the #4 (13 mm) bars exhibited a combination of
pullout and bond splitting failures. Note that both #4 (13 mm) and #7 (22 mm)
bars were tested under the sanie embedment length of 1Odb. Comparable
results for both #4 (13 mm) and #7 (22 mm) cases could have been achieved if
bond-splitting failure were avoided. Yet, the specimens, as tested, more
realistically simulate the case of a beam or slab in terms of moderate concrete
cover and, presence of shearing forces and dowel actions. Therefore, the bond
spitting failure noticed in # 7 (22 mm) cases are acceptable.
Bond stress-slip curves were generated for each beam. A typical set of
bond stress-slip curves for a beam set is presented in Fig. 7. The general
shape of the curves depended on the type of failure, however, the differences
were minimal. Two failure modes were observed: bond splitting and bond
pullout, with the majority of failures being bond splitting. Bond splitting is
well documented for steel deformed reinforcing bar (7). It is characterized by
a small amount of slip producing the bond splitting, followed by further slip
and finally a complete loss of bond. The bond splitting is caused when a
primary crack propagates to the surface and several secondary cracks form
near the lugs of the deformed reinforcing bar. These cracks are caused by the
wedging action of the reinforcing bar being pulled through the concrete.
Typically, a larger bar with a small amount of concrete cover will fail in bond
splitting. Bond pullout only occurred in a limited number of #4 (13 mm)
epoxy coated bars. This failure is characterized by an increasing amount of
slip until the bar pulls out of the concrete and the bond strength lessens
considerably. While the general shape of the stress-slip curves varied little, the
bond stress magnitude did vary depending on the variables being tested, most
notably the reinforcing bar size. In the set of curves shown in Fig. 7, two
specimens failed in bond splitting, while the third failed in bond pullout. A
side view of a beam that failed in bond splitting is given in Fig. 8.
Statistical Analvsis of Data
An inspection of the bond stress results shows the following. Bond

stresses were higher for: 1) the #4 (13 mm) bars (all 32 tests), 2) higher
compressive strength mixture (3 1 tests), 3) control compared to bond breaker
@Seismicisolation
@Seismicisolation
848 Taber et al.
(8 out of 8 tests), 4) control compared to form release agent (7 out of 8 tests),
5) splatter over control (6 out of 8 tests), and 6) plain reinforcing bar over
epoxy coated (22 out of 32 tests). The statistical significance of these trends
was investigated using JMP, which is produced by the SAS Institute (14). The
statistical tests used for comparing the data included Dunnett’s t-Test and a
Least Squares Model.
Prior to using Dunnett’s t-Test, the data was first tested for statistical
outliers and normality. Two tests were used to check outliers, the 3*0 test,
where O is the standard deviation of the test group, and the Dixon Q test. Both
methods indicated no significant outliers. Normality was checked using a Q-Q
Plot at an a (confidence level) = 0.05. Normality indicates if the data has a
normal distribution. The data was somewhat non-linearly distributed along the
Q-Q Plot, yet because the number of tests was large enough, the Central Limit
Theorem applies and the data distribution is considered approximately normal.
From the distribution of the data, the ultimate bond stress mean of all the #4
(13 mm) reinforcing bar was 2,116 psi (14.6 MPa), while the #7 (22 mm)
reinforcing bar data had a mean of 981 psi (6.8 MPa). Because the mean
ultimate bond stresses were significantly different for the two sizes of
reinforcement, the data distribution was bimodal, or had two separate humps.
The data distribution was also fairly symmetric.
Dunnett’s t-Test compares a set of means against a control group mean.

The test reports the absolute value of the difference between the control group
and group of interest minus the Least Significant Difference (UD). The LSD
is the difference that is the limit for the difference to be considered significant.
If a value is positive, its mean is more than the LSD apart from the control
group mean so that group’s mean is significantly different than the control
group mean. The Dunnett’s t-Test is used to compare the ultimate bond stress
means instead of a simple percent difference because a statistical analysis will
account for the effect of the number of tests and the standard deviation of the
data. The confidence level chosen for this test was a = 0.05. After the initial
testing phase, a level of a = O. 10 was also used to show a more inclusive set of
results. Table 3 shows the results of the Dunnett’s t-Test. At a level of 0.05,
three tests out of 24 were found to be significantly different from the respective
controls. Two of these were the #4 (13 mm) epoxy coated reinforcing bar,
4,000-psi (28 MPa) concrete specimens with form oil and bond breaker as the
contaminant. The third was the form oil contaminated #4 (13 mm) 7,000-psi
(48 MPa) test specimens.
With the confidence interval of a = 0.10, three tests (in addition to the
previous three from the a = 0.05 test) were identified as significantly different.
One was the bond breaker contaminated #4 (13 mm) bar 4,000-psi (28 MPa)
test specimen. The other two were the #4 (13 mm) 7,000-psi (48 MPa) form
oil and bond breaker test specimens. While these three tests are considered
@Seismicisolation
@Seismicisolation
significantly different, this must be tempered with the fact that this confidence
level is higher than what is usually used in analysis.
A comparison of just the means of the control specimens by Dunnett’s

t-Test showed that the following were significantly different at the 0.05 level:
bar size (uncoated) and bar size (coated). A Least Squares Model analysis
verified the significance.
In Table 3, the significantly different treatments are the #4 (13 mm)

bar, epoxy coated, 4,000 psi (27.6 MPa) and 7,000 psi (48 MPa), and #4
uncoated at 7,000 psi (48 MPa) tests contaminated with form oil and bond
breaker. This trend shows a possible interaction of the form oil and bond
breaker contaminants with the smaller bar size, weaker concrete and epoxy bar
coating. A Least Squares Model was used to investigate the significance of
those interactions. With a confidence interval of 0.05, the interaction of the
form oil contaminant with bar size, concrete strength level and coating
condition was significant.
EXPERIMENTAL AND ANALYTICAL OBSERVATIONS
Several observations were made during this test program. Some of

them are as follows:
The majority of beams failed in bond splitting; the remaining specimens
failed in bond pullout. The bond pullout occurred primarily in #4 epoxy
coated bars.
The #7 reinforcing bar gives a lower bond stress than the #4 bars for the
same embedment length.
On the macroscopic level, Fig. 6 shows the ultimate bond stresses are of
the same magnitude within a given bar size.
The wedging action caused by the reinforcing bar lugs is thought to have
produced the compacted powder on the lugs, previously described by
several investigators, notably Park and Paulay (7). As the bond fails, the
concrete is compacted into a powder by the lugs, which collects on the lug
faces. Fig. 9 shows the compacted powder on the reinforcing bar lugs of
an epoxy coated #7 (22 mm) bar.
From Table 3, of the three cases that differences in means were considered
significant at the 0.05 level, it appears as the reinforcing bar size decreases
and the surface condition changes to epoxy coated and the concrete
strength is reduced, the ultimate bond strength is more likely to be affected
by form oil and bond breaker contaminants. This trend was seen as the
most likely to affect the ultimate bond strength since a lower concrete
strength and epoxy coating weaken the bond stress while acting
independently of each other. The concrete splatter tested does not seem to
affect the ultimate bond strength at either confidence level.
@Seismicisolation
@Seismicisolation
850 Taber et al.
CONCLUSIONS
The following conclusions are drawn from the experimental and

statistical studies of this investigation:
In 13 out of 16 situations tested at 01 = 0.05, a comparison of the means
analysis showed that the ultimate bond stress was not significantly affected
by the form oil or bond breaker contaminants. None of the situations were
significantly affected by the concrete splatter.
Two types of failure mechanisms were observed in this study, bond pullout
and bond splitting. Most of the failures were by bond splitting, which is
attributed to a bearing of the concrete against the lugs of the reinforcing
bar, resulting in the splitting of the concrete. The amount of slip associated
with this failure is considerably less than observed in a bond pullout
failure. The pullout failure occurred with a few of the #4 (13mm) epoxy
coated reinforcing bars.
As the bar size and concrete strength decrease and the surface is epoxy
coated, the specimens appear to be more susceptible to a loss of bond
strength caused by bond breaker and form oil contaminants.
FüTUFLE RESEARCH
One purpose of this research was to provide a credible basis for future
research to be performed with the goal of eventually gathering enough data for
the development of appropriate specifications that consider altering the current
requirements of cleaning contaminated steel reinforcing bar prior to placement
of concrete. Studies that focus on other types of contaminants, setting times,
amount of contaminant coverage, reliability studies or factors of safety should
be undertaken.
ACKNOWLEDGMENTS
For their help with funding, the authors wish to thank the University
Transportation Center at UMR, McCarthy Construction, Associated General
Contractors, Site Advancement Foundation, Industry Advancement Fund,
Southern Illinois Construction Advancement Program, St. Louis Construction
Training and Advancement Foundation, and the Erectors and Riggers
Association of St. Louis.
@Seismicisolation
@Seismicisolation
REFERENCES
1. AC1 30 1-96, Standard Specifications for Structural Concrete, American

Concrete Institute, Fannington Hills, Michigan, 1996.
2. AC1 3 11,ACI Manual of Concrete Inspection, SP-2, ACI, 1992.
3. Suprenant, B. and Malisch, W. “How Clean Must Rebar Be?” Concrete

Construction. June 1998, pp. 517-523.
4. Darwin, D., Hadje-Ghaffari, H., Choi, O. and McCabe, S., “Bond of

Epoxy-Coated Reinforcement: Cover, Casting Position, Slump, and
Consolidation,” ACI Structural Journal. Vol. 91, No. 1. January-February
1994, pp. 59-68.
5. Kemp, E., Brezny, F. and Unterspan, J., “Effect of Rust and Scale on the
Bond Characteristics of Deformed Reinforcing Bars,” Journal of the
American Concrete Institute, ACI. September 1968, pp. 743-756.
6. Johnston, B. and Cox, K., “The Bond Strength of Rusted Deformed Bars,”
Journal of the American Concrete Institute, Vol. 37. September 1940, pp.
57-72.
7. MacGregor, J., Reinforced Concrete-Mechanics and Design, 3rd Ed.,

Prentice Hall, 1997.
8. Park, R. and Paulay, T., “Chapter 9: Bond and Anchorage,” Reinforced

Concrete Structures, John Wiley and Sons, New York, 1975.
9. RILEM 7-II-128, “RC6: Bond Test for Reinforcing Steel. 1. Pull-Out

Test.” RILEM technical recommendations for the testing and use of
construction materials, E & FN Spon, U.K. 1994, pp.102-105.
10. RILEM 7-II-28 D, “RC5: Bond Test for Reinforcing Steel. 1. Beam Test.”
RILEM technical recommendations for the testing and use of construction
materials, E & FN Spon, U.K. 1994, pp. 96-101.
11. ASTM A944-95, “Standard Test Method for Comparing Bond Strength of
Steel Reinforcing Bars to Concrete Using Beam-End Specimens,” 1997
Annual Book of ASTM Standards, Vol. 1.04, American Society for Testing
and Materials, West Conshohocken, PA. 1997, pp. 514-517.
12. ASTM A615lA 615M-96a, “Standard Specification for Deformed and

Plain Billet-Steel Bars for Concrete Reinforcement,” 1997 Annual Book of
@Seismicisolation
@Seismicisolation
852 Taber et al.
ASTM Standards, Vol. 1.04, American Society for Testing and Materials,
West Conshohocken, PA. 1997, pp. 301-305.
13. ASTM 775-96, “Standard Specification for Epoxy-Coated Steel

Reinforcing Bar,” 1997 Annual Book of ASTM Standards, Vol. 1.04,
American Society for Testing and Materiais, West Conshohocken, PA.
1997, pp. 301-305.
14. JMP IN Statistical Discovery Software, Computer Program. SAS Institute,

Inc. Version 4,2001.
@Seismicisolation
@Seismicisolation
Table 1. Test Program
i
Specimen IRebar surface Rebar size
condition
#4 (13 mm)
I Concrete splatter
i Control
1-1
7 ksi (48 MPa)
Form-oil
Bond-breaker
Uncoated ~~
I Concrete splatter
1 -i1
Control
Form-oil
4 ksi (28 MPa)
Bond-breaker
#7 (22 mm)
I Concrete splatter I
u 7 7 c (3)
U77F (3)
U77B (3)
u 7 7 s (3)
c44c (3)
C44F (3)
C44B (3)
c44s (3) #4 (13 mm)
c 4 7 c (3)
C47F (3)
C47B (3)
Coated ~
I Concrete splatter
1-11
Control
Form-oil
4 ksi (28 m a ) Bond-breaker
#7 (22 mm:
I Concretesplatter I
Control
@Seismicisolation
@Seismicisolation
854 Taber et al.
Table 2. Concrete Mixture Proportions per Cubic Meter
@Seismicisolation
@Seismicisolation
Table 3. Dunnett’s t-Test Results
14
r
?
!
@Seismicisolation
@Seismicisolation
856 Taber et al.
f Figure 1. Test Specimen (dimensions in mm)
f Figure 2. Form-Oil Contaminated Rebar
@Seismicisolation
@Seismicisolation
Concrete
Splatter Rebar
f Figure 3. Concrete Splatter Contaminated Rebar
f Figure 4. Test Setup
@Seismicisolation
@Seismicisolation
858 Taber et al.
?Figure 5. Restraint Device (a) 13 mm (#4) rebar (b) 22 mm (#7) rebar
@Seismicisolation
@Seismicisolation
1
2,250
2,000
a
n
,750
5
rn
E ,500
ti ,250
8 "
,000
8
m
6 %
750
4
-E5
500
250 2
O O
U44 C44 u47 c47 u74 c74 u77 C77
Beam ID
ImControl Form-Oil O Bond Breaker NConcrete Splatter I
f Figure 6. Bond Strength Test Results
@Seismicisolation
@Seismicisolation
860 Taber et al.
Bond Clip (mm)
O 1 2 3 4 5 6
2,500
2,250 16
2,000 14
500 4
"
O 0.05 0.1 0.15 0.2
Bond Clip (in)
f Figure 7.Typical Bond Stress-SlipCurve
@Seismicisolation
@Seismicisolation
'f
Figure 8. Side View of Failed Specimen by Bond Splitting
f Figure 9. Compacted Powder on Rebar Lugs
@Seismicisolation
@Seismicisolation
SP 209-46
Seismic Behavior of Eccentric R/C

Beam-Column-Slab Connections
Under Sequential loading in Two
Principal Directions
by B. Burak and J. K. Wight
Svnopsis: The performance of beam-to-column connections subjected to reversed

cyclic loading depends on several variables. In this investigation, the effects of ec-
centric spandrel beams, wide normal beams, column section aspect ratio, loading
in two principal directions, and slab participation on the inelastic behavior of
beam-to-column connections are studied. The experimental program features three
3/4-scale exterior reinforced concrete beam-column-slab subassemblies. The ma-
jor design variables for the specimens are the eccentricity of the spandrel beam
with respect to column, beam and column section aspect ratios, and joint shear
stress level. Results of the experimental program, including load-displacement re-
sponse, beam plastic hinge rotation, joint distortion, and bond deterioration are
presented in this paper.
Keywords: beam-to-column connections; eccentricity;

joint shear distortion; loading in t w o principal
d irection s; se ismic be havior
863
@Seismicisolation
@Seismicisolation
864 Burak and Wight
Burcu Burak, AC1 member, is a PhD. Candidate at the University of Michigan,
Ann Arbor. She is an associate member of Committee 352. Her research
interests include the seismic behavior and performance-based design of
James K. Wight, FACI, is a Professor of Civil Engineering at the University of

Michigan, Ann Arbor. He is the Chairman of AC1 Committee 3 18 and member
of Committees 352 and 445. His research interests include earthquake resistant
design and rehabilitation of reinforced concrete and composite structures.
INTRODUCTION
Seismic behavior of beam-to-column connections significantly

influences the earthquake resistance of reinforced concrete moment-resisting
frame structures. The inelastic behavior of beam-to-column joints has been
studied since 1960’s; however, the effect of some variables on the performance
of connections needs further study. This research program focuses on the effect
of eccentric spandrel beams, connections with normal beams wider than
columns, column section aspect ratio, loading in two principal directions, level
of joint shear stress, and slab participation.
The experimental program involves testing of three 3/4-scale exterior

reinforced concrete beam-column-slab connections under reversed cyclic
loading. Lateral load was applied in two principal directions, one plane at a
time, to determine the effect of prior loading on performance of the connection
and to examine the effects of biaxial loading. The first specimen was designed
as a control specimen following the recommendations of ACI-ASCE Committee
352 (1) and AC1 318-99 (2). In the other specimens, the eccentricity of the
spandrel beam with respect to the column, beam and column section aspect
ratios, and the level of joint shear stress were modified to examine their effect
on joint behavior.
The eccentricity between the centerline of the spandrel beam and the
columii centroid was selected as a design variable because some buildings with
eccentric spandrel beams exhibited extensive damage under earthquake loading
(3) and previous tests (4,5) indicated early deterioration of joint shear strength
for these types of structures. Tests by Raffaelle (4) demonstrated that the joint
shear strength reduces due to the torsion developed in eccentric connections
under lateral loading parallel to the exterior edge of the subassembly. Therefore,
Raffaelle and Wight proposed a more conservative effective joint width
@Seismicisolation
@Seismicisolation
definition than that of the current codes. This expression fits the data for tests of
two-dimensional joint specimens. However, if the three-dimensional joint
subassembly is properly modeled, this definition could be conservative.
Therefore, the following equation for the effective joint width, b,, proposed by
the second author to ACI-ASCE Committee 352, was used in the design of the
specimens:
where, bb = beam width

h, = thickness of the column in the direction of load being considered.
The major design variable for the normal beam was taken to be its width,
which is larger than the width of the supporting column for the last test
specimen. In a wide-beam structure, some of the beam longitudinal
reinforcement is anchored in the column core and the remainder is anchored in
the spandrel beams. If the spandrel beam is unable to transmit the applied torque
from the wide-beam to the column, the full moment strength of the wide-beam
cannot be developed. Prior tests with “unloaded” spandrel beams have
demonstrated satisfactory behavior for such wide-beam connections. This test
will investigate how prior loading in the spandrel beam direction will affect the
anchorage conditions for the portion of the wide-beam reinforcement that must
be anchored in the spandrel beam. The remaining.torsiona1strength and stiffness
of the spandrel beam will control whether the full moment capacity of the
normal beam can be developed at a reasonable level of story drift.
The main design variable for the column is the section aspect ratio
(width vs. depth). The results from the tests by Raffaelle (4) indicated that the
joint shear strength is influenced by the column section aspect ratio, which
controls the effective joint width that can be considered to resist shear forces.
The column section aspect ratio also affects the anchorage for both the spandrel
beam bars that pass through the joint and the normal beam bars that terminate in
the column.
Slab participation was investigated by comparing test results for these

three-dimensional specimens to results for previously tested two-dimensional
specimens, without slab and transverse beam. The experimental data include
information on joint shear distortion, lateral load-displacement response, beam
plastic hinge rotation, reinforcement strains inside and adjacent to the
connection, bond deterioration, and effective slab width for bending in both the
spandrel and normal beam directions.
@Seismicisolation
@Seismicisolation
866 Burak and Wight
Description of Specimens
The experimental program for this investigation features the testing of

three Y4-scale exterior reinforced concrete beam-column-slab subassemblies.
The specimens consist of two spandrel beams, a normal beam and floor slab.
The spandrel beams, which are narrower than the column, are flush with the
exterior face of the column. The spandrel beam centerline is eccentric to the
centroidal axis of the column, whereas the normal beam makes a concentric
connection with the column. Sufficient slab and spandrel beam dimensions are
included on each side of the normal beam to evaluate slab participation and
spandrel beam contribution to the seismic behavior of the connections. The
“strong column-weak beam” philosophy (M, > 1.O) is used in the design to have
beam plastic hinging at large displacements, rather than joint failure or column
plastic hinging. Specimen dimensions and the values for design variables are
given in Table 1.
Specimen 1 has a square column and the normal and spandrel beams
have typical dimensions. The slab is reinforced with the minimum shrinkage
and temperature reinforcement. The design of the remaining specimens was
completed after the testing of the first one and some parameters were modified
to examine their effect on connection performance.
In Specimen 2, the spandrel beam eccentricity is almost doubled

(Table 1). Moreover, it has a rectangular column with a section aspect ratio
of 1.5. The depths of the spandrel and normal beams are also increased.
Specimen 3 represents a wide-beam structure. Except for the normal

beam, all other members have the same properties with the second specimen.
The horizontal shear introduced into the joint region by the normal beam is
E
increased to a value near the code (2) limit of 15 Aj, psi (1.25 E
A,, MPa).
Test Setup and Loading Sequence
A schematic drawing of the test setup is given in Fig. 1. In this figure,

the solid lines demonstrate loading in the normal beam direction, while the total
figure, including the broken lines, illustrate loading in the spandrel direction.
The subassemblies are supported in the testing apparatus in a statically
indeterminate fashion for loading in the spandrel direction, and in a determinate
@Seismicisolation
@Seismicisolation
fashion for loading in the normal beam direction. Columns and beams are pin
supported at their mid-heights and mid-spans, where inflection points are likely
to occur during seismic loading.
All specimens are subjected to quasi-static cyclic loading. An axial load

of approximately five percent of the axial capacity of the column was applied.
The specimens were subjected to lateral loading in two principal directions, one
plane at a time. The specimens were loaded initially in the spandrel direction
(Fig. 2-a), then they were rotated 90 degrees and loaded in the normal beam
direction (Fig. 2-b). Loading in the second direction enables the determination
of the effect of prior loading on performance of the connection. The top of the
column was displaced horizontally, following a predefined displacement
history, to simulate the effects of inelastic loading during an earthquake. In each
direction, twenty cycles of lateral displacement were applied to each specimen,
starting with 0.5% story drift, and reaching a maximum story drift of 5%. Each
cycle to a new drift level is performed twice to evaluate the loss of strength and
stiffness of the specimens during the repeated cycles. Occasional small
amplitude displacement cycles are interspersed into the displacement history to
explore the stiffness loss in the specimens at low drift levels. Loading in the
spandrel beam direction gives a direct correlation between these three-
dimensional specimens and the prior testing of planar specimens with no slab
and normal beam. This correlation was used to investigate slab participation.
Instrumentation
Each specimen was instrumented with load cells, potentiometers

(displacement transducers), clinometers, and strain gages in order to capture
relevant data during testing. Load cells and displacement transducers are used in
the actuator to record column shear force and displacement, and at the vertical
beam end supports to monitor beam shear forces. Linear potentiometers and
clinometers were placed on the exterior and interior faces of the specimens in
the connection region to observe joint shear distortion. Potentiometers were also
located at 0.75d (d = effective depth of the beam) away from the column face to
measure total beam plastic hinge rotation and the concentrated rotation at the
beam-column interface. One potentiometer was placed at the end of the bottom
column to monitor movements in the test setup.
Electrical resistance strain gages were attached to the reinforcing bars at

a number of locations prior to casting the specimens. These gages were located
in and adjacent to the connection region to record the strains in the longitudinal
and transverse reinforcing bars of columns, spandrel beams, and normal beam.
Longitudinal column bars were instrumented with strain gages at top, bottom,
and mid-height of the joint region. Strain gages were placed on rectangular and
@Seismicisolation
@Seismicisolation
868 Burak and Wight
diamond shape column transverse reinforcement to record strains at interior and
exterior faces of the connection for both loading directions. Both spandrel and
normal beam longitudinal bars were instrumented at beam-column interfaces
and dl2 away from the column face to observe the spread of yielding into the
plastic hinging region. The penetration of yielding inside the slab is examined
by strain gages on beam flange bars. Strain gages were also attached to the
beam transverse reinforcement inside the beam plastic hinge region.
Two experiments have been completed and a third specimen will soon
be tested.
Lateral load versus story drift responses of Specimen 1 for loading in

both directions are given in Fig. 3. In this figure, slight pinching can be noticed
in the hysteresis loops for the cycles below 3.0% drift because of diagonal
cracking in the joint region. At larger displacement cycles, during which beam
flexural deformations dominated the specimen response, hysteresis loops
became wider.
The observed cracking patterns at the exterior gnd interior faces of

Specimen 1 are given in Figs. 4 (a) and (b), respectively. For loading in the
spandrel direction, first diagonal cracking occurred in the joint region at
approximately 1% story drift. Flexural cracks in the spandrel beams were
observed at 0.5% drift. At 1% and 1.5% drifts respectively, spandrel beam
bottom and top bars started to yield at the column face. Yielding spread over the
plastic hinging region and into the slab. Strain gage readings at d 2 away from
the column face and at the spandrel beam slab bars exceeded the yield strain by
2% drift.
For loading in the normal beam direction, additional diagonal joint

cracks formed after 1% story drift. Flexural cracking of the normal beam was
observed at 0.5% drift. At the column face, yielding started at 1.5% and 2%
drifts for bottom and top bars of the normal beam, respectively. Reinforcement
strains throughout the plastic region exceeded yielding by 2.5%. However, the
spread of yielding into normal beam slab bars was not as uniform. Some slab
bars yielded at high drift levels, and some did not yield. It is believed that the
deep cracks that occurred at the top of the slab parallel to the normal beam after
the loading in the spandrel direction reduced the spread of yielding for loading
in the normal beam direction. Yielding of the column bars was not observed in
either loading direction.
@Seismicisolation
@Seismicisolation
The specimen exhibited good behavior, even at high drift levels. The
number and size of shear cracks were low, and the spalling of concrete in or
near the joint region was insignificant. Furthermore, torsional crack formation in
the spandrel beam and connection was less than expected. This minor observed
damage correlates well with the maximum measured joint shear deformation of
0.5% at 4% story drift for loading in the spandrel direction.
The moment versus rotation diagrams of Specimen 1 are given in Fig. 5.

The data obtained from the potentiometers demonstrated that for loading in the
spandrel direction, approximately half of the total beam rotation over the plastic
hinging region was caused by flexural deformation of the beam. The remaining
rotation was due to concentrated rotation at the joint face caused by shear
deformations within the joint, rebar slip, and flexural cracks formed at beam-to-
column interface. For loading in the normal beam direction, rotation due to
flexure was one-third of the total rotation. The large beam rotations and the
wide loops shown in these graphs points out that beam rotations contributed
significantly to energy dissipation, especially at high drift levels.
Prior tests (4,5) have indicated that the eccentricity of the spandrel beam
leads to unsymmetrical damage in the joint, early deterioration of the joint shear
strength and anchorage conditions for the spandrel beam reinforcement, and
excessive pinching of the load versus displacement hysteresis loops. In those
tests, pinching was mainly due to the loss of anchorage of.the spandrel beam
bars through the joint as a consequence of large flexural cracks at the column
face and diagonal cracking in the joint. Specimen 1 demonstrated a ductile
behavior without any major pinching of the hysteresis loops for loading in both
directions. In addition, the strain gage data indicated that there was no
significant rebar slip in any member. This improved performance is believed to
be due to the presence of the floor slab because none of the previous tests
included a normal beam and slab to simulate the complete structural
subassembly at an exterior connection.
Lateral load versus story drift responses for Specimen 2 are shown in
Fig. 6 . For loading in the spandrel direction, there is more pinching than for
Specimen 1. For loading in the normal beam direction, pinching was not
observed because very few additional diagonal shear cracks have formed in the
joint.
The observed cracking patterns at the exterior and interior faces of

Specimen 2 are given in Figs. 7-a and 7-b, respectively. The cracking pattern
resembles the behavior of a frame building having eccentric connections, which
was damaged in 1995 Hyogoken-Nambu Earthquake (3). In this specimen, the
shear cracks were wider and spalling of the cover concrete at the exterior face of
the joint region was observed. From Fig. 7-a, it can also be seen that in
Specimen 2 the diagonal shear cracks originated in the core region rather than
@Seismicisolation
@Seismicisolation
870 Burak and Wight
the beam-column interface, which was the case for Specimen 1. For loading in
the spandrel direction, first joint diagonal shear cracks and spandrel beam
flexural cracks formed at approximately 0.5% story drift. At 1.5% and 2%
drifts, respectively, spandrel beam bottom and top bars yielded at the column
face. The yielding spread over the plastic hinging region and also to the flange
bars placed in the slab by 2.5% drift.
For loading in the normal beam direction, additional diagonal joint

cracks were observed after 0.5% story drift with flexural cracking of the normal
beam. At the column face, yielding started at 1.5% story drift for both top and
bottom normal beam bars and spread through the plastic hinging region at the
same drift level. Yielding did not penetrate into the slab because of the cracking
pattern at the top of the slab, which was similar to Specimen 1.
The column of this specimen dissipated more energy than that of

Specimen 1. For loading in the spandrel direction, all the column bars yielded at
the top of the joint. Yielding penetrated to the mid-depth of the joint. Exterior
column bars yielded before the interior ones. For loading in the normal beam
direction, none of the column bars yielded because the moment strength of the
column was significantly larger than that of the normal beam (Table 1). For
loading in both directions, strain data indicated minor slip of the interior column
bars.
Due to higher eccentricity, the damage at the exterior face of the

connection was significantly more than that of the interior face. In addition, the
strains measured at the exterior face of the joint stirrups were higher than the
strains at the interior face. This clearly shows the additional confinement from
the slab and normal beam. The damage correlates well with the maximum
measured joint shear deformation of approximately 1.0% at 4% story drift for
loading in the spandrel direction and 0.5% at 5% story drift for loading in the
normal beam direction.
The moment versus rotation diagrams of Specimen 2 are given in Fig. 8.

For loading in the spandrel direction, approximately one-third of the total beam
rotation over the plastic hinging region was caused by flexural deformation of
the beam. For loading in the normal beam direction, rotation due to flexure was
half of the total rotation. Measured rotations are shifted to higher values after
4% story drift due to spalling of cover concrete, which affected the
potentiometer readings.
@Seismicisolation
@Seismicisolation
The experimental program involves testing of three 3/4-scale exterior

reinforced concrete beam-column-slab connections under reversed cyclic
loading. The major parameters for this investigation are eccentricity of the
spandrel beam with respect to the column centroid, width of the normal beam,
and column section aspect ratio. Lateral load was applied in two principal
directions, one plane at a time, to determine the effect of prior loading on
performance of the connection and to examine the effects of biaxial loading.
Two tests have been completed and the results to date have indicated that
including the full floor system for this type of test specimen significantly
changes the influence of the spandrel beam eccentricity on the behavior of the
joint region under lateral loading.
Results from the completed tests during this investigation demonstrated

that the floor slab and normal beam add considerable torsional stiffness to the
subassembly and delay the deterioration of the joint shear stiffness and strength.
Moreover, anchorage conditions for the spandrel beam reinforcement did not
significantly deteriorate during the tests. In Specimen 2, the diagonal shear
cracks originated in the core region; therefore, the core dimension rather than
the column width should be taken as the development length of the spandrel
beam bars. All spandrel beam bars have yielded in both tests and the full
moment capacity of the spandrel beams was reached. Subsequent cyclic
displacement tests of the specimens in the plane of the normal beam indicated
that the stiffness of the subassembly was reduced due to the prior loading in the
plane of the spandrel beam. However, the shear strength of the joint and the
anchorage conditions for the normal beam reinforcement were not significantly
affected. Although, there was a delay in yielding of the normal beam flange
bars, the measured moment capacity of normal beams were very close to the
design values at the end of each test.
ACKNOWLEDGEMENT
This research study was sponsored by the National Science Foundation

under the Grant No 98 12464. The conclusions contained in this paper are those
of the authors and do not necessarily represent the view of the sponsor.
@Seismicisolation
@Seismicisolation
872 Burak and Wight
REFERENCES
1) “Building Code Requirements for Design of Beam-Column Joints in

Monolithic Reinforced Concrete Structures (AC1 352-R91)”, ACZ-ASCE
Committee 352, American Concrete Institute, Detroit, Michigan, 1991.
2) “Building Code Requirements for Structural Concrete”, ACZ 31 8-99,

American Concrete Institute, Farmington Hills, Michigan, 1999.
3) “Special Issue Investigation Report on the 1995 Hyogen-Nambu

Earthquake”, Arai Technical Research Report, Arai-Gumi Technical
Research Institute, Hyogo, Japan, 1999.
4) Raffaelle, G.S., Wight, J.K., “Reinforced Concrete Eccentric Beam-Column

Connections Subjected to Earthquake-Type Loading”, ACI Structural
Journal, Vol. 92, No. 1, January-February 1995.
5) Lawrence, G.M., Beattie, J.H., Jacks, D.H., “Cyclic Load Performance of an

Eccentric Beam Column Joint”, Central Laboratories Report 91-25126,
Central Laboratories, Lower Hutt, New Zealand, August 1991.
@Seismicisolation
@Seismicisolation
TABLE 1: Member Dimensions and Design Variables
I Specimen I I Specimen2 I Specimen3 I
* Effective joint width includes column width and one-fourth of the difference between beam
and column widths.
= Reinforcement ratio,
M, = Moment strength ratio; the ratio of the sum of nominal moment strengths
of columns above and below the joint to the sum of nominal moment
strengths of beams framing into the joint in a single plane; slab included as
compression flange for positive bending and slab bars included as tensile
reinforcement for negative bending,
e = Eccentricity of the spandrel beam centerline to the centroidal axis of the
column,
hco,/db,berim
= Ratio of column height to beam bar diameter,
hbeam/db,co,= Ratio of beam height to column bar diameter,
vj-i = Design joint shear stress for loading in the spandrel direction,
vj-2= Design joint shear stress for loading in the normal beam direction.
@Seismicisolation
@Seismicisolation
874 Burak and Wight
Axial Load Supplied by

Post-tensioningJacks
Figure 1: Test Setup
(a) Spandrel Direction (b) Normal Beam Direction
Figure 2: Lateral Loading of Test Specimen
@Seismicisolation
@Seismicisolation
Loading in Spandrel Direction
20
r
10
3
E
o r
O
nl
a
-10s
-
-20
-6 -4 -2 O 2 4 6
Story Drift (%)
Loading in Normal Beam Direction

120
- 20
80
z
5 40
-0
m
-L
2 0
3 -40
m
-I
-80 -
............ .......................................... -....._.............____.__
1-20
-120
-6 -4 -2 O 2 4 6
Story Drift (%)
Figure 3: Lateral Load versus Story Drift Hysteresis Loops for Specimen 1
@Seismicisolation
@Seismicisolation
876 Burak and Wight
(a) Exterior Face
(b) Interior Face
Figure 4: Cracking Pattern at the Connection Region for Specimen I
@Seismicisolation
@Seismicisolation
I
I Loading in Spandrel Direction Loading in Spandrel Direction
.
...< . .
.
.
<
..< .
.
...
j j /
-0.01 4.02 -0.01 o 0.01 0.02 0.01 ,015 -0.01 -0.005 O 0.005 0.01 0.015
Total Beam Rotation (rad) Rotation due to Flexure (rad)
Loadinu in Normal Beam Direction Loading in Normal Beam Direction
...i....... : : :
........ .
, . . < .
1 -0.04 -0.03 -0.02 -0.01 O 0.01 0.02 0 . 0 3 0.04 .O15 -0.01 -0.005 O 0.005 0.01 0.015
I
Figure 5: Moment versus Beam Plastic Hinge Rotation Curves for Specimen 1
@Seismicisolation
@Seismicisolation
878 Burak and Wight
Loading in Spandrel Direction

160
120
-z 80
5 40
U
m
-
3 0
-
$ -40
m
4 -80
-1 20
-1 60
Loading in Normal Beam Direction

160
120
-z 80
40
m
-
3 0
2
al -40
U
m
-I -80
-1 20 -
............. ............... * ............. ............. ..____.._____..._____.____
-1 60
-6 -4 -2 O 2 4 6
Story Drift (%)
Figure 6: Lateral Load versus Story Drift Hysteresis Loops for Specimen 2
@Seismicisolation
@Seismicisolation
(a) Exterior Face
(b) Interior Face
Figure 7: Cracking Pattern at the Connection Region for Specimen 2
@Seismicisolation
@Seismicisolation
880 Burak and Wight
Loading in Spandrel Direction Loading In Spandrel Direction

-E I60
-5
a!
I20
<
.. .. . <<.
60
Y
r
E
-
O
o
0
-60
rn
1
C
o
~ 1 2 0
=.I80
1.04 -0.03 -0.02 -0.01 O 0.01 0.02 -O.Os5 -0.01 -0.005 O 0,005
Loading in Normal Beam Direction Loadina in Normal Beam Direction
-0.04 -0.03 - 0 . 0 2 -0.01 O 0.01 0 . 0 2 0 . 0 3 0.04
Figure 8: Moment versus Beam Plastic Hinge Rotation Curves for Specimen 2
@Seismicisolation
@Seismicisolation
SP 209-47
Strengthening Rectangular
concrete Columns Using FRP:
A New Technique
by T. Rizk, I. Mahfouz, and S. Sarkani
Svnopsis: The objective of this paper is to introduce an innovative sandwich wrap-

ping confining system (SWCS) for upgrading and repairing rectangular columns.
The system described here replaces the conventional fiber reinforced polymer
(FRP)wrapping, which has only axial stiffness, by a sandwich wrapping confining
system that has both axial and flexural stiffness. The new system is made up of two
FRP faces separated by a light incompressible core.
Experimental and numerical analyses are performed on rectangular
columns upgraded and repaired using the SWCS. Thirty rectangular columns with
different dimensions are upgraded and repaired using the SWCS and tested to fail-
ure. A numerical study based on nonlinear finite-element analysis (material non-
linearity) is used to investigate the behavior of the columns repaired using the
SWCS and to predict the experimental failure loads. Finally, the experimental and
the analytical results are compared. Unlike conventional FRP jackets, the SWCS
can be used to improve the strength, stiffness, and ductility of rectangular columns.
Kevwords: concrete; fi be r-reinforced polymers; re pair;

strengthening; upgrading
@Seismicisolation
881
@Seismicisolation
882 Rizk et al.
Tarek Rizk-Assistant Professor, College of Engineering, Lawrence
Technological University, Southfield, Michigan
Ibrahim Mahfouz-Professor and Assistant Dean, Faculty of Engineering,

Zagazeeg University, Shoubra, Cairo, Egypt
Shahram Sarkani-Professor, School of Engineering and Applied Science, The

George Washington University, Washington D.C.
INTRODUCTION
Several traditional and nontraditional techiques are used to strengthen
reinforced concrete columns. Traditional techniques include concrete and steel
jackets. While traditional techniques can increase the column's load-carrying
capacity, they have limitations. For proper installation of a concrete jacket, a
large amount of damage to the existing structure is required to provide a bond
between the reinforcement of the existing column and that of the concrete jacket.
Concrete jackets increase the dead load of the structure, and the cross-section of
the column. The latter typically leads to an increase in the column's flexural
stiffness,which may in turn result in a larger load being attracted by the column.
While the benefits of steel jackets include high stifhess and strength, large
energy absorption, and enhanced ductility, their use to strengthen columns has
disadvantages. Because steel is an isotropic material, its resistance in the axial
and hoop directions of the column cannot be tailored or optimized. This means
that steel jackets are not suitable for enhancing the strength of the columns
without increasing their stiffness. Increasing the stiffness is sometimes not
desirable because the stiffer the column in a statically indeterminate structure,
the more load it will attract. Furthermore, the efficiency of increasing the load-
canying capacity of circular columns using steel jackets is questionable because
of bonding problems. Finally, outdoor use of steel jackets may prove costly in
corrosive environments.
The problems associated with traditional strengthening techniques may
be eliminated by using fiber-reinforced polymer materials. The reason for this is
that jackets made of FRP have a high strength to weight ratio, as well as
engineered properties that can be tailored to any specific need. Due to their
orthotropic behavior, jackets made of FRP can enhance the strength and ductility
of columns without affecting their stifhess. This can be achieved by placing the
fibers along the transverse direction of the column. In addition, jackets made of
FRY have good resistance to corrosion and chemical attacks, which makes them,
unlike steel plates and concrete jackets, suitable for repairing structures subjected
to corrosive environments. Finally, FRP jackets are easier and faster to handle on
site and do not add to the dead load of a structure.
In view of the above-mentioned advantages of FRP materials, FRP
jackets have been used by several researchers to upgrade and repair column^.'^'^
For the most part, FRP jackets have even been used successfully in the field to
retrofit circular c01umns.~FRP jackets are successful when wrapped around
circular columns because the axial stifhess of the FRP exerts a uniform
@Seismicisolation
@Seismicisolation
confining pressure around the entire cross-section of the column. The confining
pressure exerted leads to significant increases in the strength and ductility of
circular columns.
The major disadvantage of FRP jackets is that they are not suitable for
strengthening rectangular columns. When conventional FRP jackets are wrapped
around rectangular columns, the pressure exerted on the flat sides is limited or
even negligible because FRP 'ackets lack the flexural stiffness necessary for the
confinement of those sidesid Only a system that has both flexural and axial
stiffness can be used to confine rectangular columns. Conventional FRP
wrapping can only achieve acceptable degrees of success for certain types of
rectangular columns, and then only by changing their shapes so they become
circular, or by rounding their edges. Rochette and Labossiere' report the effect of
varying the comer radii. They state that changing the comer radius from 25 mm
to 38 mm on a 152 mm square element increases the axial force capacity
between 6 and 16 percent. Varying the comer radii has little effect on the
ultimate axial strain. Rounding the edges of the column will have little or no
effect on rectangular columns, especially as the aspect ratio increases. In
addition, rounding the edges of the column, or changing its shape from
rectangular to circular, is usually not acceptable, either from a structural or an
architectural standpoint.
Recognizing that the majority of columns are rectangular, one can clearly
see that practical applications of conventional FRP jackets are limited. Thus,
development of a wrapping system that is applicable to rectangular columns
becomes a paramount objective. The Sandwich Wrapping Confining System
(SWCS) introduced in this study and discussed in detail hereafter overcomes the
main disadvantage of conventional FRP jackets-their lack of flexural stiffness.
The SWCS consists of two faces of FRP sheets separated by an incompressible
core. The core acts as a spacing device separating the two FRP faces and thus
creates a structure with great bending and axial stiffness. The flexural stiffness
contributes significantly to the confinement of the flat sides.
To demonstrate the efficiency of the system, experimental and numerical
analyses are performed on reinforced concrete columns with different
dimensions. The results obtained experimentally and numerically are very close
and indicate that the proposed wrapping system can be used effectively to
strengthen rectangular columns.
Jackets made from fiber-reinforced polymers have been used
successfully to strengthen circular columns. However, when FRP jackets are
used on rectangular columns, the pressure exerted on the flat sides of the
columns is limited. Because most columns are rectangular, a system applicable
to rectangular columns becomes a paramount objective. This study introduces an
innovative sandwich wrapping confining system made of two faces of FRP
sheets separated by an incompressible core. Unlike conventional FRP jackets
that have only axial stiffness, the proposed system has both axial and flexural
stiffness, making it ideal for strengthening rectangular columns. The proposed
@Seismicisolation
@Seismicisolation
884 Rizk et al.
system’s flexural stiffness contributes significantly to the confinement of the flat
sides of rectangular columns.
DEVELOPMENT OF THE SANDWICH WRAPPING CONFINING

SYSTEM
Examination of a circular column that has a filament or fiber fabric
(referred to hereafter as a thin FRP wrapping) wound around it, shows that a
uniform confining pressure is exerted around the entire circumference of the
column because of the circumferential axial membrane stiffness of the thin FRP.
As a result, the thin FRP confines the entire cross-section. But when thin FRP is
used on square or rectangular columns, the pressure exerted is limited because
the axial stiffness of the filament or fiber winding in the transverse direction of
the column (i.e., in a direction perpendicular to the axis of the column) cannot
provide any significant confinement at the flat sides? and so the greatest part of
the confining pressure falls on the comers of the column, as can be seen in
Figure 1. Only a system that has both axial and flexural stiffness in the direction
perpendicular to axis of the column can be effectively used to improve the
strength of rectangular columns. Since thin FRP wrapping possesses negligible
transverse flexural stiffness, the flat sides of rectangular columns remain
unconfined under that system.
The innovative sandwich wrapping confining system introduced in this
study is a method applied to rectangular columns for the purpose of increasing
their strength, ductility, and stiffness, either jointly or separately. The sandwich
wrapping confining system described here replaces the FRP wrapping sheets
with a sandwich wrapping system made up of two FRP faces separated by a light
incompressible core as shown in Figure 2. The core of the sandwich structure is
used as a spacing device that separates the faces, creating a structure with great
bending and axial stiffness. The flexural stiffness of the SWCS thus created is
the main factor that contributes to the confinement of the flat sides.
The efficiency of the SWCS in improving the axial strength and ductility
of rectangular columns is governed by the wrapping’s flexural and axial
stiffnesses in the transverse direction of the column. On the other hand, the
efficiency of the SWCS in improving the flexural strength and stifhess of the
column is governed by the strength and stiffness of the SWCS in the longitudinal
direction of the column (i.e., along the axis of the column). The stiffness of the
SWCS in the transverse direction of the element must be large enough to bring
about the required confining pressures at relatively low transverse strains; that is,
if the Confining system is flexible in the transverse direction of the column, very
little confining stress will be generated and only a small increase in the strength
and the deformation capacity of the concrete column will be achieved. Therefore,
the confined column’s load vs. strain response will be similar to that of an
unconfined one, regardless of the ultimate capacity of the confining system.
Because the SWCS depends on high strength and great stiffness, the
material used for the inner and outer faces of the SWCS must possess those
properties. Materials made from Fiber-Reinforced Polymers have such
properties. The versatility of FRP is an added advantage: when an FRP sheet is
@Seismicisolation
@Seismicisolation
used in the faces of the SWCS, any characteristic or mechanical property
required to satis@ certain design requirements for the column under
consideration can be achieved. This can be accomplished by proper choices of
the orientation and type of fiber, and by the number of fiber layers in each face.
For cases requiring that the column’s strength and ductility be increased without
affecting its stifhess, all fibers can be oriented in the direction perpendicular to
the longitudinal axis of the column; such requirements are usually invoked for
cases that involve retrofitting and upgrading a column so it will resist seismic
loads. For cases requiring that the column’s flexural stifhess, strength, and
ductility be increased, the fibers can be oriented along both the longitudinal and
the transverse directions of the column.
The material chosen for use in the core of the sandwich system must be
incompressible in the transverse direction and must possess relatively high
transverse shear rigidities. Various materials, such as honeycomb structures,
possess these mechanical properties. The efficiency of the sandwich system can
also be enhanced by an increase in the thickness of the core. In general,
lightweight core materials are preferable, both statically and dynamically.
EXPERIMENTAL STUDY
In order to demonstrate the efficiency of the system, an experimental
study was carried out. The experimental investigation consisted of casting thirty
concrete columns, upgrading or repairing them using the SWCS, and loading
each column to failure. The columns were divided into six groups each
consisting of five columns. Groups A and AR columns were 165 mm x 165 mm,
groups B and BR columns were 165 mm x 248 mm, and groups C and CR
columns were 165 mm x 330 mm. It can be seen that groups A and AR,groups B
and BR, and groups C and CR had aspect ratios of 1:1, 1:1.5, and 1:2,
respectively. All columns had the same height of 914 mm. The different aspect
ratios were used to study the effect of increasing the aspect ratio on the increase
in strength. Column groups A, B, and C were properly designed (each column
had 5 No. 2 grade 40 ties, an adequate number of ties according to AC1 Building
CodeI4) and were upgraded using the SWCS. The columns in groups AR, BR,
and CR had an insufficient number of ties, (each had only 3 No. 2 grade 40 ties)
and were thus repaired, rather than merely upgraded using the SWCS. Groups
AR, BR, and CR had fewer ties so as to simulate real-life applications wherein
columns may lose ties due to corrosion, inadequate design, or poor
workmanship. The longitudinal reinforcement consisted of 4 No. 2 grade 40 bars
for the columns of groups A, AR, B, and BR, while the reinforcement for groups
C and CR consisted of 6 No. 2 grade 40 bars. The reinforcements and the
dimensions of the columns are given in Figure 3. One batch of ready-mixed
concrete was used to pour all the columns to allow for comparison. The strength
of the concrete was 20 MPa. All columns were cast vertically using wood
formworks and were left to cure for 28 days.
After the columns were cured, the SWCS was applied. Two types of
SWCS systems (fibedepoxy systems) were used for this study. SWCS system 1
(Sl) was used to upgrade the columns of groups A, B, and C, while SWCS
@Seismicisolation
@Seismicisolation
886 Ritk et al.
system 2 (S2) was used to repair the columns of groups AR, BR, and CR. Both
systems consisted of an inner layer of FRP, a core, and an outer layer of FRP.
The same type of epoxy was used for both systems. The fibers used for S1 were
unidirectional glass fibers, while those used for S2 were unidirectional carbon
fibers. Tension tests were conducted on both resin-coated and cured carbon and
glass fiber-reinforced polymer strips to determine their properties. The strips
were 25 mm wide in accordance with ASTM D3039. The ultimate strength,
modulus, and ultimate strain of the glass fiber reinforced polymers (GFRP) were
434 MPa, 25.5 kN/mm2,and 1.7%, respectively. The ultimate strength, modulus,
and ultimate strain of the carbon reinforced polymers (CFRP) were 4275 MPa,
228 kN/mm2, and 0.19%, respectively. It can be seen that the CFRP used had a
high modulus. The core used for S1 was a 20 mm wooden core, while that used
for S2 was 15 mm light weight glass reinforced concrete GRC boards. The GRC
boards were cut to the required size using an electric saw as shown in Figure 4.
The compressive strength of the GRC boards was 40 MPa. A thinner core could
be used if its material were more incompressible in the transverse direction and
possessed higher transverse shear rigidities than wood and GRC do. Examples
include honeycomb cores made of FRP materials.
For the application of the SWCS, the following procedure was followed:
1. The surface of the column was cleaned and made free of any substance that
may have adversely affected the complete bond of the SWCS to the column.
Extra care was taken to make sure that the surfaces of the columns were free
of any irregularities in the form of pockets, or browses.
2. One inner layer of GFRP, in the case S1, or CFRP, in the case of S2, was
saturated with epoxy and wrapped around the column. The main direction of
the fibers was perpendicular to the axis of the column.
3. 20 mm wooden core in the case of S 1 and a 15 mm GRC core in the case of
S2 were attached to the inner FRP layer and bonded to it using the same type
of epoxy used in step (2).
4. Finally, one outer layer, of either GFRP in case S1 and of CFRP in case S2,
was wrapped around the outer face of the core. For confining purposes the
main direction of the fibers followed the perimeter of the column.
5. C-clamps were used to eliminate gaps between the SWCS and the columns.
The C-clamps were removed before the columns were tested.
After applying the SWCS, the epoxy was left to cure for a week and then
the specimens were tested. Two columns from each group were used for control
and the rest were either upgraded or repaired using the SWCS. All of the
specimens were tested to failure under static axial compressive loading using a
universal testing machine. The load was applied by the machine on the concrete
only (no load was applied on the SWCS). The axial strains were measured using
strain gauges. Three strain gauges were mounted on the faces of the columns.
TEST RESULTS AND DISCUSSION

The results of tests on the upgraded and repaired columns are provided
in Tables 1 and 2, respectively. As was previously mentioned, the repaired
columns had 3 ties distributed along the height of the column while the upgraded
@Seismicisolation
@Seismicisolation
columns had 5 ties. It can be seen from Table 1 that the ultimate compressive
strength of the upgraded columns of Group A, Group B, and Group C increased
over the control columns by about 225%, 175%, and 160%, respectively. Table 2
shows that the ultimate compressive strength of the repaired columns of Groups
AR, BR, and CR increased by about 291%, 234%, and 212%, respectively.
Figures 5, 6, and 7 show columns A4, C5, and BR3 respectively, at failure. The
curves showing measured axial load versus axial strain for a control and an
upgraded column from each group are shown in Figure 8, and similar curves for
a repaired column from each group are provided in Figure 9. The plots of axial
load versus axial strain were identical for columns of the same group.
in most cases, the outer surface of the SWCS did not show any visible
signs of damage before failure. A slight curvature in the SWCS can be seen at
failure if the column is viewed from above. Columns in Groups A and AR were
loaded even after the columns lost their ability to withstand more load, so as to
give the failure shape shown in Figure 5. Most of the columns failed near the
comers. This was verified by the finite-element model, which indicates that the
highest stresses are near the edges of the columns. As expected, the increase in
strength of both repaired and upgraded columns decreased as the aspect ratio
increased. If a larger increase in strength is required, it can be achieved by
changing the properties of the SWCS. This can be done by using more than one
FRP layer at both the inner and outer faces, or by employing higher-strength
fibers, a thicker core, or a more incompressible core with higher shear rigidity.
For the control columns, the failure was brittle and sudden, while in the
case of the upgraded and repaired columns, there was a significant increase in the
ductility, indicated by the large increase in the flat parts of the load vs. strain
curves (Figures 8 and 9). Figures 8 and 9 also show that the increase in stiffness
was negligible, as was expected. The reason for this is that the fibers were
aligned along the transverse direction of the column. In real-life applications, an
increase in stifhess may not be desirable because, in a statically indeterminate
structure, the greater the stiffness, the greater the load attracted by the column.
The major advantage of using FRP is that the load-carrying capacity of the
column can be increased with or without affecting the stiffness; this cannot be
achieved if steel jackets are used due to the isotropic behavior of steel. If an
increase in stiffness had been required, fibers would have been placed in the
transverse and longitudinal directions with respect to the column axis.
The significant increase in the load-canying capacity and ductility of
the columns indicates that SWCS is an efficient technique for upgrading and
repairing rectangular columns.
FINITE-ELEMENT MODELING OF COLUMNS REPAIRED

USING THE SWCS
The experimental investigation shows that SWCS is an efficient
technique for enhancing the strength and ductility of rectangular columns.
However, it is n e c e s s q to verifi that numerical analysis can be used to predict
the behavior of upgraded and repaired columns. The importance of the finite-
element analysis is that it can be used for the future design of columns
@Seismicisolation
@Seismicisolation
888 Rizk et al.
strengthened using the SWCS. The finite-element analysis performed in this
study focuses on predicting the failure loads of the repaired columns and on
investigating their behavior.
Because the finite-element analysis is aimed at predicting the failure
load of the repaired columns, modeling the nonlinear material response of
concrete is crucial. Two different approaches can be used: a smeared crack
model" and a discrete crack model.I6 In the smeared crack model, the effects of
cracking are averaged over an element so that an equivalent continuum is
analyzed; the model does not track individual macro cracks. In the discrete crack
model, cracks are individually modeled by revising and updating their nodal
connectivity; in essence, cracks are represented as interelement displacement
discontinuities. Discrete analysis is very time-consuming, and may also impose
significant demands on computer memory as changes in nodal connectivity lead
to a loss of the banded structure of the stiffness matrix. For this reason, discrete
crack analysis is seldom used and almost all finite-element packages, including
ABAQUSI7and ANSYS,18use smeared crack models.
This study uses a smeared crack model. The analysis is carried out using
the finite-element program ABAQUS. The eight-node, three-dimensional, solid
element, C3D8, is utilized to model the concrete. The CONCRETE option
provided by ABAQUS is used to define the properties of plain concrete outside
the elastic range. The CONCRETE model in ABAQUS uses oriented damage
elasticity concepts to describe the reversible part of the material's response after
failure. The ELASTIC option must be used to define elastic properties.
Normally, the user must assume yield stress, plastic strain at failure, ratio of
biaxial to uniaxial compression failure stress, and Poisson's ratio. The failure
stress and Young's modulus are normally available from tests on concrete
cylinders or cubes. In this study, Poissonk ratio is taken as 0.15 and the yield
stress is taken in the range of 50% to 60%of the failure stress.
The local energy release and the concrete-rebar interaction that occur as
the concrete begins to crack are of major importance in determining the response
of a concrete structure between its initial deformation, recoverable deformation,
and its collapse. ABAQUS offers an indirect technique for modeling these
effects, known in the literature as tension stiffening. This models the energy
release rate as cracks propagate, and the associated concrete-rebar interaction, as
a gradual loss of strength beyond the failure strain. The choice of tension
stiffening parameters is very important because, generally, more tension
stiffening makes it easier to obtain numerical solutions. Too little tension
stiffening will cause the local cracking failure in the concrete to introduce
temporarily unstable behavior in the overall response of the model. ABAQUS
offers two types of tension stiffening. The default type, "Type = STRAIN",
specifies the post-cracking behavior by entering the post-failure load vs. strain
relationship. The other alternative, "Type = displacement", defines post-cracking
behavior by entering the displacement, UO, at which a linear loss of strength after
cracking gives zero stress. Type = displacement is used in modeling the columns
with uo= 0.00005. Type = strain may introduce mesh sensitivity in the analysis in
the case of little or no reinforcement. The rebars are defined using the REBAR
@Seismicisolation
@Seismicisolation
option provided by ABAQUS for defining reinforcing bars in concrete. Rebars
may be defined singly or as a layer of uniformly-spaced reinforcing bars. In this
study, the rebars are defined singly. The rebar elements are superimposed on the
mesh of the concrete elements and used with standard metal plasticity models
that describe the behavior of the reinforcing material. Effects associated with the
rebarfconcrete interface, such as bond slip and dowel action, are modeled using
the tension stiffening option described above.
The CFRP faces are modeled using the built-in S4R5, a four-node shell
element with six degrees of freedom. S4R5 has three displacement and three
rotation degrees of freedom at each of its four nodes. The CFRP sheets have no
yield or hardening and the relationship between the stress and the strain is linear.
An eight-node linear brick incompatible modes element is used to model the
core. The column, outer FRP layer, core, and inner FRP layer are modeled as
fully bonded.
The results obtained from the finite-element model are compared to those
obtained experimentally in Table 3. Figures 10 and 11 show the compressive
stresses in the concrete of the columns in group BR,and the stresses in the inner
CFRP face of the columns of group CR, respectively. Figure 12 compares the
finite-element and the experimental load vs. strain curves.
The finite-element analysis makes it possible to visualize the behavior of
the columns tested experimentally. The experimental failure of the repaired and
upgraded columns was near the comers. The finite-element analysis shows that
the highest combination of flexural and axial stresses is near the edges of the
columns. The failure loads predicted by the finite-element model are close to
those obtained experimentally. The finite-element analysis results are
conservative by around 5% for the control columns and 10% for the repaired
columns. This indicates that finite-element analysis can be used to design
columns strengthened using the SWCS.
CONCLUSIONS
1. A new system (Sandwich Wrapping Confining System) for repairing and
upgrading rectangular reinforced concrete columns based on the utilization of the
sandwich structural element is introduced. This Sandwich Wrapping Confining
System (SWCS) is a technique for improving the axial, flexural, and shear
strength-as well as the ductility-of rectangular reinforced columns, which it
accomplishes with or without increasing the stiffness of the column.
2. Because SWCS has both axial and flexural stiffness in the transverse
direction of the column, it effectively improves the properties of square and
rectangular columns. The flexural stiffness contributes significantly to the
confinement of the flat sides. Since thin FRP wrapping possesses negligible
transverse flexural stiffness, considerable portions of the rectangular columns
(the flat sides) remain unconfined under that system.
3. The SWCS possesses the two most important properties for effective
Confinement of columns, namely, a high ratio of strength to weight and a high
ratio of stiffness to weight; therefore, the system can significantly increase both
@Seismicisolation
@Seismicisolation
890 Rizk et al.
the ultimate strength and the ductility of rectangular columns, as well as their
stiffness, when required.
4. The efficiency of the Sandwich Wrapping Confining System in improving
the axial strength and ductility of rectangular columns is governed by the
wrapping’s flexural and axial stiffness in the transverse direction of the column.
The stiffness of the Sandwich Wrapping Confining System in the transverse
direction of the element must be great enough to bring about the required
confming pressures at relatively low transverse strains.
5. The versatility of SWCS in increasing strength, stiffness, and ductility,
either separately or jointly, can be accomplished by proper choices of the
orientation and type of the fiber, and by the number of fiber layers in each face.
For cases requiring that a column’s strength and ductility be increased without
affecting its stiffness, all fibers can be oriented in the direction perpendicular to
the longitudinal axis of the column; such requirements are usually invoked for
cases that involve retrofitting and upgrading a column so it will resist seismic
loads. For cases requiring that the column’s flexural stifhess be increased while
its strength and ductility are also increased, the fibers are oriented along both the
longitudinal and transverse directions of the column.
6. The experimental failure loads and those predicted by the finite-element
analyses were very close. The finite-element failure loads were conservative by
around 5% for the control columns and 10% for the repaired ones. This indicates
that finite-element analysis can be used for future design of columns
strengthenedusing the SWCS.
REFERENCES
1. Harmon, T. G.; and Slattery, K. T., “Advanced Composite Confinement of
Concrete,” Advanced Composite Materials in Bridges and Structures, CSCE,
1992, pp. 229-306.
2. Howie, I.; and Karbhari, V. M., “Effect of Materials Architecture on
Strengthening EMiciency of Composite Wraps for Deteriorating Columns in the
North-East,” Infrastructure: New Materials and Methods of Repair - Proceeding
of Third Materials Engineering Conference, San Diego, 1994, pp. 199-206.
3. Nanni, A.; and Bradford, N. M., “FRP Jacketed Concrete Under Uniaxial
Compression,’’ Construction & Building Material, Elsevier Science Ltd.,
London, V.9 No.2, 1995, pp. 115-124.
4. Saadatmanesh, H.; Ehsani, M. R.,“Strength and ductility of concrete columns
externally reinforced with fiber composite straps,” ACI Structural Journal, V.
91, No. 4, 1996, pp. 434-447.
5. Rochette, P.; and Labossière, P., “A Plasticity Approach for Concrete
Columns Confined with Composite Materials,” Proceedings of Advanced
Composite Materials in Bridges and Structures II, Montreal, 1996, pp. 359-366.
6 . Soudki, H. A.; and Green, M. F., “Performance of CFRP Retrofit Concrete
Columns at Low Temperatures,” Proceedings of Advanced Composite Materials
in Bridges and Structures II, Montreal, 1996, pp. 427-434.
@Seismicisolation
@Seismicisolation
7. Mirmiran, A.; and Shahawy, M., “Behavior of Concrete Columns Confined by
Fiber Composites,” Journal of Structural Engineering, ASCE, V. 123, NO. 5,
1997, pp. 583-590.
8. Yan, X., “Seismic Retrofit of RC Circular Columns Using Prefabricated
Composite Jacketing,“ Journal of Structural Engineering, ASCE, V. 123, NO. 10,
1997, pp. 1357-1364.
9. Hams, K.; Kestner, J.; Pessiki, S.; Sause, R.; and Ricles, J., “Axial Behavior
of Reinforced Concrete Columns Retrofit with FRPC Jackets,” Proceedings of
Second International Conference on Composites in Infastructure, Arizona,
1998, pp. 411- 425.
10. Mirmiran, A.; Shahawy, M.; Samaan, M.; Echary, H.; Mastrapa, J. C.; Pico,
O., “Effect of Column Parameters on FRP-Confined Concrete,” Journal of
Composites in Construction, ASCE, V. 2, No. 4, 1998, pp. 175-185.
11. Toutanji, H., “Stress-Strain Characteristics of Concrete Columns Externally
Confined with Advanced Fiber Composite Sheets,” AC1 Materials Journal, V.96,
NO. 3, pp. 397-404.
12. Mahfouz, I.; Sarkani, S.; and Rizk, T., “Repairing and Strengthening
Reinforced
Concrete Structures Using Fiber Reinforced Plastics,” 12‘h International
Conference on Composites Materials, Paris, France, 1999.
13. Mahfouz, I.; Sarkani, S.; and Rizk, T., “Strengthening Reinforced Concrete
Structures Using Fiber Reinforced Polymers,” Structural Faults & Repair-99,
Commonwealth Institute, London, UK, 1999.
14. AC1 Committee 3 18, “Building Code Requirements for Structural Concrete
(ACI-3 18-95) and Commentary (3 18R-95), ”American Concrete Institute,
Farmington Hills, Mich., 1995.
15. Rashid, Y.R., “Analysis of Prestress Concrete Pressure Vessels,” Nuclear
Engineering and Design, V.7, No. 4, 1968, pp. 334-344.
16. Ngo, D.; and Scordelis, A. C., “Finite Element Analysis of Reinforced
Concrete Beams,” ACI Journal, V.64, No. 3, 1967, pp. 152-163.
17. Habit, H. D.; Karlson, B. L.; and Sorensen, E. P., “ABAQUS Version 5.8,
Finite Element Program,” Habit, Karlson, and Sorensen Inc., Providence, Rhode
Island 02906-4402, 1999.
18. ANSYS, Version 5.0, Swanson Analysis Systems, Inc., Houston, Penn.,
1992.
@Seismicisolation
@Seismicisolation
892 Rizk et al.
Table 1. Results of tests on square and rectangular columns upgraded using SWCS
Group A Group B Group c
165x165 165x248 165x330
(mm) (m)
Column
1% Failure inc. in
Strength
(%)
inc. in
Strength
(%)
(control) B1
(control) I 597 (control)
(control) ?I
(control) 590 (control)
900 225 B3 I 1032 I 172 C3 I 1257 160
163
Table 2. Results of tests on square and rectangular colui
Column Failure inc. in

Load Strength
(KN) (%)
CR 1 589
(control)
CR2 585
(control)
c3 1283 217
c4 1225 208
c5 1250 212
@Seismicisolation
@Seismicisolation
finite element analysis ( E M ) and experimental results
165x248
column FEM Exp.

Failure Failure
Load Load
(W
AR1 290 300
(control)
AR2 290 298
(control)
AR3 800 875
AR4 800 820
AR5 800 900
@Seismicisolation
@Seismicisolation
894 Rizk et al.
r-l Unconfined concrete
.. .. Confined concrete
.. .. .
(a) A rectangular column (b) A circular column

Fig. 1: Area of concrete confined by thin FRP jackets.
Confined reinforced concrete
Incompressible core
__/i__ /
J
One or more layers of FRP
J
7
Fig. 2: Components of the Sandwich Wrapping Confining System (SWCS).

@Seismicisolation
@Seismicisolation
Groups A, B, and C Groups AR,BR, and CR

Upgraded columns Repaired columns
I*---*I n H
165 m m 248 m 330 m
Groups Aand AR Groups B and BR Groups C and CR

4-N0.3 (9.5 iììiiì ) 4-N0.3(9.5 ïiìiiì) 6-NO.3(9.5 ïiìiiì )
Fig. 3: Reinforcement of columns upgraded and

repaired using SWCS.
@Seismicisolation
@Seismicisolation
896 Rizk et al.
Fig. 4: Cutting of glass reinforced concrete board

using electric saw.
@Seismicisolation
@Seismicisolation
Fig. 5: Column A4 at failure.
@Seismicisolation
@Seismicisolation
898 Rizk et al.
Fig. 6: Column C5 at failure.
@Seismicisolation
@Seismicisolation
1400 1
1200
1000
g
'2
800
600 Group A
-A- Group B
400
* Group c
200
O
O 0.005 0.01 0.015 0.02
Strain (rnn'nmJ
(a) Upgraded columns
900
800
700 -e- ControlA
E ::: -A-
*
ConmolB
ControlC
2 400
300
200
100
O
O 0.005 0.01 0.015
Strain ( d m m )
@) Control columns
Fig. 7: Load vs. strain curves of columns upgraded using SWCS.
@Seismicisolation
@Seismicisolation
900 Rizk et al.
----- =.'-
~
Fig. 8: Column BR3 at failure.
@Seismicisolation
@Seismicisolation
1600 1
I200
i rn
400
O
O 0.005 0.01 0.015
strain ( d m )
(a) Upgraded columns
*O0 1
e ControlAR
-A- ControlBR
-E+ ControlCR
O 0.005 0.01 0.015
strain ( d m )
(a) Control columns
Fig. 9: Load vs. strain curves of columns repaired using SWCS.
@Seismicisolation
@Seismicisolation
902 Rizk et al.
Fig. 10: Stresses in concrete of repaired columns of group BR.
@Seismicisolation
@Seismicisolation
Fig. 11: Stresses in inner CFRP face of group CR columns.
@Seismicisolation
@Seismicisolation
Rizk et al.
O Y
O 0.007 0.014
Strain ( d m )
Fig. 12: Comparison between FEM and experimental load

vs. strain curves of columns repaired using SWCS.
@Seismicisolation
@Seismicisolation
SP 209-48
Thermal Cyclic Testing on a

Concrete Sandwich Panel System
by R. K. Devalapura, J. M. Seng, M. F. McBride,
D. W. Winiarski, and B. D. Johnson
SvnoDsis: When insulated concrete sandwich panels are used in the envelope of a
building, the experior and interior are subjected to two different environments. The
exterior concrete wythe is subjected to outside weather swings in the temperature
and humidity causing thermal expansion and contraction, whereas the interior is
exposed to a controlled steady room temperature environment. Dimensional
change in the panel depends primarily on the height of the panel and the relative
change in temperature. The severity increases when the outside concrete wythe of
a tail panel is supported (and hence constrained) on the foundation allowing verti-
cal movements only at the top. If these weather cyclic movements are restricted,
the panels may experience cracking and eventually may experience a premature
failure. Therefore, the tie system used in the panels should be flexible enough to
accommodate these differential movements. This often is the most critical issue in
the service life of the building when sandwich panels are used.
There is no standard test method available to evaluate the thermal non-uni-
form cyclic behavior of insulated panel systems. The authors have followed a sci-
entific approach to evaluate these stresses by subjecting the ties to real life cycles
occurring over a period of time. The system used in this study includes a low-con-
ductivity polymer connector with extruded polystyrene rigid foam insulation. The
testing was continued until the failure of the system or to more than 100 years of
equivalent cycling (the expected service life of the building), whichever is less.
This paper focuses on the methodology developed and parameters considered in
developing the criteria for testing weather cycles. The procedure may be followed
to evaluate any given insulated panel system to predict its long-term durability.
K e w o r d s : insulated sandwich panel; panel movement;

thermal deformation; thermal durability; wall service
loa ding; weather cycles
@Seismicisolation
@Seismicisolation
905
906 Devalapura et al.
Ravi K. Devalapura, Ph.D., is an Advanced Engineer at Owens Coming
working on developing applications of insulation materials to the building
industry. He received M.S. and Ph.D. degrees in civil engineering from the
University of Nebraska in 1992 and 1995 respectively. Dr. Devalapura has
published 18 technical papers related to concrete and masonry constructions.
Jocelyn M. Seng, Ph.D., P.E., has been with Owens Coming for seven years
and is a Senior Scientist developing new product applications. She graduated
with a BS and MS from MIT in 1984 and 1985, respectively, and a PhD from
Stanford in 1992, all in Mechanical Engineering. Dr. Seng has published more
than 20 technical papers, primarily focused on composites design and analysis.
Merle F. McBride, Ph.D., P.E. has been with Owens Coming for 25 years, is a
Senior Research Associate and received his degrees from the Ohio State
University majoring in Mechanical Engineering. His technical focus has been
related to the energy performance of residential and commercial buildings. He
has been active in ASHRAE since 1973 and has published 28 technical papers.
David Winiarski is currently a Senior Research Engineer with Battelle, PNNL

in Richland, WA. He received his B.S. degrees of Arts & Science in Physics
from Oregon State University and M.S. degree in Mechanical Engineering
from Texas A&M. Mr. Winiarslu’s professional interests include energy
efficiency, sustainable building design, and energy and resource management.
Bradley D. Johnson is the business manager for commercial foam applications

division of Owens Coming. He holds a B.S.E.E. from Duke University and an
M.M. from The Kellogg Graduate School of Management. Mr. Johnson has
actively participated in the development, testing and commercialization of
several products including an insulation system for concrete sandwich panels.
INTRODUCTION
Insulated sandwich wall panels are used to provide thermal efficiency in

buildings when interior space is used for human occupancy, specialized
manufacturing operations or controlled climate storage. Insulated wall panels
typically include an insulation board and a connector tie system, which become
an integral part of the finished panel providing durable concrete surfaces inside
and out. By providing an effective thermal barrier in the building envelope,
the temperature movement is restricted through the exterior walls, resulting in
lower energy usage to maintain a controlled interior environment.
@Seismicisolation
@Seismicisolation
Buildings across the country are subjected to different environments and must
withstand low to severe weather patterns depending on their geographical
location. Sandwich panels undergo in-plane thermal deformations due to
outside temperature changes. These deformations result in bowing of the panel
when two concrete wythes act compositely to resist lateral loads. Some
locations in the country experience severe variation in weather resulting in
significant thermal movement in the exterior concrete wythe. The linear
thermal expansion and contraction of the wall panel is calculated using the Eq.
[i] below.
AL = L * a * AT PI
where,
AL = change in panel height, mm (in)
L = initial panel height, mm (in)
a = coefficient of thermal expansion of concrete, d m m ( d i n )
AT = change in temperature between exterior and interior, OC (OF)
The concrete’s thermal movement is directly proportional to its height and the
change in temperature. As the relative temperature between exterior and
interior (assumed to be steady) of the building varies, the exterior wythe
experiences expansion or shrinkage accordingly. The weather cycles may be
approximated to be seasonal with summer and winter cycles on an annual basis
with two cycles. To be precise, the temperature swings occur each day with
warmer temperature in daytime and cooler temperature in the night. So, the
exterior wall panel will experience two cycles each day of the year. The ties
used in the panel should be able to accommodate this movement in addition to
the cyclic movements that occur as the temperature swings with time. These
stresses will be additive to the gravity load stresses that are already acting on
the ties in panels with unsupported fascia.
A limited guidance is given in PCI Architectural Precast Concrete Handbook

(1) which specifies that the connectors that are intended to accommodate
differential movement between wythes should be sufficiently ductile to
withstand 5,000 cycles of reversal at an amplitude of 3.2 mm (l/X in.). This
provision is mostly applicable for architectural panels and does not reflect high
thermal movements that occur in typical sandwich panel constructions. Initial
testing with this method revealed that uniform cycling does not reflect the
severity of the weather effects. The study reported here focuses on the
evaluation of thermal deformations and factors affecting thermal movement
through a sandwich wall panel. The methodology developed was based on a
rational and scientific approach to reflect the real-life weather cycles that occur
in insulated wall panels during their service life.
@Seismicisolation
@Seismicisolation
PARAMETERS FOR ANALYSIS
A typical commercial building was considered for this study with various
parameters described in the following sections. The study included critical
variables that affect the thermal transmission through the wall system. To
cover the most commonly used panel sizes, a total of seven different thickness
configurations were identified as shown in Fig. 1. The wall configurations are
defined by stating the thickness of each layer starting with the exterior layer of
concrete, then the insulation layer, and finally the interior layer of concrete.
All panel constructions chosen here were thought to be representative of
commonly used panel thickness used in tilt-up and precast construction. The
outside concrete wythe of minimum 50.8 mm (2 in.) was selected to be
conservative to determine if failures occur around the tie by concrete pullout or
spalling mechanism. Wall panels in all four cardinal orientations were
considered. Five different U.S. cities were selected to cover a broad spectrum
of extreme climate variations around the country. These locations were
selected from the weather data developed by the National Oceanic and
Atmospheric Administration (NOAA) (2). This federal agency gathers
weather data from nearly 4,775 stations around the country as shown in Fig. 2.
The heating and cooling degree-days in Fig. 2 are calculated using the
following equations.
HDD18 = C(18 - Tave,day),whenTave,day<18
CDD10 = Z(Tav,&y- lO),when Tavqday >10 [31

where, Tave,day = average daily temperature,O
C
Determining the exterior surface temperature of the sandwich wall is a

complex calculation which is controlled by the ambient air temperature, the
solar heat gain, the wind speed, the construction of the wall in terms of the
thickness of the concrete layers and the foam insulation, and the interior
surface temperature. The solar heat gain depends upon the solar absorptivity
of the surface, the orientation, and the weather location. For purposes of this
analysis, two levels of solar absorptivity were selected. A value of 0.91
represented bare concrete surface and 0.65 represented concrete painted black.
These were two extreme conditions that could arise in the field. All the above
envelope parameters considered for the thermal analysis are given in Table 1.
@Seismicisolation
@Seismicisolation
THERMAL ANALESES
The thermal analyses were divided into four steps.

1. Define the sandwich wall constructions and their associated material
properties.
2. Calculate the daily interior and exterior wall temperatures for each wall
construction in five locations in the United States for one year using
computer simulation.
3. Use the interior and exterior wall temperatures to calculate daily
temperature gradients through the wall sections.
4. Calculate the expansion or contraction of the exterior wythe due to the
temperature gradients it was experiencing due to thermal cycling.
The first step was to define seven sandwich wall constructions as shown in Fig.
1. The material properties needed to perform the thermal .calculations were
thickness, density, specific heat, and thermal conductivity. The values used for
the material properties are given in Table 2.
The second step in the analysis was to simulate the thermal performance of the
seven different wall constructions with two levels of solar absorptivity in five
U.S. locations using the Buildings Loads Analysis and System
Thermodynamics Program (BLAST) (3). The number and location of the
weather stations Battelle Pacific Northwest National Laboratories (PNNL)
used encompass the range of heating and cooling degree-days (HDD and
CDD) across the continental U.S. The HDD and CDD for the five cities
selected are given in Table 3. There were 70 runs completed by PNNL using a
prototypical retail building. The one story slab-on-grade building had a floor
area of 1,115 m2 (12,000 ft?) and followed the internal load schedules from
ASHRAE Standard 90.1 (4) with 3,890 hours of operation per year. The
results for each BLAST run consisted of the hourly temperatures of the interior
and exterior wall surface temperatures for each of the four cardinal orientations
for an entire year. Thus, the 70 BLAST runs generated a total of 280 unique
cases.
The third step was to model the hourly temperature profiles within each wall to
determine the temperature gradient across the insulation. This analysis was
completed using a one-dimensional finite-difference program specifically
written for this study. The interior and exterior surface temperatures predicted
by the BLAST program were the boundary conditions for this analysis for each
hour of the year. To perform this, each panel was divided into one-quarter
inch sections for purposes of the calculations. The stability criteria for the
analsis reauired 240 iteratinns each hniir in nrder tn achipvp rnnvprupnrp
@Seismicisolation
@Seismicisolation
The temperature gradient across the insulation was the final result of the
calculations for each hour of the year. Once the hourly temperature gradients
were determined, the maximum and minimum daily temperatures were
determined. These daily temperature gradients formed continuous cycles for
an entire year, which becomes the fatigue cycle of interest. Typical
temperature gradient cycles across insulation for a period of one year are
depicted in Fig. 3. The final step was to convert the daily temperature
gradients into daily deflections. This was accomplished using Eq. [i], and
assuming a panel height of 12.2 m (40 ft) and using a coefficient of thermal
expansion of 6.0 x 10“ mm/mm ( d i n ) for concrete. A sample of the generated
deflections in the panel for the Fig. 3 example is shown in Fig. 4. The
calculated deflections can be used in an Instron to drive actual fatigue on the
test specimens. Table 4 lists the maximum deflections for all of the analysis.
Having calculated the expansion and contraction cycles of various wall panels
under various environmental conditions, the performance of a tie system could
be physically tested in simulation.
THERMAL CYCLIC TESTING
The purpose of this experimental work is to evaluate cyclic fatigue stress of a

tie connector system to simulate stresses induced by differential thermal
expansion and contraction of the two layers of concrete in a sandwich wall
panel. Thermal cyclic test was performed on the insulated system to evaluate
its durability when exposed to a long-term exterior weather cycles. Results can
be used to evaluate the life of system, and to qualify acceptable design
materials for ties.
Based on preliminary tests, the rate of loading was selected to be

approximately 4 seconds per cycle to make sure the tie in the specimen did not
generate heat. The testing was continuously loaded until the failure of the
specimen or 4 days whichever comes first. This four-day uninterrupted testing
represents a total of 118 years of daily thermal cycles, a length of time in
excess of the life expectancy of the typical structure.
Concrete specimens of 152 mm (6 in.) square with one tie at the center were
used in this test due to the ease in handling and headroom restrictions in the
Instron machine. The specimens were cast similar to actual panel in the field -
with the “outer” wythe poured first, extruded polystyrene board laid on top, the
tie inserted through the foam and the “inner” wythe poured on top. The
specimens were allowed to cure for 28 days to achieve maximum strength and
@Seismicisolation
@Seismicisolation
uniform relative humidity. The specimens were cast in 6.4 mm (% in.) thick
steel tube forms with threaded boit for fixing to the machine (see Fig. 5).
During the test, only the interior concrete wythe was supported and held rigidly
to the base of the testing machine. The outer wythe that was connected to the
Instron head through a bolt is cycled following the daily weather deflections
developed earlier (as shown in Fig. 6). The testing was conducted without
interruption by repeating the annual weather cycles, which were programmed
into the Instron machine. The test specimens were cycled vertically (in-plane
movement) without allowing for rotation. The Instron machine recorded
displacement and the force required to achieve the displacement. For safety
reasons, the sample was secured to the machine in such a way that failure was
not readily observed. The failure criterion for the test was a significant
reduction (usually to zero) in the load or displacement recorded.
The objective of the paper was two fold; first, to develop realistic weather
profiles that can be used to predict non-uniform cyclic movements in insulated
wall panels, and second to test a insulated panel system to determine its
durability for use in such a weather condition.
What led the research team to pursue this project was the significant difference
between the stresses a tie system endures and the uniform cyclic testing
methodology as outlined PCI Handbook (i), which requires passing 5,000
cycles at 3.2 mm (1/8 in.) deflection. Several tie manufacturers cite this test in
their literature. However, while this test is acceptable for the architectural
concrete panels for which it was designed, it is inadequate to simulate the
stresses a tie system used in a concrete sandwich panel encounters. This is
most dramatically shown when the observer views the graph of displacements
in Fig. 4 for Phoenix. Notice that the daily cycles are superimposed on the
annual weather cycle. These multiple fiequencies lead to the non-uniform
nature of the expansion and contraction of the outer wythe of a concrete panel.
The test criteria developed by the authors for the non-uniform weather cycle
analysis is first time in the industry. The scientific methodology allowed
researchers to develop an accurate database of real daily temperature variations
around the country. The data profiles and the methodology for their
development can be used by industry to study many related areas. The data
sets that describe the interior and exterior temperatures for the variety of wall
panel constructions studied may be used to support thermal performance
claims and as an acceptable method to qualiQ systems for long-term durability
@Seismicisolation
@Seismicisolation
for weather exposure. The methodology to create such data sets could be used
to study other construction types for their thermal performance.
Finally, a physical test methodology that produces reliable and repeatable

results using standard criteria was developed. This physical test could be used
to test a range of materials and components for use in insulated concrete wall
panels and is superior to the current accepted uniform cyclic test because it
more closely simulates the stresses that occur in-situ due to weather. The
method can be used to test any insulated panel system with small to full-scale
specimens.
ACKNOWLEDGEMENT
Authors would like to thank Mr. Joe Nemeth, Mr. Don Wise, Jr., and Mr. Chns
Gill of Owens Coming Integrex Testing Systems, Granville, OH who
conducted the experimental work in the laboratory. Also, help from Ms. Shari
L. Kraber, Consultant at Stat-Ease, Inc. Minneapolis, MN in the design of
experiments analysis is appreciated.
REFERENCES
(1) Architectural Precast Concrete Handbook, Second Edition,

Precasflrestressed Concrete Institute, Chicago, IL, 1989, pp. 340.
(2) Owenby, J.R., Ezell, D.S., Heim, Jr., R.R., “Annual Degree Days to
Selected Bases Derived from the 1961- 1990 Normals”, U S . Department of
Commerce, National Oceanic and Atmospheric Administration, National
Climatic Center, Ashville, NC, July, 1992.
( 3 ) Buildings Loads Analysis and System Thermodynamics, Version 3 .O, Blast

Support Office, University of Illinois at Urbana-Champaign, 1980.
(4) ASHRAE Standard 90-1-1999, ‘Energy Standard for Buildings Except

Low-Rise Residential Buildings’ American Society of Heating,
Refrigerating and Air-conditioning Engineers, Inc., Atlanta, GA, 1999.
@Seismicisolation
@Seismicisolation
Table 1. All envelope parameters considered for analysis
I Envelope variable
~
Parameters for analysis
P
Bismarck, SD
Geographic location
Minneapolis, MN
St. Louis, MO
Ft. Worth, TX
Phoenix, AZ
5 1-38-152 (2-1.5-6)
Wall thickness configuration (in.)
51-51-152 (2-2-6)
(Concrete-Insulation-Concrete) 51-76-152 (2-3-6)
51-102-152 (2-4-6)
76-5 1-152 (3-2-6)
51-51-254 (2-2-10)
51-51-51 (2-2-2)
East
Cardinal orientation
West
North
South
0.91 (plain concrete)
Solar absorptivity
0.65 (painted black)
(25.4 mm = 1 inch)
@Seismicisolation
@Seismicisolation
Table 2. Materials properties used in the analysis
Table 3. Heating and cooling degree days for five geographic locations
Dallas 1255 (2259) 3659 (6587)

Phoenix 750 (1350) 4681 (8425)
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation
Wall Type Exterior Insulation Interior
1 51mm I ~ m m 1 152mm
3 51 mm 76mm 1 152 mm
5 I 76 mm I 51mm 1 152 mm
7 51mm I simm I 51 mrn
Fig. 1 Seven wall configurations selected for analysis
6000
5000
4000
O
3 3000
n
u
2000
1000
O
0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0
z % z 3 5 ? 3 ì ? z g g ~
WD18
(National Oceanic and Atmospheric Administration)
Fig. 2. U.S. weather data published by NOAA
@Seismicisolation
@Seismicisolation
Phoenix. South. Wall Tvpe 2. Solar Abs. = 0.91
30.00
20.00
u,
0 1u.00
2
& 0.00
E
g -10.00
*
E
g -20.00
a
-30.00
-40.00
O 30 60 90 120 150 180 210 240 270 300 330 360
Days
Fig. 3. Typical annual temperature difference generated

from finite difference analysis
Phoenix. South. Wall Tme 2. Solar Abs. = 0.91
O 30 60 90 120 150 180 210 240 270 300 330 360

Days
Fig. 4. Typical panel deflections generated for entire year
@Seismicisolation
@Seismicisolation
Cyclic load
from instron 1 Steel tube used
for casting
Inside concrete
wythe
152 mm (6in.) -
Outside
concrete wythe
Fig. 5.
- Foam insulation
Rigid fixed support

___
Schematic diagram of test specimen and Set-up
Fig. 6. A specimen under test in Instron
@Seismicisolation
@Seismicisolation
CONVERSION FACTORS-INCH-POUND TO SI (METRIC)"
To convert from to multiply by
Length
inch millimeter (mm) 25.4Ef
foot meter (m) 0.30488
Yard meter (m) 0.9244E
mile (statute) kilometer (km) 1.609
Area
square inch square centimeter 6.451
square foot square meter (m2) 0.0929
square yard square meter (m2) 0.8361
Volume (capacity)
ounce cubic centimeter (cm3) 29.57
gallon cubic meter (m3)$ 0.003785
cubic inch cubic centimeter (cm3 16.4
cubic foot cubic meter (m3) 0.02832
cubic yard cubic meter (m3)$ 0.1646
Force
kilogram-force newton (N) 9.807
kip-force newton (N) 4448
pound-force newton (N) 4.448
P m u r e or stress
(force per area)
kilogram-force/square meter pascal (Pa) 9.807
kip-forceísquare inch (ksi) megapascal (MPa) 6.895
newton/square meter (N/m2) pascal (Pa) 1 .WOE
pound-force/square foot pascal (Pa) 47.88
pound-forcdsquare inch (psi) kilopascal (kPa) 6.895
Bending moment or torque

inch-pound-force newton-meter (Nm) 0.1130
foot-pound-force newton-meter (Nm) 1.356
meter-kilogram-force newton-meter (Nm) 9.807
@Seismicisolation
@Seismicisolation
920
To convert from to multiply by
Mass
ounce-mass (avoirdupois) gram (g) 28.34
pound-mass (avoirdupois) kilogram (kg) 0.4536
ton (metric) megagram (Mg) I .000E
ton (short, 2000 Ibm) megagram (Mg) 0.9072
Mass per volume

pound-mass/cubic foot kilogradcubic meter (kg/m’) 16.02
pound-madcubic yard kilogradcubic meter (kg/m3) 0.5933
pound-mass/gallon kilogradcubic meter (kg/m3) 119.8
Temperatures
deg Fahrenheit (F) deg Celsius (C) tC (tF - 32)/1 .8
deg Celsius (C) deg Fahrenheit (F) tp = 1.8k + 32
* This selected list gives practical conversion factors of units found in concrete technology. The reference
source for information on SI units and more exact conversion factors is “Standard for Metric Practice” ASTM E
380. Symbols of metric units are given in parentheses.
t E indicates that the factor given is exact.
# One liter (cubic decimeter) equals 0.001 m’ or 1000 cm’.
8 These equations convert one temperature reading to another and include the necessary scale corrections. To
convert a difference in temperature from Fahrenheit to Celsius degrees, divide by I .8only, Le., a change from 70
to88Frepresentsachangeof 1 8 F o r 18/1.8= 1OC.
@Seismicisolation
@Seismicisolation
A Cerezo, V., 151
abrasion, 497 chloride diffusion, 701
abutments, 681 chloride ion, 279
admixture dosage, 435 circular columns, 133
admixtures, 319,347,593 circular tank, 389
Aguilar, C., 319 closed joint, 415
Akita, H., 405 columns, 65
Alcocer, S . M., 65,531 composite section, 89
Amin, N. R., 207 composite structures, 661
anhydrite, 593 composites, 187
Arenes-Oliva, G. A., 593 compression, 297,733
Arnaud, L., 151 compressive, 609
artificial neural network, 757 compressive strength, 1,593
Astudillo de la Véga, E., 369 concrete, 297, 319, 389,405, 791, 881
Attiogbe, E. K.,701 concrete frame building, 231
Awad, K. W., 435 concrete repair, 497
axial load, 757 concrete temperature, 435
confined concrete, 531
B construction specification, 839
Baron, O., 279 contaminants, 839
Barron, J. M., 641 Corinaldesi, V., 169
basic creep, 461 corrosion, 701
basic shrinkage, 475 corrosion of steel reinforcement, 279
Bayard, O., 369 cracked reinforced concrete, 169
beam-column, 757 creep, 461
bem-column connections, 661, 863
bearing pad, 553 D
Belarbi, A., 717,839 dam hydraulic structures, 497
bending, 297 de Paula, J. E, 279
bent, 133 deformation analysis, 5 13
Bergin, M., 593 degradation of concrete, 279
Bernier, G., 369 density, 53
biaxial tensile behavior, 369 design code, 389
binder, 151 Devalapura, R.K.,905
Bitrán, D., 65 development parcel, 109
boat section ramp, 109 digital image correlation, 369
bond strength, 839 Dimig, J., 269
bridge, 553,621 drying creep, 461
bridges, 133, 187 drying shrinkage, 3 19
building diaphragms, 641 ductility, 89
building evaluation, 207 Duntemann, J., 269
building performance, 23 1 durability, 169
Burak, B., 863 Durán-Hernández, R., 65,531
Buttenvorth, J. W. 89 dynamic analysis, 771
C E
capacity protection, 681 earthquake, 389,553
Carino, N. J., 247 earthquake engineering, 65,771
cement compound composition, 593 earthquake-resistant structures, 661
cementitious materials, 8 11 eccentricity, 863
@Seismicisolation
@Seismicisolation
922 Index
EI Chabib, H., 297 Hong, S. G., 513
EI Naggar, M. H., 297 Hooper, R., 8 11
El-Azazy, S., 133 Horiguchi, T., 53
electrochemical impedance spectroscopy, hospitals, 65
279 hot mixing, 593
epoxy mortar, 497 hot-weather, 47
explosive spalling, 53 hot weather concrete, 347,435
extrapolation, 319 housing, 65
Hueste, M. B. D., 641
F hybrid structures, 661
Fakih, Rabih, 435 hybrid, 791
FEMA 273,641
FEMA 3 10 evaluation, 23 1 I
ferro-silicate slag, 81 1 Idriss, R. L., 621
fiber optic sensors, 621 Imperial Smelting Furnace, 8 11
fiber reinforced concrete, 53 inclusion, 369
fiber reinforced polymer composites, 247 infrared thermography, 247
fiber reinforced polymers, 187, 881 Ingham, J. M., 89
fiber-reinforced, 297 instrumentation, 531
fiber-reinforcement, 79 i insulated sandwich panel, 905
fibres, 369 interaction curve, 757
Figueira, David, 207
final curing, 1 J
finite element modeling, 247 jacketing, 531
fire resistance, 53 Jin, Y., 347
flared, 133 Johnson, B. D., 905
flexural, 609 joint shear deformation, 661
flexure, 733 joint shear distortion, 863
Flores, Leonardo, 65 Joukoski, A., 279,497
fly ash, 1,47, 347
footing flexibility, 553 K
foundation, 269 Kachlakev, D. I., 187
fracture process, 791 Kausel, E. A., 247
freezing and thawing, 29 Kawaguchi, N., 415
Kianoush, M.R., 389
G Kindervater, L., 717
galvanized reinforcement, 169 Knight, Brian T., 23 1
Garcia, Carlos, M., 279 Koide, H., 405
Giaccio, G., 733 Kormann, A. C. M., 497
Gómez-Soberón, J. M.V., 461,475 Kuga, H., 415
Gopalakrishnan, B., 553 Kukes, R. W., 609
gypsum, 593 Kunnath, S. K., 771
H L
Hamidi, M., 389 lapped joint, 415
Han, S. M., 405 Lardner, K., 81 1
Haque, M.E., 757 lateral analysis, 109
heat generation, 435 Lawler, John S., 791
heat loss, 435 Lee, S. G., 5 13
heat of hydration, 47 Lew, H. S., 771
hemihydrate, 593 Liang, X., 661
high early strength, 435 lid structure, 109
high performance concrete, 621,733 life safety performance, 231
high temperature, 1,733 lightweight aggregates, 29
high volume fly ash, 169 lightweight concrete, 29
high-strength, 29 limited ductility, 5 13
high-strength concrete, 53 Lin, T.D., 53
historical buildings, 65 Lin, Y., 89
Hoff, G. C., 29 liquid containing, 389
@Seismicisolation
@Seismicisolation
loading in two principal directions, 863 precast floor system, 571
long haul, 435 prestress losses, 621
López-Bátiz, O., 65, 571 prestressed concrete, 62 1
Luna, R., 717
R
M RC column, 5 13
Mahfouz, I., 881 RCS frames, 661
marine exposure, 29 reactive powder concrete, 369
masonry, 65 recycled aggregate, 461
Mazen, H., 435 recycled concrete aggregate, 461,475
McBnde, M.E, 905 recycled concrete, 475
McGrath, C., 811 recycled-aggregateconcrete, 169
Megally, S., 681 Redlinger, Leanne R.,23 1
Mehta, P. K., 47 reinforced concrete, 65, 269, 531, 757, 839
membrane, 1 reinforcing bars, 839
microfibers, 791 repair and retrofit, 187
Miller, B. D., 701 repair, 65,531, 881
Miltenberger, M.A., 701 residual strength, 415
mineralogy, 717 retarder, 435
mixture proportions, 319 reversed cyclic torsion, 415
modulus of elasticity, 319 Reyes, C., 65
Moffat, B. S., 109 Richardson, D. N., 839
moment resisting frames, 571 Rivera-Torres, J. M., 1
moment, 757 Rivera-Villarreal,R., 1
monitoring, 621 Rizk, T., 881
Moriconi, G., 169 rocking, 553
Momson, C., 811 Rogalla, E., 269
mortar curing, 717
S
N sacrificial shear keys, 681
Nada, H. M., 133 Saeki, N., 53
Nehdi, M., 297 Saiidi, M., 133, 553
non-destructive evaluation (NDE), 621 Sales, A., 279
nondestructive testing, 247 Sanders, D. H., 133
nonlinear analysis, 641 Santos, R.P., 497
nonlinear static procedures, 771 Sarkani, S., 881
notch, 405 schools, 65
secondary flexure, 405
P Seible, E, 68 i
Palmer, S. D., 609 seismic, 133, 389
panel movement, 905 seismic analysis, 641
Parra-Montesinos, G. J., 661 seismic behavior, 863
Pereira, i? N., 497 seismic load, 109
performance-based design, 661,77 1 seismic performance, 68 1
performance evaluation, 641 seismic retrofit, 207
physical properties, 151 self-consolidating,297
pier, 553 Seng, J. M., 905
pile cap, 269 setting time, 435,593
piles, 297 Shah, Surendra P., 791
plastic, 297 shake table, 133,553
Plé, O., 369 shear capacity, 89
polymer-modified concrete, 497 shear friction, 681
polypropylene fibers, 53 shear sliding, 681
polystyrene lightweight concrete, shear strength, 531
89 shear stress, 531
Portella, K. E, 279,497 shear studs, 89
portland cement, 1,47 shives, 151
portland pozzolan cement, 3 19 shrinkage, 475,791
pozzolans, 1 shrinkage by drying, 475
@Seismicisolation
@Seismicisolation
924 Index
shrinkage prediction models, 3 19 thermal effects, 701
shrinkage-reducing admixtures, 3 19 Thompson, C.L., 231
Siddharthan, R.,553 Tittarelli, E, 169
silica fume, 1, 497 Tomon, M., 405
Silva, P.E, 681 torsional rigidity, 415
Silva-Olivera, H., 571 torsional stiffness, 571
slag, 1,347 toughness, 733
slippage, 839 Tso, W. K., 389
slump loss, 593
spectral response, 717 U
spectroradiometer, 7 17 uniaxial tension, 405
spillways, 497 Uniform Building Code, 207
splitting tensile, 609 updating, 319
Stames, M. A., 247 upgrading, 881
statistical evaluation, 3 19 V
steel beam, 89
Videla, C.,319
steel fibers, 733
vegetable particles, 151
steel strain, 553
volumelsurface ratio, 3 19
steel-fiber, 497
story drift, 771 W
strengthening, 65, 231, 531, 881 WIC, 435
stress-strain, 297 wall service loading, 905
structures, 187 Wan, V.,207
strut-and-tiemodel, 269, 513 water retaining, 389
superplasticizer, 1 weather cycles, 905
superplasticizer, 435 White, E. P.,109
sustainable construction, 169 Wight, J. K., 661, 863
T wind load, 109
Winiarski, D. W., 905
Taber, Lawrence H., 839
workability retention, 435
Takano, T. 53
workability, 791
Taylor, William W., 609
temperature rise, 435 Y
tensile strength, 405 Yazdani, N., 347
tension softening, 405 Yoda, T., 415
test procedure, 405
testing, 621 Z
thermal cracking, 47 Zampini, D., 791
thermal deformation, 905 Zayed, A. M., 593
thermal diffusivity, 53 Zerbino, R., 733
thermal durability, 905
@Seismicisolation
@Seismicisolation
@Seismicisolation
@Seismicisolation

@seismicisolation

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

@seismicisolation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

@seismicisolation

Uploaded by

Copyright:

Available Formats

@Seismicisolation

Printed in the United States of America

Editorial production: Bonnie L. Schmick

Library of Congress catalog card number: 2002113749

MARINE EXPOSURE OF HIGH-STRENGTH LIGHTWEIGHT CONCRETE

PERFORMANCE OF HIGH-VOLUME FLY ASH CONCRETE IN HOT WEATHER

EFFECT OF FIBER REINFORCEMENT ON RESIDUAL PROPERTIESOF HIGH-

SHEAR STUD PERFORMANCE OF LIGHTWEIGHT POLYSTYRENE

PARCEL 6 DEVELOPMENT, CENTRAL ARTERYITUNNEL PROJECT, BOSTON,

BEHAVIOR OF FLARED-COLUMN BENTS UNDER SEISMIC LOADING

MECHANICAL, THERMAL, AND ACOUSTICAL PROPERTIES OF CONCRETE

SUSTAINABLEAND DURABLE REINFORCED CONCRETE CONSTRUCTION

EXPERIMENTAL RESULTS AND VALIDATION OF THE DESIGN TECHNIQUE

INNOVATIVE SEISMIC RETROFIT SCHEME FOR 215 FREMONT STREET

QUANTITATIVEINFRARED THERMOGRAPHY FOR QUALITY CONTROL OF

DESIGN OF PILE CAPS WITH STRUT-AND-TIEMODELS

STUDY OF THE INFLUENCE OF CEMENT TYPE, CEMENT CONTENT AND

SELF-CONSOLIDATING CONCRETE PILES CONFINED IN FRP TUBES

ANALYSIS OF SHRINKAGE REDUCING ADMIXTURES ON DRYING

SUBSTITUTION OF FLY ASH, SLAG, AND ADMIXTURES IN FDOT

BIAXIAL TENSILE BEHAVIOR OF THE REACTIVE POWDER CONCRETE

BEHAVIOR OF CONCRETE LIQUID CONTAINING STRUCTURES SUBJECTED

THREE MISUNDERSTANDINGS IN UNIAXIAL TENSION TEST

EVALUATION OF RESIDUAL STRENGTH OF REINFORCED CONCRETE

CREEP OF CONCRETE WITH SUBSTITUTION OF NORMAL AGGREGATE BY

COMPARATIVESTUDY OF THE MECHANICAL PROPERTIES AND

STRUT AND TIE MODELS FOR STRENGTH AND DEFORMATION OF

SEISMIC PERFORMANCE OF A RC BUILDING WITH COLUMNS

SHAKE TABLE STUDIES OF EFFECTS OF FOUNDATION KEXIBILITY ON

EXPERIMENTAL STUDY ON THE BEHAVIOR OF REINFORCED CONCRETE

INFLUENCE OF THE VARIATION OF CEMENT SOURCE ON HOT WEATHER

EVALUATION OF IN-PLACE FLEXURAL STRENGTH CONCRETE USING

MONITORING PRESTRESS LOSSES IN HIGH PERFORMANCE CONCRETE

PERFORMANCE-BASED EVALUATION OF RECTANGULAR RC

BEHAVIOR OF RCS BEAM-COLUMN-SLAB SUBASSEMBLIES UNDER

THERMAL EFFECTS ON CHLORIDE DIFFUSION IN CONCRETE

SPECTRAL RESPONSE OF MORTAR SURFACES DURING CURING

THE PERFORMANCE OF FIBER REINFORCEMENT IN CONCRETES

REINFORCED CONCRETE BEAM-COLUMN DESIGN: AN ARTIFICIAL

ASSESSMENT OF STRUCTURAL SYSTEMS EVALUATION METHODS BASED

HYBRID FIBERS AND THE DEVELOPMENT OF CRACKING IN CONCRETE

FERRO-SILICATE SLAG FROM ISF ZINC PRODUCTION AS A SAND

EFFECT OF REINFORCING BAR CONTAMINATIONON STEEL-CONCRETE

SEISMIC BEHAVIOR OF ECCENTRIC R/C BEAM-COLUMN-SLAB

STRENGTHENING RECTANGULAR CONCRETE COLUMNS USING FRP:

THERMAL CYCLIC TESTING ON A CONCRETE SANDWICH PANEL SYSTEM

Effect of Curing in a High

Kevwords: compressive strength; final curing; fly ash;

in September 2000, the Strategic Development Council of the U. S.

This Vision expects that aggregates quarries, cement manufacturing,

Environmental issues will have a leading role in the sustainable

The appropriate use of a suitable CCM and a HRWRA is expected to

Capillary voids and micro cracks, which are generally connected or

The permeability of cement paste has also been related to concrete

Many researches are going on using ternary cementitious materials to

Several techniques have been applied to study the pore structure of

The aggregates used were crushed limestone from Monterrey, N. L.,