2008 Volume 45, No. 1

Engineering
Journal
American institute of steel Construction
first Quarter 2008 Volume 45, No. 1
1 Message from the editor

3 Acknowledgment
5 Block shear equations revisited...Again
howard i. epstein and Lance J. Aleksiewicz
13 designing Compact Gussets with
the uniform force Method
Larry s. Muir
21 Limit state response of Composite Columns
and Beam-Columns
Part ii: Application of design Provisions
for the 2005 AisC specification
roberto t. Leon and Jerome f. hajjar
47 investigation of flange Local Bending
under flexible Patch Loading
Lyle P. Carden, Gokhan Pekcan and Ahmad M. itani
57 Capacity design of Vertical Boundary
elements in steel Plate shear Walls
Jeffrey W. Berman and Michel Bruneau
73 effects of Nonverticality on steel framing systems—
implications for design
Andrea e. surovek and Justin Johnson
87 Current steel structures research
reidar Bjorhovde
93 suggested reading from other Publishers
96 errata
www.aisc.org
1Q2008EJ_cover.indd 1 4/7/2008 9:20:49 AM

ENGINEERING
JOURNAL
AMERICAN INSTITUTE OF STEEL CONSTRUCTION
Dedicated to the development and improvement of steel construction,

through the interchange of ideas, experiences and data.
Editorial staff
Editor: Cynthia J. Duncan
Associate Editor: Bo Dowswell
Research Editor: Reidar Bjorhovde
Managing Editor: Keith A. Grubb, P.E., S.E.
Production Editor: Areti Gertos
officers
Rex I. Lewis, Chairman
Puma Steel, Cheyenne, WY
David Harwell, Vice Chairman
Central Texas Iron Works, Inc., Waco, TX
Stephen E. Porter, Treasurer
Indiana Steel Fabricating, Inc., Indianapolis, IN
Roger E. Ferch, President
American Institute of Steel Construction, Chicago
David B. Ratterman, Secretary & General Counsel
John P. Cross, Vice President
AISC Marketing, LLC., Chicago
Louis F. Geschwindner, Vice President
Roberta L. Marstellar, Vice President
Scott L. Melnick, Vice President
The articles contained herein are not intended to represent official attitudes,
recommendations or policies of the Institute. The Institute is not responsible for any
statements made or opinions expressed by contributors to this Journal.
The opinions of the authors herein do not represent an official position of the
Institute, and in every case the officially adopted publications of the Institute
will control and supersede any suggestions or modifications contained in any
articles herein.
The information presented herein is based on recognized engineering principles
and is for general information only. While it is believed to be accurate, this
information should not be applied to any specific application without competent
professional examination and verification by a licensed professional engineer.
Anyone making use of this information assumes all liability arising from such use.
Manuscripts are welcomed, but publication cannot be guaranteed. All manuscripts
should be submitted in duplicate. Authors do not receive a remuneration. A “Guide
for Authors” is printed on the inside back cover.
ENGINEERING JOURNAL (ISSN 0013-8029) is published quarterly. Subscribe to Engineering Journal by visiting our web site
Subscriptions: Members: one subscription, $20 per year, included in dues; www.aisc.org or by calling 312.670.5444.
Additional Member Subscriptions: $15 per year. Non-Members U.S., Canada, and
Mexico: $40 per year, $110 for three years, single copy $15. International Members Copies of current and past Engineering Journal articles
and Non-Members: $90 per year; $250 for three years; single copy $25. Published are available free to members online at www.aisc.org/
by the American Institute of Steel Construction at One East Wacker Drive, Suite epubs.
700, Chicago, IL 60601.
Periodicals postage paid at Chicago, IL and additional mailing offices. Non-members may purchase Engineering Journal article
Postmaster: Send address changes to ENGINEERING JOURNAL in care of downloads at the AISC Bookstore at www.aisc.org/
the American Institute of Steel Construction, One East Wacker Drive, Suite 700, bookstore for $10 each. Starting with the First Quarter
Chicago, IL 60601. 2008, complete issue downloads of Engineering Journal
Copyright 2008 by the American Institute of Steel Construction. All rights reserved. are available for $15 each at www.aisc.org/bookstore.
No part of this publication may be reproduced without written permission.
An archival CD-ROM of past issues of Engineering
Journal is available by calling 800.644.2400.
1Q2008EJ_cover.indd 2 4/7/2008 9:20:49 AM

MESSAGE FROM THE EDITOR:
As you may have noticed, since the completion of the 2005 AISC Specification for Structural Steel Buildings and publication
of the 13th Ed. AISC Steel Construction Manual, we have had issues of the Engineering Journal containing articles providing
background and discussion of topics contained in those AISC documents. This quarter there is one additional article that will
provide background and insight into the provisions for composite columns and beam-columns in Chapter I of the 2005 AISC
Specification. This article entitled, “Limit State Response of Composite Columns and Beam-Columns, Part II: Application of
Design Provisions for the 2005 AISC Specification” is a sequel to Part I published in the 4th Q 2007 issue. As before, we have
identified this article with the header, Spec/Manual Reference.
Cynthia J. Duncan
Editor
ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 1
001-004_editor_acknow_2008_1Q.indd 1 3/27/08 12:16:34 PM

Acknowledgment
All AISC Engineering Journal articles are peer reviewed prior to publication for accuracy, content, and style. AISC thanks the
following engineers for their voluntary review assistance to the Engineering Journal Review Board throughout 2007.
Farid Alfawakhiri Maria Garlock Theodor Krauthammer C. Mark Saunders

American Iron and Steel Institute Princeton University University of Florida Rutherford & Chekene Consulting
Engineers
Barry L. Barger Louis F. Geschwindner Geoffrey L. Kulak
Southern Iron Works, Inc. American Institute of Steel University of Alberta Stephen P. Schneider
Construction Kramer Gehlen & Associates Inc.
Michael Barker Keith Landwehr
University of Wyoming Subhash C. Goel Schuff Steel Company Benjamin Schafer
University of Michigan Johns Hopkins University
Jeffrey W. Berman Michael E. Lederle
University of Washington John L. Gross Opus Architects & Engineers, Inc. Robert E. Shaw
National Institute of Standards and Steel Structures Technology Center,
Omer W. Blodgett Judy Liu
Technology Inc.
The Lincoln Electric Company Purdue University
Michael A. Grubb Donald R. Sherman
Michael Brennan LeRoy A. Lutz
Bridge Software Development University of Wisconsin, Milwaukee
LD Astorino Computerized Structural Design,
International, Ltd.
S.C. James A. Swanson
Roger L. Brockenbrough Jerome F. Hajjar University of Cincinnati
R.L. Brockenbrough & Associates, James O. Malley
University of Illinois at Urbana-
Inc. Degenkolb Engineers William A. Thornton
Champaign
Cives Engineering Corporation
Michel Bruneau William McGuire
Ronald O. Hamburger
University of Buffalo Cornell University Raymond H.R. Tide
Simpson Gumpertz & Heger Inc.
Consulting Engineers Wiss Janney Elstner Associates, Inc.
Michael P. Culmo David L. McKenzie
CME Associates, Inc. S.P. International, Inc. Emile W.J. Troup
Linda M. Hanagan
Pennsylvania State University Consultant
Robert Driver Ronald Meng
University of Alberta Lynchburg Steel Company, LLC Chia-Ming Uang
Allen J. Harrold
Butler Manufacturing Company University of California, San Diego
Carol J. Drucker James A. Milke
Drucker Zajdel Structural University of Maryland Elmar Upitis
Richard A. Henige
Engineers, Inc. Upitis & Associates, Inc.
LeMessurier Consultants, Inc. Heath E. Mitchell
Donald O. Dusenberry PCS Structural Solutions Asif Usmani
Steve Herth
Simpson Gumpertz & Heger, Inc. University of Edinburgh
Contech Bridge Solutions, Inc. Michael G. Moffitt
Consulting Engineers
W. Steven Hofmeister National Steel Bridge Alliance Brian Uy
W. Samuel Easterling University of Western Sydney
The Thornton-Tomasetti Group, Inc. Larry Muir
Virginia Polytechnic Institute &
State University John D. Hooper Cives Steel Company Jules Van de Pas
Magnusson Klemencic Associates Computerized Structural Design,
Bruce R. Ellingwood Thomas M. Murray S.C.
Georgia Institute of Technology Ahmad M. Itani Virginia Polytechnic Institute &
University of Nevada State University Amit H. Varma
Sherif El-Tawil Purdue University
University of Michigan Nestor Iwankiw R. Shankar Nair
Hughes Associates Teng & Associates, Inc. Robert D. Weber
Don Engler R.D. Weber & Associates, Inc.
BDS Steel Detailers, Inc. Matthew A. Johann Clarkson W. Pinkham
Ove Arup & Partners S.B. Barnes Associates Michael A. West
Mohammed M. Ettouney Computerized Structural Design,
Weidlinger Associates Inc. Richard C. Kaehler Clinton O. Rex S.C.
Consulting Engineers Computerized Structural Design, Stanley D. Lindsey & Associates
S.C. Donald W. White
John W. Fisher Kimberly Robinson Georgia Institute of Technology
Lehigh University Lawrence A. Kloiber Star Seismic
LeJeune Steel Ronald Ziemian
Theodore V. Galambos Charles W. Roeder Bucknell University
University of Minnesota Dean Krause University of Washington
Consultant

Block Shear Equations Revisited…Again
HOWARD I. EPSTEIN and LANCE J. ALEKSIEWICZ
S hortly after block shear was first identified as a possible

failure mode for coped beam connections, design equa-
)
Pbs = ( 0.3 Anv + 0.5 Ant Fu (1)
tions to account for it were incorporated into allowable stress where

design (ASD) provisions. These equations never changed, Fu = ultimate strength of the material
partly due to ASD not being updated since 1989. However, Anv = net shear area
load and resistance factor design (LRFD) treatment of block Ant = net tensile area
shear changed with each new Specification. Over the years, Over time, this concept has become more broadly applied
it was suggested that the effect of eccentricity was miss- to many other connection applications. This equation (with
ing from block shear equations. On the surface it appears the symbols Av and At used for net shear and tension areas,
that the effect of eccentricity on the block shear strength respectively) was first incorporated into allowable stress de-
of connections, as suggested by previous investigators, has sign (ASD) in the 1978 AISC Specification (AISC, 1978).
now been incorporated into the latest unified Specification. The 1989 AISC Specification (AISC, 1989), the last revision
For many connections, however, nothing has changed. It is of ASD, included the same provisions for block shear.
the conclusion of this paper that additional important cases The block shear equations in the load and resistance fac-
need to be shown in Commentary Figure C-J4.2 of the 2005 tor design (LRFD) specifications have changed with each
AISC Specification for Structural Steel Buildings (AISC, edition. Both the first edition (AISC, 1986a) and the second
2005), hereafter referred to as the AISC Specification, for edition (AISC, 1993) contain two equations for the deter-
which block shear equations now incorporate a new factor to mination of the block shear rupture design strength, φRn,
account for connection eccentricity. In particular, as a mini- given by
mum, angles connected by only one leg or tees connected by
their flanges should also be included with other connections φRn = φ [0.6 Fy Agv + Fu Ant ] (2a)
for which block shear capacities are now reduced.
BACKGROUND: CODE TREATMENT UNTIL 2005 and φRn = φ[0.6 Fu Anv + Fy Agt ] (2b)
In 1978, destructive tests on coped beams with bolted web where, in addition to the symbols in Equation 1,
connections were performed with some exhibiting what has φ = 0.75
become known as block shear as the failure mode (Birkemoe Fy = specified minimum yield strength
and Gilmor, 1978). They proposed a design equation in which Agv = gross area subject to shear
the ultimate shear strength is applied to the net shear area Agt = gross area subject to tension
and ultimate tensile strength to the net tension area. Block
shear occurs when the web, for this case, develops its ulti-
mate strength along the perimeter bolt holes and a “block”
of this web begins to fracture. Figure 1a shows this block
shear path for a coped beam. The equation that Birkemoe
and Gilmor proposed was:
Howard I. Epstein is professor, department of civil and

environmental engineering, University of Connecticut,
Storrs, CT.
Lance J. Aleksiewicz is structural engineer, Odeh (a) Coped Beam (b) Tension Connection
Engineers, Inc., North Providence, RI.
Fig. 1. Block shear failure paths (AISC, 2005).
005-012_Epstein_Aleksiewicz_2008_1Q.indd 5 3/27/08 12:17:45 PM

Equation 2a represents block shear strength determined by The third edition of the AISC LRFD Specification (AISC,
rupture on the net tensile section combined with shear yield- 1999) used the following equations for the determination of
ing on the gross section on the shear plane(s). Equation 2b the block shear rupture design strength, φRn:
represents block shear strength determined by rupture on
the net shear area(s) combined with yielding on the gross φRn = φ[0.6 Fy Agv + Fu Ant ] ≤ φ[0.6 Fu Anv + Fu Ant ]
tensile area. These equations are based on the work of Ricles
and Yura (1983) as well as that of Hardash and Bjorhovde
when Fu Ant ≥ 0.6 Fu Anv (3a)
(1985). Except for slight differences in notation, these equa-
tions did not change from the first to the second edition of
the AISC LRFD Specification, but, for some connections, and
which equation governed did change.
In the first edition of the LRFD Specification (AISC, φRn = φ[0.6 Fu Anv + Fy Agt ] ≤ φ[0.6 Fu Anv + Fu Ant ]
1986a), these equations were found only in Chapter J of the
Commentary where it stated that “the controlling equation is when 0.6 Fu A nv > Fu Ant (3b)
one that produces the larger force.” The Commentary went
on to explain that since block shear is a fracture or tearing In Equation 3a, the term 0.6Fy Agv shall not be taken greater
phenomenon and not a yielding limit state, the proper formu- than 0.6FuAnv, and in Equation 3b, the term Fy Agt shall not be
la is the one in which the fracture term is larger than the yield taken greater than Fu Ant. These two provisions didn’t exist in
term. For ductile steels where Fu is considerably larger than the second edition LRFD. Equation 3a is the equivalent of
Fy, this may be true for both equations. For steels having Fy using yielding of the shear plane and rupture of the tension
as an appreciable portion of Fu, this may not be true for ei- plane. The additional provision, however, requires that the
ther equation. The Commentary went on to state that “where yield strength of the gross shear area be less than the ulti-
it is not obvious which failure plane fractures, it is easier just mate strength of the net shear area. Conversely, Equation 3b
to use the larger of the two formulas.” In fact, the tables in implies rupture of the shear plane and yielding of the tension
the first edition of the LRFD Manual (AISC, 1986b) stated plane. Similar to Equation 3a, however, the third edition lim-
that the equation to be used is the one producing the larger ited the yield strength of the gross tension area to less than or
block shear strength. equal to the ultimate strength of the net tension area.
The block shear equations in the second edition of the AISC
LRFD Specification (AISC, 1993) were found in Chapter J BACKGROUND: COPED BEAMS VERSUS
of the Specification, as opposed to the Commentary. While ANGLE TENSION MEMBERS
the formulas were the same, the change in the second edition
was contained in a check of the relative fracture strengths, “The block shear failure mode is not limited to the coped
Fu Ant, as compared to 0.6Fu Anv, or since Fu is common to ends of beams.” This statement first appeared in the first
both terms, Ant compared to 0.6Anv. The Specification then edition of the LRFD Specification (AISC, 1986a) and then,
stated that when Fu Ant ≥ 0.6Fu Anv, use Equation 2a and when subsequently, in the 1989 ASD Specification (AISC, 1989).
Fu Ant < 0.6Fu Anv, use Equation 2b. The Commentary stated The examples shown in the Commentary included tension
that “the proper equation to use is the one with the larger connections for angles as well as gusset plates (see Figure
rupture term.” 1b). Prior to their inclusion, some argued that structural
In both the first and second LRFD editions, the Commen- engineers should have recognized block shear as a possible
tary gave two extreme examples showing which of the two failure mode for angles, despite the fact that many textbook
limiting states (shear yield/tension fracture or shear fracture/ examples did not consider block shear for angles in tension,
tension yield) was appropriate. One of the examples had a even when it clearly was the governing failure mode.
tension area much larger than the shear area while the other It was some angles found in the wreckage of the Hart-
example reversed these areas. The later interpretation of us- ford Civic Center roof, which had failed in block shear, that
ing the limiting state with the larger rupture term was cer- served as the impetus to investigate block shear in angles.
tainly justified on the basis of these examples. The same Initial finite element investigations (Epstein and Thacker,
could not easily be said of the earlier treatment. 1991) were able to accurately determine block shear as the
The first edition LRFD Specification specified to always failure mode for these Civic Center angles. That study also
use the larger strength found from Equation 2a or 2b, while investigated various similar geometries. Most importantly,
the second edition either led the designer to the same equa- the study also showed that there probably should be a sub-
tion or possibly to the equation that yields the smaller design stantial difference in the way in which block shear is treated
strength. The effect was that, for some connection geome- for coped beams (where the load is applied to the connection
tries, the second edition produced more conservative results in the plane of the web, which is also the block shear path)
(Epstein, 1996a).
6 / ENGINEERING JOURNAL / FIRST QUARTER / 2008

versus angles (where the load is applied eccentric to the For LRFD, U could similarly be a factor for each tension
failure plane). The first edition LRFD Specification (AISC, area. Even for the simple case of pure tension fracture, not
1986a) included block shear equations in the Commentary block shear, of a member with eccentricity, such as an angle,
(pages 186–188) where angles were shown. U is used to calculate an effective net area. Another way of
Extensive tests of angles to determine effects of connec- treating the behavior is to think of it as combined tension
tion geometries and eccentricities were conducted at the Uni- and bending. Using AISC interaction equations, an equiva-
versity of Connecticut (Adidam, 1990; Epstein, 1992). The lent reduction factor can be found (Epstein and D’Aiuto,
tested specimens were back-to-back angles with two rows 2002). Block shear is usually initiated with tension fracture.
of bolts in one leg of each angle. Among the specimens, the At the point where this occurs, the tension stresses, resulting
connection geometry was varied by using staggered and un- from the load eccentricity, are more than the average load
staggered bolt patterns, and alternating the position of the divided by the net area. Incorporating U into the block shear
lead bolt in the staggered bolt patterns from the inside gage equations produces far more satisfactory results, not only for
line to the outside gage line. It was concluded that the AISC angles, but for flange connected tee sections as well (Epstein
ASD and LRFD Specifications provided inadequate safety and Stamberg, 2002).
factors for the block shear failure of angles. To alleviate this Before proceeding further, it should be pointed out that
problem, it was suggested that the shear lag factor, U, be there has been disagreement with the idea that the block
included in the tension terms of the code equations for block shear equations need modification to account for eccentric-
shear capacity. ity. Grondin (2005) wrote that, “Although angles in tension
The U factor, in essence the efficiency of the connection, represent a different case since both in-plane and out-of-
is used to reduce the net area of angles in tension because plane eccentricities are present, the writer believes that these
not all of the tension area is effective in carrying the load. eccentricities are sufficiently small and the ductility of steel
As the length of the outstanding leg increases, the eccentric- sufficiently large to minimize the effect of these eccentrici-
ity increases, and U decreases. The study of angles showed ties on the block shear capacity of angles in tension.” Anoth-
that as the outstanding leg increased in size, the capacity of er study (Kulak and Grondin, 2001) concluded that, “Further
the angle in net tension and the block shear capacity was re- research is required to investigate the effect of out-of-plane
duced. These conclusions were also verified using nonlinear eccentricity on block shear failure.” However, no change was
finite element models (Epstein and Chamarajanagar, 1996). recommended in the approach for angles, in part, because 12
The results justified applying the correction factor, U, to the of the 15 block shear failures previously documented (Ep-
block shear design equations. Another study investigated stein, 1992) involved connections with staggered gage lines.
the efficiency of angles in tension rolled from high strength Stagger, however, was shown not to have an appreciable ef-
steel (Gross, 1994). Gross contended that the block shear fect on capacity.
equations for ASD and LRFD were not sufficient for high
strength steel. BACKGROUND: TEES AND OTHER SECTIONS
While the LRFD equations for block shear changed with
A limited number of tests on tee sections in tension produced
each new edition, the changes did little to address the under-
surprising results (Epstein, 1996b). The tees were connected
lying problem of the eccentricity of the load to the plane in
through the flange and the stem was the outstanding element.
which block shear occurs. Over the years, several research-
A previously undocumented alternate failure path in block
ers have suggested modifications of the block shear equa-
shear was discovered during the testing (see Figure 2). This
tions to account for the reduction in capacity with increasing
led to a full investigation of structural tees in tension. Finite
eccentricity. As was suggested previously (Adidam, 1990;
element analyses were then performed, and they supported
Epstein, 1992), the simple empirical incorporation of the
this new block shear failure mode of structural tees in ten-
“shear lag” factor, U, into the tension path for block shear is
sion (Epstein and McGinnis, 2000). The results compared
probably all that is needed. For ASD, the resulting equation
very well with previous tests in replicating the newly dis-
would be
covered alternate block shear failure path. It was found that
)
Pbs = ( 0.3 Anv + 0.5UAnt Fu (4) the moments caused by the eccentricity were not equal to
the axial load multiplied by the eccentricity. The reactions at
where the first and last bolts along a line of bolts created an oppos-
U = 1 ⫺ x–/ l (5) ing moment, which reduced the moment caused by the ec-
centricity. Again, it was concluded that the shear lag factor,
where
U, should be incorporated in the tension terms of the block
x– = connection eccentricity
shear equations.
l = connection length

A series of tests on 50 structural tees in tension were per- the block shear capacity of a connection. This equation is
formed (Epstein and Stamberg, 2002). The specimens were certainly similar to what was proposed previously (Epstein,
bolted through the flange, with the stem outstanding, and 1992). Note that shear rupture-tension yield is no longer
had varying eccentricities and connection lengths. The ge- considered. Many past research studies noted that block
ometries of the specimens were chosen such that a progres- shear failures usually initiated with a tension fracture.
sion from net section failure to block shear failure would The definition and limits for the variable, Ubs, went through
occur. The tests exhibited behavior that matched what was revisions during the development of the 2005 Specification.
indicated by finite element models. As connection lengths In an earlier treatment, the following equation was used to
decreased or eccentricities increased, or both, the efficien- calculate Ubs:
cies of the connections decreased. The previously obtained
Ubs = 1, if e/l ≤ 3 (7a)
theoretical results (Epstein and D’Aiuto, 2002) for these tests
were compared to the values calculated using the then cur- or
rent (second edition) Specification. Even though the theory Ubs = 1 – e/l , if e/l > 3 (7b)
was originally based on block shear in tees, it was found to
where
agree with the net tension failure results as well. In addition
e = eccentricity of the force tending to cause block
to incorporating a reduction factor in block shear equations,
shear rupture in the plane of the connection
it was also suggested that, as a simplified design approach,
faying surface
a reduced lower bound for the shear lag factor, U, may be
l = length of the block subject to block shear
appropriate for net section failures.
rupture
THE 2005 SPECIFICATION TREATMENT The variable e is equivalent to x– in Equation 5. Note that,
however, l is not the same as in the Equation 5. The differ-
The shear lag coefficients have been significantly reduced
ence is that the term l in Equations 7a and 7b is equal to the
in the 2005 AISC Specification for Structural Steel Build-
length of the block subject to block shear and in the equation
ings, which combines ASD and LRFD into one document
for U it is equal to the length of the connection. Therefore,
(AISC, 2005). Block shear design capacity is given by the
for the same connection, Ubs will be larger than U because
basic equation for tension rupture-shear yield as
the length of the block subject to block shear will be greater
than the length of the connection for standard connections.
φRn = φ[0.6 Fy Agv + U bs Fu Ant ] (6)
In a subsequent treatment (which became the one adopted
in the 2005 Specification) it was stated that Ubs is equal to
In Equation 6, the term 0.6Fy Agv shall not be taken greater 1.0 when the tension stress is uniform, and equal to 0.5 when
than 0.6Fu Anv, which represents the rupture-rupture limit the tension stress is nonuniform. This Ubs does not vary ac-
state. This, in essence, is Equation 3a with the exception cording to any parameter of a connection, such as connec-
of Ubs. Ubs is a new term that was added to the block shear tion length or eccentricity. This definition certainly simpli-
design equation to account for the effect of eccentricity on fies calculations.
Fig. 2. Block shear failure path for tee sections.

THE EFFECT OF THE BLOCK SHEAR REVISIONS to shear lag. Also, eccentricity causes bending which creates
ON DESIGN CAPACITIES nonuniform tensile stresses. Any connection that has either
of these characteristics (angles have both) should, therefore,
In this section, the data from previous research on angles at
use the value of 0.5 for Ubs. The capacities for the 2005 Spec-
the University of Connecticut (Epstein, 1992) will be used
ification in Table 1 include 0.5 for Ubs.
to examine the effect of the latest revision on block shear
Figure 3 shows the relationship between PF (professional
capacities (Equation 6). The majority (35) of the 38 geom-
factor, which is the ratio of the failure load determined by
etries tested in the 1992 study showed block shear failures
destructive testing divided by the corresponding design ca-
and only those are used herein for the comparisons. Since,
pacity based on measured material and geometric properties)
with the 2005 Specification, ASD and LRFD treatments have
versus 1 − e/l for the 1999 and 2005 Specifications for the
coalesced, only LRFD will be presented in the following
data in Table 1. Least square “trend lines” (straight line fits
(Aleksiewicz, 2004).
of the data) are used for comparison. The 1999 data is similar
Table 1 shows the material properties and geometries of
to that of previous Specifications in that many Professional
the single angle specimens that were tested in the 1992 in-
Factors are below 1.0 and as the eccentricity increased or
vestigation (Epstein, 1992). In this table, specimen numbers
the connection length decreased, the trend for Professional
are the same as used in 1992 and material properties of yield
Factors decreased.
and ultimate were as measured. The bolt patterns refer to
If the earlier treatment for Ubs in Equation 7 were used,
the number of bolts in the two gage lines and whether or
there would be no difference in results from 1999 except for
not they were staggered (+ or – denotes stagger and which
specimens #1, 3, 5, and 6, shown in Table 1, because only
gage line had the lead bolt). For more specific details of the
these had values of 1 – e/l < 0.67. In Figure 3, because the
specimens, the reader is referred to the original paper (Ep-
factor Ubs = 0.5 has been applied to all block shear tension
stein, 1992).
areas for the 2005 data, clearly all professional factors (PF)
Also presented in Table 1 are the capacities, as calculated
shown are increased when compared to the 1999 data. The
from the third edition of the LRFD Specification (LRFD,
trend line for 2005 clearly shows a marked improvement.
1999), as well as those from the latest Specification (AISC,
So, it appeared that the approach settled upon for block
2005). Equation 6, from the 2005 Specification, uses Ubs
shear had been satisfactorily addressed. However, in the
either equal to 1.0 (for uniform tensile stresses) or 0.5 (for
2005 Specification Commentary Section J4.3 on block shear
nonuniform tensile stresses). Therefore, Ubs is determined by
(see Figure 4), it becomes readily evident that there is no
the type of connection, not the specific parameters of a con-
change for angles because “angle ends” are shown in the
nection such as connection length (l) or the amount of ec-
category of Ubs = 1.0 (Figure 4a).
centricity (e or x–). Tension members not connected through
It is not apparent if there is a change for tees, which also,
all of their elements develop nonuniform tensile stresses due
as previously noted, have capacities (net tension and block
Fig. 3. 1998 LRFD vs. 2005 (with Ubs = 0.5) block shear treatment for tested angles.

Table 1. Geometry, Material Properties and Block Shear Capacities
Bolt Fy Fu Ptest 1999 LRFD 2005 LRFD

# Member 1 − e/l
Pattern ksi ksi kips kips PF kips PF
1 6X6XCc 2/2+ 51.9 73.9 0.640 182.5 157.9 0.87 111.8 1.22
2 6X6XCc 2/2- 51.4 77.0 0.730 204.2 182.8 0.84 134.8 1.14
3 6X6XCc 2/2 51.0 75.5 0.640 188.7 150.8 0.94 107.7 1.31
4 6X6XCc 2/3- 53.0 77.2 0.820 242.7 208.1 0.87 160.0 1.14
5 6X6XCc 3/2+ 49.3 73.6 0.640 204.9 175.1 0.88 129.1 1.19
6 6X6XCc 2/3 51.4 75.0 0.640 259.7 202.4 0.96 159.6 1.22
7 6X6XCc 3/3 51.6 74.8 0.784 237.1 194.3 0.92 151.6 1.17
9 6X4XCc 2/2+ 51.0 72.4 0.796 202.7 154.9 0.98 109.7 1.39
10 6X4XCc 2/2- 46.8 68.2 0.847 203.9 164.1 0.93 121.5 1.26
11 6X4XCc 2/2 50.3 71.0 0.796 194.2 144.8 1.01 104.2 1.40
12 6X4XCc 2/3- 55.5 80.0 0.898 247.1 216.9 0.85 167.0 1.11
13 6X4XCc 3/2+ 50.5 70.2 0.796 189.1 172.8 0.82 129.0 1.10
14 6X4XCc 2/3 49.4 68.9 0.796 219.8 192.2 0.86 152.8 1.08
15 6X4XCc 3/3 46.5 64.9 0.878 218.6 171.1 0.96 134.0 1.22
17 6X3½XCc 2/2+ 48.3 74.5 0.830 198.2 154.1 0.96 107.6 1.38
18 6X3½XCc 2/2- 52.5 76.6 0.873 198.8 184.2 0.81 136.4 1.09
19 6X3½XCc 2/2 52.1 78.2 0.830 199.3 155.3 0.96 110.6 1.35
20 6X3½XCc 2/3- 50.3 68.5 0.915 238.5 187.8 0.95 145.1 1.23
21 6X3½XCc 3/2+ 49.5 69.4 0.830 216.1 170.1 0.95 126.8 1.28
22 6X3½XCc 2/3 48.0 69.1 0.830 250.6 191.4 0.98 151.9 1.24
23 6X3½XCc 3/3 45.6 69.3 0.898 236.5 175.4 1.01 135.8 1.31
25 5X5XCc 2/2+ 44.3 62.0 0.696 154.1 114.0 1.01 84.8 1.36
26 5X5XCc 2/2- 44.6 61.5 0.772 155.8 133.2 0.88 104.2 1.12
27 5X5XCc 2/3- 45.1 63.2 0.848 194.9 153.9 0.95 124.2 1.18
28 5X5XCc 3/2+ 50.4 70.1 0.696 169.6 151.1 0.84 118.1 1.08
29 5X3½XCc 2/2+ 47.9 71.6 0.814 174.1 128.0 1.02 94.3 1.38
30 5X3½XCc 2/2- 45.0 67.8 0.860 171.8 139.8 0.92 107.9 1.19
31 5X3½XCc 2/3- 45.2 68.2 0.907 208.8 159.6 0.98 127.5 1.23
32 5X3½XCc 3/2+ 48.8 72.6 0.814 189.9 150.7 0.94 116.5 1.22
33 5X3XCc 2/2+ 42.5 59.4 0.849 149.4 109.2 1.03 81.2 1.38
34 5X3XCc 2/2- 43.1 61.0 0.887 161.5 130.2 0.93 101.5 1.19
35 5X3XCc 2/3- 42.5 62.6 0.924 187.2 148.6 0.94 119.1 1.18
36 5X3XCc 3/2+ 42.2 61.1 0.849 163.0 128.7 0.95 100.0 1.22
37 5X3XCc 1/2- 46.1 65.4 0.849 173.3 119.9 1.08 89.1 1.46
38 5x3XCc 2/1+ 44.1 61.8 0.773 126.8 95.4 1.00 66.3 1.43

shear) that are significantly reduced when only the flange is beams where there is one row of bolts (as shown in Figure 4a
connected. Further, since W shapes connected only by the wherein Ubs is 1.0). For coped beams with two rows of bolts
flanges are usually treated as half tees, one would assume (as shown in Figure 4b wherein Ubs is 0.5), they proposed
that their block shear strength would also be compromised. 0.3. In reality, the factor used for the single row coped beam
It is not clear what value of Ubs should be used for this case. connection is not that critical since the tension term is only a
A User Note in Section J4.3 of the 2005 Specification fraction of the shear term. For a two bolt connection (in one
states that, “The cases where Ubs must be taken equal to 0.5 row) on a coped beam, the tension term is as large a percent-
are illustrated in the Commentary”(Figure 4 herein). So, are age of the total as possible. But even then, the reduction in
only coped beam connections having significant in-plane ec- capacity would only be approximately 15% if Ubs = 0.5 in-
centricities affected by the new treatment for block shear? stead of the 1.0 factor given in the 2005 AISC Specification.
There is a recent study of the treatment of block shear For longer connections (three or more bolts), this decrease
equations that are presented in various design standards is much less.
(Driver, Grondin and Kulak, 2006). In this study, several What about angles (and tees)? Even for one row of bolts
standards were compared for many tests of gusset plates, an- in an angle or the stem of a tee, it was reported (Orbison,
gles, tees and coped beams. It was concluded that, except for Wagner and Grondin, 1999) that the smaller the edge dis-
gusset plates, there should be a reduction factor (comparable tance, the smaller the professional factor. Their tests were for
to Ubs) applied to the tension term of the block shear equa- fairly long connections (four bolts) in only one row. There-
tion. (It was also proposed to modify the shear term with a fore, the tension terms for these tests are a small percentage
combination of yield and ultimate stresses). of the shear term. Even so, reduced edge distance reduced
In particular, for coped beams, Driver et al. (2006) pro- the professional factors, somewhat. The reason for this is
posed a reduction factor of 0.9 for the tension term in coped fairly obvious. It is not that the stresses along the tension
path are nonuniform; it is that their magnitude is increased
because of the greater eccentricity of the load and, therefore,
the increased effect of the resulting moment on the tension
stresses. With only one row of bolts, however, similar to
coped beams, it can be argued that using Ubs = 1.0 (as shown
in Figure 4a) is satisfactory since, again, the tension term is
a small fraction of the total.
But what happens with two rows of bolts in one leg of an
angle? Note that all of the tests shown in Figure 3 had two
rows of bolts. Further note that most of these tests were not
included in the study of Driver et al. (2006) because there
was stagger between the gage lines. Clearly, the tension term
for these tests represents a larger percentage of the block
shear capacity than when there is only a single row of bolts.
So, the specifying of Ubs as significantly less than 1.0 ap-
pears to be appropriate.
As far as tees are concerned, Driver et al. show a reduction
factor of 0.9. However, it does not appear that this was meant
to apply to both situations of web-(referred to as “stem” by
Driver et al.) connected and flange-connected tees because
they do not include the latter in their list of investigations.
For stem-connected tees, there is in-plane eccentricity, but
for flange-connected tees there is out-of-plane eccentricity
and the block shear failure for these usually extends into
the web. For the flange-connected case, based on many tests
(Epstein and Stamberg, 2002), it does not appear that the
proposed reduction is sufficient. For tees that fail in block
shear along the alternate path, the tension term is a signifi-
cant fraction of the total block shear. Driver et al. did not
report on these block shear failures when presenting their
Fig. 4. Values of Ubs as shown in Figure C-J4.2 of the 2005 AISC “unified” block shear equation.
Specification Commentary (AISC, 2005).

CONCLUSIONS Driver, R.G., Grondin, G.Y. and Kulak, G.L. (2006), “Uni-
Block shear treatment by AISC has undergone several revi- fied Block Shear Equation for Achieving Consistent Re-
sions since 1978, where it first appeared, through the lat- liability,” Journal of Constructional Steel Research, Vol.
est, 2005 equations. In the opinion of the authors, either the 62, Issue 3, pp. 210–222.
definition of Ubs must change or the figures in the Commen- Epstein, H. (1992), “An Experimental Study of Block Shear
tary need to include more cases where Ubs is not equal to Failure of Angles in Tension,” Engineering Journal, AISC,
1.0. As a minimum, two gage lines in one leg of an angle Vol. 29, No. 2, 2nd Quarter, pp. 75–84.
as well as tees connected by their flanges and, probably, W Epstein, H.I. and Chamarajanagar, R. (1996), “Finite Ele-
sections connected by their flanges should be added to the ment Studies for Correlation with Block Shear Tests,”
cases in Figure 4b. The statement in the 2005 Specification Computers and Structures, Vol. 61, No. 5, pp. 967–974.
that, “Where the tension stress is uniform, Ubs = 1.0; Where Epstein, H.I. (1996a), “Effects of the Latest LRFD Block
the tension stress is nonuniform, Ubs = 0.5 in., in conjunc- Shear Code Change,” Engineering Journal, AISC, Vol.
tion with the limited cases shown in the Commentary, is not 33, No.1, 1st Quarter, pp. 30–33.
particularly helpful in pointing to cases where the use of
Ubs = 1.0 may not be conservative. It is not that the tension Epstein, H. (1996b), “Block Shear of Structural Tees in Ten-
stresses need to be nonuniform, but the magnitude of the sion - Alternate Paths,” Engineering Journal, AISC, Vol.
stresses, as influenced by the eccentricity of the load, is what 33, No. 4, 4th Quarter, pp. 147–152.
reduces the efficiency of the connection. Epstein, H. and D’Aiuto, C. (2002), “Using Moment and
Axial Interaction Equations to Account for Moment and
REFERENCES Shear Lag Effects in Tension Members,” Engineering
Adidam, N. (1990), “Analysis of Block Shear Experiments Journal, AISC, Vol. 39, No. 2, 2nd Quarter, pp. 91–99.
for Structural Steel Angles in Tension,” Thesis presented Epstein, H. and McGinnis, M. (2000), “Finite Element Mod-
to the University of Connecticut, in partial fulfillment of eling of Block Shear in Structural Tees,” Computers and
the requirements for the degree of Master of Science. Structures, Vol. 77, No. 5, pp. 571–582.
Aleksiewicz, L.J. (2004), “A Review of the Capacity of Epstein, H. and Thacker, B. (1991), “The Effect of Bolt Stag-
Angles in Tension,” Thesis presented to the University of ger for Block Shear Tension Failures in Angles,” Comput-
Connecticut, in partial fulfillment of the requirements for ers and Structures, Vol. 39, No. 5, pp. 571–576.
the degree of Master of Science. Epstein, H. and Stamberg, H. (2002), “Block Shear and Net
AISC (1978), Specification for the Design, Fabrication and Section Capacities of Structural Tees in Tension: Test Re-
Erection of Structural Steel for Buildings, American Insti- sults and Code Implications,” Engineering Journal, AISC,
tute of Steel Construction, Inc., Chicago, IL. Vol. 39, No. 4, 4th Quarter, pp. 228–239.
AISC (1986a), Load and Resistance Factor Design Specifi- Grondin, G. (2005), Discussion of “Using Moment and
cation for Structural Steel Buildings, American Institute Axial Interaction Equations to Account for Moment and
of Steel Construction, Inc., Chicago, IL. Shear Lag Effects in Tension Members,” Engineering
AISC (1986b), Manual of Steel Construction, Load and Re- Journal, AISC, Vol. 42, No. 1, 1st Quarter, pp. 45–50.
sistance Factor Design, 1st Edition, American Institute of Gross, J.M. (1994), “A Study of Block Shear Failure in Bolt-
Steel Construction, Inc., Chicago, IL. ed Connections in High-strength Steel Angles,” Master of
AISC (1989), Specification for the Design, Fabrication and Science Thesis, Bucknell University, Lewisburg, PA.
Erection of Structural Steel for Buildings, American Insti- Hardash, S. and Bjorhovde, R. (1985), “New Design Criteria
tute of Steel Construction, Inc., Chicago, IL. for Gusset Plates in Tension,” Engineering Journal, AISC,
AISC (1993), Load and Resistance Factor Design Specifica- Vol. 22, No. 2, 2nd Quarter, pp. 77–94.
tion for Structural Steel Buildings, American Institute of Kulak, G.L. and Grondin, G.Y. (2001), “AISC LRFD Rules
Steel Construction, Inc., Chicago, IL for Block Shear – A Review,” Engineering Journal, AISC,
AISC (1999), Load and Resistance Factor Design Specifica- Vol. 38, No. 4, 4th Quarter, pp. 199–203.
tion for Structural Steel Buildings, American Institute of Orbison, J.G., Wagner, M.E. and Grondin, G.Y. (1999),
Steel Construction, Inc., Chicago, IL. “Tension Plate Behavior in Single-row Bolted Connec-
AISC (2005), Specification for Structural Steel Buildings, tions Subject to Block Shear,” Journal of Constructional
American Institute of Steel Construction, Inc., Chicago, IL. Steel Research, Vol. 49, Issue 3, pp. 225–239.
Birkemoe, P. and Gilmor, M. (1978), “Behavior of Bearing Ricles, J. and Yura, J. (1983), “Strength of Double-row Bolted
Critical Double-angle Beam Connections,” Engineering Web Connections,” Journal of Structural Engineering,
Journal, AISC, Vol. 15, No. 4, 4th Quarter, pp. 109–115. ASCE, Vol. 109, No. 1, ST1, pp. 488–490.

Designing Compact Gussets
with the Uniform Force Method
LARRY S. MUIR
β
Vc =
I n 1991 an AISC task group endorsed the uniform force
method (UFM) as the preferred method for determining the
forces that exist at gusset interfaces. Since that time it has
r
P (1d)
where
been included in the AISC Manual of Steel Construction.
The UFM provides a standardized way to obtain economical,
(α + e ) + ( β + e )
2 2
r= (2)
statically admissible force distributions for vertical bracing c b
connections. One criticism of the method is that it sometimes

results in oddly shaped or disproportionately large gusset )
α = tan θ ( β + eb − ec (3)
plates. To overcome this perceived limitation of the UFM,
designers have been seeking out alternate methods.
In order to satisfy the relationship between α and β, the
This paper demonstrates that removing one unnecessary
designer is often forced to use either an oddly shaped or dis-
geometrical constraint from the formulation of the UFM will
proportionately large gusset plate. Alternately, moments can
allow greater freedom in gusset geometry, while maintaining
be introduced at the connection interfaces. Neither approach
the efficiencies that result from the method. A new formula-
is ideal.
tion of the UFM is presented, and the strengths and weak-
nesses of other proposed design methods are also explored.
ALTERNATIVES TO THE UNIFORM
FORCE METHOD
THE UNIFORM FORCE METHOD
Any viable alternative sought to replace the UFM should
The uniform force method has been included in the AISC
meet the following criteria: (1) it must provide a clear
Manual of Steel Construction since 1992. The UFM was
procedure to satisfy equilibrium and conform to the basic
originally proposed by Thornton (1991) and was based on
assumptions made during the analysis and design of the
observations from Richard’s (1986) research. In the com-
main members (the most important criterion); (2) since the
monly accepted form, the UFM produces the following force
UFM readily accommodates a wide range of geometries and
distribution:
boundary conditions, any alternate method should also be
able to accommodate such situations; and (3) it must result
α
Hb = P (1a) in economical designs.
r
Several alternatives to the UFM have been proposed. Chief
among the alternatives are the KISS Method, the parallel
eb
Vb = P (1b) force method and the truss analogy method. None of these
r methods suffer from the constrictive relationship between α
and β that exists in the UFM. In other words, these methods
ec can be used with any gusset geometry and do not force the
Hc = P (1c)
r use of oddly shaped or large gusset plates. The strengths and
weaknesses of these methods will be explored. In all of the
discussions the work-point of the brace is assumed to be lo-
cated at the intersection of the centerlines of the beam and
the column, since this is the typical case.
Larry S. Muir is president of Cives Engineering Corporation

and chief engineer of Cives Steel Company, Roswell, GA.
013-020_Muir_2008_1Q.indd 13 3/27/08 12:18:27 PM

The KISS Method If the former approach is taken, rotational equilibrium
of the beam and column will not generally be satisfied. A
KISS (Figure 1) is an acronym for “keep it simple stupid,”
moment connection will then have to be added between the
and the method is simple, as the name implies, and fool-
beam and the column. Since this will normally be a field-
proof, though uneconomical. The method involves delivering
welded connection, this is considered to be an uneconomical
the entire horizontal brace component directly to the beam
alternative in most parts of the country.
through the beam-to-gusset connection and the entire vertical
In terms of applicability to a variety of boundary condi-
brace component directly to the column through the column-
tions, the parallel force method suffers from a major short-
to-gusset connection. To satisfy equilibrium, moments must
coming. Since the forces at both the beam-to-gusset and
be introduced. At the beam-to-gusset the moment is equal to
the column-to-gusset interfaces are assumed parallel to the
Heb, and at the column-to-gusset the moment is equal to Vec.
brace force, a horizontal component will always exist at the
The KISS Method satisfies two of the three criteria for a
column-to-gusset connection. When framing to a column
viable alternative to the UFM. It satisfies equilibrium and the
web, this presents a significant design challenge, which will
design and analysis assumptions, and it is universally appli-
usually be overcome by the addition of column stiffening
cable to all geometries and boundary conditions. However,
local to the connection, further reducing the economy of the
the presence of the large moments at the connection inter-
method.
faces makes it an uneconomical choice in practice.
The parallel force method only satisfies one of the three
criteria for a viable alternative to the UFM. It satisfies equi-
The Parallel Force Method librium and the design and analysis assumptions, but it is not
as economical as the UFM and is not suited to connections
In the parallel force method, sometimes referred to as the
made to column webs.
component method (Figure 2), the reactions of the gusset at
the beam and column interfaces are assumed to act parallel
The Truss Analogy Method
to the brace force. Since the forces are parallel, they obvi-
ously do not intersect at a common point, as is the case with The truss analogy method (Figure 3) determines the force
the UFM. Therefore, in order to maintain rotational equilib- distribution on the gusset by modeling the interface forces
rium, two choices are available. Either the magnitude of the as a pinned “truss” node located at the center of the brace-
parallel forces are set so that they balance each other about to-gusset connection. The truss analogy method suffers the
the work-line of the brace, or moments are added at the beam same problem as the parallel force method when attaching
and/or column interfaces. The additional moments, though to column webs. Additionally, the truss analogy method
lesser in magnitude than the KISS method, adversely impact can result in counterintuitive and uneconomical force
the economy of the connection. distributions. This is illustrated in Figure 3 where the gusset-
Fig. 1. KISS method. Fig. 2. Parallel force method.
013-020_Muir_2008_1Q.indd 14 3/27/08 12:18:27 PM

to-column connection delivers only a horizontal component moments to be eliminated from the interfaces the forces ap-
to the column. A formalized treatment of the equilibrium plied to each element must intersect at a single point. These
requirements for the beam and column has never been pre- points of intersection are referred to as control points.
sented and is therefore left to the designer. Often moments
are required at all of the connection interfaces in order to The Beam
satisfy equilibrium.
It is easiest to begin with the beam (Figure 4), since the loca-
The truss analogy method satisfies none of the criteria for
tion of its control point is evident. The three forces applied
a viable alternative to the UFM.
to the beam are the horizontal component of the brace, H, the
beam-to-gusset force, Vb2 + H b2 , and the beam-to-column
A GENERALIZED UFM
force, Vb2 + H c2 . The horizontal component of the brace
Since none of the alternatives investigated appear to provide is resisted along the centerline of the beam and intersects
better results than the UFM, it is advantageous to make ad- the beam-to-column force at the point (ec, 0). Therefore,
justments to the formulation of the UFM to make it more the beam-to-gusset force must also pass through this point.
applicable to compact gussets. From this we find that
The goal of the UFM was to derive a procedure to ob-
tain statically admissible force distributions, which would Vb eb
= (4)
produce no moments at the connection interfaces and would Hb α
be applicable to a wide range of geometries and boundary
conditions. However, the procedure includes an additional
The Gusset
constraint that unnecessarily limits its applicability. The
force at the gusset-to-column interface, Vc2 + H c2 , is forced The three forces applied to the gusset (Figure 5) are the
to pass through a point that lies a distance, eb, above the brace force, P, the beam-to-gusset force, Vb2 + H b2 , and
work-point. the gusset-to-column force, Vc2 + H c2 . In order to eliminate
Since there is a perceived problem with the UFM that moments at the interfaces, these three forces must intersect
can be overcome by removing this constraint, it is advanta- at a single point. Since the slope of the brace force, 1 tan (θ),
geous to eliminate it from the method. In order to do so, and the slope of the beam-to-gusset force, eb α, are known,
the problem must first be defined. There are essentially three the intersection can be determined. The gusset control
elements involved: the beam, the column, and the gusset. point is:
The brace is neglected since it is assumed to carry only axial
force and is not part of the indeterminate system. Each of ⎛ eb ec tan( θ ) eb ec ⎞
⎜ e tan( θ ) − α , e tan( ) − α ⎟
the three members is subjected to three forces. In order for ⎝ b b θ ⎠
Fig. 3. Truss analogy method. Fig. 4. Beam free body diagram.
013-020_Muir_2008_1Q.indd 15 3/27/08 12:18:28 PM

The Column the interfaces can be derived. Since the column must be in
equilibrium, the following can be established:
The three forces applied to the column (Figure 6) are the
vertical component of the brace, V, the column-to-gusset
force, Vc2 + H c2 , and the beam-to-column force, Vb2 + H c2 .
Knowing that the gusset-to-column force must pass through ∑F y
= 0 = P cos( θ ) − (Vb + Vc ) (7)
the gusset control point, the slope of the gusset-to-column
force is: ∑F x
= 0 = Hc − Hc (8)
Vc β ⎡ eb ⎛ tan(θ)(e b + β) − ec⎞ ⎤
=
Hc eb ⎢⎣ β ⎜⎝
1−
α ⎟⎠ + 1⎥
⎦
(5) ∑ M = 0 = H (e c b )
+ β − P cos( θ )ec (9)
From this cos( θ )ec

From this, since the column-to-gusset force and the beam-to- Hc = P
column force must intersect at the centerline of the column, (e b
+β ) (10)
the slope of the beam-to-column force is:
= ⎜
)
Vb eb ⎛ tan( θ ) ( eb + β − ec ⎞ (6) To satisfy the requisite geometry for the beam-to-gusset and
α ⎟
H c ec ⎝ ⎠ beam-to-column forces, the following must be true:
The point of intersection of the column-to-gusset force and

the beam-to-column force, the column control point, is:
( ( )
⎡ eb sin( θ ) eb + β − cos( θ )ec
Vb = ⎢
) ⎤⎥ P (11)
⎢
⎣
α eb + β( ) ⎥
⎦
⎛ )
⎛ tan( θ ) ( eb + β − ec ⎞ ⎞
⎜ 0, eb ⎜ ⎟⎟
⎝ ⎝ α ⎠⎠ The remaining forces are apparent:
H b = P sin( θ ) − H c (12)
Force Distribution
Vc = P cos( θ ) − Vb (13)
Having established the geometrical constraints required to
eliminate moments at all connection interfaces, the forces at
Fig. 5. Gusset free body diagram. Fig. 6. Column free body diagram.
013-020_Muir_2008_1Q.indd 16 3/27/08 12:18:29 PM

With the geometry and force distribution established, a 27.75
new form of the UFM has been derived without the some- α= + 0.5 = 14.375 in.
2
what arbitrary constraint on the location of the column con-
trol point. Without this constraint, α and β can be set to any
convenient values. This removes the need to consider the 13
moments caused by α – and β–, where α– is the actual distance β = = 6.5 in.
2
from the face of the column flange to the centroid of the
–
gusset-to-beam connection, and β is the actual distance from
the face of the beam flange to the centroid of the gusset-to- Hc =
( ) ( ) 100 = 21.7 kips
cos 55o 7
column connection. (12 + 6.5)
However, there may still be a need to redistribute the ver-
tical reaction delivered to the beam, Vb. This counteracting
( )( ( )
⎡ 12 sin(55o ) 12 + 6.5 − cos(55o ) 7 ⎤ ( ))
force is referred to as ∆Vb. ∆Vb can be introduced into this Vb = ⎢ ⎥ 100 = ( )
new formulation easily to produce the full spectrum of force ⎢
⎣ (
14.375 12 + 6.5)( ) ⎥
⎦
distributions that can exist in the connection while maintain-
ing column-to-gusset and beam-to-column connections free = 50.3 kips
of moments. It is assumed that moments at the column-to-
gusset and beam-to-column connections are uneconomical
and therefore undesirable.
( )
H b = 100 sin 55o − 21.7 = 60.2 kips
Of course the introduction of ∆Vb disrupts the established
equilibrium and adjustments must be made. The adjustment
involves introducing a moment at the beam-to-gusset inter-
( )
Vc = 100 cos 55o − 50.3 = 7.06 kips
face. This moment can be calculated as:
( )
M b = H b eb − Vb − ∆Vb α (14)
Summing moments on the beam about the beam control
point produces:
Column Moment Vb α − H b eb = 50.3(14.375) − 60.2(12) ≈ 0 kip-in.
A moment gradient will exist in the column whether using

the original formulation or the new formulation of the UFM
presented in this paper. Using the original formulation, the
moment will be zero at the intersection of the top of steel
elevation and the centerline of the column. In the new for-
mulation, the moment may be either positive or negative
throughout the section of the column bounded by the con-
nection or the moment may be zero at some section similar
to the original formulation. In either case the maximum mo-
ment the column will be subjected to can be determined as:
{ ( (
M c = max Vc ec , Vc ec − H c eb + β ))} (15)
Since the choice of column section will usually be gov-

erned by buckling and the column is restrained from buck-
ling local to the brace connection, it is normal practice to
neglect this moment. For this reason, the moment internal to
the column is not mentioned in the AISC Steel Construction
Manual (AISC, 2005) discussion of the UFM.
An Example
The forces on the connection shown in Figure 7 will be
calculated to demonstrate the new formulation. Fig. 7. Example.
013-020_Muir_2008_1Q.indd 17 3/27/08 12:18:30 PM

Summing moments on the gusset about the work-point pro-
duces: Table 1. Comparison of the Traditional and the
Modified Uniform Force Methods
( ) ( )
Vb α +ec − H b eb + Vc ec − H c eb + β = 50.3(14.375 + 7) − Traditional UFM Modified UFM
Parameters without with without with
60.2(12) + 7.06(7) − 21.7(12 + 6.5) ≈ 0 kip-in. –
α –
α ∆Vb ∆Vb
α 19.4 19.4 14.4 14.4
–
α – 14.4 – –
Summing moments on the column about the beam-to-column
connection produces: β 6.5 6.5 6.5 6.5
Vb 37.2 37.2 50.3 37.2
( )
P cos( θ )ec − H c eb + β = 100 cos(55o )(7) − 21.7(12 + 6..5) Hb 60.2 60.2 60.2 60.2
Vc 20.2 20.2 7.06 20.2
≈ 0 kip-in.
Hc 21.7 21.7 21.7 21.7
∆Vb – – – 13.1
Mb – 188 – 188
Note that:
( ) ( )
tan( θ ) β +eb − ec = tan(55o ) 6.5 + 12 − 7 = 19.4 ≠ α
OTHER PRACTICES THAT CAN REDUCE THE
GUSSET PROFILE
For completeness the vertical coordinate of the column con-
Having eliminated the geometrical constraints on gusset size
trol point can be calculated as:
from the UFM, attention can be turned to other steps that can
be taken to reduce the gusset profile.
( )
⎛ tan( θ ) eb + β − ec ⎞
yccp = eb ⎜ ⎟
⎝ α ⎠ The Whitmore Section
The Whitmore section is commonly accepted to be an area,
( )
⎛ tan(55o ) 122 + 6.5 − 7 ⎞
= (12) ⎜ which extends at a 30o angle from the edges of the brace-
⎟ = 16.2 in.
⎝ 14.375 ⎠ to-gusset connection along the length of the connection.
The area beyond this section is assumed to be ineffective in
terms of gross tension yielding and compression buckling
It may be noted that for this case the term Vb is significantly of the gusset. It is common practice to try to include all of
larger than would be obtained using the traditional UFM. As the allowed Whitmore section within the gusset, but it is not
is the case with the traditional UFM, a ∆Vb can be introduced a requirement to do so. By allowing the edges of the gusset
to manipulate the distribution of vertical force. Taking ∆Vb plate to encroach on the Whitmore section, the profile of the
equal to 13.1 kips produces the same distribution of vertical gusset can be reduced.
force that is obtained from the UFM when all parameters
except α are held constant. Weld Size
As can be seen from Table 1, which presents a comparison
It is common practice to attempt to limit fillet weld sizes to
of the traditional UFM to the modified UFM, each can be
those that can be applied in a single pass, usually c in. This
modified to produce identical results. This is to be expected
greatly enhances connection economy, since the number of
since each must satisfy equilibrium. The primary advantage
passes required to complete a weld increases disproportion-
to the new formulation is that it eliminates the need for the
– and β–. Also the new formulation makes it easier ately with the leg size. To maintain a single pass weld, the
modifiers α
gusset plate dimensions, particularly at the beam-to-gusset
to overcome the perceived limitations of the UFM.
connection, are often increased. The gusset profile can be
reduced by allowing multiple pass welds to be used, but only
with increased fabrication costs.
013-020_Muir_2008_1Q.indd 18 3/27/08 12:18:31 PM

Bolt Type Hb = shear force on the beam-to-gusset connection
If reducing the gusset profile is of paramount concern, the Hc = axial force on the beam-to-column and gusset-to-
strongest possible bolt configuration should be employed. column connections (assumes no transfer force)
Slip-critical connections should be avoided since they will
Mb = moment on the beam-to-gusset connection
require more bolts and therefore a larger gusset profile. Like-
wise, if the threads will be excluded from the shear plane, V = vertical component of the brace load
which is usually the case for heavily loaded bracing con-
Vb = shear force on the beam-to-column connection and
nections, then the “X-value” for the bolts should be used.
axial force on the beam-to-gusset connection
Providing a detail that places the bolts in double shear at the
brace-to-gusset connection also helps to reduce the gusset Vc = shear force on the gusset-to-column connection
profile.
∆Vb = change in the distribution of vertical load
CONCLUSIONS α = distance from the face of the column flange or web

to the centroid of the gusset-to-beam connection
The UFM, as currently presented in the Manual, contains an
unnecessary constraint on the location of the column control β = distance from the face of the beam flange to the
point. This constraint often gives designers the perception centroid of the gusset-to-column connection
that the method is ill suited to the design of compact gusset –
α = actual distance from the face of the column flange
plates. to the centroid of the gusset-to-beam connection
By eliminating the unnecessary constraint in the new for- (This term is not required in the new formulation.)
mulation, force distributions can be derived that consist of –
only shear and axial forces at the connection interfaces. The β = actual distance from the face of the beam flange
new formulation also simplifies the UFM by eliminating the to the centroid of the gusset-to-column connection
– and β–. (This term is not required in the new formulation.)
need for α
By manipulating the term ∆Vb, designers can obtain the
full spectrum of force distributions that can exist in the con- REFERENCES
nection while maintaining column-to-gusset and beam-to- AISC (2005), Steel Construction Manual, 13th Edition,
column connections free of moments. American Institute of Steel Construction, Inc., Chicago,
IL.
NOTATION Richard, R.M. (1986), “Analysis of Large Bracing Con-
eb = one-half the depth of the beam nection Designs for Heavy Construction,” National Steel
Construction Conference Proceedings, American Institute
ec = one-half the depth of the column of Steel Construction, Inc., Chicago, IL.
yccp = vertical coordinate of the column control point Thornton, W.A. (1991), “On the Analysis and Design of
Bracing Connections,” National Steel Construction Con-
P = brace load
ference Proceedings, American Institute of Steel Con-
H = horizontal component of the brace load struction, Inc., Chicago, IL.
013-020_Muir_2008_1Q.indd 19 3/27/08 12:18:31 PM

013-020_Muir_2008_1Q.indd 20 3/27/08 12:18:31 PM
Spec/Manual Reference
Limit State Response of Composite Columns

and Beam-Columns
Part II: Application of Design Provisions
for the 2005 AISC Specification
ROBERTO T. LEON and JEROME F. HAJJAR
T he strength of composite beam-columns, including steel

sections encased in concrete (also known as SRC) and
steel sections filled with concrete (CFT), as presented in
The designer is free to use any reasonable assumptions, in-
cluding a nonlinear strain distribution in step 2 and nonlinear
material properties in step 3, so long as those assumptions
the 2005 AISC Specification for Structural Steel Buildings, are supported by analyses, test data, or other documentation.
ANSI/AISC 360-05 (AISC, 2005a), hereafter referred to as In general, the analysis is run with an assumption of a linear
the 2005 AISC Specification, must be computed based on strain distribution in step 2 and simplified uniaxial material
first principles of mechanics and reasonable models for the relationships in step 3.
stress-strain characteristics of the materials. Chapter I of the Figure 1 illustrates the strain-compatibility procedure.
2005 AISC Specification provides two options to fulfill this Figure 1a shows the subdivision of the cross-section into a
requirement: a general approach labeled the strain-compat- series of square “fibers” representing four different materi-
ibility method and a simplified approach labeled the plastic als: unconfined concrete outside the ties, confined concrete
stress distribution method. inside the ties (shaded), reinforcing bar steel, and rolled
The strain-compatibility method is conceptually similar to shape steel. Figure 1b shows a linear strain distribution with
conducting cross section analysis of a reinforced concrete an arbitrary location of the neutral axis. Figure 1c shows
section, and requires the designer to: the material properties. For each material, it shows both the
nonlinear stress-strain curves as may be assumed for the
1. Subdivide the cross section into a large number of areas
strain-compatibility analysis and the bilinear rigid-plastic
(termed “fibers”);
curves (dashed lines) that would be assumed for a simplified
2. Assume a strain distribution across the cross-section and analysis; for example, a plastic strength calculation. Figure
a location of the neutral axis; 1d shows the stress distributions for the situation usually
used to compute the ultimate strength of the cross-section, in
3. Compute stresses based on the assumed stress-strain other words, when the strain in the concrete, εc, has reached
relationships for the different components (unconfined 0.003 and the steel strain in the extreme fiber of the steel
and confined concrete, reinforcing bars, and the steel shape, εs, has exceeded its yield strain. The concrete stress
shape); blocks are shown separately for clarity. Note that for this
case the unconfined stress-strain relationship is shown up to
4. Integrate the stresses over the cross section to obtain the the top edge of the section. Figure 1e shows a stress distri-
total axial load and the moment about the plastic neutral bution when the concrete strain has exceeded 0.003 and the
axis (or other commonly assumed axis); steel shape and reinforcing bars are well into the strain-hard-
ening range. In this case the confined concrete stress-strain
5. Iterate steps 2 through 4 to obtain an axial load-moment relationship is used, but limited to the confined section; the
interaction surface for the column. unconfined section has spalled off. Figure 1f shows the typi-
cal stress distribution used for the plastic stress distribution
case. Note that this stress distribution does not correspond
exactly to any “real” stress distribution but is considered as
a conservative approximation for a section that can sustain
Roberto T. Leon is professor, school of civil and envi-
concrete strain on the order of 0.005 and steel strains larger
ronmental engineering, Georgia Tech, Atlanta, GA and
than 0.01.
former chair, AISC Task Committee 5 on Composite
Construction. In Figure 2, three qualitative curves are shown for the
stress distributions shown in Figures 1d through f. The first
Jerome F. Hajjar is professor, department of civil and envi-
curve is for the strength envelope consistent with a concrete
ronmental engineering, University of Illinois, Urbana, IL.
strain of 0.003 at the extreme fiber (Figure 1d) while the
021-046_Leon_Hajjar_2008_1Q.indd 21 4/7/2008 9:51:22 AM

Fig. 1. Development of composite beam-column cross-section strength.
Fig. 2. Interaction diagrams based on the stress distributions shown in Figure 1d through 1f.

second curve is for a concrete strain of approximately 0.005 limited ductility required for the steel to yield significantly
at the same location (Figure 1e). In the latter case, the effect (εs > 0.005) while the concrete compressive strength has not
of the unconfined concrete has been eliminated and only the decreased below that given by an assumption of a uniform
confined concrete core is contributing. Figure 2 also shows stress of 0.85f c′ for SRCs and rectangular CFTs (0.95f c′ for
the corresponding strength envelope assuming bilinear rigid- circular CFTs) over an effective depth similar to that used in
plastic stress-strain curves for the materials and calculated conventional reinforced concrete design. These assumptions
only for the position of the neutral axis at four discrete points are well within those made in Section 10.2 and 10.3 of the
(Figure 1f) rather than continuously. The latter corresponds ACI Code (ACI, 2005), and thus this approach is deemed to
to the simplified approach (plastic stress distribution meth- satisfy both the current steel and concrete design specifica-
od) used in the 2005 AISC Specification. tions. Note that the AISC Steel Construction Manual (AISC,
It should be noted that the strain-compatibility method dis- 2005b) indicates that the 0.95 factor for circular CFTs is
cussed here is the only one currently applicable to cases with for the uniform compression case only. However, the cali-
biaxial loading if one wants to generate interaction points in brations were conducted assuming 0.95 for any interaction
addition to the anchor points on the coordinate axes of the condition and thus that value is used in this paper for all cal-
interaction diagram. For that case the procedure is analogous culations for circular CFTs (Kim, 2005). The confinement
to that described in Figure 1, except that the section will not resulting from hoop stresses in circular tubes could justify a
be aligned along its principal axes as currently shown in Fig- larger factor in many practical cases, but other checks on the
ure 1a but at some other angle. The calculations will become column slenderness and the eccentricity of the load would be
more laborious and there are only approximate solutions required. The 0.95 factor was selected as a reasonable lower
available for the case of CFT columns, which can be treated bound value that would not require such further checks.
as reinforced concrete columns with distributed reinforce- The simplified or plastic stress distribution method is anal-
ment and distance between extreme bar layers equal to the ogous to the strain-compatibility method used to determine
section depth. However, the use of the interaction equations the strength of a reinforced concrete column, but rather than
of AISC 2005 Specification Chapter H (AISC, 2005a), dis- solving for a large number of points along the interaction
cussed below, provides an alternative procedure in which diagram, it relies on linear interpolation between four points
only the anchor points on the coordinate axes of the interac- for major axis bending for encased composite sections or
tion diagram need to be computed. five points for minor axis bending of encased composite sec-
The procedure discussed here is meant for monotonic tions and all filled composite sections (Figure 3). Aside from
loading cases; for seismic design further attention to con- these anchor points, the remaining points are approxima-
finement and local buckling phenomena will be needed to tions to the exact interaction curve. Designers can use the
sustain the strength envelopes shown in Figure 2 under large strain-compatibility method to check the plastic stress dis-
cyclic deformations. The strain compatibility method is now tribution method, but similar results will only be obtained if
embedded in a number of commercial structural analysis and the assumed stress-strain curves for the strain compatibility
design software packages for reinforced concrete sections method are the same as those for the plastic method.
and is accessible to most engineers. Thus the 2005 AISC
Specification explicitly endorses the use of these advanced
design tools. However, this approach is time-consuming and
not always useful for preliminary design.
The plastic stress distribution method proposed in Chap-
ter I of the 2005 AISC Specification is based on the plastic
stress distributions shown by the dashed lines in the material
properties in Figure 1. This method is intended to provide a
design-oriented approach that captures the essential features
of the strain compatibility one, but without the associated
complexities (Roik and Bergmann, 1992). This approach is
described in detail in this paper, which begins with a discus-
sion of the axial compressive strength of different composite
cross sections and then moves on to the design of composite
beam-columns. The materials are assumed to be elasto-plas-
tic, with no attempt to include deformation capacity (ductili- Fig. 3. Interaction diagrams for composite beam-columns:
ty) in the calculations. The assumption is that the confinement a) encased composite sections, strong axis,
required by the AISC Specification for encased shapes and and filled composite sections; and
that provided by the tubes in CFTs is sufficient to provide the b) encased composite sections, weak axis.

PRELIMINARY SIZING OF COMPOSITE SEC- may be used in the last term instead of 0.85 to account
TIONS FOR AXIAL COMPRESSION for the effects of confinement [see Equation (AISC
2005 I2-13)]. For encased sections and rectangular
For systems in which the composite columns are assumed to
concrete-filled tubes (RCFTs), an approximation of
carry primarily gravity loads, or for in assessing composite
Equations (AISC 2005 I2-4) and (AISC 2005 I2-13)
beam-column strength, the design process outlined below
can be used to find the preliminary size of the column as
as per the plastic stress distribution method may be used to
follows (and a similar formula is shown for circular
compute the design axial compressive strength. This may
concrete-filled tubes):
be used with the required strength, Pu, to assess the axial
strength of the member.1 See Leon, Kim and Hajjar (2007) Po
for a summary of all design equations discussed in this pro- Ag ≈ ((SRC and RCFT)
cedure. φ β ρFy + 0.3ρFyr + 0.8P5 fc′(1 − 1.3ρ 
o
)
Ag ≈ ((SRC and RCFT)
1. Select the steel shape and reinforcing bars yield strengths,
Ag ≈ φ
P
β ρFoy + 0.3ρFyr + 0.(SRC
8 5 fc
′ (1and
− 1RCFT)
. 3ρ )
(CCFT)
Fy and Fyr, respectively, and the concrete compressive
strength, f c′. φcβ ρFy + 0.95 fc′Po(1 − ρ  )
Ag ≈  (CCFT) (CCFT)
2. Select a steel ratio, ρ, for the column. This ratio refers to

φcβ ρFy + 0.95 fc′ (1 − ρ 



)
the area of the steel shape only, As, to the gross area of
concrete, Ag. The influence of the rebar will be ignored 5. Assume a preliminary section size and reinforcement
in this design procedure because the AISC Specification based on Ag calculated above and begin the checking
does not consider them in the calculations of the steel procedure. For filled composite sections, first check local
area. For encased composite sections in gravity systems, buckling of the steel tube as per AISC (2005a).
reasonable and economic sizes result from assuming ρ is
in the range of 8 to 12%. For filled composite sections, 6. Determine the coefficient C1 or C3 from Equation (AISC
the range for ρ is typically 6 to 10%. 2005 I2-7) for encased composite sections or Equation
(AISC 2005 I2-15) for filled composite sections:
3. Select a slenderness ratio for the column. Most com-
posite columns are not very slender, so the reduction in  As 
the nominal axial strength due to length effects is often C1 = 0.1 + 2   (AISC 2005 I2-7)
 Ac + As 
smaller than that for regular steel columns of the same
length. Most composite columns in gravity frames will ≈ (0 .1 + 2ρ) ≤ 0.3
have a slenderness parameter λ = Po Pe between 0.5
and 1.0 (corresponding to reductions for length effects  As 
of 80 to 65%). This reduction value will be termed β and C3 = 0.6 + 2   (AISC 2005 I2-15)
 Ac + As 
a value of 0.7 is recommended for initial trial designs.
≈ (0 .6 + 2ρ) ≤ 0.9
4. Calculate the required gross area of the concrete based on
Equation I2-4 from ANSI/AISC 360-05 (AISC, 2005a) 7. Compute the equivalent stiffness (EIeff ) from Equation
[termed in this paper Equation (AISC 2005 I2-4)]: (AISC 2005 I2-6) for encased composite sections or
(AISC 2005 I2-14) for filled composite sections:
Po
= As Fy + Asr Fyr + 0.85 Ac fc′ , kips (kN)
φc β EIeff = EsIs + 0.5EsIsr + C1EcIc, kip-in.2 (N-mm2)
(AISC 2005 I2-4)
(AISC 2005 I2-6)
In the design of SRCs, one can assume that for this cal-
culation As ≈ ρAg, Asr ≈ 0.3ρAg and Ac = (1 − 1.3ρ)Ag. EIeff = EsIs + 1.0EsIsr + C3EcIc, kip-in.2 (N-mm2)
For circular concrete-filled steel tubes (CCFT), assume
(AISC 2005 I2-14)
As ≈ ρAg, Asr = 0, and Ac = (1 − ρ)Ag. For CCFTs, 0.95
1
For a complete set of notations, see Appendix A.

8. Compute the elastic Euler buckling load, Pe, from Equa- Fy = 50 ksi, Fyr = 60 ksi and f c′ = 8 ksi. Assume the column
tion (AISC 2005 I2-5) for buckling about the axis that is continuously braced about the minor axis. In Examples 1
provides the lower buckling strength: through 3, all the material and similar limitations in AISC
Sections I1.2, I2.1a and I2.2a are satisfied. Checking of
Pe = π2(EIeff) / (KL)2, kips (kN) (AISC 2005 I2-5) those limits will be illustrated in Example 4.
9. Calculate the squash load for the columns from Equa- 1. Select an initial steel ratio, ρ, of 10% for the column.
tion (AISC 2005 I2-4) or (AISC 2005 I2-13), where C2
2. Assume β = 0.7.
is 0.85 for rectangular tubes and 0.95 for circular pipes:
3. Calculate the required gross area:
Po = As Fy + Asr Fyr + 0.85Ac fc′ , kips (kN)
Po 4, 000
(AISC 2005 I2-4) Ag ≈ =
φcβ ρFy + 0.3ρFyr + 0.85 fc′(1 − 1.3ρ  ) (0.75)(0.7) (0.1)(50) + 0.3(0.1)(6
Po = As Fy + Asr Fyr + C2 Ac fc′ , kips (kN) Ag ≈ 599 in.2
Po Po 4 , 0 00 4, 000
Ag ≈ A =
≈ =
(AISC 2005 I2-13) )
φcβ ρFy + 0.3ρFyr + 0.85 fc′(1 − 1.3ρ g φ(0cβ.75
 c )
0.3f (′0(1.1−)(160
ρ)(F0y.7+) 0(.30ρ.1F)(yr50+)0+.85 . 3)ρ+(0.85

( 0).(75
8)()(00.87
.7))(0.1)(50) + 0.3(0.1)(6

This load corresponds to Ag the stress
≈ 599 in.2distribution shown in Ag ≈ 599 in. 2
Part (a) of Table 1, or Point A in Tables 2 through 5. This

is the maximum axial compressive load that the cross-
section can carry if strain-hardening of the steel and ad- 4. Assuming a square column, this gross area will roughly
ditional strength due to the confinement of the concrete require a 24 in. × 24 in. column. The steel section will
are ignored. require an As ≈ 59.9 in.2 to achieve the desired steel ra-
tio. Select a W14×211 (As = 62.0 in.2) buckling about its
10. For design, this maximum strength needs to be adjusted major axis and assume 16-#8 bars distributed along the
to account for length effects through either Equation perimeter of the section providing a rebar reinforcement
(AISC 2005 I2-2) or (AISC 2005 I2-3). This adjustment ratio of approximately 2.2% (see Figure 4). Many of
is based on the ratio of Po /Pe for the governing axis of these reinforcing bars will be needed to maintain con-
buckling; this ratio corresponds to the slenderness ratio, finement and rebar spacing requirements, and will not
λ2c, used in previous versions of the AISC Specification:
(a) When Pe ≥ 0.44Po:
  Po 
 

Pn = Po  0.658 e   , kips (kN)
P
(AISC 2005 I2-2)
 
 
(b) When Pe < 0.44Po:
Pn = 0.877Pe, kips (kN) (AISC 2005 I2-3)
11. For design, the value of Pn is then adjusted by the appro-

priate resistance factor, φc, or safety factor, Ωc, for LRFD
or ASD:
φc = 0.75 (LRFD) Ωc = 2.00 (ASD)
Design Examples
Example 1: Encased Concrete Column (SRC)

Design an encased composite column to resist a required
axial strength of 4,000 kips (LRFD) with a KL = 24 ft. Use Fig. 4. SSRC section for Example 1.

necessarily be continuous through the joint due to the 8. Calculate the squash load for the column from Equation
presence of framing beams. In this example, only the (AISC 2005 I2-4):
four corner bars, located at a distance of 9.63 in. from the Po = As Fy + A sr Fyr + 0.85 Ac fc′
column centerline, will be assumed as continuous and
= (62.0)(50) + (33.16)(60) + (0.85)(511)(8)
used in the strength calculations [Asr = 4(0.79 in.2)]. For
this section: = 6,760 kips
As = 62.0 in. 2
Is = Ix = 2,660 in.4 9. Adjust for length effects:

Zx = 390 in.3 Pe 14, 400
Asr = 3.16 in.2 =
Po 6, 760
Ac = Ag – As – Asr = (24)(24) – 62.0 – 3.16 = 511 in.2
e1 = 9.63 in. (distance from the column centerline to = 2.13 > 0.44, use ANSI/AISC 360-05 Equation I2-2
reinforcing bars) Po 6, 760
Isr ≈ (Asr)(e1)2 = 293 in.4 = = 0.469
Pe 14, 400
wc = 148.1 lb/ft3, assumed Po
Pe
Ec = w 1.5
fc′ ( ksi) ( AISC) = 57,000 fc′ (psi) ( ACI) Pn = Po (0.658) = 6,760(0.658)0.469 = 5,560 kips
c Po
Ec ==(148.1 lb/ft 3 )1.5 8 ksi = 5,100 P

ksi P
= Po (0.658) e = 6,760(0.658) 0.469
n = 5,560 kips
Assume #3 ties at 12 in.
ρst = 0.22 in.2/12 in. = 0.0183 in.2/in. > 0.009 in.2/in. 10. Finally, for design, the value of Pn is adjusted as:
(ANSI/AISC 360-05, Sect. I2.1a) φ P = (0.75) (5,560) c n
5. Determine the coefficient C1 from Equation (AISC 2005 = 4,170 kips (LRFD) > 4,000 kips o.k.
I2-7):
The associated ASD strength is:
 As   62.0  Pn /Ωc = (5,560 kips/2.00) = 2780 kips (ASD)
C1 = 0.1 + 2   = 0.1 + 2  511 + 62.0  = 0.316 but C1 ≤ 0.3 so C1 = 0.3
 c
A + A s This should be compared to the required strength based
  62.0  on ASD load combinations.
 = 0.1 + 2  511 +
62. 0  = 0.316 but C1 ≤ 0.3 so C1 = 0.3
s
The final design is shown in Figure 4.
6. Compute the equivalent stiffness, EIeff, from Equation
(AISC 2005 I2-6) for encased shapes: Example 2: Circular Filled Concrete Column (CCFT)
EIeff = Es Is + 0.5Es I sr + C1 Ec I c Design a concrete-filled steel tube column to carry a factored

axial
24 4 load of 1,500 kips (LRFD) with an effective length
EIeff = (29, 000)(2, 660) + (0.5)(29, 000)(293) + (0.316)(5, 100) KL = −182,ft. 660 − 293
Use Fy = 42 ksi, Fyr = 60 ksi and f c′ = 5 ksi.
0.5Es I sr + C1 Ec I c EIeff = Es Is + 0.5Es I sr + C1 Ec I c  12
 24 4
  24 4
 ρ, of approximately 8% for the
0)(2, 660) + (0.5)(29, 000EI)(eff293=) +121
(0,.×316
(29 006)(
010 )(5kip 0))2+ (0.5−)(229
0-in.
2,,1p660 , 000−)(293) + (0.316)(5, 100) 1. Select
, 660 a steel
− 2, 660 − 293ratio,

 12   column.
12
6
kipp-in.2 EIeff = 121 × 10 6 kipp-in.2
2. Assume β = 0.7.
7. Compute the elastic Euler buckling load, Pe, from Equa- 3. Calculate the required gross area:
tion (AISC 2005 I2-5):
Po 1, 500
Ag ≈ =
Pe =
π EIeff
2
=
(
π 121 × 10
2 6
) = 14, 400 kips
φcβ ρFy + 0.95fc′(1 − ρ 
  ( 0). 75 )( 0 .7
7 ) 
 ( 0 . 08)(42) + 0.95(5)(0.92) 
( KL ) ) )
( 24 (12 
2 2
Po πDD 2 1, 500
  Ag ≈ A ≈ 370 in. 2
= P ⇒ D = 21.7 in. 1, 500
Ag = ≈ (0.75)(0.77) 4(o0.08)(42) + 0=
φcβ ρFy + 0.95fc′(1 − ρ g φ

) c 
   . 95 (
)
5 )( 0 . 92 ) 
β ρFy + 0.95fc′(1 − ρ  (0.75)(0.77) (0.08)(42) + 0.95(5)(0.92) 
 
πDD 2
Ag ≈ 370 in.2 = ⇒ D =A21.7 ≈ in. in.2 = πD
370
D2
⇒ D = 21.7 in.
4 g
4

4. The computed D exceeds the largest diameter avail- 9. Adjust for length effects:
able for circular hollow sections of 20 in. Assuming the Pe 10, 700
same ρ and D = 20 in., the required steel area is =
Po 2, 290
As ≈ 25.1 in.2 This is about halfway between the areas for
a HSS 20.00 × 0.500 and a HSS 20.00 × 0.375. Select = 4.67 > 0.44 use ANSI/AISC 360-05 Equation I2-2
the HSS 20.00 × 0.375 and check. For this section: Po 2, 290
= = 0.214
As = 21.5 in.2 Pe 10, 700
Is = Ix = 1,040 in.4 Po
Zx = 135 in.3 Pn = Po (0.658)

Pe
= 2,290(0.658)0.214 = 2,090 kips
t = 0.93(0.375) = 0.349 in. = HSS design thickness
Ac = π[20 − 2(0.349)]2/4 = 293 in.2 10. Finally, for design, the value of Pn is adjusted as:
Ec = (148.1 lb/ft3)1.5 5 ksi = 4,030 ksi
φcPn = (0.75)(2,090)
D 20 E  29, 000 
= = 57.3 < 0.15   = 0.15  = 104 = 1,570 kips (LRFD) > 1,500 kips o.k.
t 0.349  Fy   42 
The associated ASD strength is:
D 20 E  29, 000 
= = 57.3 < 0.15   = 0.15  = 104
104 o.k. Pn/Ωc = (2,090 kips/2.00) = 1,050 kips (ASD)
t 0.349  Fy   42 
This should be compared to the required strength based
5. Determine the coefficient C3 from Equation (AISC 2005 on ASD load combinations.
I2-15) for filled tubes:
In the 3rd Ed. AISC LRFD Manual of Steel Construction
 As   21.5  (AISC, 2001), based on the 1999 AISC LRFD Specification
C3 = 0.6 + 2   = 0.6 + 2  293 + 21.5  = 0.737 ≤ 0.9 so Cfor = 0.737
 Ac + As  Structural Steel Buildings, the tabulated design strength
3
  21.5  for this section was 1,680 kips; or a difference of –6.5%

= 0.6 + 2  = 0.737 ≤ 0.9 so C3 = 0.737
As   293 + 21.5  in strength for the 2005 AISC Specification over the 1999
provisions. If the effective length of the column is 40 ft, the
6. Compute the equivalent stiffness, EIeff, from Equation ratio of Po /Pe is 1.06, Pn = 1,470 kips, and φPn = 1,100 kips
(AISC 2005 I2-14) for concrete-filled tubes: based on the 2005 AISC Specification. The value of
EI eff = E s I s + C3 Ec I c φPn = 1220 kips in the 3rd Ed. AISC LRFD Manual of Steel
4 Construction, or a difference of –9.8%.
 π  20 − 2(0.349)  
EI = ( 29 , 000
EI eff = E s I s +EICef3f Ec=I c E I + C E I )(1, 040 ) + (0. 737)( 4 , 030 )     To check the values given for circular columns in the cur-
eff s s 3 c c  64 rent AISC Manual (AISC, 2005b), consider the design of a
  4 HSS 18.000 × 0.500 pipe column filled with 4 ksi concrete
4
π  20 − 2(0.349  )π 20 − 2(0.349)  
= )50 .4 ×.1737
0 6 ki4p- in.2)  
EI eff = (29, 000EI
EI = (+
)(1eff, 040 29,(0 000)(1)(, 040
, 030
) + (0.737)(464 , 030)    and an effective length of 24 ft. Repeating the steps above,
eff     64 the key values are:

EI eff = 50.4 × 1EI 6
0 kip- = in50
2
. .4 × 10 6 kip-in.2 1. Compute C3:
eff
 As   25.6 
C3 = 0.6 + 2   = 0.6 + 2  229 + 25.6  = 0.801 ≤ 0.9 so C3 = 0.801
 cA + A s
7. Compute the elastic Euler buckling load, Pe, from Equa-
 As   25.6 
tion (AISC 2005 I2-5): C3 = 0.6 + 2 = 0.801 ≤ 0.9 so C3 = 0.801
 A + A  = 0.6 + 2  229 + 25.6 
π 2 EI eff π 2 (50.4 × 10 6 )  c s
Pe = = = 10, 700 kips 2. Compute the equivalent stiffness, EIeff, from Equation
( KL )2 (18 × 12)2
(AISC 2005 I2-14), with Isr = 0, for concrete-filled tubes:
EI eff = Es Is + C3 Ec I c
8. Calculate the squash load for the column from Equation EI eff = Es Is + C3 Ec I c
(AISC 2005 I2-4): Ec = (145 lb/ft 33 )1.5 4 ksi = 3,490 ksi
ECc3 Ec I=
EI eff = Es Is + EI = (145
E I +lb/ft
C E )1.5I 4 ksi = 3,490 ksi
Po = As Fy + 0.95 Ac fc′ = (21.5)(42) + (0.95)(293)((5) eff c s s 3 c c π 18 − 2(0.465)  44
  
2(0.465)  
Ec = (145 lb/ft EI = ( 29 , 000 )(
= 3,490 985 ) + (0.801 )(3,
, ksi) π 18 − 64
490 )
3 1.5
eff) = (145
4 ksi lb/ft ) )ksi = 3,490
3 1.5
E
EI = ( 29 , 000 )( 985 4+ks(0i .801 )(3,
, 490 
Po = 2,290 kips c
eff  4 64
π 18 − 2(0.465 4
 
π 18 − 2(0.465)  
) 
EI eff = (29, 000EI)(ef985 +29.(0
=) (40 ×.801
2, 00010)(66)( 3,, 490
kip-in.  .801
)22(0  
EI =
EI efefff = 40.2 × 10 kip-in. 
f 985 ) + )( 64   64
3,
, 490 )

 
EI eff = 40.2 × 10 6
EI eff kip-in.
2
= 40.2 × 10 6 kip-in.2

3. Compute the elastic Euler buckling load, Pe, from Equa-

tion (AISC 2005 I2-5):
θ=
0.0260 K c − 2 K s
+
( 0.0260 K c
+ 2K s ) 2
+ 0.857 K c K s
π 2 EI eff π 2 (40.2 × 10 2 ) 0.0848 K c 0.0848 K c
Pe = = = 4,780 kips
( KL )2 (24 × 12) 2
The correct expression for ZsB is
4. Calculate the squash load for the column from Equa-
tion (AISC 2005 I2-13), with Asr = 0 and Ac = π[18 − Z sB =
d 3 − h3 3  θ
)
sin ( θ 2  +
( 2π − θ 

)
2(0.465)]2/4 = 229 in.2 12 )
 θ − sin θ ( 2 π − θ − sin ( 2 π − θ  )
Po = As Fy + 0.95 Ac fc′ = (25.6)(42) + (0.95)(229)((4)
As shown in Table 5 of this paper, ZsB can be approximated
Po = 1,950 kips by
d 3 − h 3 ( 4 / 3)
5. Adjust for length effects:
Z sB ≈
6
sin ( θ 2 )
Pe 4, 780
= = 2.45 > 0.44 use Equation I2--2 Also note that Tables 4 and 5 for CFTs carry the fifth point
Po 1, 950
(E), which was dropped from the AISC Manual tables. Fi-
Po 1, 950 nally these tables also differ from the ones in the CD-Rom in
= = 0.408
Pe 4, 780 that the intermediate bars at the centroid of the sections are
Po not considered herein.
Pn = Po (0.658)
Pe
= 1,950(0.658)0.408 = 1,640 kips The generation of axial load-flexure interaction diagrams
is a well-documented but tedious procedure (Figure 1), in-
φcPn = (0.75) (1,640) = 1,230 kips (LRFD) volving the selection of a neutral axis location and a con-
[= 1,230 kips in AISC Manual (AISC, 2005b)] o.k. trolling strain (generally 0.003 for the concrete in compres-
sion). Assuming a linear distribution of strain (Figure 1b), a
One can also compute the associated ASD strength as: corresponding stress distribution can be found based on the
Pn /Ωc = (1640 kips/2.00) = 820 kips (ASD) stress-strain characteristics assumed for the materials (Fig-
[= 821 kips AISC Manual, (AISC, 2005b)] o.k. ure 1c) and the level of strain (Figures 1d through f). Sum-
ming the forces and moments about the assumed reference
ANALYSIS OF COMPOSITE SECTIONS axis gives a single combination of axial load, P, and moment
FOR COMBINED AXIAL COMPRESSION strength, M. Moving the location of the neutral axis slightly
AND FLEXURAL LOADS will lead to a different combination of axial load and mo-
ment strength, and the interaction surface can be generated
The calculation of the flexural strengths, which follows, is by moving the location of the neutral axis methodically (in
keyed to the stress distributions shown in Table 1. Table 2 other words, giving the rounded curves in Figure 3). This is a
shows the applicable formulae for the case of an encased process that is most easily carried out with the aid of charts,
shape (SRC) for strong axis bending and Table 3 those for spreadsheets, or subroutines directly embedded within com-
weak axis bending. Note that in order to keep the tables mercial structural analysis software packages.
simple, only some cases are addressed directly. In particu- It is important to recognize some characteristics of dia-
lar these are cases where continuous reinforcing bars are grams such as that shown in Table 1, particularly the dif-
grouped near the corners of the SRC sections. Modifications ference between the assumed neutral axis and the reference
to the tables are discussed in the text; for a more general axis about which moments are calculated. The latter is an
case, see Appendix B in Viest, Colaco, Furlong, Griffis, Leon arbitrary choice, usually taken at the centroid of the sym-
and Wyllie (1997). Similar tables are given for rectangular metric section. For that case, it is easy to see that insofar as
concrete-filled tubes (Table 4) and circular concrete-filled the summation of axial forces is concerned, the contributions
sections (Table 5). Note that for Point B in Table 5 changes of the longitudinal bars and steel flanges in Figure 5 cancel
have been made to the table as it appears in the AISC 13th each other out, leaving only the contributions of the web of
Ed. Steel Construction Manual CD Companion accompany- the steel shape and the rectangular concrete block. Tables 2
ing the AISC Manual (AISC, 2005b). These changes correct through 5 make extensive use of these simplifications, and
what appear to be a typographical error in the computation what may appear to be missing terms in some of the equa-
of θ and a discrepancy in the computation of ZsB. The correct tions shown are actually the result of cancellation of terms
expression for θ is as discussed above.

A careful choice of the location of the neutral axis leads • Point C: intermediate case, with the neutral axis located
to the rapid generation of an interaction surface very close at a distance hn below the centroid. This approach as-
to the more refined one described above through the use of sumes that the interaction diagram (Figure 3a) is roughly
the expressions in these tables. The location of those points symmetrical about the axial load at the balance point
is shown in Figure 3, corresponding to the positions in Table (Point D) up to a moment equal to the plastic strength
1. Because of the concavity in the curve between Points A of the section (Point B). Thus Point C has been selected
and B for the case of minor axis bending of SRC sections, an as an arbitrary but convenient point given that it has the
additional Point E is used. same moment as Point B and twice the axial strength
The four points in Figure 3a and Table 1 correspond to: of the balance point. This choice considerably simplifies
calculations without appreciable error for symmetrical
• Point A: pure axial load case, with the cross-section un-
sections.
der uniform compression corresponding to εc = 0.003.
• Point D: balance point, or point of maximum moment,
• Point B: pure flexure case, with all steel in tension and
corresponding to the neutral axis at the centroid as this
compression yielding, ignoring concrete tensile contri-
gives the largest flexure contribution from the concrete
bution, and εc = 0.003. The neutral axis is located at a
portion.
distance hn above the centroid.
Fig. 5. Calculation of axial load and flexural strength for a given position of the NA.
Fig. 6. Axial strength at Point C obtained by adding cases (b) and (c) from Table 1.

The fifth point (Point E), used for SRC bent about its weak compression area can be represented by a rectangular
axis and concrete-filled tubes (Figure 3b), is computed by distribution across the entire section, simplifying the
selecting an arbitrary position of the neutral axis between calculation of the axial force:
Points A and C. For SRC bent about the weak axis, this point
PC = 0.8 5 fc′(h1h2 − As − Asr ) = 0.85 fc′(h1h2 − As − Asr )
is usually taken with the neutral axis at the flange tips.
In the computations that follow, as with the prior exam-
ples, the materials are assumed as rigid perfectly-plastic and 3. The axial force at Point D corresponds to one-half of that
assumed to have reached high strains so that the elastic con- at Point C, as the stress blocks corresponding to Point D
tribution is small. The design does not include any explicit will result from subtracting an area equal to hn h2 from
checks to ascertain that these large strains can be achieved. the axial force at Point C. Subtracting an additional hn
It is assumed that the requirements for local buckling and h2 from Point D will lead to Point B, in other words, the
transverse reinforcement implicitly satisfy this requirement. zero axial load case. Thus:
All the steel is assumed to be yielding in tension or com- PD = 0.425 fc′(h1h2 − As − Asr ) = 0.425 fc′Ac
pression and strain hardening is ignored. The concrete is as-
sumed to reach its strength at a strain of 0.003, and its non- PB = 0
linear stress distribution is assumed to be well-represented
by an equivalent rectangular block with a stress at 0.85f c′ . 4. The moment at Point D corresponds to the summation
There is currently some discussion of whether this is the best of all plastic section moduli times their yield stress, with
representation for high strength concrete, but this assump- the exception that the concrete contribution is halved be-
tion has provided reasonable results for composite columns cause only the portion in compression contributes. The
with concrete strengths up to approximately 10 ksi. The dif- moment at Point D is thus:
ferences in performance between confined and unconfined
concrete are ignored, except that for the case of concrete- )
M D = Z s Fy + Z r Fyr + 2 Zc ( 0.85 fc ′
filled circular pipes, where the stress can be increased from
0.85 f c′ to 0.95f c′ as mentioned earlier. where
(d − 2t f ) t w2
Steel Reinforced Concrete (SRC) Major Axis Bending Zs = + b f t f (d − t f ), or as given in Parrt 1 of the AISC Manua
The development of the unfactored interaction diagram for a (d − 2t4f ) t w22
(Rd − 2t f ) t w + b f t f (d − t f ), or as given in Parrt 1 of the AISC Manua
= given
Z s as
SRC column bent about its major axis (Table 2) requires the or
Zrs == ∑
Z Asr ein
4i , Part
wh+e1rre
bof t the
(d AISC ),Manual
− t f al or asbe
given
r of binarPar
s rt 1 of the AISC Manua
fR fis the tot num
i 4
following steps (Roik and Bergmann, 1992): i =1
R
Zr = ∑ R 2
h1h2A e , wherre R is the total number of bars
sri i where R is the total number of bars
1. Point A is the squash load for the column, Po, obtained = ∑
Z rc = 1 As−
Z i =
i =4
1 2
ei , wh
ri Z
s
− Zerrre R is the total number of bars
by setting all the materials at their plastic axial strength. hh
Thus: Z c = h11h222 − Z s − Z r
Zc = 4 − Z s − Zr
4
P A = Po = As Fy + Asr Fyr + 0.85 fc′Ac
In these formulas Asri is the area of reinforcing bar i and
2. The axial force at Point C is obtained next by adding ei is its distance from the plastic neutral axis.
the stress distributions from Cases (b) and (c) in Table
1 and integrating the resulting stresses across the cross 5. To calculate the moments at B and C, another mathemat-
section. The summation is purely a mathematical artifice ical trick is used. The stress distribution for Point C is
to obtain the axial load at C, since Case (b) corresponds subtracted from that of Point B. Most of the forces can-
to the case of no axial load while Case (c) corresponds cel out, leading to the stress blocks shown in Figure 7.
to the axial load needed for Point C. The resultant stress Because these remaining stress blocks result in a zero
blocks from this sum are shown in Figure 6. net moment about the centroid, the moments at B and C
must be equal. In addition, since we know from Step 2
As can be seen from Figure 6, all the forces in the steel (above) what the value of PC is, the distribution shown
section and reinforcing bars cancel each other out when in Figure 7 allows the value of hn to be calculated. In the
computing the resultant axial force, leaving only the con- calculations for Figure 7, the steel stress is decreased by
crete portions as the axial force resultant. As only axial 0.85f c′ for consistency with the uniform stress used in
force is to be computed using this diagram, it is possible Step 2. In addition, one must check that hn < (d/2 – tf )
to move the concrete compression block from Case (b) to insure that the location of hn is within the web of the
to a location below that of Case (c). Thus the concrete steel section. Finally, note that there are no reinforcing

bars within the 2hn zone near the middle of the 8. Finally, if the location of hn is outside the steel shape, the
beams; if there were, the force in those bars, equal to expressions for hn and Zs become:
Asr(2Fyr -0.85f c′ ), must be subtracted from the numerator
of the expression for hn. Thus: hn =
)
0.85 fc′( Ac + As − 2Fy As
MC = MB
(
2 0.85 fc′ h1 )
Z sn = Z sx
PC = 0.85 fc ′A (from Step 2)
( )
PC = 2hn 0.85 fc′h1 + t w (2 Fy − 0.85 fc′ ) (from Figure 7) Steel Reinforced Concrete (SRC) Minor Axis Bending
0.85 fc′h1h2 d  The procedure for determining the axial load-flexure interac-
hn = ≤  − tf 
2(0.85 fc′h1 + t w (2 Fy − 0.85 fc′ ))  2  tion diagram for minor axis bending is the same as that for
major axis bending with two exceptions:
6. Once hn has been obtained, the moments at B and C can 1. There are only two possible locations of hn (either within
be calculated using either of the given stress distributions. or outside the steel section – see Table 3).
Alternatively, this moment can be obtained by subtract-
ing the contribution of the portions within the central 2hn 2. Another location of hn is needed to determine Point E.
region from the maximum moment (MD). Thus: A convenient location to choose is the tip of the flanges.
This is the case shown in Table 3 as Point E. The ex-
M C = MB = M D − Z sn Fy − 2 Zcn (0.85 fc′ ) pressions derived for the cases where the plastic neutral
axis is within the steel section (Points B and C) are valid
Z sn = t w hn 2
for the calculation of the values at Point E, except that
Z cn = h1hn 2 − Z sn hn = bf /2.
N
Z = ∑As r ei , where N is the numbeer of bars within 2hn
r i =1
i
Rectangular Concrete-Filled Steel Tube (RCFT)
7. If step (5) resulted in the location of hn not being in The procedure for determining the axial load-flexure interac-
the web, the next assumption is that it will be within tion diagram for rectangular concrete-filled tubes is similar
the flange. For this case, the expressions for hn and Zs to that described above for SRC, and the resulting values are
become: shown in Table 4.
hn =
( )
0.85 fc′ Ac + As − db f − 2 Fy As − db f( ) Circular Concrete-Filled Steel Tube (CCFT)
 ( )
2  0.85 fc′ h1 − b f + 2 Fy b f 
 The resulting values for CCFT sections are shown in Table 5.
Note again that in the definitions of θ and Zs for Point B,

Z sn = Z s − b f
d
2 (− hn
d
2 )( d
)
+ hn , − t f ≤ hn ≤
2
d
2
changes have been made to correct errors in the table that
appear in the AISC 13th Ed. Steel Construction Manual CD
Companion accompanying the AISC Manual (AISC, 2005b).
Fig. 7. Moment at Points B and C: stress resultants corresponding to the subtraction of cases
(b) and (c) in Table 1 and leading to the calculation of hn.

Stability Considerations Pr
(b) For < 0.2
Once the cross-sectional strength has been established, this Pc
interaction surface needs to be reduced to account for: (a) sta-
bility effects and (b) design strength as opposed to nominal Pr  M rx M ry 
+ +  ≤ 1.0 (AISC H1-1b)
strength. To account for stability, the usual column formula 2Pc  M cx M cy 
has been used in conjunction with an equivalent moment of
inertia. This is straightforward, although it results in a sub- where
stantial difference in the approach to stability from that given, Pr = required compressive strength, kips (N)
for example, in ACI 318 (ACI, 2005) for composite columns. Pc = available compressive strength, kips (N)
The next step, the reduction from nominal to design loads, Mr = required flexural strength, kip-in. (N-mm)
is not so simple due to the fact that as the failure shifts from Mc = available flexural strength, kip-in. (N-mm)
tension yielding at low axial loads to compression at loads x = subscript relating symbol to strong axis bending
above the balance point, it will seem that the overall factor y = subscript relating symbol to weak axis bending
for the member should change. In reinforced concrete design
For this case, the safety and resistance factors from Section
this is achieved by changing the resistance factor from 0.90
I4 are applicable:
to 0.65 as the strain in the extreme tensile fiber goes from
0.0005 to the yield strain of the steel. This factor is applied to φc = 0.75 (LRFD) Ωc = 2.00 (ASD)
both the moment and axial force components. AISC has cho-
φb = 0.90 (LRFD) Ωb = 1.67 (ASD)
sen not to use that approach and to retain separate resistance
factors and safety factors for axial loads and flexure. This, Similar equations may be used for the case of axial tension
and the desire to provide simplified approaches for design, plus flexure.
has resulted in three separate approaches to checking the
strength of a composite beam-column: (1) an approach based Method 2: Full Plastic Strength Approach Based on
on the use of the existing interaction formulas in AISC 2005 Polygonal Interaction Envelope
Specification Chapter H (AISC, 2005a); (2) a more complex
This approach requires the calculation of the full interaction
approach based on the complete polygonal interaction dia-
diagram or a reduced set thereof (for example, the four or five
gram; and (3) a simplified version of the polygonal approach
points shown in Tables 2 through 5). The axial load values
that uses only one intermediate point. A description of these
approaches follows.
Method 1: Approach Based on AISC 2005 Chapter H

For this case, only the nominal axial strength including sta-
bility effects, Pn, and the nominal flexural strength of the
section, Mn, need to be computed. This makes this approach
particularly useful for biaxial bending cases where the engi-
neer does not want to compute additional interaction points
for combined axial force and flexure. The anchor points (for
uniaxial flexure) can be based on the equations for Points A
and B given in Tables 2 through 5, with the stability reduc-
tion included in Pn. The interaction Equations H1-1a and
H1-1b from Chapter H (AISC, 2005a) are then used directly
(Figure 8):
Pr
(a) For ≥ 0.2
Pc
Pr  M rx M ry  Fig. 8. Schematic representations of

+ +  ≤ 1.0 (AISC H1-1a) Methods 1 and 2 for checking
Pc  M cx M cy 
axial load and flexure interaction.

are then reduced to take into account stability effects and Pcb = axial compressive strength at balanced moment,
also reduced by the following safety and resistance factors: Mcb, kips (N)
Mr = required flexural strength, kip-in. (N-mm)
φc = 0.75 (LRFD) Ωc = 2.00 (ASD)
Mc = available flexural strength, kip-in. (N-mm)
φb = 0.90 (LRFD) Ωb = 1.67 (ASD) Mcb = balanced moment, kip-in. (N-mm)
x = subscript relating symbol to strong axis
The corresponding interaction diagram is shown in Figure 8, bending
where the subscript d indicates that the values are the design y = subscript relating symbol to weak axis bending
ones (in other words, including a slenderness reduction and
the resistance factor). The number of checks and equations Similar equations as those given for Method 1 may be used
needed increase substantially as the number of points used to for the case of axial tension plus flexure.
define the envelope increases. The complete set of equations One issue with using a different set of resistance and safety
needed for the case of checking the resistance between the factors for flexure and axial force is the possibility that upon
five points is too lengthy to be included here and thus only a the application of the stability reduction coupled with the
simplified case will be illustrated. For a possible polygonal resistance or safety factors, the resulting available strength
approach to the interaction diagram using Points A, B, and D envelope may fall outside the cross section strength envelope
only, the checks then become: in the area immediately below the balance point (Figure 9).
A simple way has not yet been determined for including this
If Pn < PD added strength without encountering this potential uncon-
servative design area within the context of accounting for
M rx M ry stability using the current AISC Specification.
and if ≤ 1 and ≤ 1 then
M cx M cy
Method 3: Simplified Approach Based on Polygon
M rx M ry
+ ≤1 (1) Values
M cx M cy
To maintain some of the substantial strength gains from the
strain compatibility approach but to simplify the design pro-
M rx M ry cess, a third approach has been proposed. In this approach,
otherwise if > 1 and ≤ 1 then
M cx M cy a third anchor point, Cd, is used in addition to points Ad and
Bd as seen in Figure 10. The new Point Cd is derived from the
Pr M cbx − M rx M ry
+ + ≤1 (2) flexural design strength of the member (Mn from Point B)
Pcb M cbx − M cx M cy and the corresponding axial strength from Point C (See Fig-
ure 3), with appropriate reduction taken to account for slen-
M rx M ry derness effects and resistance or safety factors as per Leon
otherwise if ≤ 1 and > 1 then et al. (2007). Similar equations as those given for Method 2
M cx M cy
may be used for the case of axial tension plus flexure.
Pr M rx M cby − M ry
+ + ≤1 (3)
Pcb M cx M cby − M cy
M rx M ry
otherwise if > 1 and > 1 then
M cx M cy
Pr M cbx − M rx M cby − M ry
+ + ≤1 (4)
Pcb M cbx − M cx M cby − M cy
If Pr ≥ Pcb
Pr − Pcb M rx M ry
+ + ≤1 (5)
Pc − Pcb M cbx M cby
where
Pr = required compressive strength, kips (N)
Pc = available compressive strength, kips (N) Fig. 9. Possible breaching of ultimate strength envelope
by design envelope due to application of resistance
and stability reduction factors.

Design Examples Point A in Table 2 (M = 0)

Determine the available compressive strength and moment
Example 3: SRC Beam-Column strength.
In this example, a simplified axial load-moment envelope Ac = (324 − 14.1 − 2.40) = 308 in.2
is developed for an 18 in. × 18 in. SRC beam-column with
an embedded W14×48 section. Additional reinforcement Po = AsFy + AsrFyr + 0.85Ac fc′
consists of four #7 corner bars and #3 ties spaced at 12 in.
Materials used include Fy = 50 ksi, Fyr = 60 ksi, and fc′ = Po = (14.1)(50) + (2.40)(60) + 0.85(308)(3)
3 ksi. Flexure is about the major axis, the effective length is = 1,630 kips
24 ft, and buckling is restrained about the minor axis.
 As   14.1 
Limitations: C1 = 0.1 + 2   = 0.1 + 2 
 Ac + As   308 + 14.1
1) Normal weight concrete 10 ksi > fc′ > 3 ksi; in this case = 0.188 ≤ 0.3
fc′ = 3 ksi o.k.
overall depth − 2 (cover) − 2(stirrup) − bar diameter
2) Fyr < 75 ksi; in this case, Fyr = 60 ksi o.k. e1 =  
 2
3) The cross-sectional area of the steel core shall comprise 18 − (2)(1.5 ) − (2)(0.375) − 0.875
at least 1% of the total composite cross section. e1 =   = 6.69 in.
 2
As = 14.1 in.2 > (0.01)(324 in.2) = 3.24 in.2 o.k.
Isr ≈ ∑ (A sri )(e )
i
2
= 2(1.20 )(6.69)2 = 107 in.3
4) Concrete encasement of steel core shall be reinforced
hh 3  (18)(18)3 
with continuous longitudinal bars and lateral ties or spi- I c =  1 2  − Is − Isr =  12  − 484 − 107 = 8, 160 in.
4
rals. The minimum transverse reinforcement shall be at  12 

least 0.009 in.2 of tie spacing:  h1h2 3
 (18)(18)3 
I =
ρst = 0.22 in. /12 in. = 0.0183 in. /in. > 0.009
2 2 c 2 − I
in.12/in. s sr− I =  12  − 484 − 107 = 8, 160 in.4
 
o.k.
( )
1.5
Ec = 148.1 lb/ft 3 3 ksi = 3, 122 ksi
5) The minimum steel ratio for continuous longitudinal
reinforcing, ρsr, shall be 0.004: EI eff = Es Is + 0.5Es I sr + C1 Ec I c
EI eff = Es Is + 0.5Es I sr + C1 Ec I c EI eff =E (29I + 0.5Es I sr) + (C01.E
s s, 000)(484 5)( I c , 000)(107) + (0.188)(3,122 )(8, 160)
c 29
Asr (4 × 0.6) eff
ρsr = = = 0.0074 > 0.004 o.k. = )(20 ) + ()0(8.5, 160
)(29), 000)(107) + (0.188)(3,122 )(8, 160)
Ag 324 EI eff = (29, 000)(484) + (0.5)(29, 000 EI)(efef107
ff
+.4(0
29 , 0x.00
10)(484
188 6
)(kip-in.
3,122 2
EI eff = 20.4 x 10 kip-in.

6 2
EI eff = 20.4 x 10 kip-in. 6 2
Note that if buckling had not been prevented about the minor
axis, the Is to be used in the computation of Ieff would have
been Iy rather than Ix.
π 2 EI eff π 2 (20.4 × 10 6 )
Pe = = = 2, 430 kips
( kL )2 (24 × 12)2
Po 1,630
= = 0.67 < 2.25
Pe 2,430
or
Pe 2, 430
= = 1.49 > 0.44
Po 1,630
∴ Use Equation (AISC 2005 I2-2)
Fig. 10. Simplified interaction diagram for LRFD design.

  Po 
 Before computing Mb, MD must be computed:
 
( 0.67) 
Pn = Po 0.658 e  = (1,630 kips)  0.658

( )
P 
= 1,230 kips M D = Z s Fy + Z r Fy r + 2 Zc 0.85 f c′
   
 
φcPn = (0.75)(1,230) = 923 kips (LRFD)

Zr = ∑A sri
 h2 
( )
 2 − c  = ∑ Asri ei = 4(0.6)(6.69)
Pn/Ωc = (1,230 kips/2.00) = 615 kips (ASD) = 16.1 in.3

 h h2   (18)(18) 2 
Point B in Table 2 (PB = 0) Zc =  1 2  − Z s − Z r =   − 78.4 − 16.1
 4   4 
Determine the
Determine locatio n of hof h
location
Determine location of hnn n n = 1,360 in.3
 d 13.8 in. 
1) A
1) bove th
Above the lange oorr hd> 13.8
e fflange
1) Above the flange or  hnn> n =2
= in.   == hhnn > > 6.90 in
= h > 6.90 in.
in.. M D = (78.4)(50) + (16.1)(60) + 2 (1, 360) ( 0.85)(3 )
 2 2 2   nn
) )
0.85 fc′( Ac + As − 2 Fy As = 6, 620 kip-in.
hn0=.85 fcc′( Acc + Ass − 2 Fyy Ass
hnn = 2 ( 0.85f ′h
2 ( 0.85fcc′h11 c 1 ) )
M B = M D − Z sn Fy − 2 Zcn 0.85 fc′ ( )
= =
)( )(
0.805.(835.0( 3.0308308 + 14.1
+ 14.1 − 2−( 50 ) ) )( )( ) )
2 ( 5014.14
1 .1
= (6, 620) − (8.36)(50) − 2(434) ( 0.85)(3)
2( 0.85 ( ( ) ) ) )) )
2 ( 0.853.0( 3.018(18 = 5,650 kip-in.
= −6.41 in. < 6.90 in., so hn is not outside the steel φbMB = (0.90)(5,650) = 5,090 kip-in.
section MB/Ωb = (5,650/1.67) = 3,380 kip-in.
Fo hhn within the flange 
2)2) For (  2
d
)
− t f < hn ≤
d  or 6.31 in.< h ≤ 6.90 in.
2   n Point C  (MC = MB; PC = 0.85 fc′Ac)
Fo hhor 6.31in.<0h.8n5≤fc6.90
2) Fo
2)
hn =
( (( 
′ Acin.
dd
 22 s f )) )
+ −A (
− tt−f d<b<hf hnn−≤≤2 Fy  Ad d  or
or
22  
)
−d6.31
s 6.31
in.<hhnn ≤≤66.9.900P
b f in.< )
inin..= A ( 0.85 f ′
C c c
hhnn ==
(( 2  0 . 85 ())
f ′ h
()(
− b +
00.8.855fcf′c′ AAc c++AAs s −−ddcbbf f −1−22FFyfy AAs s −−dydbbff f 2 F
))
b 
( ) ) )
= 308 in. ( 0.85 ( 3.0 ksi = 7785 kips
2
=
2200.08.85
 ((
(3ff)′c(′ 308
.585 c ))
hh11−+−b14
 
)
bf f .1++−22(F1Fy3ybb.f8f)(8.03  φPC = 0.75(785)(0.658)0.67 = 445 kips
d
00.8.855((33))((308
 308++14 2  0 . 85 (
))
3 ) ( 1
14.1.1−−(1(133.8.8)()(88.0.033  8 −
 8 .)03 + 2 (5 0 )( 8 . 03) 
 Cd Ω = (784)(0.658)
P 0.67
2.00 = 296 kips
==  

.85((33))((1188−−88.03
2200.85 )) .03 ++22((5500)()(88.03.03))

M C = M B = 5, 6650 kip-in.
φM C = φM B = (0.9)5, 650 kip-in. = 5,0090 kip-in.
−
2(550) (14.1 − (13.8)(8.03) 
 )  M C Ω = M B Ω = 5, 650 1.67 kip-in. = 3,3880 kip-in.

)
2  0.85(3) (18 − 8.03 + 2(50)(8.03) 
Point D
= 6.16 in., so hn is in the web
)
PD = c
A ( 0.85 fc′
Since h n is within the web and thus no rebars are present 2
Since hwithin
n
is w h nin(Athsre=w0):
ith eb an d t hus no re ba rs
r ar e p re se nt w ith in h(
n
( A sr
= )
0) : ) =
)
308 in.2 ( 0.85 ( 3.0 ksi
0.85fc′ Ac 2
hn =
)
2  0.85fc′ ( h1 − t w + 2 Fy t w  = 393 kips
=
)
0.85 ( 3 (3308) φPD = 0.75(393)(0.658)0.67 = 223 kkips
d
)
2  0.85(3) (18 − 0.340 + 2(50)(0.340)  PD Ω = (393)(0.658)0.67 2.00 = 148 kips
d
= 4.996 in. M D = 6,620 kip-in.

Z sn = t w hn2 = (0.340)(4.96)2 = 8.36 in.3 φM D = (0.9)6,620 = 5,960 kip-in.
Z cn = h h − Z sn = (18)(4.96) − 8.37 = 434 in.
2
1 n
2 3
M D Ω = 6,620 1.67 = 3,960 kip-in.
The results are summarized in Figure 11.

Example 4: RCFT Beam-Column Point A in Table 4 (MA = 0)

Determine the available compressive strength and flexural
Develop a simplified axial load-moment envelope for a
strength.
HSS16 in.×16 in.×s in. filled with f c′ = 4 ksi concrete. The
effective length of the member is 24 ft. Assume A500 Grade In the following calculations the symbol “≈” is used to
B (Fy = 46 ksi and Fu = 58 ksi). Note that the design thickness indicate that the effect of the corner radii is not being ac-
for a 0.625 in. nominal value is 0.581 in. counted for exactly; this effect is small and can generally be
neglected.
Limitations:
Po = AsFy + AsrFyr + 0.85Ac fc′
1) The cross-sectional area of the steel core shall comprise
at least 1% of the total composite cross-section. Ac ≈ 256 − 35.0 = 221 in.2
As = 35.0 in.2 > (0.01) (16)2 = 2.56 in.2 o.k. Po = (35.0)(46) + 0.85(221)(4)
Note that ρ = 35 = 0.137, or 13.7% which is very high. = 2,360 kips
256
 As   35 
C3 = 0.6 + 2  = 0.6 + 2 
2) The slenderness of the tube wall is:

 Ac + As   (221 + 35)
= 0.873
¥ b ´ 16 (2) 0.581 E 29,000
= 56.7 Is = 1,370 in.
4
¦§ t µ¶ = 0.581 = 25.5 < 2.26
Fy
= 2.26
46
 d4   (16)4 
≈ − Is =  12  − 1,370 = 4, 090 in.
4
I 
¥ b´ 16 (2) 0.581 E 29,000  12 
c
¦§ t µ¶ = = 25.5 < 2.26 = 2.26 = 56.7 o.k.

0.581 Fy 46
Ec = (148.1 lb/ft 3 )1.5 4 ksi = 3, 605 ksi
SRC: 18x18, W14×48, 4-#7 bars, fc′ = 3 ksi, Fy = 50 ksi, KL = 24 ft
Fig. 11. Interaction diagrams for column in Example 3.

EI eff = E s I s + 0.5Es I sr + C3 Ec I c φbMB = (0.9) (10,200) = 9,180 kip-in.

= (29, 0000)(1, 370) + (0.873)(3, 605)(4, 090) MB/Ωb = (10,200/1.67) = 6,110 kip-in.
= 52.6 × 10 kip-in. 6 2
Point C in Table 4 (MC = MB; PC =0.85 fc′Ac)
π 2 EI eff
π 2 (52.6 × 10 6 )
Pe =
( KL ) 2
=
(24 × 12)2
= 6, 2260 kips PC = Ac ( 0.85 fc′ )
Po 2,360 PC = 221 in. ( 2
) ( 0.85) ( 4.0 ksi ) = 751 kips
= = 0.38 < 2.25
Pe 6,260 φPC = (0.75)(751)(0.658)0.38 = 480 kips
d
Pe 6, 260 PC Ω = 751(0.658)0.38 2.0 0 = 320 kips

or = = 2.65 > 0.44 d
P0 2,360 M C = M B = 10,200 kip-in.

∴ Use Equation (AISC 2005 I2-2) φM C = (0.9)(10, 200) = 9, 180 kip-in.
M C Ω = 10, 200 1.67 = 6,110 kiip-in.
  Po 
 

Pn = Po 0.658 Pe   = (2,360) 0.658(0.38)

  Point D in Table 4
 
= 2,010 kips PD =
Ac ( 0.85 fc′ )
2
φcPn = (0.75) (2,010) = 1,510 kips (LRFD)
=
( 221 ) ( 0.85) ( 4.0 ksi )
Pn / Ωc = (2,010/2.00) = 1,010 kips (ASD) 2
= 376 kips
Point B in Table 4 (PB = 0)
φPD = (0.75)(376)(0.658)0.38 = 241 kips
De ter
termin
Determine
De mineee llllocation
oocatio
cationnn oooof ff hhhn n d
De ter min ocatio f hnn
00..8855 ffc′′ A hh2 PD Ω = (376)(0.658)0.38 2.00 = 160 kips
0.85 fcc′ A Accc h22
d
hhn ==  ≤
≤
hnn = 22  00..85 ffc′′ hh1 + 44tt w F  22 ≤ M D = 10,600 kip-in.
85 +
2  0.85 fcc′ h11 + 4t ww Fyy  2 F y 
φM D = (0.9)(10, 600) = 9, 540 kip-in.
00..85 (4 )(221 )
=
= 2 0.85(4)[16 − (2)(0.0581) .85
85((44 )(
)(221
221)) M D Ω = 10, 600 1.67 = 6,350 kiip-in.
= )(46 )) 
22 0.85(4)
0.85(4)[[16 16 − − (2 (2)( 581)]]] +
)(00..581) + 444 (((000...581
+
581)(
581)(46
46 ) 
((16
16 − − ((22))((00..581
581)) )) The results are summarized in Figure 12.
= 22..39
= 39 in. ≤ ≤ (16 − (2)(0.581)) = = 7.42 in. in.
= 2.39 in. in. ≤ = 77..4242 in.
222
Example 5: CCFT Beam-Column
Before computing MB, MDmust be computed. Determine the interaction diagram for a 20 in. diameter,
Z s = 200 in. 3 a in. thick circular concrete-filled tube column with a
KL = 13 ft. Assume Fy = 42 ksi and fc ′ = 5 ksi.
Zc =
h1h22
− 0.192ri ≈  
3
(
  16 − ( 2)(0.581  3 
  = 817 in.3 )
4  4 
 
M D = Z s Fy + 2 Zc 0.85 fc′ ( )
= (200)(46) + 2 (817)(0.85 × 4)
= 10,600 kip-in.
Z sn = 2t w hn2 = 2 (0.581)(2.39) 2 = 6.64 in.3
Z cn = h1hn2 = [16 − (2)(0.581)](2.39)2 = 84.8 in.3
M B = M D − Z sn Fy − 2 Zcn 0.85 fc′ ( )
= 10, 600 − (6.64)(46) − 2 (84.8)(0.85)(4)
= 100,200 kip-in.

Basic geometrical properties: The slenderness of the tube wall is:

d = 20 in. E 
D
t = 0.349 in. = 57.3 < 0.15  s  = 104 o.k.
t  Fy 
h = d − 2 t = 19.3 inn.
d −t Point A in Table 5 (MA = 0)
rm = = 9.83 in
2 Determine the available compressive strength using the
As = 2πrm t = 21.6 in.2 properties determined above.
πh 2 C2 = 0.95
Ac = = 293 in.2
4
  As  
Ag = Ac + As = 315 in.2 C3 = min  0.9, 0.6 + 2    = 0.74
  Ac + As  
Ec = (148.1 )1.5
5 = 4,030 ksi
π EIeff = EsIs + EsIsr + C3Ec Ic = 351,000 kip-ft2
Is =  d 4 − h 4  = 1,040 iin.4
64   Po = AsFy + AsrFyr + C2 Ac fc′ = 2,300 kips
4
πh π 2 EI eff
Ic = = 6,810 in.4 Pe = = 20,500 kips
64
( KL )
2
I g = I s + I c = 7,8850 in.4 c
The cross-sectional area of the steel core shall comprise at Po 2, 300

= = 0.112
least 1% of the total composite cross-section. Pe 20, 500
As
= 0.07 > 0.010 o.k.
Ac + As
RCFT: 16 × 16 × s, fc′ = 4 ksi, Fy = 46 ksi, KL = 24 ft

 Po
 φbMB = (0.9)(6,870) = 6,180 kip-in.
P
If Pe > 0.44 Po , Pn =  0.658 e  Po = 2,190 kips
  MB/Ωb = (6,870/1.67) = 4,110 kip-in.
φcPn = (0.75)(2,190) = 1,640 kips Point C in Table 5

Pn/Ωc = (2,190 kips/2.00) = 1,100 kips PC = 0.95fc′Ac = 1,390 kips
Point B in Table 5 (PB = 0) M C = M B = 6,870 kip-in.
From definitions of Point B in Table 5: Slenderness reductions on the axial strength and application
of resistance and safety factors to axial and flexural strength
K c = fc′ h 2 = 1,860 kips should be taken as per Example 4 and Leon et al. (2007).
K s = Fy rm t = 144 kips
Point D in Table 5
0.0260 K c − 2 K s ( 0.0260 K + 2K s ) 2
+ 0.857 K c K s
θ=
0.0848 K c
+
c
0.0848 K c
= 2.19
PD =
( 0.95 f ′) A
c c
= 696 kips
2
θ=
0.0260 K c − 2 K s
+
( 0.0260 K c
+ 2K s ) 2
+ 0.857 K c K s
= 2.19
Z c = h 3 6 = 1, 200 in.3
0.0848 K c 0.0848 K c
= 2.19 rad ( )
Z s = d 3 6 − Z c = 133 in.3
Z cB =
h 3 sin 3 ( θ 2 ) = 842 in. 3
M D = Z s Fy +
) = 8,440 kip-in.
Z c ( 0.95 fc′
6 2
d −h
3 3
Z sB ≈
6
sin( ) ( θ 2 = 116 in.3
4 /3
) Slenderness reductions on the axial strength and application
of resistance and safety factors to axial and flexural strength
should be taken as per Example 4 and Leon et al. (2007).
M B = Z sB Fy +
(
Z cB 0.95 fc′ ) = 6,870 kip-in. The results are summarized in Figure 13.
2
CCFT: 20 × a, fc′ = 5 ksi, Fy = 42 ksi, KL = 13 ft

CONCLUSIONS Es = modulus of elasticity of steel, which shall be taken

as 29,000 ksi (210 MPa)
This paper presents the background on the step-by-step
procedures available in the 2005 AISC Specification (AISC, EIeff = effective stiffness of composite section, kip-in.2
2005a) for computing composite column and beam-column (N-mm2)
strength, including accounting for stability effects on mem-
fc′ = specified minimum concrete compressive strength,
bers subjected to biaxial flexure plus axial compression.
ksi (MPa)
The new procedures highlighted in the 2005 AISC Speci-
fication are discussed, including the use of a plastic stress Fy = yield strength of steel section, ksi (MPa)
distribution method that accounts for the beneficial effects
Fyr = specified minimum yield strength of reinforcing
of the concrete to the strength of the cross-section. Several
bars, ksi (MPa)
examples are given for calculating the design strength of
encased composite sections (SRC) as well as filled compos- h1 = depth of the section
ite sections, both rectangular (RCFT) and circular (CCFT).
h2 = width of the section
ACKNOWLEDGMENTS Ic = moment of inertia of the concrete section,
in.4 (mm4)
This research was sponsored by the National Science
Foundation under Grant No. CMS-0084848, the American Is = moment of inertia of steel shape, in.4 (mm4)
Institute of Steel Construction, the Georgia Institute of Tech-
Isr = moment of inertia of reinforcing bars, in.4 (mm4)
nology, the University of Minnesota, and the University of
Illinois at Urbana-Champaign. The authors would like to K = effective length factor determined in accordance
thank the members of AISC Specification Task Committee with Chapter C
5 on Composite Construction for their contributions to this
L = laterally unbraced length of the member, in. (mm)
research. The assistance of Tiziano Perea, Cenk Tort, Mat-
thew Eatherton, Arvind Goverdhan and William Jacobs with N = number of longitudinal reinforcing bars
verifying the design examples is gratefully acknowledged.
tf = flange thickness, in. (mm)
APPENDIX A tw = depth of steel section, in. (mm)
wc = weight of concrete per unit volume (90 ≤ wc ≤ 150,
NOMENCLATURE
lb/ft3 or 1,440 ≤ wc ≤ 2,450, kg/m3)
As = area of the steel section, in.2 (mm2)
REFERENCES
Ac = area of concrete, in.2 (mm2) = Agross – A s – A sr
ACI (2005), Building Code Requirements for Structural
Asr = area of continuous longitudinal reinforcing bars,
Concrete, ACI 318-05, American Concrete Institute,
in.2 (mm2)
Farmington Hills, MI.
Asr = area of a longitudinal bar, in.2 (mm2) AISC (1999), Load and Resistance Factor Design Specifica-
Asrn = area of longitudinal bars within the 2hn region, tion for Structural Steel Buildings, American Institute of
in.2 (mm2) Steel Construction, Chicago, IL.
AISC (2001), LRFD Manual of Steel Construction, 3rd Edi-
Agross = total area of member, in.2 (mm2) = h1 h2
tion, American Institute of Steel Construction, Chicago,
bf = flange width, in. (mm) IL.
c = cover to centroid of longitudinal bars, in. (mm) AISC (2005a), Specification for Structural Steel Buildings,
ANSI/AISC 360-05, American Institute of Steel Con-
d = depth of steel section, in. (mm) struction, Chicago, IL.
ei = eccentricity of bar i with respect to centroid, AISC (2005b), Steel Construction Manual, 13th Edition,
in. (mm) American Institute of Steel Construction, Chicago, IL.
Ec = modulus of elasticity of concrete = wc 1. 5 fc′ ksi Kim, D.K. (2005), “A Database for Composite Columns,”
(= 5000 fc′ , MPa) M.S. Thesis, School of Civil and Environmental Engi-
neering, Georgia Institute of Technology, Atlanta, GA.

Leon, R.T., Kim, D.K., and Hajjar, J.F. (2007), “Limit State
Response of Composite Columns and Beam-Columns:
Formulation of Design Provisions for the 2005 AISC
Specification,” Engineering Journal, AISC, No. 4, 4th
Quarter, pp. 341–358.
Roik, K. and Bergmann, R. (1992),. “Composite Columns,”
in Constructional Steel Design, Dowling, P., Harding, J.E.
and Bjorhovde, R. (eds.), Elsevier Science Publishers,
New York, pp. 443–470.
Viest, I.M., Colaco, J.P., Furlong, R.W., Griffis, L.G., Leon,
R.T., and Wyllie, L.A., Jr. (1997), Composite Construc-
tion: Design for Buildings, McGraw-Hill, New York, NY.






Investigation of Flange Local Bending
Under Flexible Patch Loading
LYLE P. CARDEN, GOKHAN PEKCAN and AHMAD M. ITANI
W hen a rigid patch or line load is applied to a beam, the

flange is only able to deform as allowed by deforma-
tion of the web, thus the capacity of the web controls the de-
in the column flange assumes that a tensile line load is ap-
plied to the column flange from the beam flange. A yield
line pattern in the column flange and a uniform stress dis-
sign. However, in applications where a patch load is flexible, tribution from the beam flange is assumed in order to derive
it is possible for the flange to bend around the web, which an equation to calculate the ultimate capacity of the flange
may limit the capacity of the beam. A flexible patch load is (Graham, Sherbourne and Khabbaz, 1959). However, this
considered one where the load is able to deform as the loaded equation, which is based on the need to limit tensile stresses
member deforms, resulting in a more uniform distribution of in the connection of the beam flange at the column web, is
applied stress than when a load is applied through a rigid conservative if applied to a compressive patch load. The con-
element. A recent study by the authors (Carden, Pekcan and servatism is in part because a patch load is able to engage a
Itani, 2005) has shown that flange bending may be the criti- larger region in resisting bending of the flange than a linearly
cal limit state in applications such as bridge falsework where applied load from an adjoining flange. In addition weld frac-
timber posts bear on unstiffened steel beams. The resulting ture is not the concern resulting from flange bending in this
damage mechanism illustrated in Figure 1 has the potential application.
to lead to instabilities and collapse of bridge falsework dur- Two methods that account for the effect of patch loading
ing construction. from timber and steel posts on beams were developed else-
Past studies on the effect of patch loading (Roberts and where (Carden et al., 2005) and are summarized in what fol-
Rockey, 1979; Roberts and Markovic, 1983; Elgaaly, 1983; lows. In the first, the interaction between the flange bending
Roberts and Newark, 1997; Shahabian and Roberts, 1999; capacity and the post compression capacity was considered.
Graciano and Edlund, 2003) have generally assumed rigid If the flange is significantly weaker than the post, it will tend
patch loads which have not resulted in flange bending other to dictate the strength of the joint region. On the other hand,
than that required for deformation of the web. The case of a
flexible patch load provided by a timber post has not been
considered, thus there is currently no design methodology to
assess flange bending capacity under this loading condition.
The AISC Specification for Structural Steel Buildings
(AISC, 2005), hereafter referred to as the AISC Specifica-
tion, allows for the calculation of flange bending capacity in
beam-column joints. The ultimate flange bending capacity
Lyle P. Carden is a structural engineer, Martin & Chock,

Inc., Honolulu, HI.
Gokhan Pekcan is assistant professor, department of civil
and environmental engineering, University of Nevada,
Reno, NV.
Ahmad M. Itani is professor, department of civil and envi-
ronmental engineering, University of Nevada, Reno, NV. Fig. 1. Flange bending in sill beam
(courtesy of J. Lammers, Caltrans).
047-056_Carden_Pekcan_Itani_2008_1Q.indd 47 3/27/08 1:34:07 PM

if the post compression strength is notably lower than the FLANGE-POST CAPACITY
beam flange capacity then the post will dictate the overall
response. If the flange and post have similar strengths then Interaction Method
the joint capacity will tend to be smaller than the capacity
The flange bending capacity, Rf, due to a flexible patch load
of either component. The second method uses an effective
similar to that associated with bearing of a timber post can
bearing area for the post, which accounts for flange bend-
be described as
ing capacity in a secondary manner by reducing the effective
area of the post depending on the thickness of the flange.
R f = β t f 2 Fyf (1)
The first method was found to be most effective for predict-
ing the flange-post bearing joint strength with a timber post;
however, with a steel post, which is axially rigid, the sec- where
ond method leads to more accurate assessment of the joint β = a constant determined from a yield line analysis
strength. (Carden et al., 2005)
The above methods were developed assuming a concen- tf = thickness of the flange
tric loading, without the use of any blocking, which is some- Fyf = specified minimum yield stress of the flange
times used in attempts to strengthen the joint region. This
Expressions for β were developed assuming a uniform and
paper describes a series of experiments and finite element
triangular stress distribution for the patch load, as shown in
analyses to investigate the effectiveness of the above meth-
Figure 2. However, for a nominal 12 × 12 in. post, a β value
odologies for predicting the flange-post joint capacity; con-
equal to 18 was found to be appropriate for calculating the
sidering eccentricities between the centroids of the beam and
flange bending capacity for a range of beams with flange
post, and the use of blocking. The methodologies are then
thicknesses between 0.44 and 0.94 in. subjected to concen-
developed into a form appropriate for design using load and
tric patch loading.
resistance factor design (LRFD) and allowable stress design
(ASD) procedures.
(a) (b)
Fig. 2. Patch load assuming: (a) a uniform stress distribution and (b) a triangular stress distribution.

In order to calculate the capacity of the flange-post bear- k1 = distance from the center of the web to the edge
ing region it is also necessary to consider the compression of the fillet
capacity of a short timber post which is given by Fyp = specified minimum yield stress of the post
tp = wall thickness of the round hollow steel section
Pp = Fc′Ap (2)
The 2tp allows for the transfer of axial load on two sides of
the round hollow steel post section.
where
Ap = nominal cross-sectional area of the post Strength of Blocking
Fc′ = specified minimum compression stress of the
Falsework design typically avoids the use of stiffeners, as
timber blocking after applicable modification
beams are reused and construction is temporary. However,
factors are applied (AFPA, 1996)
in some cases timber blocking is used in an attempt to stiffen
As the compression capacity in the bearing joint region is the beam flanges. Generally timber blocks sized between
considered over a short length, the post stability factor in 4 × 4 in. and 6 × 8 in. are used, the capacity of which can be
this calculation is equal to 1.0. It is noted that stability of calculated based on the axial compression capacity of a short
the post should be considered as a separate limit state in the timber member. However, experimental and analytical stud-
design of the post. ies showed that the full capacity of the blocking is generally
The strength of the joint region can then be determined not effective, particularly for a steel post, thus the capacity
from the interaction of the flange bending and post compres- of the timber blocking, Pb, can be given by
sive strength using the following relationship
Pb = γ Fc ′Ab (5)
2 2
⎛ Pu ⎞ ⎛ Pu ⎞
⎜R ⎟ + ⎜P ⎟ ≤1 (3) where
⎝ f ⎠ ⎝ p⎠ γ = blocking effectiveness factor
Ab = combined cross-sectional area of the blocking
where on both sides of the web
Pu = applied axial load in the post
Bearing Area Method

This method focuses on the post capacity for calculating the
ultimate load, with the impact of flange bending inherent but
not explicitly considered in the formulation. The method is
found to be more effective for a steel post where the post is
axially rigid and thus does not allow the flange to deform
before failure of the post, unlike a more flexible timber
post (Carden et al., 2005). For this model an effective cross-
sectional area of the steel post is considered to carry the
entire axial load, as shown in Figure 3, and can be calculated
using similar assumptions to those used in the web yielding
equation given by the AISC Specification (AISC, 2005). The
effective area of the post is determined by the width of the
post that is effective in carrying the load multiplied by the
thickness of the post walls. An equation for the capacity of a
steel post is given by
⎣ (
Pp = ⎡⎢ α t f + tbp ) + 2k ⎤⎦⎥ 2t F
1 p yp (4)
where
α = constant which depends on the slope of the
stress gradient assumed through the flange Fig. 3. Effective area of steel post for calculating
tbp = thickness of the base plate of the post the ultimate post load.

EXPERIMENTAL STUDIES obtained from the two experiments, as timber corbels (mem-
bers located under the sill beam in typical bridge falsework
Experimental Setup and located between the beam and reaction plate in experi-
ments similar to that shown in Figure 4) were included in
A series of experiments were performed to study different
these experiments and governed the failure response.
limit states in a flange-post bearing joint region. A total of
15 experiments focused on quantifying the flange-post joint
Experimental Results
capacity; 13 with timber posts and two with steel posts. The
experimental setup with a timber post and steel beam is The force measured in the actuator was plotted against the
shown in Figure 4 as an example of one configuration. The axial displacement at the slider end of the post for each ex-
flanges were restrained at the ends of the 48 in. long beam periment. A typical force-displacement curve for a beam con-
segments by the steel frame shown in the figure, prevent- centrically loaded through a timber post without any block-
ing lateral instability of the beams for studying the flange ing is shown in Figure 5a, and that for a beam with blocking
and post limit states. Loads were applied to the end of the is shown in Figure 5b. These figures show that the stiffness
48 in. long posts through a slider to limit lateral deformation initially increases as any gaps due to lack of fit between the
at the end of the post, which was attached to a displacement
controlled hydraulic actuator. The instrumentation consisted
of displacement transducers and strain gages, as well as a
machine-vision camera that allowed the measurement of
strains and deformations. Three different beam sections were
used, including ASTM A572 Gr. 50 HP12×53 and HP14×73
beams, and ASTM A992 W14×90 beams. Number 2 Douglas
Fir 12 × 12 in. timber members were used for the posts and
corbels. An 18-in.-diameter a-in.-thick A500 Gr. B (42 ksi)
round hollow steel section with a 2-in.-thick base plate
was used to simulate a steel post. In some of the experiments,
6 × 8 in. Number 2 Douglas Fir timber blocking was also
placed between the flanges on both sides of the web. The
effects of eccentric patch loading were also investigated
experimentally by introducing eccentricities ranging from
1/12 to 6 of the flange width. The failure load of the joints
between the beams and steel posts could not be directly
(a)
(b)
Fig. 4. Experimental setup with a timber post and steel beam Fig. 5. Force displacement curve for HP14×73 beams:
with blocking and an eccentricity between (a) concentrically loaded with no blocking and
the centroid of the beam and post. (b) concentrically loaded with blocking.

various components close, after which the maximum stiff- similar to the load at which yielding is first observed, and
ness is reached. The stiffness decreases and load then starts thus, is an appropriately defined ultimate load.
to drop as crushing of the post and plastic deformation in the Along with the subassembly experiments, component ex-
flange occurs. Comparison of Figures 5a and 5b shows that periments were performed to determine the axial capacity of
blocking resulted in slightly improved capacity. three timber posts, three timber blocks, and a steel post. The
The ultimate load was defined as the load at which the resulting axial capacity of the timber post, based on an aver-
stiffness reduces to 50% of the initial stiffness, where the age from the experiments, was 352 kips. The average axial
initial stiffness was defined as the stiffness between 25% and capacity of the timber blocks was 163 kips and the capacity
75% of the ultimate load. Accordingly, the ultimate load, de- of the steel post was 888 kips. In addition, eight coupon tests
termined iteratively, is shown by the dashed lines in Figure 5 were performed on samples of steel from the beams result-
and is typically within 10% of the maximum load where post ing in an average yield stress of 53 ksi and a range between
crushing and flange bending occurred. 50 and 58 ksi.
A mixture of suspended lime and water was painted onto For the beams with timber posts, the capacity of the flange-
each beam and was observed to flake off once the steel start- post joint region was calculated using Equations 1, 2, 3 and
ed to yield. For each of the beams, the observed load where 5, when blocking was used, using the actual strengths of the
flaking occurred on the beam flange was recorded, with flak- post, blocking and steel material from the component experi-
ing typically located initially at the edge of the fillet between ments. Equation 1, with a β value of 18, was used to esti-
the flange and web under the post. Figure 6 compares the mate the flange bending capacity. The timber blocking was
load at which flaking was observed with the estimated ulti- assumed to be 100% effective, thus a γ value of 1.0 was as-
mate load based on the relative stiffness of the beams with sumed in Equation 5 and the effective blocking capacity was
timber posts. With the steel posts, flaking was not observed directly added to the flange bending capacity. The blocked
in the beams until after the posts had already yielded and or unblocked flange capacity was then combined with the
thus is not included in the figure. Comparisons for differ- timber post strength from Equation 2 using Equation 3. The
ent beam sizes and different configurations with and without experimental data show that there is a relatively small reduc-
blocking, and with different eccentricities between the post tion in capacity when the post is eccentric to the centroid
and beam are plotted in Figure 6. There are only 12 points of the beam. Thus it is assumed that the above equations
for 13 experiments with timber posts as flaking was not ob- apply equally when there is an eccentricity equal to or less
served at the appropriate time in one of the beams due to than 6 of the width of the flange. The resulting calculated
premature failure of the timber post. As can be seen in the capacity of the flange-post region for the beams with timber
figure, observed and estimated ultimate force is different by posts is compared in Figure 7 to the estimated ultimate force
no more than 13% for each beam, with an average difference from the experimental data, based on a 50% reduction in
of 2%. The comparison therefore indicates that the estimat- stiffness. The estimated forces are approximately equal to or
ed ultimate load, based on a 50% reduction in stiffness, is less than the observed forces, at between 71% and 105% of
Fig. 6. Comparison of ultimate force observed due to flaking Fig. 7. Comparison of calculated ultimate force to experimental
on the flange and ultimate force calculated at a 50% reduction ultimate force based on a 50% reduction in stiffness.
in initial stiffness from force displacement curve.

that observed, for all members. The stagger in the observed at a strain of 10% to reflect the reduction in strength due to
capacities is mainly attributed to the variability of material crushing. Based on measured values from experiments, the
properties, particularly for the timber members. The lime elastic modulus was equal to 550 ksi. The multi-linear model
flaking pattern observed in the beams after considerable in- was found to compare better to the experimental data than
elastic flange bending was very similar to the yield pattern pre-defined material models in ABAQUS. The 7.5 × 5.5 in.
assumed in developing Equation 1. Therefore, this method blocking elements were modeled with a similar model using
is appropriate for predicting the capacity of the flange-post a 4.0 ksi initial yield stress and a 330 ksi elastic modulus, as
joint region with a timber post. determined from the component experiments.
The effective post bearing area method was used to esti- Loads were applied axially to the top of the post in dis-
mate the capacity of the flange-post joint region with a steel placement control until after the maximum load in the sys-
post, based on Equation 4. The observed failure mode in the tem was reached. The interfaces between the beams and
experiments with the steel post was consistent with the effec- other components were modeled with surfaces which are as-
tive bearing area assumed for calculating the ultimate load, sumed to be connected by maintaining a constant geometric
with yielding and crippling of the post occurring in the re- relationship between adjacent nodes. This is similar to a con-
gion where the post was bearing on the beam in line with the tact model with a high friction coefficient between nodes,
web. Unfortunately, the calculated ultimate load could not be but was found to be significantly more computationally ef-
directly compared to the experimental data as timber corbels ficient with comparable results. The nodes under the beams
were also used in experiments with the steel posts and these were completely restrained, while at the end of the post, the
affected the ultimate load of the system. Thus, finite element nodes were restrained to allow axial deformation only. Both
analyses were used for comparison with the calculated ulti- flanges at the ends of the beam were also restrained to pre-
mate load for cases with steel posts. vent out-of-plane deformation of the flanges.
For each configuration considered, the axial force in the
FINITE ELEMENT ANALYSES post was plotted against the axial displacement at the free
end of the post or patch load. The ultimate load was defined
Models from the force-displacement curve at a point where the tan-
gential stiffness reduced to 50% of the initial stiffness, con-
A series of finite element models of the different experi-
sistent with the experimental data.
mental configurations were developed and verified using the
experimentally recorded response data. These finite element
models allowed a larger range of beams and configurations
to be considered, without material variability affecting the
results. HP12×53, HP14×73, HP12×89 and W14×90 beams
were modeled with timber posts (Figure 8a), and W14×90,
HP14×117 and W14×120 beams were modeled with steel
posts (Figure 8b).
The beams, posts and blocking were modeled with linear
three-dimensional eight node elements in ABAQUS (Hibbett,
Karlson and Sorensen, 2003), meshed typically as shown in
Figure 8. The steel members were modeled with a plastic
isotropic material using an expected yield strength of 55 ksi
for the A572 Gr. 50 and A992 steel beams (approximately
equal to the 53 ksi average strength measured from coupon
tests) and 46 ksi for the A500 Gr. 42 steel posts. The timber
posts were modeled with a multi-linear material model to fit
the observed force-displacement relationships resulting from
both material and geometric nonlinearities. Experiments on
the 11.5 in. square Douglas Fir timber posts resulted in a cal-
culated nominal post strength of 2.7 ksi, which was assumed
as the first-yield stress in the finite element model. The stress
(a) (b)
was assumed to increase linearly by 20% to allow for mate-
rial and geometric stiffening up to 2 times the yield strain. Fig. 8. Finite element models for: (a) a timber post and beam and
The stress was then assumed to gradually decrease to zero (b) a steel post and beam with blocking.

Results from Analyses of the experimental load. Therefore, there is good correlation
between the experimental and finite element results, with the
Comparisons of the force-displacement curves from the finite
variability attributed primarily to the variability in the capac-
element model and experimental data for two typical beams
ity of the different posts and blocking.
with and without blocking are shown in Figures 9a and 9b,
The finite element results show that blocking resulted in
respectively. The displacement from the experimental data
a 20 to 70% increase in the flange-post joint capacity when
was offset so that extrapolation of the maximum stiffness
using a timber post, depending on the flange bending capac-
passed through the origin, to allow comparison with the fi-
ity. With a steel post, blocking allowed only a 20 to 25%
nite element results. The comparisons show that the initial
increase in the joint capacity. An eccentricity between the
slope and maximum load were similar between the finite ele-
centroids of the flange and post equal to a maximum of 6 of
ment and experimental models. A comparison between the
the flange width, resulted in a relatively small reduction in
ultimate loads calculated from the finite element model and
the flange-post capacity of 10 to 15% when using a timber
experimental data is given in Figure 10. The load obtained
post. With a steel post, each eccentric case resulted in an al-
from the finite element analysis is between 76% and 121%
most identical flange-post capacity to the concentric case.
The calculated flange-timber post capacity based on the
interaction equation (Equation 3), as a ratio of the capacity
calculated from the finite element analysis, is plotted in Fig-
ure 11a, for different beam sizes, with and without blocking
and with different eccentricities. The blocking was consid-
ered to be 100% effective (γ = 1.0) with the timber posts, as
described previously. Figure 11a shows that for the beams
with timber posts, the calculated capacity is between 80%
and 105% of the capacity calculated from the finite element
model, with just two exceptions. These exceptions are for
the larger beams with a large eccentricity where the calcu-
lated capacity is up to 119% of that from the finite element
model. This level of eccentricity is unlikely to occur in a
real situation; therefore, these cases are not of particular con-
cern. Furthermore, when appropriate factors and nominal
strengths instead of expected strengths are used in design,
these cases are expected to be conservative as well. In prac-
tice, the maximum eccentricity should be limited to 6 of
(a)
(b)
Fig. 9 Comparison of force-displacement curve for HP14×73

beams: (a) concentrically loaded with no blocking and
(b) concentrically loaded with blocking from experimental data Fig. 10. Comparison of ultimate force from experimental data
and finite element analysis. and finite element analysis.

the flange width. The calculated capacity based on the inter- effective blocking capacity (Equations 4 and 5) and the fi-
action equation appears to be relatively accurate and gener- nite element analyses are shown in Figure 11b. The method
ally conservative for all cases including those with timber is shown to be accurate and conservative, with calculated
blocking. capacities between 81% and 105% of those from the finite
With the steel posts, the ultimate capacity of the flange- element model. The use of a 30% effectiveness factor for
post joint region is calculated using the effective bearing area the timber blocking gives a ratio of calculated to finite ele-
method (Equation 4). As the stiffness of the post is greater ment capacity with the timber blocking relatively consistent
than the stiffness of the blocking, the effectiveness of the to that without the timber blocking.
blocking is reduced to 30% (γ = 0.3). While the blocking can
be loaded further and the force carried by the joint continues DESIGN OF FLANGE-POST BEARING
to increase, significant flange bending and post yielding oc- JOINT REGIONS
curs resulting in significant permanent deformations in both
beam and post. Thus, the effectiveness of timber blocking Design Strength with a Timber Post
with a steel post is limited. Comparisons between the cal-
It is noted that the actual strength of the posts and beams
culated capacity using the effective post bearing area and
were used in the comparisons discussed earlier. Clearly, the
use of minimum specified stresses along with application of
load factors and resistance factors for LRFD, or allowable
stresses for ASD, will result in additional conservatism in
the design equations. In LRFD format, for a steel beam with
a patch load from a timber post, the factored applied load
in the post, Pu, should be less than a combination of flange
bending capacity, including blocking, and the post compres-
sion capacity for a short length of post, such that
1
−
⎛ 1 1 ⎞
2
Pu < ⎜ + ⎟ (6a)
⎝ φRnf φPnp 2 ⎠
2
where
φRnf = design strength of the flange including
blocking
φPnp = design strength of the post
(a)
The design strength of the flange, including blocking (when
used), is given by
φRnf = φb 18t f 2 Fyf + λ φc Fc′Ab (6b)
where
φb = resistance factor for flange bending, equal to
0.90 (AISC, 2005)
φc = resistance factor for compression in the block-
ing, equal to 0.90 (AFPA, 1996)
λ = time effect factor, equal to 1.0 for a typical
falsework duration
The design strength of the post is given by
φPnp = φc Fc′Ap (6c)

(b)
Fig. 11. Comparison of finite element analysis with calculated where

ultimate force using: (a) the interaction equation (Equation 3) and φc = resistance factor for compression in the post,
(b) the effective bearing area (Equation 4). equal to 0.90

In ASD format, the applied stress in the post, fc, should be In ASD format the flange-post connection region for a steel
less than the combination of allowable effective stress from post should be designed such that
flange bending capacity, Fcf, and allowable stress in the post,
F′cp, such that fcp < Fcpb
′ (9a)
1
−
⎛ 1 where
1 ⎞
2
fc < ⎜ 2 + 2 ⎟ (7a) fcp = applied compression stress in the gross cross-

⎝ Fcf Fcp′ ⎠ sectional area of the post
The effective stress from flange bending capacity, includ- R

fcp =
ing the capacity provided by timber blocking, can be deter- (
⎣ f bp )
⎡5 t + t + 2 k ⎤ 2t
1⎦ p
(9b)
mined from
18t f2 Ff where
Fcb′ Ab
Fcf = + (7b) R = applied force in the post
Ap Ap
The allowable stress in the post including blocking is given
by
where
Ff = allowable flange stress, equal to 22 ksi for A36 ⎛ ⎞
0.3Fcb′ Ab
steel or 30 ksi for A572 Gr. 50 or A992 steel Fcpb = Fcp ⎜ 1 + ⎟ (9c)
beams if the allowable stress safety factor (1.67) ⎜
⎝ Fcp ⎣ (f bp )
⎡ 5 t + t + 2k ⎤ 2 t ⎟
1⎦ p⎠
is applied as defined by the AISC Specification
(AISC, 2005) where
F c′b = allowable compression stress in the timber Fcp = 28 ksi for an A500 Gr. B round hollow steel sec-
blocking (AFPA, 2001) tion based on an allowable stress safety factor of
F′cp = allowable stress in the post, due to post com- 1.50, as for web yielding (AISC, 2005)
pression over a short length of the post, with ap- F′cb = allowable stress in the timber blocking (AFPA,
plicable modification factors, is taken directly 2001)
from the ASD specifications (AFPA, 2001)
SUMMARY AND CONCLUSIONS
Design Strength with a Steel Post
The capacity of the beam-post joint region for an unstiffened
The design capacity of the flange-post bearing joint region beam supporting or supported by a timber or steel post, simi-
with a steel post can be most accurately determined using the lar to those found in bridge falsework, may be governed by
effective bearing area. Therefore, using the LRFD format, a combination of flange bending and post crushing or yield-
design of the flange-post region should ensure that ing. A series of experiments and finite element analyses were
conducted to investigate potential failure mechanisms and
Pu < φRn (8a) to quantify the flange-post joint capacity. Also considered
in the study was the effect of: i) blocking placed between
where the top and bottom flanges, and ii) eccentricity between the
Pu = factored applied load centroids of the beam and post. Accordingly, the following
φRn = design strength of the post in bearing, including observations were made:
the strength provided by blocking 1. The ultimate capacity of the joint is defined at a point
φRn is given by where the stiffness of the force-displacement curve re-
duces to 50% of the initial stiffness. This definition of the
ultimate capacity provided consistent correlation with
⎣ ( )
φRn = φcp ⎡5 t f + tbp + 2 k1 ⎤ 2t p Fyp + φcb 0.3Fc′Ab (8b)
⎦ the experimental observations; namely onset of flange
bending.
where
2. Depending on the size of the beam, timber blocking
φcp = resistance factor for compression in the post,
is found to increase flange-post joint capacity by 20
which, to be consistent with that used for web
to 70%.
yielding, equal to 1.00 (AISC, 2005)
φcb = resistance factor for the blocking, equal to 0.90

3. When used under a steel post with large beam sections, REFERENCES
timber blocking is less effective resulting in a maximum
AFPA (1996), Load and Resistance Factor Design – Manual
likely increase of 25% in joint capacity.
for Engineered Wood Construction, American Forest and
4. An eccentricity between the flange and a timber post, Paper Association, Washington, DC.
which should be limited to 6 of the flange width results AFPA (2001), National Design Specification for Wood
in a reduction of flange-post strength of 10 to 15%. Construction, American Forest and Paper Association,
Washington, DC.
5. The effect of eccentricity, which should be limited to 6 AISC (2005), Specification for Structural Steel Buildings,
of the flange width, on the joint capacity when steel posts American Institute of Steel Construction, Inc., Chicago,
are used is negligible. IL.
Two methods were developed for predicting the flange-post Carden, L.P., Pekcan, G. and Itani, A.M. (2005), “Recom-
joint capacity; namely, the interaction method and the effec- mendations for the Design of Beams and Posts in Bridge
tive bearing area method. The interaction method accounted Falsework,” Research Report CCEER 05-15, Center for
for a combination of flange bending and post crushing ca- Civil and Earthquake Engineering Research, University of
pacities. Based on the experimental and analytical observa- Nevada, Reno, NV.
tions, this method is found appropriate and recommended Elgaaly, M. (1983), “Web Design Under Compressive
for use in calculating the capacity of joints that consist of Edge Loads,” Engineering Journal, AISC, 4th quarter,
a timber post bearing onto a beam flange. Blocking can be pp. 153–171.
considered “100% effective” (γ = 1.0 in Equations 5 and 7b) Graciano, C. and Edlund, B. (2003), “Failure Mechanism of
with respect to increasing the flange bending capacity when Slender Girder Webs with a Longitudinal Stiffener under
a timber post is used. The effective bearing area method is Patch Loading,” Journal of Constructional Steel Research,
recommended for calculating the flange-post joint capacity Vol. 59, No. 1, pp. 27–45.
with a steel post and is derived based on assumptions similar
Graham, J.D., Sherbourne, A.N. and Khabbaz, R.N. (1959),
to those used in the calculation of web yielding capacity. It is
Welded Interior Beam-to-Column Connections, American
noted that timber blocking has a relatively low stiffness when
Iron and Steel Institute, Inc., Chicago, IL.
used under a steel post. Therefore, a blocking effectiveness
factor of 30% (γ = 0.3 in Equation 5) is recommended based Hibbett, Karlson and Sorensen, Inc. (2003), “ABAQUS –
on the experimental and analytical findings. Finally, design Finite Element Program,” Hibbett, Karlson & Sorensen,
equations are presented in both LRFD and ASD format for Inc., Fremont, CA.
calculating flange-post capacity with and without blocking. Roberts, T.M. and Markovic, N. (1983), “Stocky Plate Gird-
ers Subjected to Edge Loading,” Proc. Instn. Civ. Engrs.,
ACKNOWLEDGMENTS Vol. 75, No. 2, pp. 539–550.
The authors would like to extend their gratitude to the Roberts, T.M. and Newark, A.C.B. (1997), “Strength of
California Department for Transportation for funding of Webs Subjected to Compressive Edge Loading,” Journal
this study under contract number 59A0445, with special of Structural Engineering, Vol. 123, No. 2, pp. 176–183.
thanks to John Lammers and Peter Lee for their assistance Roberts, T.M. and Rockey, K.C. (1979), “A Mechanism So-
and direction. lution for Predicting the Collapse Loads for Slender Plate
Girders when Subjected to In-Plane Patch Loading,” Proc.
Instn. Civ. Engrs., Vol. 67, No. 2, pp. 155–175.
Shahabian, F. and Roberts, T.M. (1999), “Buckling of Slen-
der Web Plates Subjected to Combinations of In-Plane
Loading,” Journal of Constructional Steel Research, Vol.
51, No. 2, pp. 99–121.

Capacity Design of Vertical Boundary
Elements in Steel Plate Shear Walls
JEFFREY W. BERMAN and MICHEL BRUNEAU
D esign requirements now appear in the 2005 AISC Seis-

mic Provisions for Structural Steel Buildings (AISC,
2005b), referred to herein as The Provisions, for steel plate
procedures for two example SPSW configurations. The re-
sulting design loads are then compared with the VBE design
loads as determined by nonlinear pushover analysis.
shear walls (SPSWs) that are designed such that their web
plates buckle in shear and develop diagonal tension field ac- CURRENT VBE DESIGN PROCEDURES
tion when resisting lateral loads. Energy dissipation and duc-
Three methods for ensuring capacity design of VBEs
tility during seismic events is principally achieved through
and HBEs for SPSWs are described in the commentary
yielding of the web plates along the diagonal tension field.
of The Provisions, two of the methods are linear and one
Consistent with capacity design principles, The Provisions
is nonlinear. These are: the combined linear elastic com-
require that the vertical and horizontal boundary elements
puter programs and capacity design concept (LE+CD); the
(VBEs and HBEs) of SPSWs, as shown in Figure 1, be
designed to remain essentially elastic with the exception of
plastic hinging at the ends of horizontal boundary elements.
The commentary of The Provisions provides some guidance
on how to achieve this requirement. However, the methods
described in the commentary, as shown in this paper, do not
necessarily lead to VBEs that meet the requirement of es-
sentially elastic behavior under the forces generated by fully
yielded web plates.
This paper reviews the current approaches provided in
The Provisions commentary for determination of capacity
design loads for the VBEs of SPSWs and also describes how
the capacity design objective may be achieved using nonlin-
ear static analysis. Then, a new procedure is proposed that
uses a fundamental plastic collapse mechanism and linear
beam analysis to approximate the design actions for VBEs of
SPSWs for given web plates and horizontal boundary mem-
ber sizes. The proposed procedure does not involve nonlin-
ear analysis, making it practical for use in design. VBE de-
sign loads are estimated using both the current and proposed
Jeffrey W. Berman is assistant professor, department of

civil and environmental engineering, University of Wash-
ington, Seattle, WA.
Michel Bruneau is director, MCEER, professor, department
of CSEE, University at Buffalo, Buffalo, NY.
Fig. 1. General SPSW Configuration.
057-072_Berman_Bruneau_2008_1Q.indd 57 3/27/08 12:22:03 PM

indirect capacity design approach (ICD); and nonlinear the web panel be applied to the surrounding framing mem-
static analysis (pushover analysis). Each is reviewed briefly bers in the direction of α. However, without simultaneous
below and the steps in the two linear procedures that result application of the lateral loads that caused the web plates to
in significant inaccuracies are identified. Note that nonlinear yield, the frame member moment diagrams resulting from
static analysis of SPSWs with web plates modeled as a series the application of just the web plate forces will be incorrect
of strips oriented at α from the vertical, the angle of inclina- because equilibrium with the applied loads is not satisfied
tion of the tension field as calculated per The Provisions, has and the deformed shape of the system is incorrect for those
been shown to adequately represent the behavior of SPSW loads. To illustrate this, consider Figure 2, which shows a
(Driver, Kulak, Kennedy and Elwi, 1998; Berman and Bru- SPSW with the web plate represented by strip elements sub-
neau, 2004; and Berman and Bruneau, 2005; among others). jected to lateral loads (Figure 2a), and the model implied
That strip model is also used in elastic analyses of SPSW in by step 4 of the LE+CD procedure where the strips have
the two linear procedures described below. been replaced with plate yield forces, but no corresponding
lateral loads have been applied. It is clear that the moment
Combined Linear Elastic Computer Program and diagrams for the columns from these two cases will have
Capacity Design Concept (LE+CD) different shapes regardless of the dimensions, geometry, or
configuration of the SPSW. Unfortunately, calculation of
The combined linear elastic computer program and capacity
those lateral loads and displacements is complex and since
design concept as described in Commentary Section C17.4a
the procedure is meant to provide a simple linear method
of The Provisions consists of four steps (note that Ry is the
for capacity design of framing elements it is not useful to
ratio of mean to nominal yield stress of the web plate, As
recommend the addition of such calculations. However, the
is the area of a strip in the strip model representation of a
concept that the forces from the yielding web plates may be
SPSW, and Fyp is the web plate yield stress):
applied to an otherwise elastic model in lieu of using full
1. Lateral forces: Use combined model, boundary elements nonlinear analysis is promising and is the motivation for the
and web elements, to come up with the demands in the proposed capacity design procedure below.
web and boundary elements for the code required base
shear. The web elements shall not be considered as Indirect Capacity Design Approach (ICD)
vertical-load carrying elements.
The indirect capacity design approach presented in Com-
2. Gravity load (dead load and live load): Apply gravity mentary Section C17.4a of The Provisions is from the Ca-
loads to a model with only gravity frames. The web ele- nadian Standards Association steel design standard CAN/
ments shall not be considered as vertical-load carrying CSA-S16-01 (CSA, 2001) and proposes that the loads in the
elements.
3. Without any overstrength factors, design the boundary

elements using demands based on combination of the
forces from the above steps 1 and 2.
4. Boundary element capacity design check: Check the

boundary element for the maximum capacity of the web
elements in combination with the maximum possible
axial load due to overturning moment. Use the axial force
obtained from step 1 above and multiply by the over-
strength factor, Ωo. Apply load from the web elements
(RyFyp As) in the direction of α. For this capacity design
check use a material strength reduction factor of 1.0. For
the determination of the required strength of boundary
elements and their connection to the web, neither the
resistance factor (LRFD), nor the safety factor (ASD),
are applied to the strength of the web.
While this procedure may, in some cases, result in proper (a) (b)
capacity design of boundary elements, there is an inconsis- Fig. 2. Conceptual difference between: (a) SPSW subject
tency with respect to equilibrium. In step 4 the procedure to lateral load and (b) SPSW boundary frame
requires that forces equal to the expected yield strength of subject to only infill plate yield forces.

VBEs of a SPSW may be found from gravity loads com- be large and the portion of the base shear carried by the sur-
bined with seismic loads amplified by: rounding moment frame could be substantial. For example,
a recent SPSW implementation detailed in Monnier and Ha-
B = Ve / Vu (1) rasimowicz (2007) had VBEs at the bottom stories that were
W14×730, the second largest W14 shape available. Neglect-
where
ing the strength of the surrounding moment resisting frame
Ve = expected shear strength, at the base of the wall,
in the ICD procedure when that frame has such substantial
determined for the web thickness supplied
sections will certainly result in an underestimation of the ex-
= 0.5RyFyptwLsin2α, where L is the width of the
pected SPSW shear strength, Ve, and VBE design loads that
wall
are underestimated for true capacity design.
Vu = factored lateral seismic force at the base of the
wall
Nonlinear Static (Push-Over) Analysis
In determining the loads in VBEs, the amplification factor,
Nonlinear static analysis of VBE strip models has been
B, need not be taken larger than the seismic force reduction
shown to give reasonable results for HBE and VBE moments
factor, R.
and axial forces. Capacity design may be achieved by ac-
The VBE design axial forces shall be determined from
counting for the actual thickness of the web plates and the
overturning moments defined as follows:
ratio of mean to nominal web plate yield stress. However,
1. The moment at the base of the wall is BMu, where Mu is as a design tool, nonlinear static analysis is time consum-
the factored seismic overturning moment at the base of ing as several iterations may be necessary to ensure capacity
the wall corresponding to the force Vu; design of VBEs. Additional complexity results from having
to properly account for possible formation of flexural-axial
2. The moment, BMu, extends for a height H but not less plastic hinges in the HBEs. Despite these issues, nonlinear
than two stories from the base, and; static analysis of SPSW strip models generally gives accu-
rate results for VBE demands and will be used to compare
3. The moment decreases linearly above a height H to B
the adequacy of the proposed procedure below for two ex-
times the factored seismic overturning moment at one
ample SPSWs.
story below the top of the wall, but need not exceed R
times the factored seismic overturning moment at any
PROPOSED VBE DESIGN PROCEDURE
story under consideration corresponding to the force Vu.
The local bending moments in the VBE due to tension field Objective
action in the web shall be multiplied by the amplification As discussed above there are two rather simple linear pro-
factor, B. cedures for capacity design of SPSW VBEs given in the
This procedure relies on elastic analysis of a strip model commentary of The Provisions. However, those methods, for
(or equivalent) for the design seismic loads, followed by different reasons, do not necessarily achieve the goal of VBE
amplification of the resulting VBE moments by the factor capacity design. Furthermore, the nonlinear static analysis
B. Therefore, it produces moment diagrams and SPSW de- procedure, which results in a more accurate estimation of the
formations that are similar in shape to what one would ob- capacity design demands for VBEs, is tedious for broad use
tain from a pushover analysis. Similarly, the determination in design. Therefore, a need exists to develop a reasonably
of VBE axial forces from overturning calculations based on accurate and relatively efficient method for estimating the
the design lateral loads amplified by B results in axial force demands in VBEs when full yielding occurs in web plates
diagrams that are of the proper shape. However, the amplifi- for SPSW. This method should preferably involve only lin-
cation factor used in both instances is found only for the first ear computer analyses without development of the complete
story and does not include the possibly significant strength strip model, and should account for the strength of surround-
of the surrounding frame. This leads to estimates of VBE ing framing (in other words, include the strength demands
demands that are less than those required to develop full associated with hinging at the HBE ends).
web yielding on all stories prior to development of hinges The procedure proposed below to estimate VBE design
in VBEs. For example, it is conceivable that the ratio of web loads to ensure capacity design of SPSWs combines a linear
thickness provided to web thickness needed for the design elastic beam model and plastic analysis. A model of the VBE
seismic loads is larger on the upper stories than on the lower on elastic supports is used to determine the axial forces in the
stories. In these situations, the indirect capacity design ap- HBEs and a plastic collapse mechanism is assumed to estimate
proach would significantly underestimate the VBE design the lateral seismic loads that cause full web plate yielding and
loads for the upper stories and capacity design would not plastic hinging of HBEs at their ends. A simple VBE free body
be achieved. Additionally, frame members for SPSW may diagram is then used to determine the design VBE axial forces

and moments. For use in design, iteration may be necessary as factored loads (or, for the case of HBE design the maximum
certain parameters are assumed at the beginning of the process of the factored loads or web plate yielding), the required ca-
that may need revision as the design progresses. pacity of VBEs may be found from VBE free body diagrams
such as those shown in Figure 4 for a generic four-story
Plastic Collapse Mechanism SPSW. Those free body diagrams include distributed loads
representing the web plate yielding at story i, ωxci and ωyci;
Plastic collapse mechanisms for SPSWs subject to lateral
moments from plastic hinging of HBEs, Mprli and Mprri; axial
loads have been proposed by Berman and Bruneau (2003)
forces from HBEs, Pbli and Pbri; applied lateral seismic loads,
and have been shown to agree well with experimental results
found from consideration of the plastic collapse mechanism,
for ultimate capacity of single and multistory SPSWs. They
Fi; and base reactions for those lateral seismic loads, RyL, RxL,
examined two types of plastic mechanisms for multistory
RyR, and RxR. The following describes how the components
SPSW, namely, a uniform collapse mechanism and a soft-
of the VBE free body diagrams are determined. Note that for
story collapse mechanism which are shown schematically in
the purpose of this discussion lateral forces are assumed to
Figures 3a and 3b respectively. For the purpose of capacity
be acting from left to right on the SPSW of Figure 4.
design of VBEs, it is conservative to use the uniform plas-
tic collapse mechanism as it will result in larger base shear Forces from Plate Yielding
forces and larger VBE demands. Furthermore, if a soft-story
mechanism is found to be likely, it is recommended that the The distributed loads to be applied to the VBEs (ωyci and
SPSW be redesigned to develop more uniform yielding of ωxci) and HBEs (ωybi and ωxbi) from plate yielding on each
the web plates over the height. This can be achieved, even for story i may be determined as:
web plates of equal thickness over the height, by adjusting
( )
2
the sizes and moments of inertia of the surrounding HBEs ω yci = 2Fyp twi sin 2α ω xci = Fyp t wi sin α (2)
and VBEs. Therefore, the uniform collapse mechanism
shown in Figure 3a will be used in the proposed procedure
( )
2
for determination of capacity design loads for SPSW VBEs. ω ybi = Fyp t wi cos α ω xbi = 2Fyp twi sin 2α (3)
Proposed VBE Design Procedure These are found from resolving the plate yielding force,
occurring at an angle α from the vertical, into horizontal and
Free Body Diagrams of VBEs
vertical components acting along the VBEs and HBEs as
Assuming that the web plates and HBEs of a SPSW have demonstrated for a VBE in Figure 5. In that figure, ds is an
been designed according to The Provisions to resist the incremental plate width perpendicular to the tension field,
(a) (b)
Fig. 3. SPSW Collapse Mechanisms: (a) Uniform Yielding Mechanism and (b) Soft-Story Mechanism.

(a) (b)
Fig. 4. VBE free body diagrams.
Fig. 5. Resolution of plate forces applied to a VBE.

dy is an incremental VBE height corresponding to ds, and Ps, reasonable, although less accurate, to estimate the HBE axial
Px and Py, are forces from plate yielding summed over ds in forces from the horizontal component of web plate yielding
the tension field, horizontal and vertical directions, respec- on the VBEs, Psi, considering VBE lengths tributary to each
tively. Note that the inclination angle, as determined per The HBE, in other words:
Provisions, depends on the VBE cross-sectional area and
moment of inertia and will have to be assumed at the begin- hi h
Psi = ω xci + ω xci +1 i +1 (5)
ning of a design procedure and then revised once VBEs have 2 2
been selected. An initial assumption of 45ο is suggested.
Regardless of the method used, the spring forces are used
HBE Axial Forces below to determine the HBE axial forces. Note that these
As part of estimating the axial load in the HBEs, an elastic spring forces correspond to compression forces in the HBE,
model of the VBE is developed as shown in Figure 6. The and can be of significant magnitude. Physically, one can en-
model consists of a continuous beam element representing vision the SPSW VBEs as being pulled toward each other
the VBE which is pin-supported at the base and supported by the uniformly distributed forces applied by the yielding
by elastic springs at the intermediate and top HBE locations. webs, and the HBEs acting as regularly spaced “shoring” to
HBE spring stiffnesses at each story i, kbi, can be taken as keep the VBEs apart.
the axial stiffness of the HBEs considering one-half the bay The axial force component in the intermediate and top
width (or HBE length for considerably deep VBEs), in other HBEs resulting from the horizontal component of the plate
words: yield forces on the HBEs, ωxbi, is assumed to be distributed
A E as shown in Figure 7. Note that for the bottom HBE, this
k bi = bi (4) distribution is the reverse of that in the top beam. These axial
L/2
force components are then combined with the spring forces
where from the linear VBE model, resulting in the following equa-
Abi = HBE cross-sectional area tions for the axial force at the left and right sides of the inter-
L = bay width mediate and top HBEs (Pbli and Pbri respectively):
E = modulus of elasticity
This VBE model is then loaded with the horizontal compo-
nent of the forces from the web plates yielding over each sto-
(
Pbli = − ω xbi − ω xbi +1 ) L2 + P si
(6)
ry, namely, ωxci. An initial VBE size will have to be assumed

for use in this model and some iteration may be required
once that VBE size is revised. Additionally, it is reasonable
(
Pbri = ω xbi − ω xbi +1 ) L2 + P si
(7)
to neglect the rotational restraint provided by the HBEs.

This assumption has been observed to have a negligible im- where the spring forces should be negative indicating that
pact on the resulting spring forces, Psi. Note that it is also they are adding to the compression in HBEs. As mentioned
above, the axial forces from ωxbi and ωxbi+1 in the bottom HBE
may be taken as the mirror image of those shown in Figure 7,
where ωxbi is zero in that case as there is no web below the
bottom HBE. Furthermore, there are no spring forces to con-
sider at the bottom HBE location as the horizontal component
Fig. 7. Assumed HBE axial force distribution due to horizontal

Fig. 6. Elastic VBE model with HBE springs. component of plate yield forces on the HBE.

of force from web plate yielding on the lower portion of the Applied Lateral Loads
bottom VBE is added to the base reaction determined as part
The final forces necessary to complete the free body diagram
of the plastic collapse mechanism analysis, as described be-
of the VBE are the applied lateral loads corresponding to
low. Therefore, the bottom HBE axial forces on the right and
the assumed collapse mechanism for the SPSW (Figure 3a).
left hand sides, Pbr0 and Pbl0 are:
Following the derivation in Berman and Bruneau (2003) the
governing equation for that collapse mechanism is:
L L
Pbr 0 = −ω xb1 and Pbl 0 = ω xb1 (8)
2 2 ns ns ns
∑ Fi Hi = ∑ M prli +∑ M prri
i =1 i=0 i=0
HBE Reduced Plastic Moments and Corresponding ns
(12)
Shear Forces (
1
) ( )
+ ∑ t wi − t wi+1 Fyp LH i sin 2α i
i =1 2
Once the HBE axial forces have been estimated it is pos-
sible to determine the plastic moment that will develop at where
the HBE ends for the assumed collapse mechanism, reduced Fi = applied lateral load at each story to cause the
for the presence of axial load. Note that it is conservative mechanism
to assume that this reduction is negligible; however, since Hi = height from the base to each story
substantial axial loads may develop in the HBEs, resulting
in significantly reduced plastic moment capacities, it can be and other terms are as previously defined. Note that the indi-
advantageous to account for the reduced plastic moments at ces for the HBE plastic moment summations begin at zero so
the left and right HBE ends, Mprl and Mprr, respectively. that the bottom HBE (denoted HBE0) is included.
The intermediate and top HBEs will have free body dia- To employ Equation 12 in calculating the applied lateral
grams similar to that shown in Figure 8, except that there loads that cause this mechanism to form, it is necessary to
will be no plate forces acting above the top HBE. For the assume some distribution of those loads over the height of
bottom HBE, the axial forces at the HBE ends will be in the structure; in other words, a relationship between F1, F2,
the opposite direction to those shown in Figure 8 and there etc. For this purpose, a pattern equal to that of the design
will be no plate forces acting below the HBE. The reduced lateral seismic loads from the appropriate building code may
plastic moment capacity at the HBE ends, given here for the be used. This is an approximation that is simple and that has
left end, can be approximated by (Bruneau, Whittaker and been observed to provide reasonable results for SPSW. It
Uang, 1998): would also be appropriate to use the deformation pattern of
the first mode of vibration of the structure for this purpose
⎛ P ⎞ ⎛ P ⎞ (obtained from a modal analysis), but this more sophisticat-
1.18 ⎜ 1 − bli ⎟ Z xbi Fyb if 1.18 ⎜ 1 − bli ⎟ ≤ 1.0 ed approach is unnecessary given that the code specified dis-
⎝ Fyb Abi ⎠ ⎝ Fyb Abi ⎠
tribution of lateral seismic forces vertically on a lateral force
(9)
⎛ P ⎞ resisting system is meant to simulate first mode characteris-
Z xbi Fyb if 1.18 ⎜ 1 − bli ⎟ > 1.0 tics. Once a load pattern is assumed and a relationship be-
⎝ Fyb Abi ⎠
tween the applied collapse loads at each story is determined,
Equation 12 may be used to solve for those collapse loads.
where The base shear force, V, for the collapse loading is found
Fyb = HBE yield strength by summing the applied lateral loads. Horizontal reactions at
Abi = HBE cross-sectional area for story i the column bases, RxL and RxR, are then determined by divid-
Zxbi = HBE plastic modulus for story i. ing the collapse base shear by 2 and adding the pin-support
reaction from the VBE model, Rbs, to the reaction under the
Using the reduced plastic moment capacities and the HBE
left VBE and subtracting it off the reaction under the right
free body diagram shown in Figure 8, the shear forces at
the left and right ends of all HBEs, Vbl and Vbr can be found
from:
M prri + M prli
Vbri =
L
(
+ ω ybi − ω ybi +1 ) L2 (10)
(
Vbli = Vbri − ω ybi − ω ybi +1 L ) (11)
Fig. 8. HBE free body diagram.

VBE. Vertical base reactions can be estimated from over- yield. However, at each story the VBEs should be designed
turning calculations using the collapse loads as: to resist the moments generated by yielding of the web plates
at that level and the corresponding frame moments.
ns
∑FH i i EXAMPLE ESTIMATION OF VBE DESIGN LOADS

i =1
Ryl = and Ryr = − Ryl (13)
L Two examples of the proposed procedure for estimating VBE
design loads for their capacity design are described below.
Since the primary inelastic elements in SPSWs are the web
Determination of VBE Design Loads plates and the web plate strength is a function of thickness,
The moment, axial, and shear force diagrams for the VBEs the two examples explore the cases of variable and constant
are established once all the components of the VBE free web plate thickness over the height of a four-story SPSW.
body diagrams are estimated. The diagrams give minimum In the SPSW with variable web plate thickness, a different
design actions for those VBEs such that they can resist full thickness is used at each story and not limited to those avail-
web plate yielding and HBE hinging. able for common plate stock. For the case of constant web
plate thickness over the height of the SPSW, the web plates
Additional Considerations were designed for the required story shear at the first story
and that thickness is used over all four stories. Furthermore,
Though not explicitly considered in the above formulations,
that thickness was constrained to be available from common
use of the ratio of the expected yield stress to the specified
plate stock.
minimum yield stress, Ry, may be incorporated into the
procedure when determining the distributed loads from
Structure Description and SPSW Design
plate yielding and when determining HBE plastic moment
capacity. Additionally, when deep VBEs are used, the length The MCEER (Multidisciplinary Center for Earthquake En-
between VBE flanges, Lcf, may be substituted for the column gineering Research) demonstration hospital was used as the
centerline bay width, L, when applying the plate yielding prototype structure (Yang and Whittaker, 2002) for which
loads to the HBEs. Furthermore, using the schematic struc- the SPSWs were designed. For simplicity, four SPSWs were
ture shown in Figure 3 for which structural members have assumed to carry equal portions of the seismic load resulting
no width, the HBE plastic hinges are assumed to form at the from the active seismic weight of the structure of 9,800 kips
VBE centerlines, which is not the actual case. HBE hinges (2,613 kips, 2,542 kips, 2,542 kips, and 2,103 kips at the 1st
will typically form db /2 from the column face, where db is story, 2nd story, 3rd story and roof HBEs, respectively). The
the HBE depth. This can be accounted for by either including geometries of the two SPSWs are shown in Figure 9 and the
in the VBE free body diagrams the distance from the column structure is assumed to be located on Class D soil. Design
centerlines to the HBE hinge locations or by calculating the seismic loads were calculated using FEMA 450 (FEMA,
projected column centerline moment as is done for moment 2003) and the associated spectral acceleration maps. Design
frames. This calculation is not included here for simplicity short and 1-second spectral ordinates, SDS and SD1, were cal-
and because the increase in moment applied to the VBE is culated to be 1.17g and 0.44g respectively. The period of
generally small relative to the magnitude of the moments the structure was estimated using the FEMA procedures as
generated by web plate yielding and HBE hinging. Gravity 0.38 sec and using a response modification coefficient, R,
loads are another consideration that has not been included; of 7 and importance factor, I, of 1.5, the base shear for the
however, they can easily be added to the vertical components structure was found to be 2,450 kips, or 612.5 kips for each
of the web plate yield forces that are applied to the HBEs in SPSW. Distribution of the base shear up the height of the
Figure 8. They will then be accounted for in the resulting HBE structure resulted in lateral loads of 215 kips, 195 kips, 132
shear forces and VBE axial forces. Finally, this procedure kips and 71 kips, at each story from the roof down to the
will provide reasonable VBE design forces for SPSWs that level of HBE1 for each SPSW.
can be expected to yield over their entire height—typically Figure 9 shows the web plate thicknesses selected for the
shorter SPSWs. This procedure will likely be overly conser- two SPSWs designed for the above loading in accordance
vative for tall SPSWs where nonlinear time history analysis with The Provisions, and it also shows the selected HBE and
indicates that simultaneous yielding of the web plates over VBE sizes. As mentioned above, cases of variable and
the entire SPSW height is unlikely. In those situations it may constant web plate thickness, denoted SPSW-V and SPSW-C,
be acceptable to reduce the VBE axial forces obtained from respectively, have been considered. SPSW-V uses plate
this proposed procedure (following a procedure similar to thicknesses that may not be available but correspond to the
that proposed by Redwood and Channagiri, 1991) to account minimum required for the design story shear forces and
for some web plates remaining partially elastic while others SPSW-C uses the assumed minimum available plate thickness

of 0.1875 in. Note that ASTM A36 steel has been assumed depends mostly on the aspect ratio, h/L, of the bays and
for the web plates and ASTM A992 for the HBEs and VBEs. therefore, does not show substantial difference between
VBE selection was done after a first iteration of the proposed SPSW-C and SPSW-V.
procedure for evaluation of design loads with an assumed Table 2 shows the HBE spring stiffnesses, kb, used in the
tension field inclination angle of 45ο. Described below are linear VBE models for each wall and the resulting spring
the calculations for a second iteration of the proposed pro- loads, Ps, where the entry for the bottom HBE is the horizon-
cedure. Additionally, for simplicity and to provide a direct tal reaction at the pin base of the VBE. The HBE axial forces,
comparison between design loads resulting from the cur- Pbl and Pbr, reduced plastic moments, Mprl and Mprr, and shear
rent and proposed procedures, the effects of gravity loads forces, Vbl and Vb r , at each HBE end are then calculated per
and vertical ground motion were neglected (in other words, Equations 6 through 11. Results for all HBEs in both exam-
only VBE design loads resulting from the horizontal seismic ple walls are given in Table 2. For comparison purposes, the
loading will be calculated). HBE spring forces were also estimated by considering the
horizontal component of web plate yielding forces applied to
Calculation of VBE Design Loads the VBE tributary to each HBE, as given by Equation 5. The
results given in Table 2 show some deviation relative to the
The proposed procedure for evaluating VBE design loads spring forces from the VBE model, although for preliminary
was employed for the SPSWs shown in Figure 9. Calcula- design of VBEs the difference may be insignificant.
tion results for the tension field inclination angle and dis- Next, the applied seismic lateral loads at each story were
tributed loads from web plate yielding given by Equations 2 found for the assumed collapse mechanism, similar to that
and 3, are shown for both SPSW-C and SPSW-V in Table 1. shown in Figure 3a, using Equation 12. Note that it was as-
Note that the inclination angle, as given by The Provisions, sumed that those applied loads to cause collapse were in the
(a) (b)
Fig. 9. Example SPSWs: (a) SPSW-V and (b) SPSW-C.

Table 1. Distributed Loads from Yielding Web Plates
α ωyc ωxc ωyb ωxb

Wall Story
(deg) (kip/in.) (kip/in.) (kip/in.) (kip/in.)
1 46 3.37 3.49 3.26 3.37
SPSW-C
2 48 3.35 3.75 3.00 3.35

3 48 3.35 3.75 3.00 3.35
4 48 3.35 3.75 3.00 3.35
1 46 3.24 3.36 3.13 3.24
SPSW-V
2 48 3.06 3.42 2.74 3.06

3 48 2.23 2.49 1.99 2.23
4 48 1.13 1.26 1.01 1.13
Table 2. Linear VBE Model Parameters, Results and

Corresponding HBE End Actions
kb Ps (b) Ps (c) Pbl Pbr Mprl Mprr Vbl Vbr
Wall HBE
(kip/in.) (kips) (kips) (kips) (kips) (kip-in.) (kip-in.) (kips) (kips)
0 (a) -277 -283 496 -496 71500 71500 965 8
4242 -540 -564 -543 -537
SPSW-C
1 4811 4862 -5 71
2 4242 -622 -563 -622 -622 4095 4095 28 28
3 4242 -537 -563 -537 -537 4864 4864 33 33
4 19235 -277 -281 -770 216 68030 71500 915 34
0 (a) -250 -272 477 -477 70500 70500 940 19

8582 -542 -529 -569 -515
SPSW-V
1 20860 21560 86 202

2 8582 -463 -443 -585 -341 20650 23820 42 261
3 8582 -295 -281 -456 -134 22320 25000 17 305
4 8582 -71 -95 -237 95 25000 25000 22 319
(a)
Not applicable, in the linear VBE models there are pin supports at the bottom HBE locations.
(b)
Values at the bottom HBE locations are the horizontal reactions at the pin supports.
(c)
Spring forces approximated by Equation 5.
same pattern of distribution as the design lateral loads given in Figure 9. Note that SPSW strength and VBE demand are
above. Resulting lateral loads, F, are given in Table 3 and proportional to the bay width, L, and in the case considered
Table 4 gives the corresponding base shear, V, and base reac- here the bay width is large. Lower SPSW overstrength and
tions, Rxl, Ryl, Rxr, and Ryr, for each of the example walls. VBE demands may be achieved by reducing the bay width,
Axial, moment and shear force diagrams for the VBEs of however, the aspect ratio of the bay, L/h, must be greater than
the two SPSWs are shown for the left VBE of SPSW-C in 0.8 and less than or equal to 2.5, as specified in The Provi-
Figures 10a, 10b and 10c, respectively. The resulting forces sions Section 17.2b.
at the bases of the columns, where they are a maximum, are
given in Table 5 for both SPSWs. Assuming lateral brac- Comparison with Current Procedures
ing of the columns at each story, the moment-axial capac-
ity interaction equation values, given by Equation H1-1 of To judge the adequacy of both the current and proposed ap-
the AISC Specification for Structural Steel Buildings (AISC, proximate procedures for determining VBE design loads,
2005a), were 0.96 and 1.0 for the VBE sizes for SPSW-C and nonlinear static analysis of strip models of SPSW-C and
SPSW-V respectively, considering the VBE sections shown SPSW-V are used. The strip models had tension only strip

Table 3. Seismic Loading for Table 4. Base Shear and Reactions for Assumed Collapse Mechanism
Assumed Collapse Mechanism
V Rxl (a) Ryl (a) Rxr (a) Ryr (a)
F Wall
Wall Story (kips) (kips) (kips) (kips) (kips)
(kips)
SPSW-C 2041 -1298 -3107 -744 3100
1 236
SPSW-C
2 440 SPSW-V 1702 -1101 -2591 -601 2600

(a)
Positive x and y directions are right and up, respectively.
3 649
4 716
1 197
SPSW-V
2 367
3 541
4 597
Table 5. Actions at Column Bases from Proposed Procedure
Axial Moment Shear

Wall Column Side
(kips) (kip-in.) (kips)
Left 2142 71500 802
SPSW-C
Right -3115 71500 248
Left 1651 70500 624
SPSW-V
Right -2610 70500 164
(a)
Positive x and y directions are right and up, respectively.
612 612 612
462 462 462

Height (in)
Height (in)
Height (in)
312 312 312
162 162 162
0 0 0
0 1500 3000 −1 −0.5 0 0.5 1 −1000 −500 0 500 1000
Left VBE Axial (kip) Left VBE Moment (k−in) x 10
5 Left VBE Shear (kip)
(a) (b) (c)
Fig. 10. Force diagrams for the VBE of SPSW-C: (a) axial, (b) moment, and (c) shear.

elements with elastic-perfectly-plastic behavior. Flexural- the VBE force diagrams, such as moment-axial interaction
axial hinges were also added at all possible HBE and VBE in HBEs, and proper distribution of HBE axial load to the
locations. The HBE and VBE hinges were defined to have right and left VBEs.
moment-axial interaction per FEMA 356 (FEMA, 2000) and The LE+CD procedure agrees well with pushover analy-
a low strain hardening level (in other words, 0.5%) intro- sis for VBE axial force as shown in Figures 12 and 14. How-
duced to ensure numerical stability but close to zero because ever, because the procedure neglects the application of the
none of the design procedures that will be compared con- lateral loads to cause web plate yielding when evaluating the
sider strain hardening. Resulting pushover curves for the two VBE moments, the moment diagrams in Figures 13 and 15
SPSWs are shown in Figure 11. The analyses showed that all are not in agreement. Neglecting the applied loads to cause
web elements yielded and many HBEs formed hinges prior infill yielding (recall Figure 2), results in moment diagrams
to yielding of any VBEs. Note that linear analyses of the strip that do not include the significant contributions of frame
models for the design seismic loads were also performed for action under those loads, which in these cases are actually
use in the LE+CD and ICD approaches as described below. large enough to not only change the magnitude but also the
VBE design loads were calculated again using the LE+CD sign of the VBE moments. Although it appears the VBE mo-
procedure as per the commentary of The Provisions. The dis- ment diagrams from the LE+CD for the VBEs from SPSW-
tributed loads from web plate yielding given in Table 1 were V may simply be reversed, that is not the case, and those for
applied to linear models of the surrounding moment resist- SPSW-C would not agree even if they were converted into
ing frames for each SPSW and resulting VBE moments were their mirror image.
recorded. VBE axial loads were then found by multiplying Finally, the ICD approach reasonably estimates the VBE
the axial loads from the linear strip model analysis factored axial forces; however, because the overstrength is very large
by the 2.5 overstrength factor. for these SPSWs, it is not able to adequately estimate the
The indirect capacity design approach (ICD) was also used VBE moments. As shown in Figure 13 and 15, the ICD re-
to estimate the VBE design loads for each SPSW. Moments sults in VBE moment diagrams that have similar shape to
from the linear strip models were factored by B, which was those from pushover analyses but significantly underesti-
2.10 and 2.03 for SPSW-C and SPSW-V, respectively. Axial mates the values. Therefore, the proposed procedure is the
loads were found from B times the overturning moment as only one of the three methods available for estimating design
described in the commentary of The Provisions. loads for VBEs that ensures web plate yielding is able to
Figures 12 through 15 compare axial loads and moments fully develop prior to hinging in VBEs.
from the three procedures for approximating VBE design
loads (in other words, LE+CD, ICD, and the proposed pro-
cedure) with those from pushover analysis for both SPSW-
C and SPSW-V. As shown, the proposed procedure agrees
well with pushover results in terms of both VBE axial force 612
and moment. Furthermore, the proposed procedure is able to
capture the important aspects of SPSW behavior that effect
462
Height (in)
2000
312
Base Shear (kip)
1500
1000
162
Proposed Procedure
Pushover Analysis
500
LE+CD
SPSW−C ICD
SPSW−V 0
0 −3000 −2000 −1000 0 1000 2000 3000
0 0.5 1 1.5 2 2.5 VBE Axial Force (kip)
Roof Drift (%)
Fig. 12. Comparison of VBE Axial Forces from Various Methods
Fig. 11. Pushover Curves for Example SPSW. for SPSW-C.

612 612
462 462
Height (in)
Height (in)
312 312
162 162
Proposed Procedure Proposed Procedure
Pushover Analysis Pushover Analysis
LE+CD LE+CD
ICD ICD
0 0
−1 0 1 2 3 −1 0 1 2 3
Left VBE Moment (k−in) x 10
5 Right VBE Moment (k−in) x 10
5
(a) (b)
Fig. 13. Comparison of VBE moments from various methods for SPSW-C.
612
462
Height (in)
312
162
Proposed Procedure
Pushover Analysis
LE+CD
ICD
0
−3000 −2000 −1000 0 1000 2000 3000
VBE Axial Force (kip)
Fig. 14. Comparison of VBE axial forces from various methods for SPSW-V.

CONCLUSIONS REFERENCES
A procedure for estimating the design loads for VBEs of AISC (2005a), Specification for Structural Steel Buildings,
SPSWs has been proposed. The procedure does not involve ANSI/AISC 360-05, American Institute of Steel Con-
nonlinear analysis and is based on an assumed plastic collapse struction, Chicago, IL.
mechanism and linear model of one of the vertical boundary AISC (2005b), Seismic Provisions for Structural Steel Build-
elements. Moment and axial force diagrams from the pro- ings, ANSI/AISC 341-05, American Institute of Steel
posed procedure were shown to agree well with results from Construction, Chicago, IL.
pushover analyses of two example steel plate shear walls.
Berman, J.W. and Bruneau, M. (2003), “Plastic Analysis and
Furthermore, deficiencies in the two approximate methods
Design of Steel Plate Shear Walls,” Journal of Structural
for capacity design of VBEs in the commentary of the The
Engineering, ASCE, Vol. 129, No. 11, pp. 1448–1456.
Provisions were identified and they were found to result in
moment diagrams that differed significantly from those ob- Berman, J.W. and Bruneau, M. (2004), “Steel Plate Shear
served in pushover analyses. Walls Are Not Plate Girders,” Engineering Journal, AISC,
Vol. 41, No.3, 3rd Quarter, pp. 95–106.
ACKNOWLEDGMENTS Berman, J.W. and Bruneau, M. (2005), “Experimental Inves-
tigation of Light-Gauge Steel Plate Shear Walls,” Jour-
This work was supported in part by the Earthquake Engi-
nal of Structural Engineering, ASCE, Vol. 131, No. 2,
neering Research Centers Program of the National Science
pp. 259–267.
Foundation under Award Number ECC-9701471 to the Mul-
tidisciplinary Center for Earthquake Engineering Research. Bruneau, M., Whittaker, A.S. and Uang, C.M. (1998), Duc-
However, any opinions, findings, conclusions and recom- tile Design of Steel Structures, McGraw-Hill, New York,
mendations presented in this paper are those of the authors NY.
and do not necessarily reflect the views of the sponsors.
612 612
462 462
Height (in)
Height (in)
312 312
162 162
Proposed Procedure Proposed Procedure
Pushover Analysis Pushover Analysis
LE+CD LE+CD
ICD ICD
0 0
−1 0 1 2 3 −1 0 1 2 3
Left VBE Moment (k−in) x 10
5 Right VBE Moment (k−in) x 10
5
(a) (b)
Fig. 15. Comparison of VBE Moments from Various Methods for SPSW-V.

CSA (2001), Limit States Design of Steel Structures, CAN/ Monnier, A.B. and Harasimowicz, A.P. (2007), “Shear
CSA-S16-01, Canadian Standards Association, Wil- Strength,” Modern Steel Construction, AISC, January,
lowdale, Ontario, Canada. pp. 22–25.
Driver, R.G., Kulak, G.L., Kennedy, D.J.L. and Elwi, A.E. Redwood, R.G. and Channagiri, V.S. (1991), “Earthquake-
(1998), “Cyclic Test of Four-Story Steel Plate Shear Wall,” Resistant Design of Concentrically Braced Steel Frames,”
Journal of Structural Engineering, ASCE, Vol. 124, No. Canadian Journal of Civil Engineering, Vol.18, No. 5,
2, Feb., pp. 112–130. pp. 839–850.
Federal Emergency Management Agency (2000), “Prestan- Yang, Y.T. and Whittaker, A.S. (2002), “MCEER Demon-
dard and Commentary for the Seismic Rehabilitation of stration Hospitals, Mathematical Models and Preliminary
Buildings,” FEMA 356, Washington, DC. Analysis Results,” Technical Report, Multidisciplinary
Federal Emergency Management Agency (2003), “NEHRP Center for Earthquake Engineering Research, Buffalo,
Recommended Provisions for Seismic Regulations for NY.
New Buildings and Other Structures: Part 1-Provisions,”
FEMA 450, Washington, DC.

Effects of Nonverticality on Steel Framing
Systems—Implications for Design
ANDREA E. SUROVEK and JUSTIN JOHNSON
T he Specification for Structural Steel Buildings (AISC,

2005a), hereafter referred to as the Specification, allows
the engineer a great deal of freedom in selecting the type of
• Discuss implications of the current limits on when
imperfection effects may be neglected in the pres-
ence of lateral loads in the direct analysis approach of
analysis to be used when assessing the strength and stability Appendix 7.
of a structure or framing system. Chapter C states: The first section of the paper describes the types of initial
“Any method that considers the influence of second imperfections typically considered in planar frame analysis
order effects (including P-∆ and P-δ effects), flexural, and their effect on members and framing systems. This is
shear and axial deformations, geometric imperfec- followed by a discussion of how the effects of imperfections
tions, and member stiffness reduction due to residual are treated in the AISC Specification (AISC, 2005a).
stress on the stability of the structure and its elements A parametric study is presented in which the sensitivity
is permitted.” of framing systems to imperfection effects is investigated
with respect to a number of parameters, including slender-
By individually identifying each of the phenomena that ness ratios, leaning load levels, gravity-to-lateral load ratios,
affect member and system strength in a framing system, the and lateral frame stiffness, as measured by a second-order
Specification is highlighting for the engineer the importance to first-order drift ratio. Understanding the types of frames
of each of these phenomena. Consequently, an understand- that show sensitivity to imperfections provides a basis for
ing of each of these effects is beneficial in applying the new determining when imperfections can have an influence on
stability provisions of the Specification. This paper focuses their inclusion in a design approach and decisions on when
on the effects of geometric imperfections, with an emphasis they might be neglected in the analysis.
on frame nonverticality, or out-of-plumbness. The objectives In addition to the sensitivity study, a number of columns
of this paper are to: and simple frames are analyzed with and without imperfec-
• Illustrate how initial imperfections affect the strength tions using the direct analysis approach for assessing frame
of members and framing systems. stability outlined in Appendix 7 of the AISC Specification
(AISC, 2005a). The differences in the interaction checks for
• Discuss how imperfections affect the magnitude and simple columns and frames are used to discuss the current
distribution of internal member forces and moments. limits on when imperfection effects may be neglected in the
Specification in the presence of a higher lateral load.
• Demonstrate the sensitivity of different types of un-
braced frames to imperfections.
INITIAL IMPERFECTIONS
• Outline how imperfections are included in the AISC In planar frame analysis, two types of imperfections affect
Specification approaches for assessment of frame strength and stability due to the amplification of the in-plane
stability. moments:
• Frame out-of-plumbness (or nonverticality) that may oc-
cur during erection, designated by ∆0 in Figure 1.
• Member out-of-straightness which is a sweep of the

Andrea E. Surovek is assistant professor, civil and envi-
member between the member ends that occurs during
ronmental engineering, South Dakota School of Mines and
Technology, Rapid City, SD.
fabrication (one possible pattern is shown by δ0 in Figure
1). It is typically considered as a single curvature sweep
Justin Johnson is project engineer, Phelps Dodge Corpo- with the maximum imperfection at the midpoint of the
ration, Tempe, AZ.
member.
073-086_Surovek_Johnson_2008_1Q.indd 73 3/27/08 12:22:46 PM

The AISC Code of Standard Practice for Steel Buildings INCLUSION OF INITIAL IMPERFECTIONS
and Bridges (AISC, 2005b) specifies the following toler- IN AISC SPECIFICATION BASED
ances for these two imperfections: DESIGN APPROACHES
• Maximum member out of straightness, δ0, of L/1000. In the most recent version of the AISC Specification (2005a),
significant changes were made to the way in which stability
• For buildings less than 20 stories, out-of-plumbness, may be assessed in steel framing systems. In Chapter C, the
∆0, of H/500 in any shipping piece with a maximum Specification states that the required strengths for member
lean of 1 in. towards the exterior or 2 in. towards the design may be checked with member forces and moments
interior over the building height. Additional restric- obtained from one of the following analysis methods:
tions based on building height are imposed as shown
in Figure 2. • Second-order analysis using nominal frame stiffness
and a minimum lateral load of 0.002Yi, where Yi is the
Imperfections are considered in design because they amplify gravity load applied at level i.
the moments in members when second-order effects are con-
sidered. The principal effect of an initial out-of-straightness • Direct analysis, as outlined in Appendix 7 of the AISC
on an individual member is the additional of an internal mo- Specification.
ment when the axial load, P, acts through the initial out-of-
• First-order analysis, provided limitations on axial load
straightness, δ0, as shown in Figure 1. This moment reduces
levels are met, with an additional notional load, Ni,
the maximum axial capacity of a column. P-δ0 moments can
equal to
also impact adjoining members if their effect is to amplify
the end moments. In many practical cases, this effect on end
moments (when present) is minimal. ⎛ ∆⎞
N i = 2.1 ⎜ ⎟ Yi (1)
The initial out-of-plumbness (or nonverticality) also af- ⎝ L⎠
fects member strength, with the additional moment caused
by the axial load, P, acting through the nonverticality, ∆0. The where
P-∆0 moment also impacts the forces and moments in con- ∆ = first-order drift
necting elements, including connections, beams, base plates, L = story height
slabs, etc. Figure 3 shows a column in which P-∆0 moments
are transferred from the column to the adjoining beam (and,
of course, the connection between them.)
Fig. 1. Column flexure due to the axial load, P, acting through Fig. 2. AISC specified allowable erection tolerances for building
imperfections δ0 (out-of-straightness) and ∆0 (out-of-plumbness). frames.

In all cases, the initial out-of-plumbness is included, in The direct analysis approach includes the effects of the
some fashion, directly on the analysis side of the design two planar initial geometric imperfections as follows:
process. A minimum lateral load of 0.002Yi, required in both
• Out-of-straightness is explicitly accounted for by ei-
the second-order and direct analysis approaches, provides
ther altering the frame geometry or adding a notional
amplified moments and forces equivalent to those obtained
load of Ni = 0.002Yi at each level, where Yi is the total
when an L/500 out-of-plumbness is modeled directly. In the
factored gravity load acting on level i. The notional
first order approach, the effect of the initial imperfection is
load is equivalent to a uniform nominal out-of-plumb-
one factor included in the total notional load, Ni. The first-
ness of L/500.
order approach is similar to notional load approaches used
internationally, where a single horizontal load is used to • Out-of-straightness is implicitly accounted for by
account for multiple phenomena. calculating the compressive strength of the column
While out-of-straightness can have an important influence using the column curve provided in Chapter E of the
on the maximum strength of members in which the strength specification. The development of this curve includes
limit involves a nonsway failure mode, the modeling of the reduction in strength due to out-of-straightness.
member out-of-straightness within an analysis of the overall
structural system is more cumbersome than the modeling of By modeling the out-of-plumbness directly in the analysis,
a uniform frame nonverticality. In lieu of direct modeling, the increase in second-order moments due to the imperfec-
the effect of out-of-straightness on the strength is accounted tions is incorporated in the analysis results. In contrast, the
for in the axial strength term of the interaction equation giv- out-of-straightness is not measured directly as an increase in
en by Equations E3-2 through E3-4 in the AISC Specifica- second-order moment, but is accounted for as a decrease in
tion (AISC, 2005a). This is the case in all of the approaches member strength.
listed above.
The direct analysis approach of Appendix 7 meets all of FRAME SENSITIVITY TO INITIAL
the criteria listed in Chapter C. In the development of the di- OUT-OF-PLUMBNESS
rect analysis approach (Maleck, 2001; Surovek-Maleck and
White, 2004), an emphasis was placed on including in the Structural engineers can intuitively understand that for some
analysis, in as transparent a fashion as possible, both the ef- frames, such as highly redundant, laterally stiff frames, the
fects of geometric imperfections and the stiffness reduction initial imperfections will have a negligible effect on the
due to inelasticity and residual stresses. overall frame strength. This naturally leads to the questions:
Fig. 3. Transfer of forces and moments due solely to initial nonverticality (out-of-plumbness).

“How sensitive are frames to initial imperfections?” and, The frames studied were divided in two groups: axially
“When can these effects be neglected?” loaded portal frames, shown in Figure 4, and multi-bay
frames with distributed member loading. The behavior of
In order to address these questions, a parametric study was
portal frames with concentrated axial forces was examined
designed to determine:
with respect to the sensitivity of the ultimate frame strength
• The sensitivity of different frame configurations to to the inclusion of nonverticality in the model. In these stud-
initial out-of-straightness. ies, the beams were considered to be rigid and to remain
elastic; thus, the load capacity of the portal frames was en-
• Which parameters most affect the significance of the tirely dependent on the failure of the beam-column.
out-of-plumbness on the strength of the frame. Previous studies that have been performed to assess the
effects of imperfections (Wald, 1991; DeLuca, Faella and
Ultimate frame strength was determined using rigorous sec- Mele, 1993; Clarke and Bridge, 1997) have considered
ond-order, inelastic analyses (aka “plastic zone analyses”). frames that are loaded only with concentrated nodal loads.
The analysis included initial nonverticality, directly mod- The effects of relative beam-to-column stiffness, inelastic
eled in the frame geometry, as well as the effects of residual behavior of beams, and particularly, moment transferred to
stresses on the inelastic response. This analysis approach is the columns from the adjoining members due to transverse
typically used to determine the “exact” strength of a frame. beam loading were not considered. Consequently, the stud-
(Kanchanalai, 1977; White and Chen, 1993). Initial out-of- ies are limited in that, in all cases, the load capacity of the
straightness was not modeled in order to isolate the effects frame was dependent only on the column strength. In addi-
of nonverticality on the frame response. Both P-∆ and P-δ tion to the portal frames, the frames considered in this study
effects were captured by the analysis. Frame parameters included 14 single-story, multi-bay frames as well as 11
considered in the study included: multi-story, multi-bay frames with distributed beam loads.
• Column slenderness ratios (L /r). All of the frames (other than those originally studied by oth-
ers) were designed per AISC strength criteria as well as to
• Leaning load levels, that is, the axial load on pinned- meet serviceability limits. The frames were designed con-
pinned columns that do not provide stiffness to the sidering a reasonable set of parameters with regard to frame
lateral resistance of the frame. stiffness and gravity-load to horizontal-load ratios, although
a few very flexible frames were considered. Both strong-
• Gravity-load to lateral-load ratios. column, weak-beam and strong-beam, weak-column frames
were considered to assess the impact of the failure mode of
• Lateral frame stiffness (as measured by the B2 factor in the frame on the imperfection sensitivity.
the AISC Specification).
Portal Frame Studies
A number of small, stability critical portal frames in strong-
axis bending were analyzed by second-order, inelastic
analyses to assess the effects of nonverticality on sensitive
benchmark frames. These types of frames are often studied
to consider limiting cases with respect to beam-column
and frame stability (Kanchanalai, 1977). The portal frame
studied, which is oriented in strong-axis bending, is shown
in Figure 4. Parameters considered in the study included
gravity-load to horizontal-load ratios (ΣP/ΣH), as well as
base fixity, column slenderness (L /r), and amount of lean-
ing load (α). The pinned columns in frames UP and UR in
Figure 9 represent leaning columns. The beam was modeled
as rigid and of sufficient length so that overturning moments
did not impact results. Values of the parameters considered
included:
ΣP/ΣH = 10, 20, 33, 50, 100, 200
L /r = 20, 30, 40, 50, 60, 70, 80, 90, 100
α = 0, 1, 2, 4
Fig. 4. Imperfection study portal frame example.

Figure 5 illustrates the impact of horizontal-to-gravity load illustrate a particular behavior, and eight were taken from
ratios on imperfection sensitivity, measured as previous research studies. The details on all of the studied
frames are provided in Appendix A.
imperfection sensitivity = [(λ0 − λi)/λ0](100%) (2)
Single Story Frames
where
λ0 = ultimate load parameter for the perfect structure Twelve of the LRFD-designed frames were single-story
λi = the corresponding parameter with the inclusion frames ranging from one to eight bays. All of the frames,
of an H/500 imperfection except one, were designed for a low 10 psf nominal wind
load (to accentuate stability effects due to gravity load) and
In strong axis bending with a single leaning load, only those
to meet a maximum drift criterion of H/500 [using a service
frames with ΣP/ΣH > 50 showed greater than 5% imper-
load combination of 1.0D + 0.5L + 0.7W per Ellingwood
fection sensitivity. In Figure 6, the impact of leaning loads
(1996)]. The heavy gravity load was defined as 100 psf dead
on imperfection sensitivity is highlighted. For frames with
and 150 psf live, while light load was defined as 50 psf dead
larger leaning loads (2P to 4P), the imperfection sensitivity
and 50 psf live. The results for the 12 single-story frames are
is as high as 11%, but only when combined with very high
presented in Table 1.
gravity to horizontal load ratios. Column slenderness was
Only four of the frames considered had greater than 5%
not a significant parameter for lower values of ΣP/ΣH but
imperfection sensitivity. Of these, three were designed as
increased in significance for higher load ratios where the be-
weak-column/strong-beam frames in order to limit yielding
havior is dominated by axial effects. As shown in Figure 6,
in the beams at failure. In particular, two of these frames
the effects of leaning loads increases for higher ΣP/ΣH again
were designed with overstrength beams to limit or eliminate
due to axial effects; however, this effect of higher leaning
yielding in the beams at the failure load levels. All of these
load diminishes as slenderness ratios increased. It is interest-
sensitive frames exhibited extensive spread of plasticity in
ing to note that the slenderness ratio has little to no impact
the columns with no significant beam plasticity present at
on the sensitivity of frames with low ΣP/ΣH ratios.
failure. The fourth frame exhibiting imperfection sensitiv-
ity was a two-bay, pinned-base frame that was allowed to
Generalized Frame Studies
exceed drift limits by 40% and had a second order amplifica-
To extrapolate the results of the above portal frame studies tion factor, B2 , of 1.9.
to less idealized framing systems, 25 frames in strong axis All of the frames that showed greater than 5% sensitiv-
bending were analyzed with and without imperfections to ity to imperfections have B2 values greater than 1.3. While
determine their sensitivity to the inclusion of imperfections Figure 7 shows a general trend in greater sensitivity with
in the geometry. The frames were subjected to distributed respect to higher B2 values for the symmetric frames which
gravity loading. Fifteen frames were designed per AISC meet drift criteria, it indicates that there is not a direct corre-
specifications, two were modifications of the first fifteen to lation between lateral frame stiffness, as measured by the B2
Fig. 5. Imperfection sensitivity versus ΣP/ΣH, Fig. 6. Imperfection sensitivity due to leaning loads,
strong-axis bending, α = 1, Ks = 0. L/r = 20, Ks = 0.

factor, and the effect of the initial imperfection on the ulti- NEGLECTING THE IMPERFECTION EFFECT
mate strength of the frame. IN THE PRESENCE OF A LATERAL LOAD
The frames with higher B2 values all exhibited limited PER AISC APPENDIX 7
yielding in the beams and failure largely associated with plas-
Appendix 7 of the AISC Specification, which outlines the
ticity and eventually instability in the columns. In addition
direct analysis approach, allows for the notional load rep-
to the symmetric frames studied, two-bay, single-story un-
resenting initial imperfections to be treated as a “minimum
symmetrical frames with differing base fixity were analyzed.
horizontal load” in frames where the ratio of second-order
Due to the tendency of these frames to drift under gravity
drift to first-order drift, ∆2/∆1, (approximated by the B2 fac-
load, the P-∆0 moments due to the imperfections were not
tor) is less than 1.5. In other words, the notional load repre-
significant when compared to the P−∆ moments due to the
senting the imperfection may be neglected if a larger lateral
drift under gravity load, and neither exhibited a sensitivity
load is applied to the structure. While not explicitly stated in
greater than 1%, despite having B2 values of 1.5 to 2.4.
the specification, it is presumed this limit must be checked
and met at each story level.
Multi-Story Frames
Logic would suggest that since the notional loads repre-
Eleven multi-story frames were studied. Eight of these were sent physical imperfections, and since these physical imper-
two story frames previously studied by Ziemian (1990) with fections would be present regardless of the load condition,
variation in the following parameters: symmetry, base fixity the notional loads should be additive to other lateral loads.
and load level. Of these frames, none exhibited greater than However, the verification studies used to validate the direct
2.25% imperfection sensitivity. Two six-story, two-bay fixed analysis method (Maleck and White, 2003) suggest that for
base frames were designed under heavy and light gravity stability critical frames that meet the B2 limit, the unconser-
loading. Design of the heavily loaded frame was controlled vative error associated with disregarding the impact of the
by the maximum live load combination, while design of the imperfection in the presence of a higher lateral load is typi-
members in the lightly loaded frame was controlled by both cally less than 6%.
wind and gravity load combinations. In both cases, imper- In order to verify the maximum unconservative error asso-
fection sensitivity was less than 3%. ciated with neglecting the imperfection in the presence of a
In general, only those frames that were “stability critical”, larger lateral load, a small verification study was run that fo-
that is those frames where the failure was dependent on in- cused specifically on this effect. The studies were designed
stability of the columns rather than yielding of the beams, to correspond with those used to verify the direct analysis
were likely to be sensitive to the initial imperfection. Lean- method and included:
ing load levels and ΣP/ΣH ratios have a direct impact on
• individual column cases in which parameters consid-
frame sensitivity, all other parameters remaining constant.
ered fixity and gravity-load to horizontal-load ratios
While lateral stiffness has some impact on overall sensitivity
of a frame to imperfections, there was no direct correlation. • portal frames considered in the original verification
studies for the direct analysis method that exhibited
the highest unconservative error
In these studies, interaction diagrams were created based on

results from rigorous second-order plastic zone analyses and
compared to those developed from direct analysis results.
The second-order analysis results for the direct analysis ap-
proach were performed using a closed form second-order
analysis solution developed by LeMesurier (1977). LeMes-
surier’s method accounts for both P∆ and Pδ effects and is,
for all practical purposes, exact for the sidesway problems
studied in this work.
Both first-order (P versus M1) and second-order (P ver-
sus M2) interaction curves were developed, where M1 is the
maximum primary bending moment in the member due to
the applied loading, and M2 is the maximum internal second-
order bending moment, located at the member ends in all the
Fig. 7. Imperfection sensitivity for symmetric frames. problems considered.

The P versus M2 curves represent the strength of the Column Studies
member and, for the direct analysis approach, are the famil-
As anticipated, the maximum strengths (as indicated by the
iar interaction equations provided in Chapter H of the AISC
interaction check) of columns with higher ratios of gravity
Specification. The moments M2 are the moments that must
to horizontal loads (ΣP/H) were more sensitive to the exclu-
be transferred to the adjacent framing (in other words, to the
sion of the imperfection. For the fixed-guided case shown
beams and their connections at the beam-column joints) for
in Figure 8, the difference in strength between the columns
satisfaction of equilibrium at the beam-column joints. The P
with both the imperfection notional load and the lateral load
versus M1 interaction curves represent the maximum load-
and those that excluded the notional load was below 6% in
ings that can be applied to the column or frames. The first
all cases. It is interesting to note that the general trend does
order curves are of particular interest in design.
not continue to increase as lateral stiffness decreases. Simi-
The error in the design interaction curves relative to the
lar error levels were observed for pinned-fixed columns and
rigorous plastic zone solutions was measured for the col-
cantilever columns, and in no case did the unconservative
umns and frames with and without imperfections. This error
error exceed 6%.
is defined as
However, when considering the error in the moment cal-
rPZ − r culated in the columns, the error was entirely dependent on
e= (3) the gravity-load to horizontal-load ratio. The error associated
rPZ
with the resulting first order column moment was as high as
50% for a ΣP/ΣH ratio of 500. Because the error is associ-
where
ated with a high axial-load to horizontal-load ratio, the effect
rpz = radial measure to the interaction curve devel-
on the column interaction check is nominal (<6%), since the
oped using plastic zone analysis
interaction check is dominated by the axial term.
r = radial measure to the interaction curve devel-
oped using the direct analysis approach
The radial error represents a measure of the error in the inter-
action check, or an overall error in the strength as measured.
The horizontal error is a measure of the error in the first
order column moment. The maximum moment in the cases
studied occurred at the end of the column, so this represents
the error in the moment being transferred to adjoining mem-
bers and connections as well as the maximum moment in
the column.
Fig. 8. Sensitivity of fixed-guided columns to exclusion Fig. 9. Portal frames used in the verification
of imperfection in addition to horizontal load, H. of the direct analysis approach.

Portal Frame Studies were laterally supported by columns in weak axis bending.
None of the strong axis frames exhibited more than 6% error
In this study, the most unconservative of the verification
when a rigorous second-order analysis (that is, one consider-
studies for the direct analysis approach (Maleck and White,
ing both P-∆ and P-δ effects) was used. Figure 10 shows the
2003) were replicated with and without imperfections. Only
first order interaction curves for the UP_W40_G1_α1 frame
cases where B2 < 1.5 were considered. Both second-order
using a rigorous analysis, and Figure 11 shows the results for
analyses including P-δ effects and those including only P-∆
the P-∆ analysis. The complete results of the most sensitive
effects were performed. The verification frames are shown
cases are shown in Table 2.
in Figure 9; they are similar in parameters to those shown in
The verification studies were based on the error in the col-
Figure 4 with the exception that the beam stiffness was also
umn moment, only. For most practical adjoining members,
considered, as measured by the G factor found in the Com-
if a distributed load is present, the moment being transferred
mentary to Chapter C of the Specification, given by:
by the column (particularly the moment due to an imper-
IC fection) will not be a significant portion of the maximum
∑ moment on the beam or connection. However, in the rare
LC
G= (4) cases where the beam is not transversely loaded, such as the
I frame configuration shown in Figure 3, omission of the im-
∑ G
LG perfection effect may lead to a rather unconservative design
of the beam or connection. In cases where the design of con-
where necting elements may be negatively affected by omission of
IC = moment of inertia of the column this moment, such as in the case listed above, the notional
IG = moment of inertia of the girder load representing the imperfection should not be neglected
LC = unsupported length of the column regardless of the B2 factor.
LG = unsupported length of the girder
The frame designations indicate whether they are symmet- CONCLUSIONS AND RECOMMENDATIONS
ric or unsymmetric (S, U), pinned or restrained at the base
For the engineer engaged in design of steel frames, a ba-
(P, R), oriented in weak or strong axis bending (W, S), have
sic understanding of how initial imperfections affect frame
infinitely rigid (G0) or flexible beams (G1 or G3), and the
strength and behavior is beneficial, particularly when inter-
level of leaning column load (α). Only the most critical
preting the stability design provisions of the AISC Specifica-
frame studies were recreated (For complete results of the
tion. The discussed studies of columns and framing systems
original study, the reader is directed to Maleck and White,
considering the effects of nonverticality and a careful assess-
2003). The only variation from the initial study is that
ment of the current AISC provisions suggest the following:
error was only measured for interaction values where
B2 < 1.5. The highest unconservative error was reported in 1. The ultimate strength of most practical building frames
the UP_W40_G1_α1 and SP_W60_G0 frames; both frames will not be sensitive to initial nonverticality.
Fig. 10. First-order normalized interaction curves for Fig. 11. First-order normalized interaction curves for
UP_W40_G1_α1 frame P-δ (rigorous) analysis. UP_W40_G1_α1 frame, P-∆ analysis.

Table 1. Imperfection Sensitivity of Single-Story Symmetric Frames
Imperfection Number of
B2 ΣP/ΣH
Sensitivity frames
< 2% 4 1.03–1.1 60–280
2–5% 4 1.07–1.31 140–710
> 5% 4 1.3–1.93 280–1100
Table 2. Radial Error Measured in Interaction Checks for

Verification Frames for B2 < 1.5
Error with Error without
Imperfection Imperfection
Frame
P-∆ P-δ P-∆ P-δ
Designation
SP S40 G0 -4% -3% -7% -6%
SP S60 G0 -4% -2% -8% -6%
SP S20 G0 -2% -2% -6% -5%
UP W40 G1 α2 -1% 0% -8% -6%
SP W60 G0 -6% -5% -10% -8%
SP W80 G3 -2% -1% -2% -2%
2. The parameters that have the greatest effect on imper- design may be significantly impacted by the moment
fection sensitivity include gravity-load to horizontal- amplification due to the imperfection.
load ratios, frame symmetry, and amount of leaning One point to consider is: When using the direct analysis
load. However, the primary cause of initial imperfec- approach in Appendix 7, is it easier to modify the geometry
tion sensitivity in frames is the mode of failure and or calculate notional loads than to perform the calculations
whether that failure is initiated by instability of a col- necessary to determine whether the imperfection effects can
umn rather than yielding in a beam. be neglected? Consider that if a direct second order analysis
algorithm is used (that is, if B1 and B2 factors are not being
3. Frames for which imperfection effects are negligible
separately calculated), a separate first-order analysis is still
are not easily identified quantitatively, as no single
required to determine if the imperfection can be ignored. If
parameter controls the sensitivity.
the imperfection effect is small enough to neglect, inclusion
4. The provision of AISC Specification Appendix 7, in of the notional load will have a negligible effect on the final
which the effects of imperfections may be neglected in design. In short, if the impact of including imperfections is
lieu of higher lateral loads when B2 < 1.5, is shown to negligible, economy will not be lost by including these ef-
produce a maximum unconservative error of 8%. This fects, and this requires less effort than determining whether
error occurred in a highly stability-critical portal frame they can be neglected. It is also more rational, since the no-
laterally supported by a weak axis column only. When tional load represents a potential physical phenomenon that
a less rigorous P-∆ analysis is used, the maximum un- is independent of the load case.
conservative error was 10%. For practical frames, this Currently the maximum permissible imperfection is in-
maximum unconservative error will be significantly corporated into the design provisions. There is a lack of data
smaller. Most practical frames will not be governed on measured nonverticality in constructed facilities to warrant
by the behavior of a weak-axis, unbraced, laterally a reduction of this imperfection. There are very few pub-
resisting column. lished studies that include surveyed measurements of a con-
structed building (Bridge, 1998; Beaulieu and Adams, 1978).
5. If neglecting imperfections in the presence of a larger Bridge concluded that many individual columns exceeded
lateral load per Appendix 7, the engineer should be construction tolerances; however, story and global imperfec-
careful to consider the “special” cases where the beam tion tolerances were met by compensating imperfections.

More data on erected structures would be useful if modifica- Surovek-Maleck A.E. and White, D.W. (2004), “Alterna-
tions to these provisions in the form of reduced requirements tive Approaches for Elastic Analysis and Design of Steel
for imperfections were to be considered in future editions of Frames. I: Overview,” Journal of Structural Engineering,
the Specification. ASCE, Vol. 130, No. 8, pp. 1186–1196.
Wald, F. (1991), “Sensitivity of Semi-Rigid Frames to Initial
REFERENCES Imperfections,” Journal of Constructional Steel research,
AISC (2005a), Specification for Structural Steel Buildings, Vol. 18, pp. 309–316.
American Institute of Steel Construction, Inc., Chicago, IL. White, D.W. and Chen, W.F., eds. (1993), Plastic Hinge
AISC (2005b), Code of Standard Practice for Steel Build- Based Methods for Advanced Analysis and design of Steel
ings and Bridges, American Institute of Steel Construc- Frames – An Assessment of the State of the Art, Structural
tion, Inc., Chicago, IL. Stability Research Council, Bethlehem, PA, 299 pp.
Beaulieu, D. and Adams, P. (1978), “Results of a Survey on Ziemian, R.D. (1990), “Advanced Methods of Inelastic
Structural Out-of-Plumbs,” Canadian Journal of Civil En- Analysis for in the Limit States Design of Steel Struc-
gineering, Vol. 5, No. 4, Dec, 1978, pp. 462–470. tures,” Ph.D. Thesis, Cornell University, Ithaca, NY.
Bridge, R.Q. (1998), “The Inclusion of Imperfections in
Probability-Based Limit States Design,” Proceedings, APPENDIX A
1998 Structural Engineering World Congress, San Fran-
cisco, California, July. Parametric Study Frames
Clarke, M.J. and Bridge, R.Q. (1997), “Notional Load Ap- Beam and column designations used in the single and multi-
proach for the Assessment of Frame Stability,” Chap- story frames are presented below. The frame designations
ter 4, Effective Length and Notional Load Approaches are shown in Figure A.1. Steel designations used in the
for Assessing Frame Stability: Implications for Ameri- single and two-bay symmetric test frames are given in Ta-
can Steel Design, American Society of Civil Engineers, bles A.1 and A.2, respectively. The remaining frame designs
pp. 181–278. are given in Figures A.2–A.11. Table A.3 presents the B2
DeLuca, A., Faella, C., and Mele, E. (1993), “Advanced factors, analysis results, and imperfection sensitivities for
Analysis: Numerical Results and Design Guidelines for each of the test frames. The imperfection sensitivity was
Rigid and Semi-Rigid Sway Frames,” in Plastic Hinge calculated with
Based Methods for Advanced Analysis and Design of Steel
Frames – an Assessment of the State of the Art, Structural imperfection sensitivity = [(λ0 − λi)/λ0](100%)
Stability research Council, Bethlehem, PA, pp. 41–64.
where
Ellingwood, Bruce (1996), “Structural Serviceability: Re-
λ0 = ultimate load parameter for the perfect
view and Standard Implementation,” Building an Interna-
structure
tional Committee of Structural Engineers, Proceedings of
λi = the corresponding parameter with the inclusion
Structures Congress XIV, American Society of Civil En-
of an H/500 imperfection
gineers, pp. 436–442.
Kanchanalai, T. (1977), “The Design and Behavior of Beam- Details of the frames with designations beginning with U or
Columns in Unbraced Steel Frames,” AISI Project No. S may be found in Ziemian (1990).
189, Report No. 2, Civil Engineering/Structures Research For the symmetric frames:
Lab., University of Texas, Austin, TX, 300 pp. story height = 12 ft
LeMessurier, W. J. (1977), “A Practical Method of Second bay width = 25 ft
Order Analysis. Part 2: Rigid Frames,” Engineering Jour- frame spacing = 25 ft
nal, AISC, Vol. 14, No. 2, pp. 49–67.
For the unsymmetric frames:
Maleck, Andrea E. (2001), “Second-Order Inelastic and story height = 12 ft
Modified Elastic Analysis and Design Evaluation of Pla- bay widths = 20 ft and 30 ft
nar Steel Frames,” Ph.D. Dissertation, Georgia Institute of frame spacing = 25 ft
Technology, 579 pp.
Maleck (Surovek), A.E. and White, D.W. (2003), “Direct
Analysis Approach for the Assessment of Frame Stabil-
ity: Verification Studies,” Proceedings of the 2003 SSRC
Annual Technical Sessions and Meeting, Baltimore, April,
pp. 423–441.

Fig. A.1. Key to frame designation.
Fig. A.2. 2B1S-FUH.
Fig. A.3. 2B1S-PUH.
Fig. A.4. Frame 4B1S-PSH1.
Fig. A.5. Frame 4B1S-PSH2 (oversized beams).
Fig. A.6. Frame 5B1S-PSH1.
Fig. A.7. Frame 5B1S-PSH2 (oversized beams). Fig. A.8. Frame 8B1S-PSH.

Fig. A.9. Frame 2B6S-FSL.

Table A.1. Single Bay Steel Sizes Table A.2. Two-Bay Symmetric Steel Sizes
Designation Beam Column Exterior Interior

Designation Beam
1B1S-FSH W30x90 W12x53 Column Column
2B1S-FSH W27x84 W10x33 W8x40
1B1S-PSH1 W30x99 W14x43
2B1S-PSH W24x94 W8x25 W8x40
1B1S-PSH2 W24x104 W8x40
1B1S-FSL W12x58 W8x31
1B1S-PSL W12x58 W10x39
Table A.3. Results of Planar Frame Analyses

Frame Sensitivity
B2 λi λ0
Designation (%)
1B1S-FSH 1.028 1.478 1.478 0
1B1S-PSH1 1.082 1.517 1.522 2.31
1B1S-PSH2 1.247 1.322 1.354 2.36
1B1S-FSL 1.037 1.165 1.165 0
1B1S-PSL 1.082 1.449 1.425 1.67
2B1S-FSH 1.106 1.606 1.623 1.11
2B1S-PSH 1.926 1.324 1.392 5.13
2B1S-FUH 1.546 1.485 1.486 0.02
2B1S-PUH 2.42 1.359 1.37 0.75
4B1S-PSH1 1.25 1.441 1.503 4.32
4B1S-PSH2 1.354 1.496 1.616 8
5B1S-PSH1 1.314 1.427 1.492 4.57
5B1S-PSH2 1.3 1.495 1.624 8.61
8B1S-PSH 1.47 1.351 1.466 7.84
2B6S-FSH – 1.299 1.3 0.34
2B6S-FSL – 1.413 1.418 0.37
1B10S-FSL – 1.38 1.434 3.54
U-P36L – 1.144 1.145 0.07
U-P36H – 1.073 1.088 1.46
U-F36L – 1.163 1.164 0.2
U-F36H – 1.161 1.162 0.1
S-P36L – 1.226 1.227 0.03
S-P36H – 1.218 1.246 2.23
S-F36L – 1.242 1.264 1.8
S-F66H – 1.184 1.189 0.4

Current Steel Structures Research
REIDAR BJORHOVDE
S tandardization is essential to activities in all walks of

life, but probably nowhere as important as it is for all
aspects of the construction industry. The approaches vary
The projects that are presented in this paper reflect a broad
range of state-of-the-art work. A recent project in Spain has
provided a broad assessment of the methods that are used
somewhat from country to country and region to region, but to determine the moment capacity of laterally unsupported
it is a fact that quality in all respects depends on a body of beams. One German study focuses on residual stresses in
practically oriented standards. The complexity of the subject high strength steel shapes, using a new measurement tech-
cannot be overestimated, since there are so many materials nique, and another German project looks at the influence
and components and methods of fabrication and erection of the Bauschinger Effect on the deflections of cambered
that govern the quality of the final product. The various steel beams. Projects addressing seismic effects continue to be
construction institutes play key roles in all of these efforts, very prominent in the US and many other locales, and a nov-
primarily because that is where the technical expertise re- el structural system utilizing repairable “fuses” is examined
sides. However, the interaction between any institute and its in a major, multi-university and design firm US project. Full-
constituency is critical as well, and the model of cooperation scale testing forms part of several investigations, including a
that is used in North American standardization work is an ex- four-story building that has been tested to collapse in Japan.
cellent vehicle to bring all interested parties to the table, so to Finally, numerous researchers continue to develop software
speak. The AISC Committee on Specifications, for example, that may eventually be used in practice, and one such re-
is made up of approximately 45 individuals, with one third cent product from Belgium deals with analysis and design
from consultants, one third from academia and one third from of plated structures.
industry. The arrangement works extremely well, producing a References are provided throughout the paper, whenever
steel design standard that is second to none in terms of state- such are available in the public domain. However, much of
of-the-art approaches and practical usefulness. the work is still in progress, and reports or publications have
On the other hand, design manuals are the final element not yet been prepared for public dissemination.
of the research and development and standardization work.
Whether the manuals appear in the form of “hard” printed EFFECTS OF MATERIAL CHARACTERISTICS
copies or as electronic tools is immaterial: the point is that ON MEMBER BEHAVIOR
they are critical to the users of the standards that have been
Residual Stresses in Hot-Rolled Shapes of S460 Steel:
developed. The AISC Manual of Steel Construction is cur-
This project is conducted at the Technical University of
rently in its 13th edition, and is accompanied by numerous
Darmstadt in Darmstadt, Germany, with Professor Jörg
other manuals and design guides that together aim to make
Lange as the director.
the work of steel construction professionals easier. This ap-
The S460 steel grade is a high strength low alloy mate-
proach is pursued just about everywhere in the world, most
rial with a specified minimum yield stress of 460 MPa
recently in Europe, where the European Convention for
(65 ksi). There is significant interest in the use of this and
Constructional Steelwork (ECCS) has embarked on an ag-
similar high strength grades for construction, but a drawback
gressive program of design manual development. The ECCS
to their application continues to be that knowledge of the re-
program aims at having a full complement of manuals that
sidual stress magnitudes in high strength steel shapes is very
will ease the acceptance and adoption and use of Eurocodes
limited. Although the value of the modulus of elasticity is the
3 (steel structures) and 4 (composite construction) by profes-
same for all grades of steel and the levels of residual stress
sionals throughout Europe. This is a major effort, for certain,
therefore should not vary much between identical shapes in
but it is critical for industry and indeed the economies of the
different grades, there are still applications where the actual
various countries.
stress levels are needed. This is especially the case for sta-
bility issues such as column buckling, and also for girders
where fatigue and fracture may play a role.
Traditional residual stress measurements are commonly
made via destructive test methods, such as sectioning and
Reidar Bjorhovde is the Research Editor of the AISC Engi- hole drilling, to mention the two most common methods
neering Journal. (Tebedge, Alpsten and Tall, 1971; Galambos, 1998). These
approaches are very accurate but extremely costly and time-
087-092_research_2008_1Q.indd 87 3/27/08 12:24:02 PM

consuming. The Darmstadt project focuses on the use of a due to the combination of service load stresses and residual
magnetic effect (“Barkhausenrauschen”) and the fact that stresses would almost certainly not take place, and the local
the spatial structure of magnetic fields changes during de- yielding that was developed during the cambering operation
magnetization. Further, stresses in the ferromagnetic mate- would also be very limited. Of course, the latter would de-
rial change the magnetic fields significantly, and this change pend on the amount of cambering or curving, but it will be
can be quantified. With appropriate calibration tests, the val- very small for the usual fraction of the dead load camber, as
ues of the residual stresses can be determined. is common in American practice.
This testing method offers major potential for the deter- Short beam segments in S355 steel (50 ksi yield stress)
mination of uniaxial and biaxial residual stress distributions. were loaded to varying levels of strain beyond yield, and
The work continues to improve calibration techniques and then loaded in the opposite direction. These tests imposed
hence the accuracy of the stress magnitudes that are deter- various degrees of prestrain, as high as 2%, and demonstrat-
mined. It is expected that the residual stress data base for ed that the Bauschinger Effect was clearly reflected in the
such high strength steel members will be significantly en- cyclic stress-strain relationships. Using 18 m (60 feet) long
larged. The benefits for bridge construction in particular may beams that were curved to varying degrees, tests and addi-
be very useful. tional analyses showed significant differences between the
measured and computed deflections, as illustrated in Figure
Influence of the Bauschinger Effect on Deflections of
1. The agreement between tests and theory is generally very
Cambered Beams: This is another study that has been con-
good. A non-cambered beam was included for reference.
ducted at the Technical University of Darmstadt, also with
It is certainly possible to take this type of behavior into
Professor Jörg Lange as the director.
account in the analysis of a structure, but it is a question
The Bauschinger Effect is a phenomenon that affects the
whether it is strictly necessary. The study demonstrates and
response characteristics of steels that undergo strain harden-
explains the behavior that can be anticipated. On the other
ing. It is also referred to as the directionality of strain hard-
hand, the structural displacements will stabilize once the
ening (Dieter, 1986). For example, a specimen that has been
loads have been applied, and no further deflection increases
loaded in tension past the yield level and into strain harden-
should be expected. Finally, the amount of cambering that is
ing will unload elastically. Upon reloading in the same direc-
needed can take these kinds of increases into account, if it is
tion as the original load, the material will appear to have a
deemed necessary for the service performance of the mem-
level of yield stress that is the same as the level of stress at
bers and the structure.
which the original test was stopped. Reversing the sense of
the loading to zero and then into compression, the point at
STRUCTURAL MEMBER RESPONSE MODELING
which yielding commences in compression is significantly
lower than the original compressive yield stress. This is the Equivalent Moment Distribution Factors for Lateral-
Bauschinger Effect. It is completely reversible, to the effect Torsional Buckling: This research project has been con-
that if the original loading were in compression, the higher ducted at the University of Navarra in Pamplona, Spain,
yield stress would occur under this state of stress and the with Professor Miguel Serna and Dr. Aitziber Lopez as the
lower value would be found for tension. The Bauschinger directors.
Effect is important for high-deformation operations such as
plate bending or other functions where the material is alter-
nately subjected to high strains in tension and compression.
The study at the Technical University of Darmstadt started
out as an investigation of the strength and behavior of steel
and composite beams under service conditions, and specifi-
cally aimed at determining the deflections of such members
(Grages, 2007; Grages, Lange and Sauerborn, 2007). One of
the original issues was that the calculated deflections often
understated the measured values. The measurements were
made at various stages of various structures, and one of the
interesting findings was that cambered beams deflected more
than straight members. It was initially thought that this was
caused by the residual stresses in the members, although it
should have been clear that such could not be the case. The
magnitudes of the service loads and the accompanying de-
flections and rotations are very small and the response of the Fig. 1. Deflections of cambered and non-cambered beams
beams is elastic for all practical purposes. Any local yielding (courtesy of Hauke Grages).

The equivalent moment factor, Cb, has been used for many The values of the end moment ratio M/M0 are shown in the
years and by many design codes worldwide in the design of horizontal axis; the values of the equivalent moment factor
laterally unsupported beams that may fail in lateral-torsional are shown on the vertical axis. Albeit much more complex
buckling. It is a convenient conversion factor that translates than the expression used in the current codes, the research-
the actual moment distribution into an equivalent uniform ers observe that including the warping restraint leads to
distribution, for which the theoretical elastic buckling solu- significantly higher Cb-values than what is predicted by the
tion is well known. Some of the early research studies on the expressions of the 1994 AISC LRFD Specification and the
subject were conducted in the 1960s and 70s by W.J. Austin, 2000 British code BS 5950; further, including the effect of
J.W. Clark and others (Galambos, 1998). Very significant the bending restraint leads to lower Cb-values. These find-
work has been done for the past 25 years or so by Galambos, ings are demonstrated by the data shown in Figure 3. The
Trahair, Nethercot and other researchers. The Cb equation study included a broad range of comparisons with the two
that has been used in the AISC Specification (AISC, 2005) code documents, and the researchers note that the lower
for several editions is attributed to Kirby and Nethercot Cb-values for the end-restraint case indicate that the codes
(1979), and it offers the convenience of applicability to all may be unconservative for such cases. The higher values for
types of moment distributions. the warping-included cases obviously indicate conservative
The investigation by Serna and Lopez focused on the fact code criteria.
that the current Cb equation does not take into account any
lateral bending or warping restraint at the supports or sup- SEISMIC RESPONSE OF STEEL STRUCTURES
port points for the beam. Examining a wide range of beam
Controlled Rocking of Steel-Framed Buildings with
support and moment distribution cases, using finite element
Replaceable Energy Dissipating Fuses: This is a very
as well as finite difference solutions, the study incorporated
substantial investigation of the performance of a novel type
the following restraint conditions:
of structural system, currently underway as a joint project
(i) no restraint to lateral bending and warping of the University of Illinois, Stanford University, Hokkaido
University in Japan, and the California consulting firms
(ii) full restraint to lateral bending and warping Tipping & Mar Associates and GPLA. The directors are
Professors Gregory G. Deierlein of Stanford University and
(iii) restraint to lateral bending only Jerome F. Hajjar of the University of Illinois. It is supported
by funding from the NEES program of the National Science
(iv) restraint to warping only
Foundation.
As an example, Figure 2 shows a typical symmetric moment The principal aim of the project is to arrive at performance
distribution that was used for the analysis of an 8 m (26 ft and design criteria that will eliminate residual drift and also
8 in.) long IPE 500 beam. This is a 20-in.-deep hot-rolled concentrate the structural damage in the replaceable fuses of
shape with no closely comparable American counterpart; the the braced frame. Residual drift is the permanent sway that
flange width of the IPE 500 is 8 in., the flange thickness is
s in. and the web thickness is a in. Figure 3 shows the result-
ing equivalent moment factor values, as determined by the
finite difference solution, by the solution of the researchers,
and by the AISC Cb expression from the 1994 LRFD Specifi-
cation for Structural Steel Buildings (which is the same as in
the 2005 Specification for Structural Steel Buildings).
Fig. 2. Linear moment distribution used in computational example Fig. 3. Values of Cb for the IPE 500 example beam
(courtesy of M. Serna). (courtesy of M. Serna).

remains in the building following an earthquake. The revo- (iii) Providing an effective graphical interface to improve
lutionary concept of this system is that the seismic damage the interpretation of the analytical results
will be focused primarily, if not entirely, in the replaceable
shear fuses that are incorporated in the frame, along with the (iv) Offering an efficient substitute for much more complex
self-centering post-tension bars that form an integral part of finite element analysis
the system. That is, rather than focusing totally on limiting or
preventing the local or overall failure of the frame, the rock- Named EBPlate, the program is available for download-
ing concept and shear fuses aims specifically at providing ing at no cost at the websites www.cticm.com and www.
for economical repair of a damaged structure. Figure 4 illus- steelbizfrance.com . It is noted that CTICM is the French
trates the basic elements and the key behavioral aspects. steel construction institute.
The project is further enhanced by the collaboration with
researchers for a project of Hokkaido University in Japan. REFERENCES
A full-scale test of a frame with square built-up tubular col- AISC (2005), Specification for Structural Steel Buildings,
umns and wide-flange beams was recently (September 27, ANSI/AISC 360-05, Chicago, IL.
2007) tested at the three-dimensional E-Defense facility in CEN (2005), Eurocode 3: Design of Steel Structures – EN
Japan. Figure 5 shows this frame after the completion of the 1993, Comité Européen de Normalisation, Brussels,
test. Evaluations of the results are forthcoming. Belgium.
COMPUTER SOFTWARE DEVELOPMENT Dieter, George E. (1986), Mechanical Metallurgy, 3rd Edi-
tion, McGraw-Hill, Inc., New York, NY.
Software for Design of Plate Structures against Plate
Galambos, T.V. (1998), Guide to Stability Design Criteria
Buckling: This is a major software development that has
for Metal Structures, Editor, 5th Edition, Wiley-Inter-
been undertaken by the research group at the University
science, New York, NY.
of Liège in Liège, Belgium. The project director has been
Professor René Maquoi. Funded by the COMBRI project of Grages, Hauke (2007), Beitrag zur Verformungsanalyse von
the European Union, the software is intended for use in the Verbundträgern, Ph.D. Dissertation, Technical University
analysis and design of steel-plated structures for bridges. of Darmstadt, Darmstadt, Germany.
The design criteria that have been used in this software
development are those of the plate buckling requirements of
Eurocode 3, Section 1-5 (CEN, 2005). The focus is on the
following criteria:
(i) Accurate determination of elastic plate buckling stresses for
rectangular stiffened plates, covering a much wider range
of design parameters than what is currently available
(ii) Accounting for the stabilizing effects of the torsional

stiffness of stiffeners and the rotational stiffness of the
edges of the plates
Fig. 4. Self-centering frame with replaceable shear fuses Fig. 5. Full-scale frame test at the E-defense facility
(courtesy of J.F. Hajjar). near Kobe, Japan.

Grages, Hauke, Lange, Jörg, and Sauerborn, Norbert (2007), ACKNOWLEDGMENTS
“Influence of the Bauschinger Effect on the Deflection
Special thanks are due the following members of the Inter-
Behaviour of Cambered Steel and Steel-Concrete Com-
national Structural Steel Research Advisors (ISSRA) who
posite Beams,” in Steel and Composite Structures, Y. C.
provided input to this paper:
Wang and C. K. Choi (Eds.), Taylor and Francis, London,
England, pp. 133–138. Eduardo Bayo, University of Navarra, Pamplona, Spain
Kirby, P.A. and Nethercot, D.A. (1979), Design for Struc- Ulrike Kuhlmann, University of Stuttgart, Stuttgart,
tural Stability, Wiley-Interscience, New York, NY. Germany
René Maquoi, University of Liège, Liège, Belgium
Serna, M.A., Lopez, A., Puente, I. and Yong, D.J. (2006),
Luis da Silva, ECCS and University of Coimbra, Coim-
“Equivalent Uniform Moment Factors for Lateral-Tor-
bra, Portugal
sional Buckling of Steel Members,” Journal of Construc-
tional Steel Research, Vol. 62, No. 6, pp. 566–580. Additional assistance has been provided by Jörg Lange,
Tebedge, N., Alpsten, G.A. and Tall, L. (1971), “Measure- Technical University of Darmstadt, Darmstadt, Germany;
ment of Residual Stresses: A Comparative Study of Meth- Dieter Ungermann, University of Dortmund, Dortmund,
ods,” Fritz Engineering Laboratory Report No. 337.8, Le- Germany; Jerome F. Hajjar, University of Illinois, and Judy
high University, Bethlehem, PA. Liu, Purdue University.

Suggested Reading from Other Publishers
The following abstracts summarize papers published by others on the subject of steel design and construction that may be of
interest to Engineering Journal.
Reidar Bjorhovde, Research Editor of the AISC Engineering Journal, prepared the following abstracts from the Journal of
Constructional Steel Research.
From Volume 63, Number 8, 2007 of the Journal of Con- Effect of Shear Force on the Initial Stiffness of Top and
structional Steel Research (JCSR) published by Elsevier, Seat Angle Connections with Double Web Angles
Ltd.: F. Danesh, A. Pirmoz and A.S. Daryan
Extensive 3D numerical modeling of a variety of connections
Strength Design Criteria for Steel Members at Elevated
has been performed, using realistic material properties and
Temperatures
member and connection details. Effects such as connection
Jiro Takagi and Gregory G. Deierlein
slip, bolt pretension and friction forces have been incorpo-
Design equations for steel members at fire-level temperatures rated in the models. A large number of connection tests have
are evaluated by comparisons with nonlinear finite element also been conducted, in good agreement with the analytical
simulations. The AISC and Eurocode 3 equations for lateral- results. The effect of the shear force is examined for several
ly unsupported W-shape columns, beams and beam-columns levels of load, and it is shown that the shear force reduces
are evaluated for temperatures between ambient and 800 oC the initial stiffness. An equation is developed that reflects the
(1,472 oF). It is shown that the Eurocode provisions are ac- influence of the shear force on the stiffness.
curate to within 10 to 20% of the finite element results. The
AISC equations predict capacities that are as much as 100 Experimental Behavior of High Strength Steel End-Plate
percent unconservative (i.e. the design capacities are twice Connections
as large as the finite element solutions). The differences are Ana M. Girão Coelho and Frans S. K. Bijlaard
particularly large for members with intermediate slender-
Addressing issues such as high steel strength, limited ductil-
ness and for temperatures above 300 oC (572 oF). Modifi-
ity and connection rotation capacity, a large number of tests
cations are proposed for the AISC equations that improve
and analyses of end-plate connections has been conducted.
the accuracy to within 20 to 30% of the nonlinear finite
The columns and beams in the test specimens were of the
element findings. The paper also discusses the limitations
steel grade S355 (355 MPa = 50 ksi specified minimum
of member-based design criteria and examines research and
yield stress); the steel grade S690 (690 MPa = 100 ksi yield
development needs for structural fire engineering.
stress) was used for the end plates. The major contributions
of the study are: (i) the characterization of the nonlinear be-
From Volume 63, Number 9, 2007 of the Journal of Con-
havior, (ii) the validation of the use of Eurocode 3 criteria
structional Steel Research (JCSR) published by Elsevier,
for such connections, and (iii) the ductility analysis of such
Ltd:
high strength steel connections. The connections that were
tested satisfy current design provisions for stiffness and
Capacities of Headed Stud Shear Connectors in Com-
strength, and they exhibit reasonable rotation capacity. [It is
posite Steel Beams with Precast Hollow-Core Slabs
noted (Bjorhovde comment) that Eurocode 3 currently does
Dennis Lam
not have provisions for steel with yield strength larger than
The author observes that almost all shear connector tests have 460 MPa (65 ksi)].
been performed with either solid slabs or with slabs on steel
deck. However, hollow-core slabs have become very com-
mon in composite construction. A push-out test for use with
hollow-core slabs is developed, and a very large number of
tests have been conducted. A number of tests have also been
performed with solid slabs, to verify the proposed push-out
procedure. Design equations for studs in hollow-core slabs
are also presented.
093-095_suggested_2008_1Q.indd 93 3/27/08 12:24:27 PM

The Semi-Rigid Behavior of Three-Dimensional Development of design methods for orthotropic decks is dis-
Steel Beam-to-Column Joints Subjected cussed. For basic design simplified approaches, such as the
to Proportional Loading Pelikan-Esslinger method, are generally used. For assess-
Part I: Experimental Evaluation ment of the local stresses in the plating finite elements analy-
Part II: Theoretical Model and Validation sis may be useful, however, this method is not applicable
J. M. Cabrero and Eduardo Bayo in the vicinity of the welds where the material is generally
plasticized because of the residual stresses due to the weld
A number of three-dimensional extended end-plate con-
shrinkage, actual magnitudes and distribution of which are
nections have been tested for static proportional loads, with
unknowable to the designer. The classical (Wöhler’s) fatigue
bending about both major axes. It is shown that the rotational
theory is valid only within the elastic limits of the material.
stiffness increases with this type of loading. Further, as ex-
Because of these analytical difficulties and impracticality of
pected, increasing the end-plate thickness increases the flex-
refined numerical investigations in the deck design , current
ural stiffness and strength, but reduces the rotation capacity.
design specifications permit empirical fatigue safety assess-
In the analytical companion part of the project, the compo- ment by conformance with approved geometric standards.
nent method that is used in Eurocode 3 was expanded to take
Due to their light weight, durability and load carrying ca-
into account the 3D response characteristics. A new com-
pacity, orthotropic decks have a promising future being in-
ponent that is used for bending of laterally supported plates
dispensable in the super-long span bridges. Because of their
is introduced, along with complete elastic models for major
capability for being structurally integrated with the main
and minor axis bending of the connections. The expanded
members of the existing steel bridges, orthotropic decks are
method of analysis offers significant improvement for three-
also excellently suited for replacing the deteriorating con-
dimensional analysis.
crete decks. In the U.S. nearly 40% of highway bridges are
now considered “structurally deficient” or “functionally
From the July 2007 issue of Stahlbau, published by Ernst
obsolete”, with failed concrete decks accounting for about
& Sohn:
two-thirds of the deficiency cases. Several major U.S. bridges
have recently received new orthotropic decks. Special prob-
Orthotropic Decks—Past Developments and Future
lems encountered with redecking are discussed.
Outlook
Roman Wolchuk The use of orthotropic decks may be substantially increased
by further optimization of their structural properties and by
Bridges with steel orthotropic decks, first introduced in the
their cost reduction. Nearly 75% of the cost of an orthotropic
1950s in Germany, have been subsequently built in other
deck is due to the cost of fabrication and erection labor. Op-
countries, with several thousand of such structures now in
timization and cost reduction measures are suggested, such
service throughout the world. The performance record of or-
as using longer deck spans and minimizing the number of
thotropic decks has been satisfactory, excepting occasional
expensive intersections of the deck ribs with the cross beams
fatigue cracking which had occurred on some decks built in
and simplification of these details. Fatigue-safe and eco-
the 1960s and 1970s, mainly caused by inappropriate details
nomical standard designs and details could be established
and fabrication defects. Specific causes of such failures are
by further research and testing and incorporated in future
discussed, including locked-in tensile stresses at welds made
design codes, which should also include stringent fabrica-
in constrained conditions, excessive fabrication tolerances,
tion and welding rules for orthotropic decks. International
faulty welds, local stress concentrations at structural discon-
cooperation in these efforts would be welcome.
tinuities. Surfacing failures have been due to the initial use
of thin 10 to 12 mm deck plates with excessive local deck
flexibility causing stresses in the surfacing acting compos-
itely with the deck plate.

From 6 November 2007 issue of The Structural Engineer¸
published by The Institution of Structural Engineers, United
Kingdom:
Hot v. Cold Formed Hollow Sections

(Corus Tubes)
The quality and type of hollow sections is vitally important
for the overall performance of a structure. When considering
the differences in the quality and performance of both hot
finished and cold formed hollow sections, there are five main
considerations that have to be taken into account: manufac-
ture, properties, performance, certification and cost. This
article addresses these considerations for the two types of
hollow sections.
A New Design for Steel Bridge Decks Using Laser

Fabrication
S.R. Bright and J.W. Smith
Steel orthotropic decks provide a lightweight form of con-
struction, essential for weight-critical structures. However,
their cost and poor record of fatigue durability has discour-
aged their use for mainstream construction. As a result, steel
decks are generally considered an option of last resort, only
used where the minimization of self-weight is essential, such
as long-span and moveable bridges. An innovation is pro-
posed to overcome these problems and transform the design
of steel decks. The innovation is based upon the use of laser
welding to produce an enclosed “sandwich panel” profile.
The sandwich profile overcomes many of the constraints to
structural performance associated with the use of conven-
tional orthotropic steel decks.
From the December 2007 issue of Journal of Protective

Coatings and Linings, pp. 12–14:
Intumescents: Swell Coatings for Fire Protection

This article contains a brief survey of recent literature on the
subject of intumescent coatings.

ERRATA
Geometric Formulas for Gusset Plate Design

Paper by JANICE J. CHAMBERS and TONY C. BARTLEY
(3rd Quarter, 2007)
Revise the author affiliation for Tony C. Bartley on page Revise the equation in the lowest box of Fig. 9 (cont’d) on
255 as follows: page 262 as follows:
a − g sin θ B sin θcc

(H − k) − L
Tony C. Bartley is a graduate student, department of civil
L3 = +
& environmental engineering, University of Utah, Salt Lake
City, UT.
cos θ B (
cos θ B − θcc ) Le
Revise footnote a in Table 2 on pages 264 through 267,

Revise Equation 26 on page 261 as follows: replacing “radians” with “degrees,” as follows:
a − g sin θ B sin θcc
a
All units are in inches (in.) and degrees; 1 in. = 25.4 mm.
L3 = + ( )
H − k − LLe (26)
cos θ B (
cos θ B − θcc ) In the “Input” column of Ex. 4 in Table 2 on page 265,
replace “θcc = –9.0” with “θcc = 9.0”.
In the “Input” column of Ex. 12 in Table 2 on page 267,
replace “a = 14.5” with “a = 19.5”.
Limit State Response of Composite Columns

andBeam-Columns
Part 1: Formulation of Design Provisions for the 2005
AISC Speciﬁcation
Paper by ROBERTO T. LEON, DONG KEON KIM and JEROME F. HAJJAR
(4th Quarter, 2007)
Revise Equation (AISC 2005 I2-5) on page 346 as follows:
Pe = π2 (EIeff)/(KL)2 (AISC 2005-I2-5)
096_errata_2008_1Q.indd 96 3/27/08 12:28:58 PM

Guide for Authors
SCOPE: The ENGINEERING JOURNAL is dedicated to the improvement and
advancement of steel construction. Its pages are open to all who wish to
report on new developments or techniques in steel design, research, the
design and/or construction of new projects, steel fabrication methods, or
new products of significance to the uses of steel in construction. Only
original papers should be submitted.
GENERAL: Papers intended for publication must be submitted by mail to the
Editor, Cynthia J. Duncan, ENGINEERING JOURNAL, AMERICAN
INSTITUTE OF STEEL CONSTRUCTION, One East Wacker Drive,
Suite 700, Chicago, IL, 60601-1802.
The articles published in the Engineering Journal undergo peer review
before publication for (1) originality of contribution; (2) technical value to
the steel construction community; (3) proper credit to others working in
the same area; (4) prior publication of the material; and (5) justification of
the conclusion based on the report.
All papers within the scope outlined above will be reviewed
by engineers selected from among AISC, industry, design firms, and
universities. The standard review process includes outside review by an
average of three reviewers, who are experts in their respective technical
area, and volunteers in the program. The maximum number of papers
sent to a single reviewer is three per year, with a frequency of not more
than one per quarter. Papers not accepted will be returned to the author.
Published papers become the property of the American Institute of Steel
Construction, Inc. and are protected by appropriate copyrights. No proofs
will be sent to authors. Each author receives three copies of the issue in
which his contribution appears.
MANUSCRIPT PREPARATION:
Manuscripts must be provided on PC-formatted media, such as a CD-ROM,
in Microsoft Word format. A laser-quality proof must accompany your
submittal. Fonts and spacing must be suitable for easy reading and
reproduction (for the peer-review process). Do not embed photographs,
diagrams, illustrations, charts or graphs within the electronic manuscript
files. Only equations may be embedded within the flow of the text. Specific
requirements for electronic graphics are outlined below. Engineering
Journal reserves the right not to publish a submittal if suitable graphics
cannot be provided by the author.
Title and By-Line: Exact name, title and affiliation of the author or
authors are required to appear on the first page of the manuscript.
Body Text: Please restrict font usage to Times, Helvetica, Times New
Roman, Arial, and Symbol (for Greek and mathematical characters.)
Headings: All headings should be typed flush left, using upper and
lower case, with two line spaces above.
Tables: Each table should appear on its own page. Footnotes to tables
should appear below the table, identified by superscripted lower case let-
ters (a, b, c, etc.).
Equations: Whenever possible, equations should be set using Microsoft
Equation Editor or MathType (www.mathtype.com). Please set equations
using Times New Roman, Times, and/or Symbol fonts.
Captions: Captions should be typed, double-spaced, and located at the
end of the electronic manuscript. All photos and graphics must be clearly
marked to indicate their corresponding caption.
References: Should be noted clearly in the text and listed, double-
spaced, on a separate page in the following sample format.
In text: (Doe, 1992)
In Reference List:
Doe, J.H. (1992), "Structural Steel," Engineering Journal, AISC,
Vol. 100, No. 1, 1st Quarter, pp. 2–10.
Footnotes: Footnotes should be noted clearly in the text with a super-
script asterisk, and should appear at the bottom of the text page, in the
following style:
*For a detailed discussion, see...
Graphics (other than photographs): Provide a clear 81/2 in. × 11
in. laser-quality proof of each graphic element. Graphics should repro-
duce cleanly in black-and-white format. Graphics may be submitted and
reproduced in color at the Editor's discretion. Please restrict font usage
to Helvetica, Arial or Symbol, in sizes suitable for at least 50%
reduction (12 pt. minimum). Line weights must be suitable for 50%
reduction. When possible, provide each graphic element in a separate
electronic file. TIF or EPS file formats are preferred, with a minimum
resolution of 300 dots per inch.
Photographs: Provide either original photographs or high-quality
electronic files and laser-quality proofs. Electronic photographs may
be submitted as grayscale TIF or JPG images, one photograph per file.
Photographs may be submitted and reproduced in color at the Editor's
discretion. Minimum image resolution is 300 dots per inch. Photographs
should be a minimum of 4 in. wide, so the minimum image width is 1200
pixels. Detailed photographs may require resolutions up to 1200 dots per
inch. Photos embedded in word-processing documents are not acceptable.
1Q2008EJ_cover.indd 3 4/7/2008 9:20:50 AM

2008 Volume 45, No. 1

Uploaded by

Copyright:

Available Formats

2008 Volume 45, No. 1

Uploaded by

Document Information

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

2008 Volume 45, No. 1

Uploaded by

Copyright:

Available Formats

Engineering

first Quarter 2008 Volume 45, No. 1

1 Message from the editor

1Q2008EJ_cover.indd 1 4/7/2008 9:20:49 AM

Dedicated to the development and improvement of steel construction,

1Q2008EJ_cover.indd 2 4/7/2008 9:20:49 AM

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 1

001-004_editor_acknow_2008_1Q.indd 1 3/27/08 12:16:34 PM

Farid Alfawakhiri Maria Garlock Theodor Krauthammer C. Mark Saunders

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 3

001-004_editor_acknow_2008_1Q.indd 3 3/27/08 12:16:35 PM

S hortly after block shear was first identified as a possible

tions to account for it were incorporated into allowable stress where

Howard I. Epstein is professor, department of civil and

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 5

005-012_Epstein_Aleksiewicz_2008_1Q.indd 5 3/27/08 12:17:45 PM

6 / ENGINEERING JOURNAL / FIRST QUARTER / 2008

005-012_Epstein_Aleksiewicz_2008_1Q.indd 6 3/27/08 12:17:45 PM

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 7

005-012_Epstein_Aleksiewicz_2008_1Q.indd 7 3/27/08 12:17:45 PM

Fig. 2. Block shear failure path for tee sections.

8 / ENGINEERING JOURNAL / FIRST QUARTER / 2008

005-012_Epstein_Aleksiewicz_2008_1Q.indd 8 3/27/08 12:17:45 PM

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 9

005-012_Epstein_Aleksiewicz_2008_1Q.indd 9 3/27/08 12:17:46 PM

Bolt Fy Fu Ptest 1999 LRFD 2005 LRFD

10 / ENGINEERING JOURNAL / FIRST QUARTER / 2008

005-012_Epstein_Aleksiewicz_2008_1Q.indd 10 3/27/08 12:17:47 PM

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 11

005-012_Epstein_Aleksiewicz_2008_1Q.indd 11 3/27/08 12:17:47 PM

12 / ENGINEERING JOURNAL / FIRST QUARTER / 2008

005-012_Epstein_Aleksiewicz_2008_1Q.indd 12 3/27/08 12:17:48 PM

connections. One criticism of the method is that it sometimes

Larry S. Muir is president of Cives Engineering Corporation

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 13

013-020_Muir_2008_1Q.indd 13 3/27/08 12:18:27 PM

Fig. 1. KISS method. Fig. 2. Parallel force method.

14 / ENGINEERING JOURNAL / FIRST QUARTER / 2008

013-020_Muir_2008_1Q.indd 14 3/27/08 12:18:27 PM

Fig. 3. Truss analogy method. Fig. 4. Beam free body diagram.

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 15

013-020_Muir_2008_1Q.indd 15 3/27/08 12:18:28 PM

From this cos( θ )ec

The point of intersection of the column-to-gusset force and

16 / ENGINEERING JOURNAL / FIRST QUARTER / 2008

013-020_Muir_2008_1Q.indd 16 3/27/08 12:18:29 PM

Column Moment Vb α − H b eb = 50.3(14.375) − 60.2(12) ≈ 0 kip-in.

A moment gradient will exist in the column whether using

Since the choice of column section will usually be gov-

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 17

013-020_Muir_2008_1Q.indd 17 3/27/08 12:18:30 PM

18 / ENGINEERING JOURNAL / FIRST QUARTER / 2008

013-020_Muir_2008_1Q.indd 18 3/27/08 12:18:31 PM

CONCLUSIONS α = distance from the face of the column flange or web

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 19

013-020_Muir_2008_1Q.indd 19 3/27/08 12:18:31 PM

Limit State Response of Composite Columns

T he strength of composite beam-columns, including steel

ENGINEERING JOURNAL / FIRST QUARTER / 2008 / 21

021-046_Leon_Hajjar_2008_1Q.indd 21 4/7/2008 9:51:22 AM