Classificação Dos Solos

SOILS and ROCKS An International Journal of Geotechnical and Geoenvironmental Engineering
ISSN 1980-9743
Soils and Rocks

100 100
95 95
75 75
Volume 38, N. 3, September-December 2015
25
Table of Contents 25
5 5
0 PACHECO SILVA LECTURE 0
New Developments in the Control and Prediction of the Movements

Induced by Deep Excavations in Soft Soils
M. Matos Fernandes 191
ARTICLES
Use of Sugarcane Bagasse as Carbon Substrate in Permeable Reactive Barriers:

Soils and Rocks
Laboratory Batch Tests and Mathematical Modeling An International Journal of Geotechnical
R.C. Mattos, P.S. Hemsi, E.Y. Kawachi, F.T. Silva 219
and Geoenvironmental Engineering
Apparent Pressures on Multi-Propped Retaining Walls in Soils Under Drained Conditions
with Shallow Water Table
Volume 38, N. 3
L. Andrade Viana, N.M. da Costa Guerra 231
On the Interpretation of the Bidirectional Static Load Test

F. Massad 249
TECHNICAL NOTE
Revisiting Classical Methods to Identify Collapsible Soils

O.M. Vilar, R.A. Rodrigues 265
B
100
95
MS 100
95
75 75
25
Volume 38, N. 3 25
September-December 2015
2015
5 5
0 0
Soils and Rocks is an International Journal of Geotechnical and
Geoenvironmental Engineering published by
ABMS - Brazilian Association for Soil Mechanics and Geotechnical Engineering

Av. Queiroz Filho, 1700 – Torre A Sky, Sala 106 – Vila Hamburguesa
05319-000, São Paulo, SP
Brazil
SPG – Portuguese Geotechnical Society

LNEC, Avenida do Brasil, 101
1700-066 Lisboa
Portugal
Issue Date: December 2015

Issue: 400 copies and 800 online-distributed copies
Manuscript Submission: For review criteria and manuscript submission information, see Instructions for Authors at the end.
Disclaimer: The opinions and statements published in this journal are solely the author’s and not necessarily reflect the views or opinions
of the publishers (ABMS and SPG). The publishers do not accept any liability arising in any way from the opinions, statements and use of
the information presented.
Copyright: Authors and ABMS-Brazilian Society for Soil Mechanics and Geotechnical Engineering
SOILS and ROCKS
An International Journal of Geotechnical and Geoenvironmental Engineering
Editor Waldemar Coelho Hachich - University of São Paulo, Brazil
Co-editor Manuel Matos Fernandes - University of Porto, Portugal
Executive Board
Alberto Sayão Ian Schumann Martins
Pontifical Catholic University of Rio de Janeiro, Brazil Federal University of Rio de Janeiro, Brazil
Jorge Almeida e Sousa João Maranha
University of Coimbra, Portugal LNEC, Portugal
Associate Editors
H. Einstein E. Maranha das Neves Harry G. Poulos
MIT, USA Lisbon Technical University, Portugal University of Sidney, Australia
John A. Hudson Nielen van der Merve Niek Rengers
Imperial College, UK University of Pretoria, South Africa ITC, The Netherlands
Kenji Ishihara Paul Marinos Fumio Tatsuoka
University of Tokyo, Japan NTUA, Greece Tokyo University of Science, Japan
Michele Jamiolkowski James K. Mitchell Luiz González de Vallejo
Studio Geotecnico Italiano, Italy Virginia Tech., USA UCM, Spain
Willy A. Lacerda Lars Persson Roger Frank
COPPE/UFRJ, Brazil SGU, Sweden ENPC-Cermes, France
Editorial Board Members

Roberto F. Azevedo Orencio Monje Vilar R. Kerry Rowe
Federal University of Viçosa, Brazil University of São Paulo at São Carlos, Brazil Queen’s University, Canada
Milton Kanji Maria Eugenia Boscov R. Jonathan Fannin
University of São Paulo, Brazil University of São Paulo, Brazil University of British Columbia, Canada
Antonio M.S. Oliveira Eda Freitas de Quadros Laura Caldeira
University of Guarulhos, Brazil BGTECH, Brazil LNEC, Portugal
Omar Y. Bitar Vinod Garga António S. Cardoso
IPT, Brazil University of Ottawa, Canada University of Porto, Portugal
Lázaro V. Zuquette Richard J. Bathurst José D. Rodrigues
University of São Paulo, Brazil Royal Military College of Canada Consultant, Portugal
Fabio Taioli Robert Mair António G. Coelho
University of São Paulo, Brazil University of Cambridge, UK Consultant, Portugal
Tarcisio Celestino Serge Leroueil Luís R. Sousa
University of São Paulo-SC, Brazil University of Laval, Canada University of Porto, Portugal
Roberto Q. Coutinho Mario Manassero Rui M. Correia
Federal Univ. of Pernambuco, Brazil Politécnico di Torino, Italy LNEC, Portugal
Nilo C. Consoli Luis Valenzuela João Marcelino
Federal Univ. Rio Grande do Sul, Brazil Consultant, Chile LNEC, Portugal
Sandro S. Sandroni Jorge G. Zornberg António C. Mineiro
Consultant, Brazil University of Texas/Austin, USA University of Lisbon, Portugal
Sérgio A.B. Fontoura Andrew Whittle António P. Cunha
Pontifical Catholic University, Brazil MIT, USA LNEC, Portugal New
Ennio M. Palmeira Pierre Bérest António G. Correia
University of Brasilia, Brazil LCPC, France University of Minho, Portugal
Luciano Décourt Peter Kaiser Carlos D. Gama
Consultant, Brazil Laurentian University, Canada Lisbon Technical University, Portugal
Faiçal Massad He Manchao José V. Lemos
University of São Paulo, Brazil CUMT, China LNEC, Portugal
Marcus Pacheco Teruo Nakai Nuno Grossmann
University of the State of Rio de Janeiro, Brazil Nagoya Inst. Technology, Japan LNEC, Portugal
Paulo Maia Claudio Olalla Luís L. Lemos
University of Northern Rio de Janeiro, Brazil CEDEX, Spain University of Coimbra, Portugal
Renato Cunha Frederick Baynes Ricardo Oliveira
University of Brasília, Brazil Baynes Geologic Ltd., Australia COBA, Portugal
“Ad hoc” Reviewers (2014-2015)
Ana Cristina Sieira Edmundo Rogério Esquível Luís Valenzuela
UERJ, Brazil USP-S. Carlos, Brazil Arcadis, Chile
André Luís Brasil Cavalcante Eurípedes Vargas Marc Vinches
UnB, Brazil PUC-RJ, Brazil Ecole des Mines d'Alès, France
Antonio Airton Bortolucci Fernando Antonio Medeiros Marinho Marcia Maria dos Anjos Mascarenhas
USP-S. Carlos, Brazil USP-Poli, Brazil UFG, Brazil
Arsenio Negro Jr. Fernando Olavo Franciss Maria Cláudia Barbosa
Bureau, Brazil Progeo, Brazil UFRJ, Brazil
Bernadete Ragoni Danziger Francisco Chagas da Silva Filho Maurício Ehrlich
UERJ, Brazil UFC, Brazil UFRJ, Brazil
Carlos Medeiros Silva Francisco de Rezende Lopes Nelson Aoki
Embre, Brazil UFRJ, Brazil USP-S.Carlos, Brazil
Claudio Fernando Mahler Frederico Falconi Paulo Scarano Hemsi
UFRJ, Brazil ZF, Brazil ITA, Brazil
Claudio Michael Wolle Isabel Lopes Renato Pinto da Cunha
USP-Poli, Brazil IST, Portugal UnB, Brazil
David Martins Geraldo Taborda João Manuel Marcelino Mateus da Silva Sussumu Niyama
Imperial College, UK LNEC, Portugal Tecnum, Brazil
Denise Gerscovich José Camapum de Carvalho Tácio Campos
UERJ, Brazil UnB, Brazil PUC-RJ, Brazil
Eda Quadros José Renato Baptista de Lima Tiago Miranda
BGTECH, Brazil USP-Poli Univ. do Minho, Portugal
Edgar Odebrecht Kátia Bicalho Waldyr Lopes de Oliveira Filho
Geoforma, Brazil UFES, Brazil UFOP, Brazil
Edgard Poiate Leonardo Becker
Petrobrás, Brazil UFRJ, Brazil
Soils and Rocks publishes papers in English in the broad fields of Geotechnical Engineering, Engineering Geology and Geo-
environmental Engineering. The Journal is published in April, August and December. Subscription price is US$ 90.00 per year. The jour-
nal, with the name “Solos e Rochas”, was first published in 1978 by the Graduate School of Engineering, Federal University of Rio de
Janeiro (COPPE-UFRJ). In 1980 it became the official magazine of the Brazilian Association for Soil Mechanics and Geotechnical Engi-
neering (ABMS), acquiring the national character that had been the intention of its founders. In 1986 it also became the official Journal of
the Brazilian Association for Engineering Geology and the Environment (ABGE) and in 1999 became the Latin American Geotechnical
Journal, following the support of Latin-American representatives gathered for the Pan-American Conference of Guadalajara (1996). In
2007 the journal acquired the status of an international journal under the name of Soils and Rocks, published by the Brazilian Association
for Soil Mechanics and Geotechnical Engineering (ABMS), Brazilian Association for Engineering Geology and the Environment
(ABGE) and Portuguese Geotechnical Society (SPG). In 2010, ABGE decided to publish its own journal and left the partnership.
Soils and Rocks
1978, 1 (1, 2)
1979, 1 (3), 2 (1,2)
1980-1983, 3-6 (1, 2, 3)
1984, 7 (single number)
1985-1987, 8-10 (1, 2, 3)
1988-1990, 11-13 (single number)
1991-1992, 14-15 (1, 2)
1993, 16 (1, 2, 3, 4)
1994-2010, 17-33 (1, 2, 3)
2011, 34 (1, 2, 3, 4)
2012-2014, 35-37 (1, 2, 3)
2015, 38 (1, 2, 3)
ISSN 1980-9743 CDU 624.131.1

ISSN 1980-9743
SOILS and ROCKS

Publication of
SPG - Portuguese Geotechnical Society
Table of Contents
PACHECO SILVA LECTURE

M.M. Fernandes 191
ARTICLES
Laboratory Batch Tests and Mathematical Modeling


F. Massad 249
TECHNICAL NOTE
Pacheco Silva Lecture
The Pacheco Silva Lecture is delivered each two years by an important geo-
technical professional from Brazil or abroad to honour the memory of the distin-
guished Brazilian geotechnical engineer Francisco Pacheco Silva (1918-1974).
Pacheco Silva was a researcher of the Technological Research Institute (IPT/SP)
for 33 years and consultant in geotechnical engineering in several occasions. In
1947 he obtained his MSc. degree from Harvard University, where he was a stu-
dent of Karl Terzaghi and Arthur Casagrande. He worked in practically all fields
of geotechnical engineering, but his main research interests were on soil behav-
iour, laboratory testing and geotechnical instrumentation. He was also one of the
founders and past presidents of the Brazilian Association for Soil Mechanics and
Geotechnical Engineering - ABMS.
Prof. Manuel de Matos Fernandes

The 2014 Pacheco Silva Lecturer is Manuel de Matos Fernandes, 61 years, Full
Professor at the Faculty of Engineering of University of Porto (FEUP), Portugal.
He got his degree in Civil Engineering in 1976 at FEUP. The PhD thesis (1984), on
flexible earth-retaining structures, was developed at LNEC, Lisbon. He was the
main responsible for the development at FEUP of a team devoted to Geotechnics.
His areas of interest have been focused on numerical modelling of geotechnical
structures, deep excavations, Eurocodes and teaching Soil Mechanics (he is author
of two text books). In the last 25 years he has had activity of design and consul-
tancy, particularly in bridge foundations and urban excavations.
Soils and Rocks

v. 38, n. 3
New Developments in the Control and Prediction of the
Movements Induced by Deep Excavations in Soft Soils
M. Matos Fernandes
Abstract. The paper corresponds to the Pacheco Silva Lecture delivered by the author in September 2014. In the first part
of the text, a general perspective of the movements induced by deep excavations in soft soils is outlined. The general report
presented by Ralph B. Peck to the 1969 ISSMFE México conference is taken as reference and the progress achieved since
then in the control of the movements is emphasized. It is pointed out that such progress permits that in very demanding
urban scenarios involving soft soils one should work towards a performance, with regard to induced movements, similar to
(or even better than) that is achieved in comparable excavations in stiff soils. Starting from this general framework, some
issues relevant to an effective control and to a reliable prediction of the movements are treated: the pre-stress of the strut
system, the methodologies for providing support to the retaining wall before the removal of the soil, the surface settlements
induced by the consolidation of the soil in the long-term and the influence of the non-linear behaviour of the reinforced
concrete wall. Results of research on these topics are presented and discussed and general conclusions are extracted.
Keywords: deep excavations, soft soils, induced movements, soil improvement, consolidation, nonlinear reinforced concrete.
1. Introduction clayey soils, with the wall tip embedded in a firm layer un-
derlying the soft deposit. This sequence is selected from a
Deep excavations in thick deposits of soft soils have
finite element simulation but it could correspond to real ob-
been one of the major challenges that geotechnical engi-
served results. In the finite element analysis the struts were
neers have faced in the last decades.
pre-stressed.
Literature contains a large number of references to
such works in cities like Oslo (Karlsrud, 1983), Lisbon It is possible to observe that with the progress of the
(Matos Fernandes, 1985, 2007), Chicago (Finno et al., construction process the wall face at the excavated side ex-
1989), San Francisco (Clough & Reed, 1984; Koutsoftas et hibits an increasing convexity, with the point of maximum
al., 2000), Singapore (Wong et al., 1997), Shanghai (Wang displacement at increasing depth. Simultaneously, the
et al., 2005; Lui et al., 2011; Ng et al., 2012) and Taipei (Ou ground surface becomes concave, with the point of maxi-
et al., 2006; Hwang et al., 2007). mum settlement being situated at increasing distance in re-
lation to the retaining wall. This concavity arises from the
A major issue when dealing with excavations in thick
upward tangential stresses applied by the wall to the sup-
deposits of soft clayey soils is the magnitude of the move-
ported ground through its back face, therefore ensuring
ments induced in the vicinity. This subject has known, for
smaller settlements in the surface region adjacent to the ex-
the first time, a consistent approach in the well-known Gen-
cavation (Matos Fernandes, 2004).
eral Report to the México ISSMFE conference by Peck
(1969). Since then an auspicious progress in the under- Figure 2a is extracted from the sequence commented
standing and the control of the movements was achieved as above, distinguishing the lateral displacement of a given
a result of the introduction of new construction technolo- point P of the wall face when the excavation platform
gies, new structural systems and new methods of analysis. reaches the elevation of that point. So, this displacement
The fact that such excavations are becoming deeper has occurred when P was underneath the base of the exca-
and are being done under more daring and demanding con- vation. In Fig. 2b, the displacement of P at the completion
ditions is the motive to reanalyse this topic in the light of the of the excavation is shown. It can be observed that this dis-
recent results published in the literature and of the new pos- placement practically coincides with the one shown in Fig.
sibilities provided by the techniques currently available. 2a. So, it can be concluded that the displacement of P when
this point was above the current excavation platform was
2. Moviments Induced by Excavations in Soft practically null.
Clayey Soils. General Perspective This result illustrates an important characteristic of
this type of excavations with controlled construction
2.1. Introduction. Some previous matters
(avoiding overexcavation and with pre-stressed struts
Figure 1 shows the typical sequence of movements of having high axial effective stiffness). Basically, the ob-
a strutted flexible retaining wall in a thick deposit of soft served lateral displacements at each point of the wall
Manuel de Matos Fernandes, Full Professor, Faculty of Engineering, University of Porto, Portugal. e-mail: [email protected].
Invited Lecture, no discussion.
Soils and Rocks, São Paulo, 38(3): 191-215, September-December, 2015. 191
Matos Fernandes
Figure 1 - Typical sequence of the movements of a strutted flexible retaining wall in a thick deposit of soft clayey soils.
Figure 2 - Displacements underneath the current excavation base (a) and at the final excavation stage (b) of a generic point of the wall
from the sequence illustrated in Fig. 1.
occur under the excavation base, that is, before an ade- of the ones of the ground surface during the excavation
quate structural support could be given to the wall at that phase. Assuming undrained soil behaviour during the exca-
point. This fact is the key of the major difficulty in con- vation, that is null volumetric strains, the two areas marked
trolling the movements induced by deep excavations in in Fig. 3, corresponding to the integral of the horizontal
thick deposits of soft soils! wall displacements and to the integral of surface settle-
Obviously, the control of the lateral wall displacements at the completion of the excavation will be practi-
ments is relevant because such displacements are the cause cally equal.
192 Soils and Rocks, São Paulo, 38(3): 191-215, September-December, 2015.
New Developments in the Control and Prediction of the Movements Induced by Deep Excavations in Soft Soils
Before closing this section, it should be noticed that in

most of the published case histories it is very difficult to
separate the surface settlements into the three components
referred above. In addition, the cases including observation
of a time period after the conclusion of the excavation are
relatively scarce. This turns difficult to interpret and to treat
those observed results in a general way.
2.2. General report of Peck (1969) - The founder docu-

ment of the theme
Figure 3 - Integrals of the horizontal wall displacements and of The first published work presenting a scientific analy-
surface settlements at the end of the excavation from the simula- sis of the movements induced by deep excavations, and par-
tion shown in the preceding figures. ticularly by excavations in thick deposits of soft soils, is the
general report presented by Ralph B. Peck to the VIIth Con-
However, it must be taken into account that the sur- ference of the ISSMFE, in México City (Peck, 1969). The
face movements are not restricted to the ones discussed report includes the famous settlement chart illustrated in
above. As shown by Fig. 4, before the excavation, the wall Fig. 5.
installation (and other preparatory constructive operations, That chart is organized according to the type of soil.
such as demolition of ancient foundations) will typically In thick deposits of soft soils, the expected settlements may
cause non negligible surface settlements (Clough & achieve very high values: over 1% to 2% of the excavation
O’Rourke, 1990; Poh et al., 2001; Finno et al., 2002). In ad- depth. This magnitude of the expected settlements mainly
dition, the dissipation of excess pore pressures (mostly) af- arose from the dramatic limitations of the type of retaining
ter the excavation may originate further settlements by structures and construction employed until the sixties (and,
consolidation of the soil, which may achieve a considerable to some extent, from the limitations of the methods of anal-
magnitude (Finno et al., 1989; Alves Costa et al., 2007; ysis, as well). This mainly consisted of very flexible sheet
Matos Fernandes, 2010; Matos Fernandes et al., 2012; pile or soldier-pile walls with timber lagging associated
Borges & Pinto, 2013; Borges & Guerra, 2014). with cross-bracing (generally non pre-stressed).
In the present state of our understanding about the The ability of such retaining structures to control in-
phenomena involved in such works, and with the available duced movements, particularly in soft soils, was in fact
constructive solutions, the component of surface move- very modest. Then the statement of the author: “The most
ments concomitant with the excavation (among the three important variable that determines the amount of move-
described above) seems easier to be minimized and, as ment is not the stiffness of the exterior wall or the vertical
well, it is the one that our analytical models (particularly, spacing of the bracing but the characteristics of the sur-
the finite element models) more faithfully capture. This rounding soils. (...)”.
explains the relevance of analysing and discussing all the It is very stimulating to undertake a careful lecture of
issues that may contribute to minimize that component. the report today and to check the solution suggested by
That challenge will be a major goal of this work, although Peck to control the movements induced by the excavations
some attention will be focused on the third component, as in soft clayey soils. This sequence of thoughts is particu-
well. larly lucid:
Figure 4 - Movements (a) before, (b) during and (c) after of the excavation.
Matos Fernandes
did not affect the comfort in the rest of the building (Ran-
dall, 1999). The method adopted in the excavation, called
the trench method by Peck, is depicted in Fig. 7 (drawings
by Bárbara Rangel for the author of this paper), and is de-
scribed as follows by Peck (1969):
According to this procedure, a trench was
excavated by hand around the periphery of a
building at the proposed locations of the perma-
nent basement walls. The width of the trench was
as small as practicable. The sides were supported
by timber planks and large number of horizontal
trench braces or jacks extending from one side to
the other. The completely sheeted trenches were
carried to the full depth of the outside walls of the
structure. Reinforcement was then set in the tren-
ches and the concrete placed directly against the
sheeting which served as the form.
Simultaneously with the wall construction,
cross-trenches were excavated, generally along
the column lines, and were similarly timbered and
braced to the same depth as the external walls.
Concrete struts, later forming part of the lowest
basement floor, were then cast in the bottoms of
the trenches. The building columns were estab-
lished at the intersections of these struts (...). In
this manner, a complete system of cross-lot brac-
ing was established while most of the soil within
the future basement area was still in place. As the
final step, the soil was excavated (...). Hence, the
settlements were minimized.
Figure 5 - Summary of settlements adjacent to open cuts in vari-
ous soils, as function of distance from edge of excavation (Peck, This ingenious solution may be, to some extent, con-
1969). sidered pioneer in relation to the diaphragm wall technol-
ogy. On the other hand, it is fully understandable that its use
had been relatively rare when urban excavations became
(...). While this excavation is going on, the much more common by the middle of the century, due to
walls move. Whatever portion of the movement the high cost and execution time.
takes place below the bottom of the excavation
It is perhaps curious to observe that the best solution
cannot be prevented irrespective of the capacity of
pointed out by Peck to control the induced movements is a
the supports or the degree to which bracing is
procedure developed more than six decades before! And,
pre-stressed.
that the method for predicting the magnitude of the move-
In principle, the movements could be pre-
ments is a chart organized by type of soil, so independent
vented if the entire supporting structure (...) and
on the retaining structure. The same can be said about the
even the base slab for the completed structure
well-known diagrams of apparent earth pressures, dis-
could be constructed in their final positions before
cussed in the same general report, dependent just on the
the removal of the enclosed soil. (...). Such an ide-
type of soil, as well.
alized procedure can exist only in imagination. It
may be approached in practice, however (...). This permits to envisage the severe technological and
The approach suggested by the author is the so-called analytical limitations at the time of this brilliant document.
trench method, developed in Chicago in the very beginning Fortunately, these limitations were about to be overcome!
of the XXth century - period of very rapid and profound
2.3. The progress in the control of induced movements
transformation of the city - for the construction of the three
in the last decades
basements of the new Chicago Tribune Building in the
Madison Avenue (Fig. 6). After the late sixties and in the following decades
The three basements were required for housing the some remarkable new technologies were progressively in-
presses of the journal, in order that the vibration from them troduced, such as reinforced concrete diaphragm walls and
Figure 6 - Tribune building, Madison Avenue, Chicago: a) photo in 1910; b) scheme of the Trench Method applied in the excavation for
the three basements (drawing by Bárbara Rangel).
Figure 7 - Two stages of the Trench Method used in the excavation for the three basements of Tribune building, Chicago, 1901 (draw-
ings by Bárbara Rangel).
continuous pile walls, capable of better controlling induced spread use of finite element analyses - whose first applica-
movements. tions to deep excavations were published in the seventies
These solutions and the understanding of the parame- (Bjerrum et al., 1972; Egger, 1972; Palmer & Kenney,
ters affecting induced movements brought by the wide- 1972; Clough & Tsui, 1974; Clough & Mana, 1976) - al-
Matos Fernandes
lowed the execution of excavations in soft ground with

much better results in comparison with Peck’s chart.
This is well evidenced by the results presented two
decades later by Clough and O’Rourke (1990), represented
in Fig. 8. Additionally, the large scatter of the results col-
lected proves that the specific parameters of the retaining
structure and of the construction are extremely relevant
with regard to the induced movements.
Clough & O’Rourke (1990) recognize this point, stat-
ing that: “Although the zones are helpful in delineating
broad trends of performance, it is evident that the scatter in
settlement magnitude places limitations on this type of em-
pirical approach for making predictions about movement”.
Figure 9 is extracted from a database collected by
Moormann (2004) in which only a few case histories prior
to 1980 were introduced. It illustrates the maximum lateral
wall displacement, expressed as a percent of the excavation
depth, with the factor of safety against basal heave. This
factor is calculated by not considering the wall embedded
Figure 9 - Maximum lateral wall displacement versus basal heave
length, according to the Terzaghi definition. It is interesting
safety factor, for excavations in soft to medium clay (data col-
to observe that much better results appear in comparison lected by Moormann, 2004).
with the trend identified by Mana & Clough (1981). Never-
theless, the data is highly scattered: for the same value of
the safety factor, the displacements may vary more than ten The author of this paper pointed out that if such re-
times. sults are relatively common even nowadays this is possibly
The results presented above confirm that the behav- related to the old idea - which is still widely accepted in
iour of excavations in thick deposits of soft clay is very practice - that the level of achievable performance with re-
complex, depending on a large set of features and parame- gard to movement control mainly depends on the ground
ters concerning the ground, the structure and the construc- conditions.
tion. Due to this idea, the Profession has been, in fact, more
In spite of the enormous progress in the control of tolerant with excavation induced movements in soft
movements revealed by Figs. 8 and 9, it should be recog- ground. However, ancient structures and services in soft
nized that the actual displacement values shown by these ground may probably be less tolerant to further movements
figures are in many cases capable of inducing serious dam- than similar constructions over stiff soils (Matos Fernan-
age in the vicinity of the excavation. Many underground des, 2007).
works in soft clay recently carried out in our cities (some in-
volving excavation depth over 20 m), would be intolerable 2.4. Eight golden rules for movement control
if the induced movements were in accordance with many of This reflection recommends, in fact, a complete in-
the results shown above. version of the perspective commonly adopted when dealing
with this critical matter. It should be pointed out that the ex-
perience of a number of works has shown that it is possible
to achieve in soft soils levels of performance typical of
more favourable geotechnical conditions, if appropriate
structural solutions and construction sequences are em-
ployed. Namely, it is possible nowadays to deal with sur-
face settlements one order of magnitude lower than the
values considered in Peck’s chart (Wang et al., 2010).
Searching for case studies in the literature with small
induced movements, while attempting to identify the fea-
tures related with the structure and the construction which
are more or less common to them, led to the formulation of
the eight golden rules summarized in Fig. 10 (Matos Fer-
Figure 8 - Summary of measured settlements adjacent to excava- nandes, 2007).
tions in soft to medium clay collected by Clough and O’Rourke It is interesting to observe that the first letters of these
(1990) over Peck’s chart. eight rules form the word RELIABLE. In the thesaurus dic-
Figure 10 - Golden rules for a reliable movement control induced by deep excavations in thick deposits of soft ground (Matos
Fernandes, 2007).
tionary (Collins, 1984) it is possible to find the following manding urban scenarios involving soft soils one should
synonymous for that word: safe, sound, stable and work towards better or equal performance, with regard to
PREDICTABLE! induced movements, than in a similar excavation in stiff
In fact, these rules may be further considered as con- soils. This requires efforts and progresses in various fields,
ditions for obtaining reliable predictions of the induced such as, new construction technologies, structural concep-
movements: a retaining structure conceived and designed tion and methods of analysis.
according to these rules is capable of maintaining the major
part of the ground mass far from yielding. Thus, its perfor- 3. The Pre-Stress of the Strut System
mance shall be easier to predict because it will be mainly This part will be devoted to a brief note on the pre-
dependent on structural and construction features, exhibit- stressing of the strut system, since the scope of this work
ing small sensitivity to those issues whose accurate charac- corresponds to excavations in thick deposits of soft clayey
terization is more difficult. In short, it will be a reliable soils. Traditionally, the pre-stress was looked as a mere ex-
system. pedient of adjusting the strut to the wall, therefore permit-
In such cases, agreement between prediction and per- ting that the effective axial stiffness could reach a high
formance should be mainly credited to appropriate concep- percentage of the theoretical axial stiffness (the one of the
tion of the structure and to competent construction, and not structural member materializing the strut), as expressed by
so much assigned to the sophistication of the design predic- Fig. 11.
tion analyses. In this context, the applied pre-stress was typically a
Geotechnical engineers very often state that the anal- small fraction of the design load of the strut. This was based
yses performed by our structural colleagues are more reli- on the following mental model: since the theoretical stiff-
able because they deal with materials whose mechanical ness is very high, if the effective stiffness is not far from the
properties are carefully controlled. This is just one part theoretical one, the wall displacements at that point will be
(probably the smaller) of the truth! The other is that the eradicated for the rest of the construction stages. This per-
safety factors and the design practice in Structural Engi- spective neglects the very positive effect that a strong
neering ensure that plastic yielding of structural elements is pre-stress may induce on the stress state of the ground mass,
typically incipient, so linear elastic behaviour is, in general, then, on the stiffness exhibited by the soil.
a good approach. If a degree of plastic yielding similar to Figure 12 shows results from a finite element simula-
that achieved in many geotechnical problems with soft soils tion of a multi-strutted excavation in soft soil, in terms of
were allowed in conventional structures, the reliability of the total stress path of a given point of the supported soil
structural analyses would become very questionable. mass. As shown in Fig. 12a, if struts are not pre-stressed the
In the following sections, four issues related with total stress path moves progressively and almost mono-
those golden rules will be discussed by reporting some re- tonically towards the failure envelop (the active state, in the
cent research results. The key point is that in some very de- context of classical Soil Mechanics). On the contrary, with
Matos Fernandes
Figure 11 - Axial load-displacement diagram of a strut and of its

interface with the wall and effect of the pre-stress on the effective
stiffness.
high pre-stressing applied on the struts, as shown by Figure 12 - Total stress paths in a generic point of a soft clay mass
Fig. 12b, the stress path corresponds to a cyclic loading: the supported by a multi-strutted wall: a) with no pre-stress of the strut
pre-stress applied on each level of supports compensates, in system; b) with pre-stress of the strut system.
a high degree, the decompression associated with the previ-
ous excavation stage. If the stress state of each point of the
ground is kept far from the failure envelop, that point will So, the pre-stress of the strut system is capable of: i)
respond with higher stiffness in the sequent excavation increasing effective axial strut stiffness; ii) recovering
stages. some displacements occurred in the preceding excavation
Figures 13a and 13b represent, for the same analyses, stages; iii) last but not least, maintaining the state of stress
the shear stress levels in the ground mass at the end of the of the soil far from failure, then, ensuring a stiffer response
excavation (the point whose stress path was mentioned of the soil in the subsequent excavation stages.
above is marked in black). It may be observed that in the It is therefore recommended to use high pre-stress to
analysis with pre-stressed struts the stress levels are consid- control the lateral wall displacements above and below the
erably smaller. current excavation base. It should be noticed that this is fea-
Figure 13 - Shear stress levels in the ground mass at the end of the excavation: a) with no pre-stress of the strut system; b) with pre-stress
of the strut system.
sible just with a pre-stress system well-conceived and exe- Evaluating the most appropriate distribution of an-
cuted and with a reinforced concrete wall with adequate chor pre-stressing forces in a tied-back diaphragm wall, the
bending stiffness in order to get a reliable control of the authors find that the triangular at-rest diagram is not partic-
forces effectively installed (Matos Fernandes, 2010). ularly effective (in comparison with trapezoidal diagrams)
in reducing movements (see Fig. 14). They add the follow-
4. Ground Improvement. General ing comment: “The results show that the old concept of re-
Perspective and Proposal storing the initial soil pressure through pre-stress loads
and eliminating movement does not work since wall move-
4.1. The “delay” in the installation of the retaining ments occur during each excavation segment before the
structure subsequent pre-stress load can “restore” the lost earth
pressure”.
The author begs the tolerance of the reader to remind By other words: due to the inelastic behaviour of
this reflection of Peck (1969) on the question of the move- soils, restoring the state of stress does not ensure the resto-
ments: “In principle, the movements could be prevented if ration of the state of strain!
the entire supporting structure (...) and even the base slab As it was already emphasized, a real turn point in the
for the completed structure could be constructed in their fi- field of deep urban excavations was the introduction of re-
nal positions before the removal of the enclosed soil”. inforced concrete diaphragm walls. With this technology, it
This question about the delay of the installation of the was possible to install the entire retaining wall - stiff, resis-
retention system in relation to the execution of the excava- tant, impermeable, beyond the excavation bottom, with mi-
tion as the source of the inevitability of the movements, nor or modest impact on the ground - before performing the
appears very clear, as well, in the first of the series of out- excavation. This fact was extremely relevant for the prog-
standing papers of Clough and co-workers with applica- ress in the control of movements, as commented above.
tions of finite element models to excavations (Clough & The understanding of the deleterious effects of the
Tsui, 1974). mentioned delay, soon encouraged attempts to provide sup-
Figure 14 - Parametric study on the effect of magnitude and distribution of anchor pre-stressing forces on soil and wall movements
(Clough & Tsui, 1974).
Matos Fernandes
port to the wall before the excavation is made, in comple-

ment with the previous installation of the concrete periph-
eral wall.
4.2. Historical overview of the solutions providing sup-
port to the wall before the execution of the excavation
The first solution, represented in Fig. 15, was applied
in the Oslo Metro at the beginning of the 1970 decade. It
consisted of cross diaphragm wall concrete panels, acting
as buttresses, under the bottom of the excavation (Eide et
al., 1972).
In this solution: i) the cross wall panels remain under-
neath the excavation bottom due to the difficulty involved
in their demolition; ii) difficulties arise due to imperfect
cleaning of the interface between the longitudinal walls and
the cross walls. This might explain why its application has
been very rare and restricted to Scandinavian countries
(Karlsrud & Andresen, 2007). Figure 16 - Cross section of the retaining structure of the excava-
tion for the Cais do Sodré Station of Lisbon Metro, with a jet grout
In the nineties, the application of various techniques
slab below the base of the excavation (Matos Fernandes, 2007).
of soil treatment became frequent as complement of con-
ventional retaining structure solutions, particularly in deep
excavations in soft soils (Karlsrud, 1997). One of the limitations of the treatment by jet grouting
In most cases the treatment is applied in order to form is that the control of the geometry of the columns is diffi-
a continuous “slab” of improved soil under the bottom of cult, particularly in this context because they are executed
the excavation. This may be done by jet-grouting (Wang et at great depth. Bearing in mind that soft soil must not re-
al., 2005; Matos Fernandes et al., 2007; Ng et al., 2012) or main between columns, these are executed by adopting dis-
by compaction grouting (Liu et al., 2011; Tan & Wei, tance between column axes smaller than the expected di-
2012). Figure 16 illustrates the case of Cais do Sodré Sta- ameter. This can lead to outward wall displacements even
tion of the Lisbon Metro, concluded in 1994. greater than the inward ones that are to be prevented with
The use of isolated jet-grout columns in square mesh the treatment (Wong & Poh, 2000).
plan distribution under the base of the final excavation can In this context, the application of the Deep Mixing
also be found, as reported by Ou et al. (1996) and Hsieh et Method (also known by Deep Soil Mixing) seems to pres-
al. (2003). ent some advantages, with regard to the control of the ge-
When the excavation is very deep and its base is close ometry (diameter) of the treated soil columns, as well as
to a firm layer, the slab of improved soil is placed above the with regard to the magnitude of the grouting pressures, far
final excavation platform, being demolished together with below jet-grouting ones. On the other hand, the strength of
the excavated soil (Brito & Matos Fernandes, 2006; Matos the treated material is not lower than the one provided by
Fernandes, 2007). Cases with two slabs of treated soil, jet-grouting (Pinto, 2014).
above and below the final excavation level, can also be O’Rourke & McGinn (2006) report the application of
found in the bibliography, as the one illustrated in Fig. 17 Deep Mixing Method forming vertical transversal elements
(Lui et al., 2005; Tan & Li, 2011). working as buttresses in the large project of the Boston
Central Artery, as shown by the example depicted in
Fig. 18.
A very peculiar solution was conceived and con-
structed in London, in the nineties, in a very deep excava-
tion (39 m) performed inside the old Westminster railway
Station (XIXth century), as part of the construction of the
new Jubilee Line of the London Metro (Glass & Stones,
1999; Mair & Harris, 2001). Even if the soil involved (Lon-
don clay) is not a soft soil, the movement restrains were ex-
tremely severe, due to the presence of the Big Ben tower in
the vicinity of the excavation.
As shown by Fig. 19, the toe of the 1.0 m thick dia-
Figure 15 - Cross diaphragm wall panels below the base of the ex- phragm walls was supported by cylindrical “tunnelled
cavation (Oslo Metro). struts” (1.8 m in diameter), built by hand-mining methods
Figure 17 - Jet grout slabs below and above the base of the excavation in a station of the Shanghai Metro (adapted from Tan & Li, 2011).
using pre-cast concrete segments, filled with in situ rein- reinvented with an interesting particularity: the cross walls
forced concrete. These were installed along the critical are constructed up to the surface, being filled by plain
zone (in plan), underneath the base of the excavation (about (unreinforced) concrete above the final excavation base
40 m in depth) and before the removal of the soil. The con- (Ou et al., 2006; Hsieh et al., 2008; Wu et al., 2013). In the
struction utilized, as access, vertical wells (diameter of pioneer example represented in Fig. 20, one can observe
3.0 m) situated on the axis of the excavation. Within each that the reinforcement philosophy is to some extent distinct
strut, a 2.44 m diameter jacking chamber was constructed, of the works previously reported, in which the length of the
housing five pairs of hydraulic flat jacks with a total capac- excavation is much greater than its width. In fact, the use of
ity of 7640 kN. This allowed the pre-stressing of the struts, cross walls subdivides the wide excavation (width and
ensuring a good connection to the peripheral wall. One can length of the same order of magnitude) in thinner excava-
imagine the difficulty of such an operation! tions. Then, the efficacy in the control of the movements
Over the last decade, in a number of large buildings in mainly arises from exploring the so-called corner effect
Taipei, the capital of Taiwan, the Oslo Metro solution was (3D effect), which reduces the movements in the concave
Matos Fernandes
Figure 18 - Buttress of soil treated with Deep Mixing Method in the excavations for the Boston Central Artery and Tunnel Project
(O’Rourke & McGinn, 2006).
Figure 19 - Cross section of the solution applied in Westminster Station, London, showing the tunnelled struts underneath the excava-
tion bottom (Mair & Harris, 2001).
corners of the excavation. Obviously, that effect will be po- 4.3. Application of buttresses of soil treated by cutter
tentiated if the excavation and the construction of the per- soil mixing
manent buried structure are carried out, for each elemental
One recent and promising technology of ground treat-
area, in different phases over time. Some useful sugges-
ment is the so-called cutter-soil-mixing (CSM). This
tions can be found in these works with regard to the execu- method is similar to the deep mixing method/deep soil mix-
tion of the joints between the peripheral wall and the cross ing (DMM/DSM) but the device that performs the mixture
wall in order to eliminate the lime (Hsieh et al., 2008). is inspired in diaphragm wall cutter wheels rotating around
Figure 20 - Excavation with cross walls of concrete up to the ground surface: a) plan scheme; b) detail of a joint cross wall-peripheral
wall (Ou et al., 2006, Ou et al., 2011).
a horizontal axis. This means that the geometry of treated Figure 21 compares: i) the continuous horizontal sup-
mass corresponds to panels of rectangular section in plan. port, obtained by jet-grouting, compaction grouting or by
With this technique a continuous wall may be formed by other method, which can already be considered a classic so-
the construction, in an alternating sequence, of overlapping lution; ii) the proposed support by cross vertical (but-
primary and secondary panels. Secondary panels can be tresses) elements, disposed at given intervals in the longitu-
constructed immediately after completion of primary pan- dinal direction, obtained by cutter-soil-mixing. The figure
els or by cutting into panels that have already hardened. includes, for the two systems, the treatment ratio, TRr, that
Panels can be constructed in lengths ranging from 2.2 m to is, the ratio of the volumes of the treated soil and the exca-
2.8 m and wall thicknesses of 0.5 m to 1.0 m (Fiorotto et al., vated soil. It should be noted that, in the case of buttresses,
2003, 2005). their top may rest at the elevation of the base of the first ex-
In the Manuel Rocha Lecture 2009, the author of this cavation stage, since above that elevation their effect is
work pointed out this methodology as a very convenient so- null.
lution to support the wall before the excavation (Matos As suggested by Fig. 21, it would be convenient to in-
Fernandes, 2010). stall the CSM panels coinciding with the joints of the pe-
Figure 21 - Comparative structural systems of the soil treatment and expressions of the treatment ratio: a) slab of jet-grout; b) buttresses
of soil treated by CSM.
Matos Fernandes
ripheral wall, whose interval is generally 5-6 m. For this • the CSM panels may be used, while they are fluid, to in-
range, and bearing in mind the diaphragm wall thickness stall vertical soldier piles for support of the bracing sys-
commonly adopted in this type of excavations, it may be tem;
anticipated that the 3D effects (that is, the differences be- • in comparison with the other solutions, full advantage of
tween the wall displacements and stresses in the supported the constructive operations is taken, since practically all
sections and in the sections midway between supports) the penetration of the device is used to improve the
would be negligible. ground mass.
Figure 22 illustrates the typical sequence of a deep There is already a substantial amount of data concern-
excavation with the proposed treatment applied before ex- ing the mechanical parameters of the treated soil achieved
cavation starts. The treated material is removed together by CSM, as resumed by Fig. 23 (Denis et al., 2012). For
with the soil, stage by stage, in conjugation with the instal- soft soils, the Portuguese experience, which is very signifi-
lation and pre-stressing of the temporary struts. cant, permits to indicate the following values: qu = 2.5 -
The experience available suggests that this technique 3.0 MPa and Eu /qu = 400 - 500 (Matos Fernandes et al.,
could fulfil the following five essential requirements in 2010; Peixoto et al., 2012; Pinto, 2014).
comparison with other treatment technologies: Numerical finite element simulations comparing the
• small impact associated with installation, avoiding performance of the two systems illustrated by Fig. 21, as-
over-compression or stress relief of the surrounding soil;
• constituent material easy to excavate but with good
strength and stiffness, as well;
• very accurate installation position, namely with regard to
verticality, position in plan and geometry of the treated
soil;
• confidence regarding continuity in the transverse direc-
tion;
• confidence regarding the connection to the peripheral
concrete wall.
Beyond these points, already confirmed by actual ex-
perience (Pinto et al., 2012), there are other very important
advantages, namely:
• the CSM panels remaining underneath the final excava-
tion level may provide a foundation for the bottom slab
Figure 23 - Values of the uniaxial compressive strength and of the
of the permanent structure; deformation modulus of soil treated by Cutter Soil Mixing and
• the CSM treatment can seal a hole resulting from a seri- values derived from the Portuguese experience (adapted from
ous defect at a wall joint; Denis et al., 2012).
Figure 22 - Simplified sequence of a multi-strutted deep excavation with buttress of soil treated by CSM (Matos Fernandes, 2010).
suming the same mechanical properties for the treated soil inspiring: it suggests that there is a synergetic effect (that is,
and similar treatment ratios, led to the following conclu- a virtuous combination) between pre-stressing and soil im-
sions (Matos Fernandes, 2010): provement in the control of the movements. This had been
i) both solutions revealed similar efficacy with regard to the already identified by Matos Fernandes et al. (2008) with the
control of the soil and wall movements; soil improvement by a jet-grout slab.
ii) the solution employing CSM panels takes large advan-
tage as far as the wall bending moments and shear
5. The Long-Term Hydraulic Conditions and
forces are concerned; this is probably due to the fact the Settlements by Consolidation
that the vertical buttresses provide to the peripheral
5.1. Introduction
wall a more distributed support along the vertical di-
rection, the one where the structural generalized for- As it was emphasized above, the consolidation in-
ces are predominant. duced by the excavation may originate surface settlements.
A final comment on the effect of soil treatment in ad- Note that in many cases the soft soil consists of organic
vance of the excavation may be presented with the help of clayey silt, whose (not very small) permeability allows a
Fig. 24, which collects results from a series of finite ele- consolidation process during a relatively short period after
ment simulations of a deep excavation in soft clay in order the conclusion of the excavation.
to evaluate: i) the effect of the level of the total pre-stress For very thick soft deposits, extending the wall in
force applied by the struts without any soil treatment in ad- depth to the bedrock is a recommendable but expensive so-
vance; ii) the effect of strut pre-stressing combined with lution. An alternative solution consists of limiting the em-
soil treatment by CSM cross panels. The total pre-stress bedded wall length to the value required by a comfortable
force is expressed as a percentage of the integral of the safety factor against bottom heave. In general, the bottom
at-rest total horizontal stresses computed from the surface slab of the permanent buried structure incorporates a drain
to the base of the excavation. in order to avoid uplift pressures in the long-term.
The figure presents the soil displaced volume (that is, When comparing the two solutions, depicted in
the integral of surface settlements) for all the analyses per- Fig. 25, a relevant issue is related to the differences of the
formed, at the completion of the excavation. As a reference, consolidation process during and, mostly, after the comple-
the figure also includes the volume from Peck’s chart for tion of the excavation, as will be highlighted below through
maximum settlement of 1% of the excavation depth. finite element analyses (Matos Fernandes et al., 2012).
The exam of the figure permits to observe that, for the
5.2. Assumptions for the numerical study
analyses with no treated soil, the strut pre-stress is capable
of reducing the displaced volume up to about 1/2 of that ob- The selected problem corresponds to an excavation
tained when no pre-stress is applied, whereas a reduction to with an infinite length, 10 m deep and 12 m wide, per-
about 1/3 is obtained with similar pre-stress for the series of formed in soft clayey silt 30 m thick, overlying a hard and
analyses with the soil treated by CSM. This result is very practically impermeable substratum (Fig. 25). The soft soil
Figure 24 - Synergetic effect between pre-stressing and soil treatment in the control of the movements.
Matos Fernandes
Figure 25 - Cross section of the excavations considered for studying the influence of the long-term hydraulic conditions on the settle-
ments by consolidation: a) case A; b) case B.
is lightly over-consolidated from the surface down to a The excavation is carried out in a total time of 25 days
depth of 4 m and normally consolidated for larger depths. at a uniform rate. A fully coupled analysis is performed
The retaining structure consists of a 1.0 m thick rein- both during and after the excavation period.
forced concrete diaphragm wall and three levels of cross- 5.3. Discussion of the results
lot steel struts. Two solutions were considered for the re-
taining wall, as shown in Fig. 25: embedded length of 7.0 m 5.3.1. Excavation period
(case A); embedded length of 20 m, which means that the During the excavation period, because of the low per-
wall is extended down to the substratum (case B). The meability of the soft soil, the behaviour is approximately
safety factor against bottom heave for case A, calculated by undrained.
the method of Bjerrum & Eide (1956), is 1.42. Figure 26 illustrates the distribution of the excess
In order that both retaining structures could ensure pore pressure at the end of excavation for solutions A and
comparable control of the movements during the excava-
tion phase, it was assumed a 2 m thick jet-grout slab, exe-
cuted prior to the soil removal and placed immediately
below the base of the excavation.
Basically, the finite element code uses a fully coupled
formulation of the flow and equilibrium equations with soil
constitutive relations in effective stresses, according to the
Biot consolidation theory (Britto & Gunn, 1987; Borges,
1995) and the p - q - q critical state model (Lewis &
Schrefler, 1987; Borges, 1995; Borges & Pinto, 2013; Bor-
ges & Guerra, 2014).
As for the hydraulic boundary conditions, it is as-
sumed that: i) in the supported side, the water table remains
at the ground surface (which is a conservative option, as-
suming that there is a flow providing water to the ground
surface); ii) in the excavated side, the water level coincides
with the current excavation base; iii) the diaphragm wall
and the substratum are impermeable; iv) the long-term con-
ditions correspond to null pore pressures at the top of the Figure 26 - Excess pore pressure at the end of excavation: a) case
jet-grout slab. A; b) case B.
B. Excess pore pressure is defined as the difference be- water pressure increases in solution A whereas in solution
tween pore pressure at a particular moment and its initial B it decreases in a part of the ground (at larger depths) and
hydrostatic value. increases in another (close to the excavation bottom). On
The results reveal that negative values are generated the supported side, the evolution is contrary, although of
by the excavation in both cases. As could be expected, lower magnitude: there is a decrease of pore pressure in
highest negative values of excess pore pressure occur be- case A, related with downward water flow, while an in-
low the excavation base, in association with the very pro- crease takes place in case B.
nounced decrease of the total mean stress in that zone. On Horizontal displacements of the wall at the comple-
the supported side of the ground mass a tendency for gener- tion of the excavation and at the end of consolidation, for
ating negative excess pore pressures is also observed, al- both cases, are shown in Fig. 28. As could be expected, the
though with much smaller values than at the opposite side. displacements are larger in case A. This is obviously due to
This is mainly explained by the effective restraint provided the shorter embedded length, which allows a somewhat
by the jet-grout slab, which decisively contributes to avoid larger rotation of the wall around the contact with the
a ground decompression with much larger magnitude. jet-grout slab (between -10 m and -12 m in depth).
Maximum values of negative excess pore pressure on the In overall terms, the wall displacements above the ex-
supported side are around 55 kPa for case A (in a region be- cavation base level do not significantly change with the
low the wall tip) and 13 kPa for case B. consolidation. The most expressive variation is observed
below that level, where displacements reduce in case A and
5.3.2. Post-excavation period
increase in case B. For case A this evolution may be ex-
Figure 27 represents the distribution of pore pressure plained by the swelling effect of the soil underneath the ex-
at the end of consolidation. It should be noted that, as said cavation (associated to the decrease of mean effective
above, excess pore pressure is considered herein (as de- stress, as seen above) as consolidation takes place in that
fined in the computer code) the difference between pore zone. In case B the evolution is the opposite: since the mean
pressure at a particular moment and its initial hydrostatic effective stress increases (at larger depths) as the consolida-
value (before excavation), and not its final value (at the end tion evolves, a shrinking on the soil underneath the excava-
of consolidation). tion occurs, which permits a movement of the wall towards
In case A long-term steady flow contouring the wall that side. The slight increase in the pore pressure on the
tip takes place, whereas in solution B hydrostatic states on
both sides of the wall are obtained at end of consolidation.
Due to the distinct long-term pore pressure condi-
tions, the most significant conclusion extracted from the
comparison of results is that the consolidation process is
qualitatively quite different in the two cases. As could be
expected, the most significant variations of pore pressure
are observed on the excavated side. However, on that side
Figure 27 - Distribution of pore pressure at the end of consolida- Figure 28 - Wall horizontal displacements at the completion of
tion: a) case A; b) case B. the excavation and at the end of the consolidation.
Matos Fernandes
other side of the wall also contributes for the observed different in steel and in reinforced concrete walls. For the
movement evolution. bending strain level commonly attained by concrete retain-
The distribution of the surface settlements is illus- ing walls, the behaviour is no longer linear elastic. In spite
trated in Fig. 29. In case A, there are pronounced downward of this fact, the geotechnical finite element models com-
displacements on the supported ground with the consolida- monly used in design practice assume a linear elastic be-
tion, while in case B upward displacements are observed, haviour for the structural components, while offering many
although with much less magnitude. non-linear constitutive laws for the soil, some of them
The evolution in case A is mainly justified by the re- highly sophisticated. It is interesting to observe that the op-
duction of volume on the supported soil, as discussed posite occurs with regard to the soil-structure interaction
above. In case B there are two contrary effects: incremental analyses performed by structural engineers: more or less
horizontal displacements of the wall (below the excavation complex non-linear constitutive laws are used for the rein-
base) occur - which, by themselves, tend to increment sur- forced concrete whereas the foundation soil is assumed to
face settlements; however, the supported soil swells as the be linear elastic!
consolidation takes place - which tends to upraise the Figure 30 resumes this analytical gap, as this bizarre
ground surface. This second effect is stronger, provoking situation was classified by the author (Matos Fernandes,
upward displacements after the excavation period. 2010). In that paper it was stated that to overcome this gap
As a final comment, it is interesting to observe that would probably be the most challenging and critical task
the considered solutions provide similar lateral wall dis- for the near future in the field of the computational methods
placements, both at the completion of the excavation and at applied in design. Fortunately, some progress in this field
the end of the consolidation. As well, they ensure very re- may be presented in this occasion.
sembling surface settlements at the end of the excavation.
Nevertheless, the corresponding surface settlements at the 6.2. Numerical model
end of the consolidation process are very discrepant (maxi- The solution found consists of an interaction between
mum settlement for case A is about six times greater than the geotechnical computational code (Code G) and the rein-
the one for case B). forced concrete (nonlinear) computational code (Code
This highlights the great convenience of preventing, RC), as illustrated in Fig. 31.
for the long-term conditions, a steady flow of water towards For each stage of construction, the bending moment
the excavation. The settlements induced by the new distri- and axial stress diagrams in the wall, computed by Code G,
bution of effective stresses in the supported ground associ- are applied to the structure modelled by Code RC, which
ated to this flow may cause serious damage in ancient accounts for the actual steel reinforcement at each section.
constructions that in some urban areas are founded on the This code then calculates the strains and cracking in the re-
desiccated crust. inforced concrete wall and the corresponding adjusted stiff-
ness at each section, which is introduced in Code G for
6. The Influence of the Nonlinear Behaviour being used in the next construction stage.
of the Reinforced Concrete This idea was implemented using as the geotechnical
6.1. Introduction code FEMEP, developed at the Universities of Porto and
Coimbra (Cardoso et al., 2006; Almeida e Sousa et al.,
In flexible retaining structures the deformations by 2011) and as the structural code EVOLUTION, developed
bending play a critical role in the distribution of earth pres- at the University of Porto (Ferraz, 2010).
sures and of the structural stresses. The ability to experi- With the purpose of exemplifying the application of
ence deformation without stiffness reduction is quite the procedure described it was taken the example illustrated
in Fig. 32: an excavation 18 m deep and 24 m wide in a de-
posit of soft clay 27 m thick, supported by a diaphragm wall
Figure 29 - Surface settlement distribution at the completion of Figure 30 - The “analytical gap” between structural and geotech-
the excavation and at the end of the consolidation. nical analyses (Matos Fernandes, 2010).
Figure 31 - The interaction between the geotechnical code (code G) and the structural code (code RC) to account for the nonlinear be-
haviour of the reinforced concrete retaining wall.
Figure 32 - Cross section of the excavation considered for studying the influence of the non linear behaviour of the reinforced concrete
wall.
Matos Fernandes
(1.0 m thick and the toe sealed 1.0 m into the substratum) 25% in the maximum displacement was obtained, which is
and by three levels of cross-lot non pre-stressed steel struts. rather expressive.
Figure 36 collects, for the final excavation stage, the
6.3. Discussion of results results from the analysis with the nonlinear concrete behav-
Firstly, it was performed an analysis assuming the re- iour, showing in its central part (Fig. 36b) the ratio, for each
inforced concrete wall as linear elastic. This analysis: i) horizontal wall section, between the effective bending stiff-
served as comparison for the analysis in which the numeri- ness and the elastic bending stiffness. It can be observed
cal model described above was applied; ii) on the other that, in the region where bending moments and displace-
hand, its results concerning the structural stresses in the ments are more pronounced, along more than 10 m, the ef-
wall were the basis for calculating the steel reinforcement, fective stiffness ratio is about 1/3 of the elastic one.
using the conventional procedures of concrete structures. In These preliminary results seem to confirm the rele-
particular, the bending moments obtained from the finite el- vance of considering the nonlinear behaviour of reinforced
ement analysis were multiplied by 1.35. This is illustrated concrete in the geotechnical finite element analyses for this
in Fig. 33. type of excavations.
Figures 34 and 35 compare the results in terms of wall
bending moments and lateral wall displacements provided
7. Conclusions
by the analysis just mentioned above (linear elastic wall) Deep excavations in thick deposits of soft soils have
and by the analysis which considers the nonlinear behav- known, along the last decades, an outstanding evolution.
iour of reinforced concrete, with the steel reinforcement We now accomplish, in that domain, works that a few de-
shown in Fig. 33b. cades ago were unimaginable or that would require cost,
As expected, the nonlinear behaviour of reinforced execution time and nearby impact of intolerable magnitude.
concrete, implying a decrease of the wall bending stiffness, Such evolution is hardly replicable in other fields of Civil
leads to a reduction of the maximum bending moments and and Geotechnical Engineering.
to an increase of wall displacements. The influence regard- The interaction between a flexible retaining structure
ing the bending moments is extremely relevant; as far as the and the adjacent soil mass is what is called, in Structural
wall movements are concerned, an increase of the order of Engineering, a highly hyper-static problem. Then, before
Figure 33 - Design of the reinforced concrete wall: a) bending moment diagrams assuming linear elastic behaviour of the concrete; b)
steel reinforcement.
Figure 34 - Wall bending moments: a) linear elastic concrete wall; b) nonlinear concrete wall.
Figure 35 - Lateral wall displacements: a) linear elastic concrete wall; b) nonlinear concrete wall.
Matos Fernandes
Figure 36 - Results of the analysis considering the nonlinearity of the concrete wall at the completion of the excavation: a) wall bending
moments; b) ratio between the effective and the elastic bending stiffness; c) lateral wall displacements.
the application of finite element models to these problems, ments induced by the excavations. Particularly in soft soils,
the available design methods were very limited and of emit is possible today to achieve values of the surface settle-
pirical or semi-empirical nature. Contrarily to various geo- ments one order of magnitude below the most favourable
technical fields - such as spread and deep foundations, results reported by Peck (1969) in his splendid general re-
natural slopes, etc. - modern deep excavations in soft soils port.
would not be possible without the profit of an analytical The use of cutter-soil-mixing to construct vertical
tool as powerful as the finite element method. cross buttresses of treated soil at given intervals presents a
In spite of such a brilliant progress, the Profession is remarkable set of advantages - concerning the conception,
tacitly more tolerant with excavation induced movements the execution and the structural performance - in compari-
in soft ground. However, in soft ground ancient structures son with other well-known solutions involving soil treat-
and services might probably be less tolerant with regard to ment.
further movements than similar constructions over stiff The combination, in the retaining structure, of high
soils. level of strut pre-stressing with the treatment of the soft soil
More robust and reliable solutions, exploiting the produces a synergetic effect for the minimization of the
best the technology and the experience have convened, are movements.
highly recommendable. These solutions are not just a con- The settlements developed by consolidation of the
dition for minimizing the induced movements. They soil after the conclusion of the excavation are a topic de-
strongly increment the reliability of our prediction analy- serving more attention. The results presented show the
ses! great convenience of preventing, for the long-term condi-
Like in the sixties the introduction of diaphragm tions, a steady flow of water towards the bottom slab of the
walls, permitting to install the entire reinforced concrete permanent buried structure. The settlements induced by the
wall before performing the excavation, was a paramount new distribution of effective stresses in the supported
technical advance, a number of solutions have been tested ground associated to this flow may achieve various times
in order to provide support to the wall before the removal of the ones prevailing at the conclusion of the excavation.
the soil. The combination of these two expedients has per- The consideration in finite element analyses of the
mitted an auspicious progress in the control of the move- nonlinear behaviour of reinforced concrete wall may have a
non negligible influence on the computed wall displace- considering the consolidation effect. Geotechnical and
ments and a strong influence on the values of the structural Geological Engineering, 31(2):593-615.
stresses in the wall. Borges, J.L. & Guerra, G.T. (2014). Cylindrical excava-
The author would like to conclude by emphasizing tions in clayey soils retained by jet grout walls: numeri-
that a sustainable progress in this area depends on an inge- cal analysis and parametric study considering the influ-
nious exploitation and combination of the advances in the ence of consolidation. Computers and Geotechnics,
three following fields: construction technologies, concep- 5(1):42-56.
tion of the retaining structures and methods of analysis. Britto, A.M. & Gunn, M.J. (1987). Critical Soil Mechanics
via Finite Elements. Ellis Horwood Limited, England.
Acknowledgments
Brito, J.M. & Matos Fernandes, M. (2006). Terreiro do
The author expresses his deep gratitude to the Brazil- Paço Station of Metropolitano de Lisboa: Conception,
ian Association for Soil Mechanics for the kind and very design predictions and performance (in Portuguese).
th
prestigious invitation to present the 11 Pacheco Silva Lec- rd
Proc. 3 Portuguese-Brazilian Congress, v. 1, pp. 21-
ture. He is indebted to the Colleagues that collaborated in 42.
the research presented, António Cardoso (Univ. Porto), Cardoso, A.S.; Guerra, N.C.; Antão, A. & Matos Fer-
António Pedro (Univ. Coimbra), Jorge Almeida e Sousa nandes, M. (2006). Limit analysis of anchored-concrete
(Univ. Coimbra), José Cândido Freitas (Inst. Polit. Porto), soldier-piles in clay under vertical loading. Canadian
José Leitão Borges (Univ. Porto), José Mateus de Brito Geotechnical Journal, 43(5):516-530.
(Cenorgeo, Lisbon), José Menezes (Univ. Porto), Miguel
Clough, G.W. & Tsui, Y. (1974). Performance of tied-back
Ferraz (Univ. Porto), Pedro Alves Costa (Univ. Porto) and
walls in clay. Journal Geotechnical Engineering,
Pedro Pereira (Univ. Porto). Special appreciation is due to
ASCE, 100(12):1259-1273.
Alberto Sayão (PUC, Rio de Janeiro), Alexandre Pinto
Clough, G.W. & Mana, A.I. (1976). Lessons learned in fi-
(JETSJ, Lisbon), Arsénio Negro Jr. (Bureau, São Paulo),
nite element analyses of temporary excavations in soft
Artur Peixoto (Georumo) and Nuno Cristelo (Univ. Trás- nd
clay. Proc. 2 Int. Conf. on Numerical Methods in
os-Montes e Alto Douro), that provided documents, techni-
Geomechanics, Blacksburg, Virginia, v. 1, pp. 496-510.
cal advice and other data for the preparation of the lecture.
Great esteem and recognition are due to Anna Laura Nunes Clough, G.W. & Reed, M.W. (1984). Measured behavior of
(Federal Univ. Rio de Janeiro), Bárbara Rangel (Univ. braced walls in very soft clay. Journal Geotechnical En-
Porto) and Manuel Carvalho (Univ. Porto) for their inesti- gineering, ASCE, 110(1):1-18.
mable support in the presentation of the lecture. The careful Clough, G.W.; Smith, E.M. & Sweeney, B.P. (1989).
review of the paper by Sara Rios is gratefully acknowl- Movement control of excavation support systems by it-
edged. erative design. Proc. ASCE Conf. on Foundation Engi-
neering: Current Principles and Practices, v. 2,
References pp. 869-884.
Almeida e Sousa, J.; Negro, A.; Matos Fernandes, M. & Clough, G.W. & O’Rourke, T.D. (1990). Construction in-
Cardoso, A.S. (2010). Three-dimensonal nonlinear duced movements of insitu walls. Proc. ASCE Conf. on
analyses of a metro tunnel in Sao Paulo porous clay, Design and Performance of Earth Retaining Structures,
Brazil. Journal Geotechnical and Geoenvironmental Cornell University, Ithaca, New York, pp. 439-470.
Engineering, ASCE, 137(4):376-384. Collins (1984). The Collins Thesaurus in A-Z Form.
Alves Costa, P.; Borges J.L. & Matos Fernandes, M. HarperCollins Publishers, Glasgow, 432 p.
(2007). Analysis of a braced excavation in soft soils Denies, N.; Huybrechts, N.; De Cock, F.; Lameire, B.;
considering the consolidation effect. Geotechnical and Vervoot, A.; Van Lysebetten, G. & Maertens, J. (2012).
Geological Engineering, 25(6):617-629. Soil mix walls as retaining structures - mechanical char-
Bjerrum, L. & Eide, O. (1956). Stability of strutted excava- acterization. International Symposium of ISSMGE -
tions in clay. Géotechnique, 6(1):32-47. TC211 on Recent Research, Advances and Execution
Bjerrum, L.; Clausen, C. & Duncan, J. (1972). Stability of Aspects of Ground Improvement Works. Brussels, Bel-
th
flexible structures. Proc. V European Conference on gium.
Soil Mechanics and Foundation Engineering, Madrid, Egger, P. (1972). Influence of wall stiffness and anchor
th
General Report. prestressing on earth pressure distribution. Proc. V Eu-
Borges, J.L. (1995). Geosynthetic-Reinforced Embank- ropean Conference on Soil Mechanics and Foundation
ments on Soft Soils. Analysis and Design. PhD Thesis Engineering, Madrid, v. 1, pp. 259-264.
in Civil Engineering, University of Porto, Portugal (in Eide, O.; Aas, G. & Josang, T. (1972). Special application
Portuguese). of cast-in-place walls for tunnels in soft clay in Oslo.
th
Borges, J.L. & Pinto, R.G. (2013). Strutted excavation in Proc. V European Conference on Soil Mechanics and
soft soil incorporating a jet-grout base slab: analysis Foundation Engineering, Madrid, v. 1, pp. 485-498.
Matos Fernandes
Ferraz, M. (2010). Model for structural behavioural assess- Mair, R. & Harris, D. (2001). Innovative engineering to
ment of bridges. PhD thesis, University of Porto. control Big Ben’s tilt. Ingenia, Royal Academy of Engi-
Finno, R.J.; Atmatzidis, D.K. & Perkins, S. B. (1989). Ob- neering, Issue 9, pp. 23-27.
served performance of a deep excavation in clay. Jour- Mana, A.I. & Clough, G.W. (1981). Prediction of move-
nal Geotechinical Engineering, ASCE, ments for braced cuts in clay. Journal Geotechnical En-
115(8):1045-1064. gineering, ASCE, 107(6):759-777.
Finno, R.J.; Bryson, S. & Calvello, M. (2002). Performance Matos Fernandes, M. (1985). Performance and analysis of a
th
of a stiff support system in soft clay. Journal Geotech- deep excavation in Lisbon. Proc. XI ICSMFE, San
nical and Geoenvironmental Engineering, ASCE, Francisco, v. 4, pp. 2073-2078.
128(8):660-671. Matos Fernandes, M. (2004). Supported excavations. Soil-
Fiorotto R.; Stoetzer E. & Schoepf, M. (2003). CSM Cutter structure interaction analysed through tangential stress
Soil Mixing An innovation in Soil Mixing for Cut-off at the interface (in Portuguese). Soils & Rocks,
and Retaining Walls. BAUER Maschinen, Germany. 27(1):3-14.
Fiorotto, R.; Stötzer, E.; Schöpf, M. & Brunner, W. (2005). Matos Fernandes, M.; Maranha das Neves, E.; Salgado, F.;
Cutter Soil Mixing (CSM). Proc. Int. Conf. on Deep Caldeira, L.; Pina, R.; Flor, A.; Brito, J.M. & Tavares,
Mixing’05, 6 p (CD-Rom). A. (2007). Jet grouting solutions for cut-and-cover sub-
th
Glass, P. & Stones, C. (1999). Construction of Westminster way metro stations in Lisbon downtown. Proc. XIV
Station, London. Proc. of the Institution of Civil Engi- ECSMGE, Madrid, v. 2, pp. 895-900.
neers, Structures & Buildings, 146(3):237-252. Matos Fernandes, M. (2007). An overview on induced
Hsieh, H.S.; Wang, C.C. & Ou, C.Y. (2003). Use of jet movements by deep excavations in soft ground and
th
grouting to limit diaphragm wall displacement of a deep eight golden rules for their control. Proc. XIV
excavation. Journal Geotechnical and Geoenvironmen- ECSMGE, Madrid, v. 5, pp. 379-384.
tal Engineering, ASCE, 129(2):146-157. Matos Fernandes, M.; Pereira, C.D. & Guerra, N. (2008).
Search for null displacement of walls supporting deep
Hsieh, H.S.; Lu, Y.C. & Lin, T.M. (2008). Effects of joint
excavations in soft clay - A numerical experiment on a
detail on the behavior of cross walls. Journal of Geo-
real case. Proc. Int. Conf. Development Urban Areas
Engineering, 3(2):55-60.
and Geotechnical Engineering, S. Petersburg, v. 2,
Hwang, R.N.; Moh, Z. & Wang, C.H. (2007). Performance
pp. 587-592.
of wall systems during excavation for Core Pacific City.
Matos Fernandes, M.; Peixoto, A., Pinto; A., Pita, X.; Topa
Journal of Geoengineering, 2(2):53-60.
Gomes, A. & Pedro, A. (2010). The technology of cut-
Karlsrud, K. (1983). Performance and design of slurry ter soil mixing applied in urban excavations (in Portu-
walls in soft clay. NGI, Publication Nº 149, Oslo. th
guese). Proc. 5 Portuguese-Brazilian Congress,
Karlsrud, K. (1997). Some aspects of design and construc- Gramado, Rio Grande do Sul, CD-Rom.
th
tion of deep supported excavations. Proc. XIV Int. Matos Fernandes, M. (2010). Deep urban excavations in
Conf. Soil Mechanics and Foundation Engineering, Portugal: practice, design, research and perspectives.
Hamburg, v. 4, pp. 2315-2320. Soils & Rocks 33(3):115-142.
Karlsrud, K. & Andresen, L. (2007). Design of deep exca- Matos Fernandes, M.; Cardoso, A.S.; Topa Gomes, A.;
th
vations in soft clays. Proc. XIV ECSMGE, Madrid, Borges, J.L.; Guerra, N.C. & Antão, A. N. (2012). Deep
v. 1, pp. 75-99. excavations in urban areas - Finite element modelling
Koutsoftas, D.C.; Frobenius, P.; Wu, C.L.; Meyersohn, D. for three geotechnical scenarios and retaining solutions.
& Kulesza, R. (2000). Deformations during cut-and- Ribeiro e Sousa et al. (eds). Innovative Numerical
cover construction of MUNI Metro Turnback Project. Modelling in Geomechanics. Taylors & Francis, Ox-
Journal Geotechnical and Geoenvironmental Engi- ford, chapter 3, pp. 51-76.
neering, ASCE, 126(4):344-359. Moormann, C. (2004). Analysis of wall and ground move-
Lewis, R.W. & Schrefler, B.A. (1987). The Finite Element ments due to deep excavations in soft soil based on a
Method in the Deformation and Consolidation of Po- new worldwide database. Soils and Foundations,
rous Media. John Wiley and Sons, Inc., New York. 44(1):87-98.
Lui, G.B.; Ng, W.W. & Wang, Z.W. (2005). Observed per- Ng, C.W., Hong, Y., Liu, G.B. & Liu, T. (2012). Ground
formance of a deep multi-strutted excavation in Shang- deformations and soil-structure interaction of a multi-
hai soft clays. Journal Geotechnical and Geoenviron- propped excavation in Shanghai. Géotechnique,
mental Engineering, ASCE, 131(8):1004-1013. 62(10):907-921.
Liu, G.B.; Jiang, R.J.; Ng, C.W. & Hong, Y. (2011). Defor- O’Rourke, T.D. & McGinn, J. (2006). Lessons learned for
mation characteristics of a 38 m deep excavation in soft ground movements and soil stabilization from Boston
clay. Canadian Geotechnical Journal, 48(12):1817- Central Artery. Journal Geotechnical and Geoenviron-
1828. mental Engineering, ASCE, 132(8):966-989.
Ou, C.Y.; Wu, T.S. & Hsieh, H.S. (1996). Analysis of deep Journal Geotechnical and Geoenvironmental Engi-
excavation with column type of ground improvement in neering, ASCE, 127(12):1061-1069.
soft clay. Journal Geotechnical Engineering, ASCE, Randall, F.A. (1999). History of the Development of Build-
nd
122(9):709-716. ing Construction in Chicago. 2 edition, University of
Ou, C.Y.; Lin, Y.L. & Hsieh, P.G. (2006). Case record of an Illinois Press, Champaign.
excavation with cross walls and buttress walls. Journal Tan, Y. & Li, M. (2011). Measured performance of 26 m
of GeoEngineering, 1(2):79-86. deep top-down excavation in downtown Shanghai. Ca-
Ou, C.Y.; Hsieh, P.G. & Lin, Y.L. (2011). Performance of nadian Geotechnical Journal, 48(5):704-719.
excavations with cross walls. Journal Geotechnical and Tan, Y. & Wei, B. (2012). Observed behaviors of a long
Geoenvironmental Engineering, ASCE, 137(1):94-104. and deep excavation constructed by cut-and-cover tech-
Palmer, J.H. & Kenney, T.C. (1972). Analytical study of a nique in Shanghai soft clay. Journal Geotechnical and
braced excavation in weak clay. Canadian Geotechnical Geoenvironmental Engineering, ASCE, 138(1):69-88.
Journal, 9(2):145-164. Wang, Z.W.; Ng, C.W. & Liu, G.B. (2005). Characteristics
of wall deflections and ground surface settlements in
Peck, R.B. (1969). Deep excavations and tunnelling in soft
th Shanghai. Canadian Geotechnical Journal,
ground. Proc. VII ICSMFE, Mexico City, General Re-
42(5):1243-1254.
port, State-of-the-Art Volume, pp. 225-290.
Wang, J.H.; Xu, Z.H. & Wang, W.D. (2010). Wall and
Peixoto, A.; Matos Fernandes, M.; Sousa, E. & Gomes, P. ground movements due to deep excavations in Shang-
(2012). The application of Cutter Soil Mixing to an ur- hai soft soils. Journal of Geotechnical and Geoenviron-
ban excavation at the riverside of Lagos, Portugal. Proc. mental Engineering, ASCE, 136(7):985-994.
International Symposium of ISSMGE - TC211 on Re- Wong, I.H.; Poh, T.Y. & Chuah, H.L. (1997). Performance
cent Research, Advances and Execution Aspects of of excavations for depressed expressway in Singapore.
Ground Improvement Works. Brussels, Belgium. Journal Geotechnical and Geoenvironmental Engi-
Pinto, A.; Tomásio, R.; Pita, X.; Godinho, P. & Peixoto, A. neering, ASCE, 123(7):617-625.
(2012). Ground improvement solutions using CSM Wong, I.H. & Poh, T.Y. (2000). Effects of jet grouting on
technology. Proc. International Symposium of adjacent ground and Structures. Journal Geotechnical
ISSMGE - TC211 on Recent Research, Advances and and Geoenvironmental Engineering, ASCE,
Execution Aspects of Ground Improvement Works. 126(3):247-256.
Brussels, Belgium. Wu, S. H.; Ching, J. & Ou, C.Y. (2013). Predicting wall
Pinto, A. (2014). Personal communication. displacements for excavations with cross walls in soft
Poh, T.Y.; Goh, A.T. & Wong, I. H. (2001). Ground move- clay. Journal Geotechnical and Geoenvironmental En-
ments associated with wall construction: case histories. gineering, ASCE, 139(6):914-927.
Articles
Soils and Rocks

v. 38, n. 3
Use of Sugarcane Bagasse as Carbon Substrate in Permeable
Reactive Barriers: Laboratory Batch Tests and Mathematical
Modeling
R.C. Mattos, P.S. Hemsi, E.Y. Kawachi, F.T. Silva
Abstract. The combination of the lignocellulosic residue sugarcane bagasse and poultry litter as carbon and nutrients
sources, respectively, in organic permeable reactive barriers for remediation of metal laden acid mine drainage has not yet
been extensively tested and may be promising based on chemical composition. This system is being tested in the laboratory
by means of microcosm batch experiments using sugarcane bagasse with and without poultry-litter as nitrogen and phos-
phorous source. As expected based on the compositions of sugarcane bagasse and poultry litter, the experiments (dupli-
cate) combining both residues achieved the highest rates of sulfate reduction, between ~ 30 to 45 mg sulfate/L-day, and
better sustained these rates throughout the duration of the experiment (78 days). Also, pH values increased from ~ 6 to 8,
and the oxidation-reduction potential achieved < -100 mV, indicating anaerobic sulfate-reducing conditions. Mathematical
modeling performed on these experiments, with a previously-tested model, resulted in simulated results that approached
the experimental data when the sugarcane bagasse was modeled as comprising a cellulose fraction with a Contois specific
degradation rate similar to that for wood chips and leaf mulch, plus an additional easily-degradable fraction to account for
the occurrence of soluble sugars in sugarcane bagasse.
Keywords: acid mine drainage, batch tests, permeable reactive barrier, remediation, sugarcane bagasse, sulfate reduction.
are advantageous in comparison to operation-intensive and

1. Introduction
energy-demanding pump-and-treat remediation (Wilde-
Detrimental environmental impacts result from man et al., 1997, Benner et al., 1999, Voudrias, 2001,
ground and surface water contamination from acid mine Whitehead et al., 2005, Hallberg and Johnson, 2005).
drainage (AMD). In Brazil, examples include metallifer- Among other possibilities, such as constructed wetlands,
ous-mining sites in Minas Gerais (iron, gold, nickel, cop- open limestone channels, anoxic limestone drains, or verti-
per, zinc, among others) and coal-mining sites in Santa cal flow systems, in-situ passive remediation may be ac-
Catarina, among others. According to FEAM (2011), in complished using permeable reactive barriers (PRBs) (U.S.
2010, the state of Minas Gerais alone produced more than EPA, 1998, Blowes et al., 2000, Benner et al., 2002, Grou-
3.2 x 106 tons of fine residues with high AMD-releasing po- dev et al., 2003).
tential. Inadequate disposal may result in environmental The concept of a PRB is illustrated in Fig. 1. Perme-
contamination and a threat to groundwater, soil and surface able reactive barriers are an in-situ passive-remediation
water. In the carboniferous regions of Santa Catarina, cur- technology in which reactive materials are emplaced in a
rently approximately 5,000 ha of land are contaminated by subsurface trench in order to intercept the path of a contam-
mining residues including contamination of water bodies ination plume and allow the occurrence of reactions based
(Rubio et al., 2008). In this state, 809 abandoned mines on the contact between contaminants in the plume and the
have been listed up to 2010 with 192 generating AMD materials emplaced in the PRB, retaining or transforming
(Krebs, 2010). Internationally, AMD contamination is very pollutants into environmentally acceptable forms to meet
significant. For example, in the USA, thousands of areas remediation goals down gradient from the barrier.
are reported as contaminated with AMD in the western re- The majority of PRBs worldwide are filled with me-
gion alone (WGA, 1998). tallic zero-valent iron (Fe0) to promote abiotic reductive
Groundwater remediation in areas affected by AMD dehalogenation of chlorinated solvents (Gillham and
effluents, infiltrating or emanating from surface and under- O’Hannesin, 1994) or Cr (VI) reduction, among others,
ground mines, or leaching from tailings pits, may be ac- generally in industrial sites. The first field application of an
complished using in-situ passive remediation systems that organic PRB occurred in 1995 at the Nickel Rim mine site
Renata Cotrim de Mattos, MSc., Instituto Tecnológico de Aeronáutica, Praça Mal. Eduardo Gomes 50, São José dos Campos, SP, Brazil. e-mail: [email protected].
Paulo Scarano Hemsi, Ph.D., Instituto Tecnológico de Aeronáutica, Praça Mal. Eduardo Gomes 50, São José dos Campos, SP, Brazil. e-mail: [email protected].
Elizabete Yoshie Kawachi, DSc., Instituto Tecnológico de Aeronáutica, Praça Mal. Eduardo Gomes 50, São José dos Campos, SP, Brazil. e-mail: [email protected].
Flávio Teixeira da Silva, DSc., Escola de Engenharia de Lorena, Universidade de São Paulo, Estrada Municipal do Campinho s/n, Lorena, SP, Brazil. e-mail:
[email protected].
Submitted on June 25, 2015; Final Acceptance on November 4, 2015; Discussion open until April 30, 2016.
Mattos et al.
2. Theoretical Background
The organic PRB relies on the activity of a bacterial
consortium with functions of hydrolysis and decomposition
of lignocellulosic (polymer) tissues in the solid organic ma-
terials emplaced in the PRB, that are hydrolyzed to form
monomers (e.g., sugars), and fermented to produce soluble
shorter-chain organic compounds such as lactate, acetate,
propionate, butyrate, as well as hydrogen, which are, in
turn, utilizable as carbon and energy (electrons) sources by
sulfate-reducing bacteria (Fig. 2).
Microbial sulfate reduction promotes the consump-
Figure 1 - Conceptual illustration of a permeable reactive barrier
for groundwater remediation (from U.S. EPA, 1998).
tion of sulfate (SO42-) from the plume, production of hydro-
gen sulfide in solution (H2S), and promotion of alkalinity,
as illustrated by the following reaction mechanism at pH ~
in Sudbury, Ontario, Canada (Benner et al., 1999). More re- 6 to 7:
cently, mulch biowalls have been studied and implemented C 6 H10 O 5 + 0.21H 2 O + 0.26NH +4 ®
at several contaminated sites in the USA, aimed at degrad- 1424 3
ing plumes of organic contaminants such as chlorinated
cellulose (1)
. H + + 0.26 C 5 H 7O 2 N
. C 3 H 5 O -3 + 182
156
solvents, perchlorate, and explosives (U.S. AFCEE, 2008). 1
424 3 14243
lactate biomass
Thus far, the PRB application has been limited in Brazil.
2-
For example, in the state of São Paulo, only nine field in- . C 3 H 5 O + 0.78SO
156 -
®
1
424 3 3 4
stallations have been reported out of ~ 400 industrial, com- lactate (2)
mercial, or land-disposal sites currently under remediation . HCO -3 + 0.78H 2 S
. C 2 H 3 O -2 + 156
156
(CETESB, 2013). 1
424 3
acetate
Organic PRBs are based on a mixture of filling solid

materials, comprising (a) liable-carbon sources, (b) sources
of nutrients, (c) pH buffering materials, such as calcite, (d)
microbial innocula, and (e) porous media as support materi-
als, such as gravel, sand, etc. The liable-carbon sources in-
clude agricultural by-products and are aimed at providing
carbon and energy (electrons) for sulfate-reducing bacteria
(SRB).
The results of laboratory studies involving sulfate-

reducing systems for metal precipitation have demon-
strated the potential use of these systems for AMD reme-
diation, coupled to the decomposition of carbon sources in
the form of solid organic low-cost lignocellulosic residues
(Gibert et al., 2004). Examples of the solid substrates tested
include sawdust (Tuttle et al., 1969, Wakao et al., 1979),
spent mushroom compost (Dvorak et al., 1992, Hammack
and Edenborn, 1992), fresh alfalfa (Bechard et al., 1994),
leaf mulch and wood chips (Waybrant et al., 1998, Chang et
al., 2000), and corn stover (Figueroa et al., 2007, Hemsi et
al., 2010), among others.
This paper describes the experimental results from

laboratory batch tests aimed at testing the ability of reactors
using sugarcane bagasse as the liable-carbon source, com-
plemented by poultry litter as nutrient source, to remediate
a synthetic AMD solution. In addition, simulation results Figure 2 - Sequence of anaerobic transformations of solid organic
obtained with a mathematical model previously calibrated matter with emphasis on sulfate reduction (modified from Logan
for this type of remediation system are presented. et al., 2005).
Use of Sugarcane Bagasse as Carbon Substrate in Permeable Reactive Barriers: Laboratory Batch Tests and Mathematical Modeling
where Eq. 1 is a simplified heterogeneous reaction for cel- respectively. For example, urea was added (as N source) to
lulose degradation releasing soluble lactate, and Eq. 2 is the wood chips and leaf mulch in batch tests in order to lower
incomplete oxidation of lactate coupled with sulfate reduc- the C/N ratio to ~ 16 (Cocos et al., 2002).
tion and releasing acetate, alkalinity, and hydrogen sulfide. Grubb et al. (2000) tested a sugarcane bagasse from
In Eq. 1, cellulose is in fact the polymer of C6H10O5, i. e., Peru for immobilizing metals and buffering pH of a syn-
n(C6H10O5). The release of acetate and bicarbonate in Eq. 2 thetic AMD solution in batch laboratory experiments. Pure
+
provides a buffer capacity to neutralize H released in Eq. 1. bagasse was not capable of significant pH or metals (Al,
In addition, further sulfate reduction coupled to acetate oxi- As, Cu, Fe, Mn, Zn) amelioration after 180 days. Digester
dation promotes more alkalinity, as follows: sludge then was amended to sugarcane bagasse, producing
. SO 24 - + 156
. C 2 H 3 O -2 + 156
156 . H+ ® near neutral pH and achieving very high metals removal
1
424 3 (3) (Cu, Pb and Zn) after a similar period. Grubb et al. (2000)
acetate
- concluded that an exogenous labile-carbon source (glu-
. HCO +156
312 . H2 S
3
cose) was not necessary, but that the addition of digester
Hydrogen sulfide released in solution (H2S), in turn, sludge was required for remediation. The digester sludge
promotes the precipitation of metals (e.g., Fe, Cd, Cu, Ni, requirement is likely to reflect the need for nutrients N and
Pb, Zn) in the form of stable metal monosulfides: P addition to sugarcane bagasse.
The high C/N and L/N ratios of sugarcane bagasse re-
H 2 S + M 2 + ® MS (s) + 2H + (4) quire nutrients N and P from a different source. Another
+
Although two moles of H are released per mole of solid residue, poultry litter, is reported, on the other hand, to
metals precipitated, the impact of Eq. 4 on pH is alleviated be rich in N (C/N ~10-18, L/N ~2-5) and P (%P = 1.2) (de
by the fact that the number of moles of metals is generally Mattos, 2013). The comparison indicates that mixtures of
significantly lower than the number of moles of lactate, ac- sugarcane bagasse and poultry litter, with bagasse provid-
etate and sulfate. ing both liable and durable carbon and poultry litter nutri-
In Eq. 2, lactate is representative of the range of re- ents N and P, may be a promising composition for a low-
duced organic compounds utilizable by SRB. Fermentation cost sulfate-reducing reactor for AMD amelioration.
and sulfate reduction require negative oxidation-reduction
3. Materials and Methods
potential, Eh < -100 mV (Szogi et al., 2004), and pH be-
tween 6 and 9. The pH optimum for anaerobic digestion is 3.1 Sulfate-reducing batch experiments
within the similar range of pH between 6.5 and 8.2.
The kinetic rate-limiting step in these systems (Fig. 2) The batch tests involved placing the contaminant so-
is the hydrolysis of lignocellulosic materials releasing solu- lution in contact with the organic solid materials and SRB,
ble organic molecules (Logan et al., 2005, Hemsi et al., under closed, anaerobic conditions and ambient tempera-
2005, Neculita et al., 2007). Thus, the rates of fermentation, ture of 24 ± 2 °C in the dark. Each reactor was prepared
sulfate reduction and metals precipitation are critically de- with a solid-to-liquid ratio of 1:20 in terms of total mass.
pendent on the rates of solids hydrolysis. For this reason, Thus, in each reactor, 1 L of the synthetic AMD solution
emphasis must be placed on the selection of the ligno- was added to 50 g dry mass of total solids. The selected
cellulosic residues to be emplaced in the organic PRB. solid-to-liquid ratio was lower than used by Zagury et al.
Sugarcane bagasse, an abundant agro-industrial resi- (2006), which performed sulfate-reducing batch tests with
due in central-southern Brazil, has not yet been extensively a 1:4 solid-to-liquid ratio. The ratio used in this study (1:20)
tested as a solid-carbon source for environmental bioreme- was aimed at testing the possibility and extent of reme-
diation. As discussed in de Mattos and Hemsi (2012), sug- diation of the synthetic solution at a low solid-substrate
arcane bagasse is rich in liable to moderately recalcitrant content.
carbon (sugars and cellulose/polyose), but poor in the nutri- Table 1 presents the proportions of solid materials
ents nitrogen and phosphorous. A review by de Mattos employed in each of the three types of batch reactors used
(2013) indicates that sugarcane bagasse has a composition in this study. Each reactor type was prepared in duplicate.
(dry-mass basis) of ~ 40-50% cellulose, ~ 10-25% polyose, These proportions were based on those used by Zagury et
and ~ 20 to 35% lignin, carbon/nitrogen ratio (C/N) ranging al. (2006), but sugarcane bagasse and poultry litter were
from ~ 65-130 and lignin/nitrogen ratio (L/N) ranging from used in this study. Cow manure was included as a bacterial
~ 15-25. The high values of C/N and L/N denote nitro- inoculum for SRB (Choudhary and Sheoran, 2011).
gen-to-carbon deficiency for bacterial degradation and re- The sugarcane bagasse used in this research was ob-
calcitrance (Gibert et al., 2004). tained from the University of São Paulo, and consisted of a
The desirable ranges for the parameters C/N and L/N fresh sample taken from the discarded bagasse generated at
for bacterial decomposition are from 10-20 (Neculita et al., an ethanol and sugar refinery in Piracicaba/São Paulo (Bra-
2007, Khalid et al., 2011) and < ~ 20 (Duryea et al., 1999), zil).
Mattos et al.
Table 1 - Dry mass percentage of solid materials used in reactor formed batch tests for evaluating the sulfate-reducing re-
types 1 to 3. mediation of a simulated mine drainage with an initial pH
ranging from 5.5 to 6.5.
Reactor type 1 2 3
After preparation of the solution, the solution was an-
% (dry mass) alyzed in an ion chromatographer (Dionex, DX-600 IC)
Reactive materials with an anions column IonPac AS-14A and guard column
Sugarcane bagasse 50 30 0
IonPac AG-14A. The initial sulfate concentration in solu-
tion was determined as 5281 mg/L, which was considered
Poultry litter 0 20 50
similar to the expected initial concentration (Table 2).
Base materials The three types of batch reactors were prepared in du-
Sand 30 30 30 plicate, totalizing six sulfate-reduction reactors (1A, 1B,
Limestone 2 2 2 2A, 2B, 3A and 3B). As shown in Fig. 3, low-cost reactors
were assembled using 2-L polyethylene (PET) discarded
Inoculum
bottles, previously washed with a 2-% nitric acid (HNO3)
Cow manure 18 18 18 solution and rinsed with water.
Total 100 100 100 Periodically, liquid samples were collected from each
reactor (2 mL for pH/ORP and 8 mL for chromatography),
using a disposable 10-mL syringe, and pH and oxida-
The chemical composition of the synthetic AMD so-
tion-reduction potential, corrected to Eh, were determined
lution prepared for this study is shown in Table 2. Sulfate
immediately, and sulfate ion (SO42-) concentration was de-
salts were used, avoiding the toxic threshold concentrations
termined using ion chromatography (Dionex, DX-600 IC).
for Zn and Ni (Zagury et al., 2006), and adding the bacterial
Samples for chromatography were pre-treated by filtering
nutrient potassium (K). The initial sulfate concentration
using simple coffee filter papers to remove sugarcane-
was chosen as approximately 5000 mg/L. For example,
bagasse solids from the liquid followed by two consecutive
Zagury et al. (2006) used an initial sulfate concentration of
separations using 98-% hexane in a separation funnel, in
4244 mg/L. Type I deionized water (inorganics
order to separate the organic fraction of the sample (which
< 18.2 MW cm at 25 °C, total organic carbon < 10 ppb, bac-
aggregated to hexane forming a gel) from the inorganic
teria < 1 CFU/mL) (Purelab Option Q-7, Elga) and scien-
fraction of the sample, in which sulfate was determined.
tific-grade reagents and salts were used. Calcium carbonate
Finally, the inorganic fraction was filtered with syringe fil-
(CaCO3) was added when preparing the solution at
ters of acetate-cellulose (0.45 mm and 0.2 mm) mounted in
487.5 mg/L. A volume of ~ 0.37 mL 98-% sulfuric acid
series, diluted by a factor of 50 with deionized water, and
(H2SO4) was added per liter of solution to correct the initial
analyzed by chromatography.
pH to ~ 6.0. This pH of the synthetic solution may be con-
sidered too high to represent AMD, but may be representa- 3.2 Mathematical modeling
tive of a solution reaching the up-gradient side of a PRB
The mathematical model for the sulfate-reducing sys-
after passing through a layer of sand with limestone, which
tem based on solid hydrolysis has been formulated as a sim-
would have increased pH. Also, Waybrant et al. (1998) per-
Table 2 - Ion concentrations and resulting pH of the synthetic

AMD solution used for the batch tests.
Ion Concentration (mg/L)

2+
Ca 390.5
2-
CO3 584.5
Mg2+ 98.9
2+-
Ni 8.64
2+
Zn 8.41
Na2+ 472.8
K+ 2400
2+
Mn 12.46
2-
SO4 5036
Figure 3 - Batch reactors mounted on a shaking wood support,
pH 6 also showing valves used for periodic gas venting.
plification of the sequence shown in Fig. 2 (Hemsi et al., was modeled using an instantaneous-reaction algorithm,
2010), including: (i) solid-substrate hydrolysis, (ii) sulfate based on the presence of lactate (Hemsi et al., 2010):
reduction, release of hydrogen sulfide, and (iii) consump- If [ LA ] > 0 and ([ SO ] - Y3 [ LA ]) ³ 0
tion of metals based on precipitation of metal sulfides.
[ SO ] = [ SO ] - Y3 [ LA ]
3.2.1. Solid substrate hydrolysis [HS] = [HS] + Y4 [ LA ]
Hydrolysis of solid organic substrates (S) and bacte- [ LA ] = 0
rial (X) growth due to fermentation are coupled, with first- If [ LA ] > 0 and ([ SO ] - Y3 [ LA ]) < 0
order or Contois kinetics. Considering Contois kinetics as (8)
1
previously recommended by Hemsi et al. (2010): [ LA ] = [ LA ] - [ SO ]
Y3
ì d[ S] [ S] / [ X ]
= -k c [ X ] Y4
ïï dt K A + ([ S] / [ X ]) [HS] = [HS] + [ SO ]
í (5) Y3
ï d[ X ] = -Y d[ S] - r [ X ] [ SO ] = 0
ïî dt 1
dt
d
where [SO] is the concentration (mg/L) of sulfate (SO42-),

where [X] is the equivalent biomass concentration for de- [HS] is the concentration (mg/L) of hydrogen sulfide in so-
grading bacteria (mg/L), kc = kc(t) is a specific rate of de- lution (H2S), and Y3 and Y4 are mass yield coefficients based
-1
composition for Contois kinetics (d ), KA is an affinity on the reaction.
coefficient (mg/mg), Y1 is a biomass mass yield coefficient,
-1
rd is a decay rate for degrading bacteria (d ), and t is time. 3.2.3. Metal sulfide precipitation
Sulfate-reducing bacteria are capable of utilizing only sim-
Second-order kinetics could be used to model metal
ple soluble organic substrates, such as acetate, lactate, pro-
precipitation with hydrogen sulfide in solution. However,
pionate, or H2. Of all these possibilities, lactate is selected
similarly, the rate-limiting step to metal-sulfide precipita-
as the model utilizable soluble organic substrate for SRB
tion is the release of hydrogen sulfide based on sulfate re-
(Neculita et al., 2007).
duction. Thus, for the precipitation of metal-sulfide, an
Based directly on Eq. 5, the kinetics for the produc- instantaneous process formulation also was employed
tion of lactate in solution is given by: (Hemsi et al., 2010), as follows:
N d[ S ]
d[ LA ] If [HS] > 0 and ([ M] - Y5 [HS]) ³ 0
= -Y2 å i
(6)
dt i =1 dt [ M] = [ M] - Y5 [HS]
where [LA] is the concentration of lactate in solution [ MS] = [ MS] + Y6 [HS]
(mg/L), N is the number of solid decomposable organic [HS] = 0
substrates considered and Y2 is a mass yield coefficient ob-
If [HS] > 0 and ([ M] - Y5 [HS]) < 0 (9)
tained from the chemical reaction. For low values of [S],
the Contois kinetic model approaches first-order kinetics 1
[HS] = [HS] - [ M]
with respect to [S]. Also in this case, the specific rate coeffi- Y5
cient may be considered to decline with time as a result of Y6
the composition change undergone by the solid organic ma- [ MS] = [ MS] + [ M]
Y5
terial:
[ M] = 0
d[ S]
= -k(t)[ S] (7)
dt where [M] is the metal concentration (mg/L), [MS] is the
equivalent metal-sulfide precipitate concentration (mg/L),
where k(t) is a time-declining first-order specific rate coef- and Y5 and Y6 are mass yield coefficients obtained from the
ficient. precipitation reactions. When considering solution H2S,
volatilization to the gas phase may be an important process
3.2.2. Bacterial sulfate reduction
to consider if contact with the atmosphere is significant
Dual-substrate Monod kinetics limited by sulfate and (Hemsi et al., 2005).
lactate could be used to model the growth of SRB and the
3.2.4. Model parameters and solution
production of sulfide from sulfate reduction. However, af-
ter the onset of the experiments, the release of lactate in so- Previous model calibrations using published experi-
lution due to solid-substrate hydrolysis and fermentation mental data from batch and column tests without sugarcane
becomes the rate-limiting step to sulfate reduction (Logan bagasse provided values for the Contois decomposition pa-
et al., 2005, Neculita et al., 2007). Thus, sulfate reduction rameters (kc and KA) for specific organic residues as summa-
Mattos et al.
rized in Table 3. The kc and KA values depend on chemical tively. However, in reactors 3A and 3B, pH values re-
composition, structure (cellulose/lignin) and grain sizes of mained near the initial value of 6. Between 30 and 78 days,
the residues, as well as temperature, pH, Eh, nutrient availa- the pH values started to decrease. However, the pH values
bility, and other environmental characteristics. remained around 7 in reactors of types 1 and 2, reaching
The mathematical model for cellulose degradation in- values below 6 only for reactors of type 3. The behavior of
cludes a set of three adjustable parameters, which are kc, KA, pH versus time in these reactors depends on solid-substrate
and rd, and an unknown initial condition, the initial ratio decomposition, which releases H+ (Eq. 1), decreasing pH,
[S]/[X]. For the parameter calibrations shown in Table 3, and sulfate reduction, which releases acetate and alkalinity
the affinity coefficient, KA, was maintained at 30 mg/mg, (Eq. 2) and tends to increase pH. In general, the sequential
whereas the specific rate of decomposition, kc, was varied processes are expected to result in overall pH increase
reflecting higher or lower material degradability. In these (Blowes et al., 2000), as observed in batch tests of types 1
simulations (Table 3), the initial equivalent biomass con- and 2 of this study.
centration of degrading bacteria (X) was assumed as 0.5%
of the initial equivalent concentration of the solid cellulosic 4.1.2. Sulfate concentrations and rates of sulfate
substrate being degraded, and the decay rate for degrading reduction
bacteria, rd, was assumed as 10% of the specific rate of de-
composition for Contois kinetics. Maintaining KA at Figures 4b to d present the results in terms of sulfate
30 mg/mg allows for the comparison of kc adjusted for dif- concentrations versus time for each reactor type. The com-
ferent materials, as shown in Table 3. parison between these results shows that the most signifi-
2-
Initial parameter values for sugarcane bagasse were cant decline in sulfate (SO4 ) occurred in reactors 2A and
based on parameters shown in Table 3, and adjusted on the 2B between 0 and 40 days. In contrast, reactors of type 3
basis of the comparisons between simulated results and ex- showed a very slight variation in sulfate concentrations
perimental data. over time. Decreasing sulfate concentrations over time in
In terms of the model simulations, the ordinary differ- reactors of types 1 and 2 were corroborated by two aspects,
ential equations of the kinetic model (Eqs. 5 and 6) were intense sulfidric gas (H2S) odor in these reactors and a
numerically integrated in time using the Runge-Kutta- change in color of the solid mixture in these reactors, which
Fehlberg method coupled to the algorithms for the instanta- went from greenish brown to black. Reactors of type 3 did
neous reactions of Eqs. 8 and 9. not exhibit these changes in odor and color to a similar ex-
tent. Neculita et al. (2007) and Waybrant et al. (1998)
4. Results and Discussion tested poultry and sheep manures, respectively, as carbon
and nitrogen sources for sulfate-reducing batch tests, and
4.1 Sulfate-reducing batch experiments
concluded that the residues were poor substrates for sulfate
The six batch reactors used in this study consisted of reduction based on the lack of total organic carbon (TOC).
three reactor types (1, 2 and 3), as described in Table 1, pre-
After the total testing period of 78 days, the sulfate
pared in duplicate (for each type, duplicates were desig-
concentrations in solution decreased, on average, by 1170,
nated as A and B). Liquid samples from the solution in each
1600, and 1060 mg sulfate/L in reactors of types 1, 2 and 3,
batch reactor were collected every 7 to 15 days during the
respectively.
total testing period of 78 days.
The average rates of sulfate reduction over time were
4.1.1. pH
calculated for each interval between measured data, and are
Figure 4a presents the pH results in each reactor ver- shown in Fig. 5 for each reactor, with rate values (bars)
sus time. After 30 days, pH values had increased from 6 to plotted at the center of each time interval (e.g., plotted at
~ 7.5 to 8 in reactors 1A and 1B, and 2A and 2B, respec- time 67.5 days, for interval between 58 and 77 days).
Table 3 - Contois kinetic decomposition parameters based on model calibrations using experimental data (temperature = 25 °C).
-1
Material kc (d ) KA (-) References
Experiment Modeling
Wood chips 0.4 30 Waybrant et al. (1998), batch tests Hemsi et al. (2005)
Leaf mulch 0.6 30
Sawdust 0.8 30
Corn stover 0.5 30 Figueroa et al. (2007), column tests Hemsi et al. (2010)
Walnut shells 0.05 30
Figure 4 - Batch tests results: (a) pH in all reactors, (b) sulfate concentrations in reactors of type 1, (c) sulfate concentrations in reactors
of type 2, and (d) sulfate concentrations in reactors of type 3.
Figure 5 - Average sulfate-reduction rates (rsulfate) over time for batch reactors: (a) type 1, (b) type 2, and (c) type 3 reactors.
A comparison among the results in Fig. 5 reveals that may be explained as analytical measurement imprecision,
the calculated rates of sulfate reduction over time are in or actual release of sulfate by the solid materials.
closer agreement for the duplicate reactors of type 2 (2A 4.1.3. Eh
and 2B) than for either one of the other two types of reac-
tors. Also, average rates of sulfate reduction for reactors of Figure 6 presents values of Eh plotted against testing
time for different batch reactors. For all reactors, the Eh
type 2 were > 30 mg sulfate/L-day, peaking at ~40-50 mg
value was measured at 78 days. The data for these reactors
sulfate/L-day for times up to 38 days, which are greater indicate reactors of types 1 and 2 with 78-day Eh values less
than the values achieved in reactors of types 1 and 3 during than -100 mV, whereas reactors of type 3 with 78-day Eh
the entire testing period. Greater variability is observed for values of ~ 0.
reactors of types 1 and 3. Negative values of rates of sulfate The Eh in Fig. 6 is the corrected value, i.e., relative to
reduction, indicating increase in sulfate concentrations, the standard hydrogen electrode. Such values were calcu-
Mattos et al.
The bacterial activity of organic decomposition/fer-

mentation and sulfate reduction was annulated in reactor
2F, since the value of Eh, measured at 15 days from the on-
set of this reactor, was equal to +230 mV (Fig. 6), indicat-
ing aerobic conditions and the absence of sulfate reduction
that lowered Eh in reactors of type 2. Thus, the concentra-
tion of glutaraldehyde utilized was adequate to promote the
biocide effect in this abiotic control.
In contrast, the un-innoculated reactor 2E displayed a
15-day value of Eh similar to the values for reactors 2C and
2D (at 15 days), indicating that the addition of cow manure
in reactors 2C and 2D as innoculum material for SRB may
not be required, and that sulfate reducing bacteria may be
present in this system via the sugarcane bagasse and poultry
litter. In addition, as shown in Fig. 7, reactor 2E displayed a
black to dark grey color, whereas the abiotic reactor 2F re-
Figure 6 - Oxidation-reduction potential values relative to the mained of a light brown color, basically the same color of
standard hydrogen electrode in reactors 1A, 1B, 2A, 2B, 3A and the reactors right after assemblage. Also, reactor 2E emit-
3B at 78 days, in reactors 2C and 2D from zero to 28 days, and in ted a strong hydrogen sulfide-gas odor, whereas reactor 2F
reactors 2E and 2F at 15 days. did not exhibit hydrogen sulfide-gas odor.
4.2 Mathematical modeling
lated from the direct ORP measurements obtained with a
The mathematical model was used in batch mode (hy-
Ag/AgCl half-cell electrode (Hanna HI-3131B), according
drostatic conditions) in order to simulate sulfate dynamics
to (Stumm and Morgan, 1996, Szogi et al., 2004):
in batch reactors 2A and 2B, which were the reactors that
E h = ORP + 200( mV) (10) better sustained bacterial sulfate reduction in this study.
Sulfate reduction was modeled as an instantaneous process
Values of Eh less than -100 mV are indicative of sul-
limited by lactate from hydrolysis/decomposition of cellu-
fate-reducing conditions in solution, whereas values ~ 0 are
lose based on Contois kinetics.
not (Szogi et al., 2004).
In order to follow the evolution of Eh with time from 4.2.1. Initial conditions
the onset of the batch experiments, two additional reactors,
The multi-species model is a set of ordinary differen-
2C and 2D, were assembled identically to reactors 2A and
tial equations that require a set of initial conditions to
2B. The values of Eh in these reactors were measured from 0
evolve in time. The initial concentrations for sulfate
to 28 days, as shown in Fig. 6.
([SO] = 5200 mg/L), nickel and zinc ([M] = 17 mg/L), lac-
The results for reactors 2C and 2D (Fig. 6) indicate Eh tate ([LA] = 0), and dissolved hydrogen sulfide ([HS] = 0)
values declining from > 200 mV (aerobic conditions) to were assigned. The initial equivalent concentration of cel-
~ -100 mV within 7 days, i.e., denoting oxygen depletion
followed by establishment of sulfate-reducing conditions
after 7 days. Also, the results for the duplicate reactors 2C
and 2D were similar.
From 7 to 28 days, the values of Eh remained at
-100 mV, indicating sustained sulfate reduction during this
period. The final values were in agreement with the values
of Eh measured at 78 days for reactors 2A and 2B, which
also were -100 mV.
Finally, two additional batch reactors were assembled
to further study Eh. These reactors were control reactors,
designated 2E and 2F, prepared almost identically to reac-
tors of type 2, with the following differences: reactors 2E
and 2F did not contain cow manure and reactor 2F con-
tained 50 mL of a 2-% (by volume) Glutaraldehyde solu-
tion. Thus, reactor 2E was intended to be an un-inoculated
control, whereas reactor 2F was intended to be an abiotic Figure 7 - Visual aspect at 15 days for batch reactors: (a) 2E, and
control. (b) 2F.
lulose, [S], in a reactor, coming from the packed sugarcane

bagasse was unknown. The total dry mass of sugarcane ba-
gasse in type-2 reactors was 30% (by mass) of 50 g/L, i.e.,
15 g/L per reactor. Based on the compositions reviewed by
de Mattos (2013), cellulose and polyose contents may be
50-75% by dry mass in sugarcane bagasse. Discounting
cellulose that may be protected by lignin, a 50-% degra-
dable polysaccharide fraction, on a dry-mass basis, was
considered in this stydy, corresponding to initial
[S] = 7500 mg/L. The initial equivalent concentration [X]
was thus assumed = 37.5 mg/L, as per Hemsi et al. (2010).
4.2.2. Modeling results

The modeling results depend critically on the parame-
ter values assumed for degradation of cellulose in sugar-
cane bagasse. The Contois specific degradation rate (Eq. 5),
kc, was chosen from 0.5 to 1.0 d-1 with affinity coefficient,
KA, of 30 mg/mg (Table 3).
Simulation case 1, with kc = 0.5 d-1, considering cellu-
lose in sugarcane bagasse to degrade similarly to that in
wood chips or leaf mulch, resulted in significant under-
prediction of the rates of sulfate reduction for time 0-78 Figure 8 - Modeling results for simulation cases 1, 2 and 3: (a)
days. Then, simulation case 2, with kc = 1.0 d-1, considering sulfate concentrations over time; (b) rates of sulfate reduction
a degradability close to that of very fine-graded sawdust over time.
(Table 3) resulted in slight over-prediction of the rates of
sulfate reduction from 50-70 days (Figs. 8a, b).
of sulfate reduction in type-2 reactors were maintained
Better modeled results were obtained for simulation
throughout the 78-day duration of the duplicate tests. Also,
case 3, in which cellulose in sugarcane bagasse was consid-
pH values increased from ~ 6 to 8, and Eh achieved < -100
ered to be as degradable as that in wood chips or leaf mulch,
mV, indicating anaerobic sulfate-reducing conditions.
but highly-degradable sugars were included, i.e.,
Mathematical modeling performed on these experiments
7500 mg/L of cellulose with kc = 0.5 d-1 (= Case 1) plus an
with a previously-tested model resulted in better results
additional 1500 mg/L of a degradable material with when the sugarcane bagasse was modeled as comprising
kc = 5.0 d-1. This highly-degradable material may be thought cellulose with a Contois specific degradation rate similar to
of as being soluble sugars observable in sugarcane bagasse. that of wood chips and leaf mulch added to a fraction of eas-
The results in Fig. 8 indicate that the two-substrate ily-degradable substrate, to account for the presence of sol-
model (N = 2 in Eq. 6) with an easily-degradable fraction uble sugars in sugarcane bagasse. The results from the
(kc = 5.0 d-1) allowed simulating the initial high rates of sul- batch experiments performed in this study provide an indi-
fate reduction that were verified in the experiments. These cation for assessing the performance of sugarcane bagasse
results indicate that, when trying to simulate the perfor- as a carbon source for sustained sulfate-reduction in reac-
mance of a system using sugarcane bagasse, considering an tors treating AMD, and the modeling results may be useful
easily-degradable substrate in addition to cellulose may be for trying to estimate the performance of these systems.
important. However, the limited time scales in this study relative to the
actual field application must be considered.
5. Conclusions
The use of the lignocellulosic residue sugarcane ba- Acknowledgments
gasse as carbon source in organic permeable reactive barri-
The authors acknowledge the financial support from
ers for the environmental remediation of mining effluents
FAPESP, Project number 2012/08797-2. Thanks to Prof.
was tested at the laboratory bench scale. Microcosm batch
Dr. Cláudio Aguiar (Esalq/USP) for providing bagasse
experiments using sugarcane bagasse with and without
samples.
poultry-litter amendments for nutrients N and P were per-
formed. The experiments combining sugarcane bagasse References
and poultry litter (reactors type 2) achieved rates of sulfate
reduction between ~ 30 to 45 mg sulfate/L-day, outper- Bechard, G.; Yamazaki, H.; Gould, W.D. & Bedard, P.
forming experiments based on single substrates. The rates (1994). Use of cellulosic substrates for the microbial
Mattos et al.
treatment of acid mine drainage. Journal of Environ- Figueroa, L.; Miller, A.; Zaluski, M. & Bless, D. (2007).
mental Quality, 23(1):111-116. Evaluation of a two-stage passive treatment approach
Benner, S.G.; Blowes, D.W.; Gould, W.D.; Herbert, R.B. for mining influenced waters. In: 2007 National Meet-
& Ptacek, C.J. (1999). Geochemistry of a permeable re- ing of the American Society of Mining and Reclama-
active barrier for metals and acid mine drainage. Envi- tion, R.I. Barnhisel (ed), Gillette, WY.
ronmental Science and Technology, 33(16):2793-2799. Gibert, O.; de Pablo, J.; Cortina, J.L. & Ayora, C. (2004).
Benner, S.G.; Blowes, D.W.; Ptacek, C.J. & Mayer, K.U. Chemical characterization of natural organic substrates
(2002). Rates of sulfate reduction and metal sulfide pre- for biological mitigation of acid mine drainage. Water
cipitation in a permeable reactive barrier. Applied Geo- Research, 38(19):4186-4196.
chemistry, 17:301-320. Gillham, R.W. & O’Hannesin, S.F. (1994). Enhanced deg-
Blowes, D.W.; Ptacek, C.J.; Benner, S.G.; McRae, C.W. & radation of halogenated aliphatics by zero-valent iron.
Puls, R.W. (2000). Treatment of dissolved metals and Groundwater, 32(6):958-967.
nutrients using permeable reactive barriers. Journal of Groudev, S.; Nicolova, M.; Spasova, I. & Schutte, R.
Contaminant Hydrology, 45(1):123-137. (2003). Treatment of waters from a copper mine by
CETESB (2013). Companhia Ambiental do Estado de São means of a permeable reactive barrier. Fifty Years of
Paulo. Relação de Áreas Contaminadas e Realibilitadas the University of Mining and Geology “St. Ivan Rilski”,
no Estado de São Paulo. http://areascontamina- University of Mining and Geology, Sofia, 46(II):229-
das.cetesb.sp.gov.br/wp-content/uploads/sites/45/2013 231.
/11/texto-explicativo.pdf. Grubb, D.G.; Landers, D.G. & Hernandez, M. (2000). Uti-
Chang, I.S.; Shin, P.K. & Kim, B.H. (2000). Biological lization of sugarcane bagasse to treat acid mine drain-
treatment of acid mine drainage under sulphate- age. Proc. GeoEng 2000, Melbourne, Australia.
reducing conditions with solid waste materials as sub-
Hallberg, K. & Johnson, D. (2005). Microbiology of a wet-
strates. Water Research, 34(4):1269-1277.
land ecosystem constructed to remediate mine drainage
Choudhary, R.P & Sheoran, A.S. (2011). Comparative from a heavy metal mine. Science of the Total Environ-
study of cellulose waste versus organic waste as sub- ment, 338(1-2):53-66.
strate in a sulfate reducing bioreactor. Bioresource
Hammack, R.W. & Edenborn, H.M. (1992). The removal
Technology, 102:4319-4324.
of nickel from mine waters using bacterial sulfate re-
Cocos, I.A.; Zagury, G.J.; Clément, B. & Samson, R.
duction. Applied Microbiology and Biotechnology,
(2002). Multiple factor design for reactive mixture se-
37(5):674-678.
lection for use in reactive walls in mine drainage treat-
ment. Water Research, 32:167-177. Hemsi, P.S.; Shackelford, C.D. & Figueroa, L.A. (2005).
Modeling the influence of decomposing organic solids
de Mattos, R.C. (2013). Utilização dos Resíduos Lignoce-
on sulfate reduction rates for iron precipitation. Envi-
lulósicos Bagaço de Cana-de-Açúcar e “Cama de Aviá-
ronmental Science and Technology, 39(9):3215-3225.
rio” como Substratos Orgânicos em Ensaios de
Laboratório de Redução Bacteriana de Sulfato. MSc. Hemsi, P.S.; Shackelford, C.D. & Figueroa, L.A. (2010).
Thesis (in portuguese), Instituto Tecnológico de Aero- Calibration of reactive transport models for remediation
náutica, São José dos Campos, Brazil. of mine drainage in solid substrate bio-columns. Jour-
de Mattos, R.C. & Hemsi, P.S. (2012). Estimativa do poten- nal of Environmental Engineering, 136(9):914-925.
cial de utilização de resíduos agrícolas em barreiras Khalid, A.; Arshad, M.; Anjum, M.; Mahmood, T. &
reativas permeáveis com base em sua composição e Dawson, L. (2011). The anaerobic digestion of solid or-
degradabilidade. In: XVI Brazilian Congress on Soil ganic waste. Waste Management, 31:1737-1744.
Mechanics and Geotehnical Engineering, Krebs, A.J. (2010). Monitoramento de bocas de minas
COBRAMSEG 2012, Porto de Galinhas, Brazil. abandonadas com drenagens ácidas na área correspon-
Duryea, M.L.; English, R.J. & Hermanesen, L.A. (1999). A dente à Bacia Carbonífera de Santa Catarina. In: Brazil-
comparison of landscape mulches: Chemical, ian Goundwater Congress, 2010, São Luiz, Brazil.
allelopathic, and decomposition properties. Journal of Logan, M.V.; Reardon, K.F.; Figueroa, L.A.; McLain, J. &
Arboriculture, 25:88-97. Ahmann, D.M. (2005). Microbial community activities
Dvorak, D.H.; Hedin, R.S.; Edenborn, H.M. & McIntire, during establishment, performance, and decline of
P.E. (1992). Treatment of metal-contaminated water us- bench-scale passive treatment systems for mine drain-
ing bacterial sulfate reduction: Results from pilot-scale age. Water Research, 39(18):4537-4551.
reactors. Biotechnology and Bioengineering, Neculita, C.M.; Zagury, G.J. & Bussière, B. (2007). Pas-
40(5):609-616. sive treatment of acid mine drainage in bioreactors
FEAM (2011). Fundação Estadual do Meio Ambiente. In- using sulfate-reducing bacteria: Critical review and re-
ventário de Áreas Impactadas pela Mineração do Esta- search needs. Journal of Environmental Quality,
do de Minas Gerais, 2011. FEAM-GESAD-RT-3-2011. 36(1):1-16.
Rubio, J.; Silva, R. & Silveira, A. (2008). Técnicas para Voudrias, E.A. (2001). Pump-and-treat remediation of
tratamento e alternativas de reuso de águas ácidas de groundwater contaminated with hazardous waste: Can
minas de carvão. In: Proc. VI Internacional Symposium it really be achieved? Global Nest Journal, 3(1):1-10.
on Environmental Quality, ABES-RS & PUCRS, Porto Wakao, N.; Takahashi, T.; Sakurai, Y. & Shiota, H. (1979).
Alegre, Brazil. A treatment of acid mine water using sulfate-reducing
Stumm, W. & Morgan, J.J. (1996). Aquatic Chemistry: bacteria. Journal of Fermentation Technology,
Chemical Equilibria and Rates in Natural Waters. John 57(5):445-452.
Wiley & Sons, NY. Waybrant, K.R.; Blowes, D.W. & Ptacek, C.J. (1998).
Szogi, A.A.; Hunt, P.G.; Sadler, E.J. & Evans, D.E. (2004). Selection of reactive mixtures for use in permeable re-
Characterization of oxidation- reduction processes in active walls for treatment of mine drainage. Environ-
constructed wetlands for swine wastewater treatment. mental Science & Technology, 32(13):1972-1979.
Applied Engineering in Agriculture, 20(2):189-200. Western Governors’ Association, WGA (1998). Cleaning
up Abandon Mines: A Western Partnership. Electronic
Tuttle, J.H.; Dugan, P.R. & Randles, C.I. (1969). Microbial document. http://www.westgov.org/wga/publicat.
sulfate reduction and its potential utility as an acid mine Whitehead, P.G.; Cosby, B.J. & Prior, H. (2005). The
water pollution abatement procedure. Applied Microbi- Wheal Jane wetlands model for bioremediation of acid
ology, 17(2):297-302. mine drainage. Science of the Total Environment,
U.S. Air Force Center for Engineering and the Environ- 338(1-2):125-135.
ment (2008). Final technical protocol for enhanced an- Wildeman, T.R.; Gusek, J.J.; Miller, A. & Fricke, J. (1997).
aerobic bioremediation using permeable mulch Metals, sulfur, and carbon balance in a pilot reactor
biowalls and bioreactors. Washington (DC), treating lead in water. In: In Situ and On-Site Bio-
http://www.cluin.org/download/techfocus/prb/Final- remediation, Battelle Press, Columbus.
Biowall-Protocol-05-08.pdf. Zagury, G.J.; Kulnieks, V. & Neculita, C.M. (2006). Char-
U.S. Environmental Protection Agency (1998). Permeable acterization and reactivity assessment of organic sub-
Reactive Barrier Technologies for Contaminant strates for sulphate-reducing bacteria in acid mine
Remediation. Report EPA/600/R-98/125. drainage treatment. Chemosphere, 64:944-954.
Apparent Pressures on Multi-Propped Retaining Walls in
Soils Under Drained Conditions with Shallow Water Table
L. Andrade Viana, N.M. da Costa Guerra
Abstract. In the design of flexible retaining walls supported by several levels of struts, apparent design diagrams are
frequently used, particularly to predict the loads on the struts. In the case of sandy soils, the traditional design diagrams do
not consider water table. The present work deals with apparent design diagrams for cases considering water table. A brief
analysis of the problem based on published information is presented. Stress-strain finite element analyses are applied to a
case study, comparing situations without water table with others where the water table is assumed. For the cases where
water table is assumed two types of conditions are considered: those where seepage is allowed and those where it is not. A
parametric analysis is performed to study the effects on soil pressures and on apparent pressures of geometrical and
mechanical parameters of the soil and structure. From the analysis of the results a proposal for the apparent diagrams for the
design of retaining walls performed in soils under drained conditions with water table above the bottom of the excavation is
presented.
Keywords: multi-propped retaining walls, apparent earth pressures, water pressures, seepage.
1. Introduction walls supported by several levels of struts. When the

excavation is carried out below the water table, without
Numerical methods are commonly used in the design
lowering it, and the wall is impermeable, a substantial part
of multi-propped retaining walls. They allow the stress-
of pressures on the wall will be due to water. When seepage
strain step-by-step analysis of the construction procedure
is not allowed (because the wall was extended to an imper-
and therefore the prediction of the displacements induced
meable stratum) the structure has to be designed to with-
in the supported soil, the forces and bending moments on
stand the hydrostatic pressures. Strom & Ebeling (2001)
the wall and the loads on the struts. However, its use needs
proposed an apparent diagram considering both soil and
the pre-definition of several geometrical and structural
water pressures (see Fig. 1).
characteristics. In the case of the wall and struts, such
pre-definition needs the prediction of the loads installed, This diagram is based on Terzaghi and Peck’s, using
which is frequently done using apparent design diagrams. an average effective unit weight to which the pore-water
The most known and used apparent design diagrams pressure is added. The average effective unit weight, ge, can
are those proposed by Terzaghi & Peck (1967). In case of be calculated by either of two methods, in which the soil
sandy soils, the diagram is rectangular, with horizontal above the water table is considered with the moist unit
stress given by: weight, g, and the soil below the water table is considered
s H = 0.65K a gH (1) with the buoyant unit weight, g¢.
In method 1 (Strom & Ebeling, 2001), ge is the
in which Ka is the Rankine active earth pressure coefficient,
weighted average using depths H - zw and zw shown in Fig. 2:
g is the unit weight of the soil and H is the depth of the exca-
vation. g ¢( H - z w ) + gz w
ge = (2)
These diagrams were the result of monitoring the H
loads on the struts of several multi-propped flexible retain-
ing walls throughout a few decades of the twentieth cen- In method 2 (Ebeling & Morrison, 1992), ge is the weighted
tury. In most cases, for drained situations, the adopted average using A1 and A2 also shown in Fig. 2:
construction procedure allowed lowering the water table in
cases where it was above the bottom of the excavation and g ¢A1 + gA2 g ¢( H - z w ) 2 + gz w2
ge = = (3)
therefore no water pressures existed on the retaining walls. A1 + A2 H2
There is no generally accepted way to take into ac-
count in the apparent diagrams the existence of water table In the particular case of water table on the ground sur-
above the bottom of the excavation in flexible retaining face the application of methods 1 and 2 results in the same
Laís Andrade Viana, MSc, Departamento de Engenharia Civil, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, Lisboa, Portugal. e-mail: l.viana@cam-
pus.fct.unl.pt.
Nuno M. da Costa Guerra, Associate Professor, UNIC, Departamento de Engenharia Civil, Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa, Lisboa, Portu-
gal. e-mail: [email protected].
Submitted on July 16, 2015; Final Acceptance on November 2, 2015; Discussion open until April 30, 2016.
Andrade Viana & da Costa Guerra
ì 0.65K a gH , z £ zc
sH = í (5)
î 0.65K a g e H + g w ( z - z w ), z > z c
In cases where seepage is allowed around the wall
into the excavation, with steady-state seepage conditions,
there are seepage forces on the ground. The diagrams,
adapted from the previous, consider, instead of the buoyant
unit weight g¢, this unit weight plus the seepage forces. In
addition, the water pressures will no longer be hydrostatic
and u (see Fig. 1) will therefore not be gw(H - zw) and the lin-
ear distribution will be a simplification. It is possible to ver-
ify that this proposal leads to lower total pressures on the
wall when compared to the ones from the hydrostatic case,
and therefore only the hydrostatic ones will be shown in the
paper, for comparison with the results obtained from the
Figure 1 - Apparent diagrams for sandy soils for no water table numerical analyses.
(Terzaghi & Peck, 1967) and for water table above the bottom of The diagrams described above have not been con-
the excavation (Strom & Ebeling, 2001) and proposed correction firmed either by monitoring or numerical analyses. There-
of this method, given by Eq. 5. fore, the present work intends to numerically obtain appar-
ent diagrams, for cases considering the presence of the
water table in the soil. It is somehow interesting that a nu-
merical procedure is felt useful to obtain diagrams of the
same type as others that were probably only developed be-
cause of the lack of such advanced techniques, at the time.
2. Modelling
Stress-strain finite element analyses were performed
for this study, using the finite element program Plaxis
(2014). The problem consists of a symmetrical excavation
supported by a multi-propped retaining wall and plane
strain conditions are assumed.
Five scenarios were analyzed and they are schemati-
cally shown in Fig. 3:
• Scenario A: no water table, the wall does not reach the
rigid impermeable stratum.
• Scenario B: water table at depth zw; the wall does not
Figure 2 - Geometric parameters for determining the effective
reach the rigid impermeable stratum and the water level
unit weight for partially submerged soils. Angle a is the angle de-
fining the failure wedge. is lowered inside the excavation by continuous pumping.
Seepage modelling is performed.
• Scenario C: no water table, the wall reaches the rigid im-
solution and the average effective unit weight is equal to the permeable stratum and it can not rotate at the toe.
buoyant unit weight. • Scenario D: water table at depth zw; the wall reaches the
rigid impermeable stratum and it can not rotate at the toe.
The resulting diagram implies the consideration up to
No seepage is allowed.
a certain depth, zc, given by:
• Scenario E: water table at depth zw; the wall reaches an al-
most impermeable stratum and the water level is lowered
g -ge
z c = 0.65K a H + zw (4) inside the excavation by continuous pumping. Seepage
gw modelling is performed, even though seepage occurs
mainly in the stratum with very low permeability.
of a total pressure that is lower than the one that would be The excavation was modelled in alternating stages of
considered if there were no water table, which does not excavation and installation of strut levels. The excavation
seem to make much sense. Therefore, a proposal that seeks stages also consider that the water level is lowered inside
to rectify this aspect is also shown in Fig. 1 and the horizon- the excavation (scenarios B, D and E) and the calculation of
tal pressures take the following values (see Andrade Viana, the associated seepage (scenarios B and E). It should also
2014): be mentioned that the soil was assumed homogeneous and
Apparent Pressures on Multi-Propped Retaining Walls in Soils Under Drained Conditions With Shallow Water Table
Figure 3 - Schematic representation of scenarios A, B, C, D and E.
isotropic. If real conditions of the soil involved heterogene- stiffnesses are required: E50,ref, Eur,ref and Eoed,ref. As average
ity and (or) anisotropy, the groundwater flow could have a values for various soil types, Eur,ref = 3E50,ref and Eoed,ref = E50,ref
significant effect on the pressures on the wall (Kaiser & are suggested as default settings (Plaxis, 2014).
Hewitt, 1982).
3. Case Study
The wall and the struts were admitted linear elastic.
The soil was modelled by the ”Hardening Soil Model” The case study has the following values for the geo-
(Schanz et al., 1999). The HSM is an elastoplastic type of metrical parameters: B/2 = 5 m; e = 0.4 m; H = 10 m;
hyperbolic model, formulated in the framework of shear D = 6 m; De = 4 m; h1 = 1 m; h2 = h3 = h4 = 3 m; zw = 0 m. The
hardening plasticity. The model also involves compression excavation was modelled in stages, for a total of nine
hardening to simulate irreversible compaction of soil under stages. It was performed in four levels of excavations and
primary compression (Plaxis, 2014). the wall is supported by three levels of struts.
Limiting states of stress are described by means of the The soil was considered as a sandy soil with drained
shear strength angle, f¢, the effective cohesion, c¢, and the behaviour. In scenarios A and C the moist unit weight was
3
dilatancy angle, y. Soil stiffness is described by using three set as 20 kN/m . In scenarios B, D and E the water table was
different input stiffnesses: the secant triaxial loading stiff- considered on the ground surface and the saturated unit
3
ness to half of the tensile strength, E50, the triaxial unloading weight was also taken equal to 20 kN/m .
and reloading stiffness, Eur, and the oedometer loading stiff- The chosen stiffness parameters intend to portray the
ness, Eoed. These stiffnesses depend on stress-state accord- behaviour of a medium sand: E50,ref = 25000 kPa,
ing to the relations: Eur,ref = 75000 kPa, Eoed,ref = 25000 kPa and m = 0.5. It was as-
sumed a shear strength angle, f¢, of 30°, a reference effec-
m
æ s¢3 + c¢ref cot f¢ ö tive cohesion, c¢ref, of 1 kPa and a dilatancy angle, y, of 1°.
E 50 = E 50 , ref ç ÷ (6) The at rest earth pressure coefficient, K0, was taken equal to
ç p¢ref + c¢ref cot f¢ ÷
è ø 0.5. The soil was considered with homogeneous hydraulic
m characteristics (permeability coefficient in the horizontal
æ s¢3 + c¢ref cot f¢ ö
E ur = E ur , ref ç ÷ (7) direction equal to the permeability coefficient in the verti-
ç p¢ref + c¢ref cot f¢ ÷ cal direction, kx = ky). This coefficient was taken equal to
è ø -3
10 m/s. In scenario E the soil below the wall was consid-
m
æ s1¢ + c¢ref cot f¢ ö ered with a very low permeability coefficient, in both direc-
E oed = E oed , ref ç ÷ -9
(8) tions, of 10 m/s. All other parameters were considered
ç p¢ref + c¢ref cot f¢ ÷
è ø equal to the ones of the soil above.
Interfaces were modelled by joint elements and its
in which m is a power that expresses the dependence of the strength and stiffness properties are defined by the strength
soil stiffness with the stress state in the soil, p¢ref is a refer- reduction factor, Rinter (Plaxis, 2014), which was set as 0.67.
ence stress, usually taken equal to 100 kPa, s¢1 is the major The wall was considered built before the excavation,
principal stress and s¢3 is the minor principal stress. In order with elastic characteristics of a reinforced concrete wall
2
to characterize the soil stiffness the following reference with a bending stiffness, EI, of 160000 kNm /m. Given that
it is an ”infinite” wall, since the analysis is in plane strain and, thus, of the surface of the supported soil, are particu-
conditions, a null Poisson’s ratio, n, was adopted. larly high. It is interesting to notice that the displacements
The wall weight per unit area was considered equal to for this scenario (B) are greater than the ones for scenario E,
10 kN/m/m. where the wall toe can move and water pressures on the
The stiffness of the struts was chosen by performing right side are hydrostatic and larger than in scenario B.
preliminary calculations using scenarios A and C, so that Figure 6 shows the plastic points obtained after the
pressures on the wall and apparent pressures inferred from last excavation stage for all scenarios that were analised. It
the maximum loads on the struts roughly correspond to the is possible to observe that in scenarios A and C there are
apparent diagrams of Terzaghi and Peck with horizontal few failure points which occur specially in the soil-struc-
stresses given by Eq. 1. The same characteristics were ture interface. In scenario B failure points occur mainly in
adopted in scenarios B, D and E. Thus, the struts were as- the soil mass, showing the beginning of a sliding surface; it
sumed as linear elastic with axial stiffness, EA, of should be noted that this occurs even though the global
120000 kN/m. safety factor against the hydraulic uplift is, in this case, rel-
Figure 4 shows the distribution of hydraulic head in atively high (FS = 1.43). In scenario D failure points occur
the soil mass after the last excavation stage in scenarios B especially in the soil-structure interface and in a band of
and E. The distributions of hydraulic head correspond to the soil inside the excavation, without any hint of a sliding sur-
expected and it is possible to note that the lines defined by face. In scenario E there is, again, an intermediate situation
color changes correspond to equipotential lines. between scenarios B and D, with failure points in soil-
Figure 5 shows the displacements obtained for the structure interface and also the beginning of a potential
wall and for the surface of the supported soil mass after the sliding surface in the soil on the right side of the wall.
last stage of excavation. It can be observed that the dis- Figure 7 shows the results obtained after the last stage
placement values corresponding to scenarios with water (B, of excavation for total, effective and pore water pressures
D and E) are substantially higher than those associated to on the wall. It also shows the apparent pressures inferred
scenarios without water (A and C). In particular, in scenario from the maximum loads on the struts. The apparent dia-
B, wherein the wall is not based on rigid impermeable stra- gram of Terzaghi and Peck (Eq. 1) and the apparent dia-
tum and seepage is allowed, the displacements of the wall gram given by Eq. 5 are also shown in Figs. 7(a) and 7(b),
for comparison with the results obtained.
Scenarios A and C have equal total and effective pres-
sures as there are no pore water pressures. In scenario B
pore water pressures deviate progressively from hydro-
static pressures, while in scenarios D and E they are equal
to hydrostatic pressures. In fact, in scenario E, the seepage
is negligible and therefore the pore water pressures are very
similar to hydrostatic pressures.
Figure 4 - Distribution of water head after the last excavation Figure 5 - Displacements of the wall and of the surface of the sup-
stage. ported soil mass after the last excavation stage.
Figure 6 - Failure points after the last excavation stage.
Figure 7 - Pressures on the flexible retaining wall after the last excavation stage (symbols) and apparent pressures inferred from the max-
imum loads on the struts (lines; Fig. 7(a)).
In scenarios A and C the distribution of total stresses the wall at the same depth, while, from that same depth, the
on the wall approaches the apparent diagrams of Terzaghi apparent pressures become greater than the distribution of
and Peck. Indeed, as referred, the axial stiffness adopted for total stresses on the wall.
struts was chosen so that the distribution of pressures on the In scenario D total stresses on the wall are also
wall for these scenarios would be approximately constant greater than the apparent diagram given by Eq. 5. How-
in depth and equal to the value obtained from Eq. 1. ever, it can be noted that in this scenario the distribution of
In scenario B total stresses on the wall are greater than total stresses on the wall is closer to this apparent diagram
those corresponding to the apparent diagram given by than in scenario B. Thus, contrary to what is verified in the
Eq. 5. This diagram is, up to a certain depth, similar to the situation in which the existence of water level is not con-
obtained distribution of effective stresses, despite this ap- sidered, the existence of a rigid impermeable stratum
parent diagram being established in terms of total stresses. shows considerable influence on the distribution of total
It can be seen that the apparent pressures corresponding to pressures on the wall. It can be seen that in this scenario
the last level of struts are very close to the ones correspond- the apparent pressures corresponding to the last level of
ing to the second level of struts. In fact, the apparent pres- struts are much lower than the distribution of total stresses
sures corresponding to the last level of struts are, up to a on the wall at the same depth. This is probably due to the
certain depth, lower than the distribution of total stresses on fact that the wall is embedded in a more competent stra-
tum, which is responsible to absorb a significant percent- one where all dimensions of the problem were halved and
age of the pressures. one other where they were doubled. For the analysis of the
In scenario E there is an intermediate situation be- variation of the mechanical characteristics of the soil, only
tween scenarios B and D, with total stresses on the wall that two soils were studied: the soil that was considered in the
remain greater than the apparent diagram given by Eq. 5. It case study (soil 1, in bold in the Table 1), and a soil with
is interesting to note that, when compared with scenario D, better mechanical characteristics (soil 2): f¢ = 40°, y = 8°,
in scenario E the apparent pressures corresponding to the E50,ref = 90000 kPa, Eur,ref = 270000 kPa and
last level of struts are much closer to the distribution of total Eoed,ref = 90000 kPa. For soil 2, the at rest earth pressure coef-
stresses on the wall. It is also interesting to notice that effec- ficient was taken equal to 0.8, assuming an overconso-
tive stresses are greater at the upper part of the wall and lidated material.
much lower at the bottom than in cases where soil is consid- In the second stage of the parametric analyses (right
ered dry. part of Table1), different depths of the water table, zw, and
Figure 7 shows that the diagrams proposed in the lit- different embedded lengths of the wall, for both soils - 1
erature are not suitable approximations for the distribution and 2 - were considered. In these analyses scenarios B, D
of pressures on the wall or the apparent pressures for the and E were studied.
cases where water table was considered. Therefore, this As previously presented, the main objective of this
justifies further research and, in particular, the parametric paper is to propose apparent diagrams that can be used in
analysis presented below. the design of flexible retaining walls with water table above
4. Parametric Analysis the bottom of the excavation. For such purpose, an overall
analysis of all calculations is needed. The procedure fol-
4.1. Cases lowed is next presented as an example, and the same steps
Based on the case study, parametric analyses of sev- were taken for all the other analyses.
eral geometrical and mechanical aspects were performed. 4.2. Example: analysis of the variation of the embedded
A brief overview of all the analyses performed is presented length of the wall, D
in Table 1.
Those analyses were performed in two stages. In the As an example of the adopted procedure, the analysis
first stage (left part of Table 1) the parametric analyses con- of the variation of the embedded length of the wall is pre-
sisted in the variation of one parameter at a time. In these sented for all scenarios A to E. The pressures on the flexible
analyses scenarios A, B, C and D were studied. For exam- retaining wall after the last excavation stage and apparent
ple, when the analysis of the variation of the embedded pressures inferred from the maximum loads on the struts
length of the wall was performed, only this parameter took are, for these cases, shown in Fig. 8.
different values: the one corresponding to the case study it- In scenarios A and C the total stress distributions on
self (in bold in the Table 1), a value above and another be- the wall do not show major differences between the results
low it. A similar procedure was performed for the axial obtained for each adopted value for the embedded length of
stiffness of the struts, the wall bending stiffness, the width the wall. However, for very small values of D, in scenario C
of the excavation, the scale of the problem and the unit it is possible to observe a change of the pressure distribu-
weight of the soil. In the parametric analysis of the scale of tions that show greater values for lower values of length D
the problem, two additional calculations were performed: for greater depths and the reverse more superficially. In
scenario B the distribution of total stress on the wall does
not have a continuous evolution. Also, for D = 8 m, appar-
Table 1 - Cases analysed in the parametric analyses. ent pressures are lower than the ones for D = 6 m; however,
for D = 3 m, apparent pressures at the lowest level signifi-
Case A B C D B D E cantly increase. This is due to the fact that in this case, the
hydraulic uplift safety factor is very low. Also, it can be
D [m] 3/6/8 2/6/8 3/6/8 2/6/8
seen that for the lowest value of D, apparent pressures in-
EA [kN/m] 0.2 EACS / EACS / 5EACS EACS ferred from the strut loads are very large for scenario B. It
2
EI [kNm /m] EICS /8 / EICS / 8EICS EICS should however be noticed that these correspond to unusual
B [m] 6/10/20 10 situations with very low hydraulic uplift safety factors. In
scenario D, in the middle of the excavation height, the
Scale ´0.5 / ´1 / ´2 ´1
greater the embedded length of the wall, the greater the
Soil - S 1/2 1/2 stress. In the last meters of the excavation height there is an
g [kN/m ] inversion of this relationship. The apparent pressures show
3
17.5/20/25 20
greater difference between the results corresponding to
zw [m] - 0 - 0 0/2.5/5/7.5/10
each embedded length of the wall the greater the depth. In
EACS and EICS are the values adopted in the case study. scenario E the pressure distributions do not show major dif-
Figure 8 - Analysis of the variation of the embedded length of the wall: pressures on the flexible retaining wall after the last excavation
stage (symbols) and apparent pressures inferred from the maximum loads on the struts (lines).
ferences between the results obtained for each adopted does not have great influence, the opposite of what is veri-
value for the embedded length of the wall, except, however, fied in Scenario B).
for the results corresponding to the embedded length of the Results represented in the dimensionless way shown
wall with very small value (2 m). It is possible to note that by Eq. 9, as performed in Figs. A.1 to A.4, show approxi-
in this case the pressures are globally higher than those cor- mately constant values of b, which means that this is an ad-
responding to other cases. In this case this is caused by the equate way of presenting the results. These figures, as in
large water pressures on one side of the wall and the consid- Fig. 9, include the same proposed diagram that will be pre-
erably lower ones at the other side. sented in Section 5.
The results obtained after the last excavation stage for
the pressures on the wall and for the apparent pressures (in- 4.4. Analysis of the variation of the depth of the water
ferred from the maximum loads on the struts) were con- table (zw)
verted into dimensionless quantities by determining the
Parametric analyses to study the influence of the vari-
ratio between the obtained pressures and the ones from
ation of the depth of the water table were also performed
Terzaghi and Peck’s diagram (without water table):
and the results of the dimensionless pressures in scenario B
sH are shown in Fig. 10 as an example.
b= (9)
0.65K a gH As expected, the more shallow the water table, the
greater the pressures. The lower the zw, the greater the dif-
and are shown in Fig. 9. This figure includes a proposed di- ference between the distribution of total pressures on the
agram that will be presented in section 5. wall and the apparent pressures inferred from the strut
Figure 9 shows that the results can be reasonably nor- loads, specially at the depth corresponding to the last strut
malized using Eq. 9, resulting in values of b approximately level. For zw/H = 1 soil 1 leads to greater normalized pres-
constant for each set of calculations. The same representa- sure on the wall than the soil 2. However, for water table
tion will be used for the other results obtained in the para- closer to the ground surface, there is a reversal of this rela-
metric analysis. tion.
4.3. Analysis of the variation of other geometric and me- Scenarios D and E were also studied in order to ana-
chanical aspects lyse the influence of the depth of the water table variation
and the results of the dimensionless pressures are shown in
The same procedure followed in subsection 4.2 was
Appendix B in Figs. B.1 and B.2, respectively. As ex-
adopted for the different parameters shown in Table 1 and
pected, it can be seen that shallower water tables result in
the results for the dimensionless pressures are shown in Ap-
greater pressures on the wall and greater apparent pres-
pendix A in Figs. A.1 to A.4. Scenarios B and D are the sub-
sures. As in previous figures, Figs. 10, B.1 and B.2 include
ject of the analysis; however, scenarios A and C are also
proposed diagrams (now depending on the water table
studied for comparison.
depth) that will be presented next.
It can be briefly mentioned that in the case where high
water table with steady-state seepage is considered (Sce- 5. Proposal for Apparent Pressures
nario B), the parameters that show greater influence on
pressures on the retaining wall and on apparent pressures For cases without water table, scenarios A and C
using the loads on the struts are: embedded length of the (Figs. A.1 and A.3), the dimensionless horizontal stresses,
wall, width of excavation, mechanical characteristics of the b, are reasonably approximated by 1.00, corresponding to
soil and unit weight of the soil (which is more relevant in the diagram of Terzaghi & Peck (1967) (Eq. 1). Moreover,
Soil 1); the parameters that have less influence than the pre- this value remained globally adequate in most of the analy-
vious ones are: axial stiffness of the struts, bending stiffness ses (see Figs. 9(a), 9(c), 10(a), A.1, A.3, B.1(a) and B.2(a)),
of the wall and scale of the problem. showing the suitability of the traditional diagram of Terza-
In the case where high water table with hydrostatic ghi and Peck as a conservative diagram for the results of the
pressures is considered (Scenario D), the parameters that pressures.
show greater influence on pressures on the retaining wall For cases where water table is taken into account, sce-
and on apparent pressures using the loads on the struts are: narios B, D and E, the results obtained for the total pres-
embedded length of the wall, bending stiffness of the wall sures on the wall and apparent pressures inferred from
(in contrast to what is verified for Scenario B), mechanical maximum loads on the struts allow to propose apparent dia-
characteristics of the soil and unit weight of the soil (which, grams. These diagrams are also shown in Figs. 9, 10, A.2,
as for Scenario B, is more relevant in Soil 1); the parame- A.4, B.1 and B.2 and are detailed in Fig. 11. Such diagrams
ters that have less influence than the previous ones are: ax- admit that apparent pressures should not be less than the
ial stiffness of the struts, scale of the problem (which have ones from the diagram of Terzaghi and Peck. They consider
greater influence on the apparent pressures using the loads Terzaghi and Peck’s value at the ground level and increase
on the struts), but mostly the width of excavation (which linearly down to a depth of 0.4H, after which a constant dia-
Figure 9 - Analysis of the variation of the embedded length of the wall: Dimensionless pressures on the flexible retaining wall after the
last excavation stage (symbols) and apparent pressures inferred from the maximum loads on the struts (lines). Note the difference be-
tween the horizontal scales of the graphics on the left and on the right.
Figure 10 - Scenario B - analysis of the variation of the embedded length of the wall to different depths of the water table: dimensionless
pressures on the flexible retaining wall after the last excavation stage (symbols) and apparent pressures inferred from the maximum loads
on the struts (lines).
Figure 11 - Proposal for apparent design diagrams.
Figure 12 - Proposed values of bbase as a function of the depth of

gram given by a horizontal pressure of bbase 0.65 KagH is
the water table.
proposed.
In the particular case of water table on the ground sur-
face, it appears from Figs. 9, A.2 and A.4 that the results of behaviour of the soil, for two soil conditions, for different
the dimensionless pressures can be approximated by geometries (embedded length of the wall, width of the ex-
bbase = 3.1 for scenario B and 2.75 for scenarios D and E. cavation, scale of the problem), different soil unit weights
These values show a considerable influence of the water ta- and different stiffnesses of the wall and of the struts. Also,
ble on total pressures and on the loads on the struts. It several positions of the water table was considered.
should be noted that in the course of the parametric analysis The obtained pressures were normalized using the
performed, a few cases of very large apparent pressures stress of Terzaghi and Peck’s apparent diagram for sands
corresponding to the last strut level were obtained. These and the normalized results could be reasonably represented
few cases mainly correspond to situations where hydraulic by a proposed apparent design diagram for cases in which
uplift safety factors are very low and beyond the scope of the water level is assumed, either when seepage is allowed
usual design. Such cases involve the same type of phenom- and when hydrostatic pressures are considered. The excep-
ena also present in undrained situations with poor basal stations correspond to cases with low hydraulic uplift safety
bility conditions (Terzaghi & Peck, 1967; Bjerrum et al., factors. The proposed apparent design diagram shows a
1972; Peck et al., 1974). Therefore, these situations were very large influence of the water table on the pressures on
ignored for the proposal of the apparent pressure distribu- the wall, increasing them significantly.
tion diagrams presented. References
Based on the parametric analysis of the variation of
the depth of the water table that was performed for scenar- Andrade Viana, L. (2014). Pressões Sobre Cortinas de
ios B, D and E (Figs. 10, B.1 and B.2) a chart with the corre- Contenção em Solos com Nível Freático Elevado. Mas-
ter’s Thesis, UNL, Faculdade de Ciências e Tecnologia,
sponding values of bbase (Fig. 12) for different values of zw is
Universidade Nova de Lisboa, Lisboa.
further proposed. The resulting diagrams are shown in Figs.
Bjerrum, L., Frimann Clausen, C. & Duncan, J. (1972).
10, B.1 and B.2.
Earth pressure on flexible structures - A state of the art
6. Conclusions report. In: Proc. 5th European Conference on Soil Me-
chanics and Foundation Engineering, Madrid, volume
From the developed study it can be concluded that the 2, p. 169-196.
existing proposals in the literature for the apparent dia- Ebeling, R.M. & Morrison, E.E. (1992). The seismic design
grams corresponding to situations in which the water level of waterfront retaining structures. Technical Report
is considered and seepage is allowed, as well as the situa- ITL-92-11, U.S. Army Engineer Waterways Experi-
tions in which the water level is considered with hydrostatic ment Station.
pressures, are not suitable approximations for the distribu- Kaiser, P.K. & Hewitt, K.J. (1982). The efect of groundwa-
tion of pressures on the wall or the apparent pressures. In- ter flow on the stability and design of retained excava-
deed, in all calculations the results achieved were greater tions. Canadian Geotechnical Journal, 19:139-153.
than the values corresponding to these diagrams. Peck, R.B., Hanson, W.E. & Thornburn, T.H. (1974).
The calculations were performed for a relatively wide Foundation Engineering. John Wiley and Sons, New
range of situations, assuming homogeneous and isotropic York, 2nd edition.
Plaxis (2014). Plaxis 2D Manual. Brinkgreve et al., eds. Strom, R.M. & Ebeling, R.M. (2001). State of the practice
Delft. in the design of tall, stiff, and flexible tieback retaining
walls. Technical Report ITL TR-01-1, US Army Corps
Schanz, T., Vermeer, P.A. & Bonnier, P.G. (1999). The of Engineers. Engineer Research and Development
hardening-soil model: Formulation and verification. In: Center.
Proc. Beyond 2000 in Computational Geotechnics, Terzaghi, K. & Peck, R.B. (1967). Soil Mechanics in Engi-
Balkema, Rotterdam, p. 281-290. neering Practice. John Wiley & Sons, Inc., 2nd edition.
Appendix A
Figure A.1 - Scenario A - analysis of the variation of the: embedded length of the wall; stiffness of the struts; stiffness of the wall; width
of excavation; scale and unit weight of the soil. Dimensionless pressures: total stresses on the flexible retaining wall after the last excava-
tion stage (symbols) and apparent pressures inferred from the maximum loads on the struts (lines).
Figure A.2 - Scenario B - analysis of the variation of the: embedded length of the wall; stiffness of the struts; stiffness of the wall; width
tion stage (symbols) and apparent pressures inferred from the maximum loads on the struts (lines). Note the difference in the horizontal
scale in the bottom right graphic.
Figure A.3 - Scenario C - analysis of the variation of the: embedded length of the wall; stiffness of the struts; stiffness of the wall; width
Figure A.4 - Scenario D - analysis of the variation of the: embedded length of the wall; stiffness of the struts; stiffness of the wall; width
Appendix B
Figure B.1 - Scenario D - analysis of the variation of the embedded length of the wall to different depths of the water table.
Dimensionless pressures: pressures on the flexible retaining wall after the last excavation stage (symbols) and apparent pressures in-
ferred from the maximum loads on the struts (lines).
Figure B.2 - Scenario E - analysis of the variation of the embedded length of the wall to different depths of the water table.
Dimensionless pressures: pressures on the flexible retaining wall after the last excavation stage (symbols) and apparent pressures in-
ferred from the maximum loads on the struts (lines).
F. Massad
Abstract. The paper deals with the bidirectional test, aiming at interpreting its results, showing the factor that governs the
upward movement and presenting approximate formulas to find the equivalent load-settlement curve of a conventional
compressive test. The factor is the product c’.k = (1 - c).k, where c’ and c are the Leonards-Lovell coefficients, related to the
elastic shortening of the shaft under bidirectional and conventional static loading tests, respectively, and k is the relative
stiffness of the pile-soil (shaft) system. The paper shows that when this product is constant the properly normalized upward
curve is invariant, independently on shaft resistance distribution. Nomograms are presented to quickly determine c’ for two
patterns of shaft friction associated to weak upper layers over deep more resistant soils. Under this condition, it is shown
that c > c’, i.e., the load causing the downward movement in the conventional test induces greater elastic shortening than in
the bidirectional test. Therefore, additional movement must be added to the measured displacement in the bidirectional
test, which constitutes the basis of the proposed approximate formulas to find the equivalent curve of the conventional test.
To validate these findings a mathematical model is used together with five case histories, comprising short to long piles. In
one case a conventional compressive loading test was also available.
Keywords: bidirectional test, elastic shortening, equivalent curve, compressive loading test.
1. Introduction this curve is in the evaluation of pile bearing capacity, us-

ing, for instance, the Davisson Offset Limit (Davisson,
The Brazilian hydrodynamic expansive cell was de- 1972 cited by Fellenius, 2015).
veloped by Silva (1983 and 1986) and since the 1980’s has The objectives of this paper are: a) to show that the
been used in static load tests chiefly in bored piles. Its use product c’.k = (1 - c).k, where c is the Leonards-Lovell co-
spread worldwide after Osterberg (1989) and is known as efficient, related to the elastic shortening of the shaft under
“O-Cell Test” or the bidirectional test. One or more expan- conventional test, and k the relative stiffness of the pile-soil
sive interconnected cells are placed on the tip of a metal (shaft) system, governs the normalized upward movement
frame and introduced into the shaft, generally near the pile of a bidirectional test; and b) to present approximate formu-
toe, and concreted together with the pile. The activation of lae to determine the equivalent load-settlement curve of a
the cells takes place hydraulically causing its expansion, conventional test in a simpler and more rational way.
pushing the shaft upward and the toe downward. The up-
ward and downward movements can be measured at the 2. Shortening of Piles Under Compressive
level of the cells (bottom and top) with “tell tales” and at the Loadings
top of the pile with dial gauges or displacement transducers.
The analysis will be initiated evaluating the shorten-
The reaction system is provided by the pile shaft and
ing of piles subjected to both download conventional test
the test is limited to the exhaustion of one of the pile capac-
and bidirectional test.
ity, tip or friction. Its execution is rapid and very high loads
may be applied when associated with several cells. 2.1. Shortening of piles during a download conventional
The issue has been the subject of analysis by Alonso test
& Silva (2000), trying to simulate the equivalent conven-
tional static load test. The usual procedure to obtain an To estimate the shortening of vertical piles, under ax-
equivalent conventional test curve consists in adding the ial compressive loading at the pile head (Po), not necessar-
shaft and toe applied loads which cause the same measured ily at failure, the following expression may be used (see the
displacement, up and down. As the load causing the down- list of symbols):
ward movement in the equivalent test induces greater elas- Qp Al
tic shortening than in the bidirectional test, additional De = + c× (1)
Kr Kr
movement must be added to the measured displacement.
Many procedures were proposed accounting for pile com- where Qp and Al are toe and shaft loads, respectively, so
pression to construct the equivalent download curve, like that:
those presented by Alonso & Silva (2000), Loadtest (2001);
Kwon et al. (2005); and Kim et al. (2012). The relevance of Po = Q p + Al (2)
Faiçal Massad, Dr. Eng., Full Professor and Geotechnical Consultant, Escola Politécnica of the São Paulo University, Av. Prof. Almeida Prado 271, trav. 2, São Paulo, SP,
Brazil. e-mail: [email protected].
Submitted on October 7, 2015; Final Acceptance on November 11, 2015; Discussion open until April 30, 2016.
Massad
Kr is the pile stiffness, with height h, cross sectional area S simple forms of distribution of fu can be obtained rapidly
and modulus of elasticity E, given by: using the nomograms prepared by Leonards & Lovell
E× S (1979) or the equations proposed by Fellenius (1980),
Kr = (3) shown in Fig. 2. Note that h1 and h2 are the thicknesses of
h
the softer and stiffer layers, respectively, and fu1 and fu2 are
In Eq. 1 c is the Leonards & Lovell (1979) coefficient, the corresponding shaft frictions.
i.e., the ratio of the average value of the transferred lateral
2.2. Shortening of piles during a bidirectional test
load (the hatched area of Fig. 1-a over pile height) and the
total shaft load (Al), i.e.: For upward loads in the bidirectional test Eq. 1 chan-
ges to:
Al - Al
c= (4) Al
Al De = c¢ × (5)
Kr
The coefficient c depends on the distribution of the
unit shaft friction (f). If the shaft load is fully mobilized where c’ is given by Eq. 4 but related to Fig. 1-b. A similar
(Al = Alr), then c = 0.5 for fu = const along depth and c = 2/3 nomogram may be constructed for c’, as shown in Fig. 3
for fu increasing linearly with depth. Values of c for other with the associated equations. Note that now h1 and h2 are
Figure 1 - Load distribution -a) Conventional test; and b) - Bidirectional test.
Figure 2 - Nomograms for head-down loads (conventional loading test).
the thicknesses of the stiffer and softer layers, respectively, download conventional test is used. It is based on Modified
and fu1 and fu2 are the corresponding shaft frictions. Com- Cambefort Relations (Fig. 4) and consideration is made of
paring Figs. 1-a and 1-b, on one hand, and Figs. 2 and 3, on many aspects of load transfer phenomena, like the progres-
the other hand, it can be seen that: sive failure, due to pile compressibility, and the residual
stresses due to driving or subsequent loadings. As the piles
c + c¢ =1 (6)
in this paper are supposed to be cast in place and submitted
Note also that in general c’ < c, that is, the elastic to a single loading, residual stresses will be ignored.
compression of piles under O-Cell upward loads is smaller A coefficient that measures the relative stiffness of
than the corresponding value for the download test. This is the pile-soil (shaft) system was introduced and is defined as
due to the fact that the O-Cell upward loads mobilize first follows:
the deeper and more resistant soils and later the upper softer Alr
layers. k= (7)
K r × y1
3. Mathematical Simulation of Bidirectional
where Alr is the ultimate shaft load; y1, the pile displace-
Tests. The Upward Curve Invariance with
ment, of the order of a few millimeters, required to mobilize
c’.k full shaft resistance. The coefficient k may be associated to
3.1. The mathematical model the term (mh)2 of Randolph & Wroth (1978). The model
gave a further insight on pile behavior and led to a pile clas-
For the simulation of bidirectional tests a mathemati- sification, with respect to the k values: “short” or rigid
cal model developed by the author (Massad, 1995) for
Figure 3 - Nomograms for upward loads (bidirectional test).
Figure 4 - Modified Cambefort Relations: a) shaft and b) toe.
Massad
(k £ 2); intermediate (2 £ k £ 8); and “long” or compressible

(k ³ 8).
3.2. Simulation of the Bidirectional test: invariance of

the upward curves with c’.k
To simulate the bidirectional test this model was
changed to incorporate a soft upper layer (fu2) over a deep
more resistant soil (fu1). For each layer along the shaft a rela-
tion as shown Fig. 4-a holds. Using subscripts 1 and 2 to
distinguish them, it follows that:
f u1 ³ f u 2 and Alr = Alr1 + Alr 2 (8)
For both layers the value of y1 (see Fig. 4-a) was as-
sumed to be the same, so that the k value (Eq. 7) may be as-
sociated to the whole subsoil along the pile shaft. Note that
Kr in Eq. 7 refers to the pile height above the O-Cell.
Ten cases of the bidirectional tests were simulated,
varying the relations fu2/fu1 and h2/h, as displayed on Fig. 5.
For each case the soil below the O-Cell was the same, obey-
ing the relation of Fig. 4-b, with the following parameters: Figure 5 - Analyzed cases.
R¢ × S p R¢
y 3 = y1 besides = 0.80 and =5 (9)
Kr × k R c¢ × Alr
Po 4 = Alr and y 4 = y 1 + = y 1 × (1 + c¢ × k)
where Sp is the cross sectional area of the pile toe. Kr
(10)
Figures 6-a, 7-a and 8-a show, respectively, results y
then 4 = 1 + c¢ × k
for a rigid pile (k = 1 and Case 5), an intermediate pile (k = 5 y1
and Case 3) and a long pile (k = 10 and Case 10). Note that
the loads and movements are normalized with respect to Alr Eqs. 10 still holds for cases of linearly increasing
and y1, respectively. The range 0-3 corresponds to the initial maximum unit skin friction with depth: in this case, c’ = 1/3
pseudo elastic line of Fig. 4-a, with inclination B; the range (see Fig. 3-b).
3-4 refers to the progressive mobilization of shaft resis-
tance, from bottom to top, up to yh = y1 (yh is the pile top The full mobilization of fu1 of the deep stiffer layer ini-
movement); point M corresponds to the fully mobilization tiates at points 3 of Figs. 6-a, 7-a and 8-a. The following re-
of fu1 of the deep stiffer layer. For the unloading the analo- lation holds:
gous ranges are indicated through the points 3’and M’.
y3
At the O-Cell level, the following relations hold for =1 (11)
point 4: y1
Figure 6 - Bidirectional test simulation for a rigid pile (k = 1, Case 5).
Figure 7 - Bidirectional test simulation for an intermediate pile (k = 5, Case 3).
Figure 8 - Bidirectional test simulation for a long pile (k = 10, Case 10).
Moreover, the ratio Po3/Alr depends only on c’.k, as In the attached Appendix more cases are presented
shown in Figs. 9 and 10. Note that the case of linearly in- showing that the mentioned invariance with respect do c’.k
creasing maximum unit skin friction with depth (c’ = 1/3) is still holds, even for other transfer function besides the
included in Fig. 9. Cambefort Relation for the shaft friction.
Taking also into account the last equation of Eq. 10, it It is worth mentioning that the movement at pile top
follows that the normalized upward curves of the bidi- (yh) reaches the value yh = y1 at shaft failure load (Al = Alr), as
rectional tests are approximately invariant with respect to can be seen in Figs. 6-a, 7-a and 8-a. Moreover, taking into
c’.k, independently on the distribution of shaft resistances. account Eqs. 5 and 7, it follows:
This conclusion is confirmed by the plots of Figs. 11-a Alr De s
(c’.k @ 0.5), 12-a (c’.k @ 2.3) and 13-a (c’.k @ 4.7). Note that c¢ × k = c¢ × = max (12)
K r × y1 y1
the Arabic numbers associated to the curves refer to the
cases of Fig. 5. To stress this point, Fig. 14 was prepared that is, the ratio between the maximum elastic shortening of
showing different distribution of shaft resistances for cases the pile shaft and the value of y1 governs the upward curve
7 and 10 of Fig. 13, but with almost the same c’.k, i.e., 4.9 of a bidirectional test.
and 4.8, respectively.
4. Mathematical Simulation of the Equivalent
The conclusion about the invariance of the normal-
ized upward movements related to c’.k is maintained even
Curve of the Conventional Test
when one compares the cases of Fig. 15, with different val- The same mathematical model was also applied to
ues of k, and also the cases of Fig. 16, whose skin friction simulate an equivalent download conventional test.
distribution is presented in Fig. 17. Note that in the latter Figs. 6-b, 7-b and 8-b show the equivalent curves for the
cases the product c’.k is close together. cases of a rigid pile (k = 1), an intermediate pile (k = 5) and a
Massad
Figure 9 - Po3/Alr versus c’.k for h2/h £ 0.75. Figure 10 - Po3/Alr versus c’.k for h2/h ³ 0.75.
Figure 11 - Bidirectional test simulations on short piles. Arabic numbers refer to the cases of Fig. 5.
Figure 12 - Bidirectional test simulation for intermediate piles. Arabic numbers refer to the cases of Fig. 5.
long pile (k = 10), respectively. The notable points 3, M and

c × Alr R¢ × S p × y 1
4 (for loading) and 3’and M’ (for unloading) are indicated y 4 = y1 + +
in these plots. They have the same meaning as in the Kr Kr
(13)
bidirectional tests, but with the progressive mobilization of y R¢ × S p
shaft resistance going from top to bottom. then 4 = 1 + c × k +
y1 Kr
For point 4, the following relationships hold:
Figure 13 - Bidirectional test simulations for long piles. Arabic numbers refer to the cases of Fig. 5.
Figure 14 - Distribution of unit skin friction with depth: a) case 7

and b) case 10 of Fig. 13-a.
Figure 16 - Bidirectional tests with different skin friction distribu-
tion (see Fig. 17).
Figure 15 - Bidirectional tests with different values of k, but with Figure 17 - Distribution of unit skin friction with depth of cases of
the same product c’.k. Fig. 16.
Po 4 = Alr + R¢ × S p × y 1 below the O-Cell was supposed to be the same (Eqs. 9), the
downward curves of the conventional tests are invariant
Po 4 R¢ × S p (14)
then =1 + with c’.k, as shown in the plots of Figs. 11-b, 12-b and 13-b,
Alr Kr × k taken separately.
From Eqs. 13 and 14 it follows that if k and c are con- Finally, as Po4 is a function of k (Eq. 14), different
stants, then c.k is also a constant and the coordinates of values of k lead to distinct curves Po - yo. This fact can be
point 4 coincide, regardless the values of fu2/fu1 and h2/h as- confirmed by comparing together Figs. 11-b, 12-b and
sociated to c or to c’ = 1 - c (Eq. 6). Additionally, as the soil 13-b.
Massad
5. Approximate Formulas to Determine the The differences are greater for long piles as compared to
Equivalent Curve of the Conventional Test short piles. As a mater of fact, for very rigid or very short
piles, Kr is very large and both, De and Q’p/Kr approach to
To derive approximate formulas for the equivalent zero and Eqs. 17 and 18 reduce to yo = y’p = yf and
curve, considerations will be made using Fig. 18, supposing Po = Al + Q’p: the usual procedure is valid.
that the upward movement was measured at the top of the If the movement of the pile head is available, instead
O-Cell and the load reached Alr. Let P be a point of the up- of that of the top of the O-Cell, the procedure is analogous,
ward curve with the coordinates yf and Al. as displayed in Fig. 19. And finally, if the movements at the
Next, the elastic shortening of the shaft and y’p, an ap- pile head and on top of O-Cell are available, the measured
proximate measure of the pile head displacement, are com- De is used instead the value of Eq. (15).
puted as follows: It is worth highlighting that cases like those of Figs.
c¢ × Al 14 and 17 conduct to the same values of c and c’. In fact, for
De = (15)
Kr the pair h2/h = 0.1 and fu2/fu1 = 0.7 (Case 7 Fig. 14-a) it fol-
lows from Fig. 3-a c’ = 0.49 than c = 0.51, and for h2/h = 0.9
y ¢p = y ¢f - De ³ 0 (16) and fu2/fu1 = 0.7 (Case 10 Fig. 14-b), c’ = 0.48 than c = 0.52.
The same conclusion arises from Fig. 17: for Case 3
Equation 16 is an approximate value of y’p because c
c’ = 0.31 than c = 0.69 (see Fig. 3-a), very close to c’ = 0.33
and c’ depend on the amount of shaft friction mobilized
and c = 0.67 for the case of linearly increasing shaft resis-
during the loading. In a similar problem involving unload-
tance with depth (see Fig. 3-b). The conclusion is that dif-
ing in a conventional test, Massad (2001) showed that, for
ferent distribution of shaft resistance may lead to the same
practical purposes, c may be approximated to the value cor-
elastic shortening and so the same equivalent curve: the key
responding to the maximum unit skin friction. It will be val-
factor is the elastic shortening of the shaft.
idated later on.
To simulate the download conventional test, y’p is set- 6. Practical Applications
tled as the toe movement; it is associated to Q’p, as indicated
Applications will be made to five case histories, com-
in Fig. 18.
prising short to long piles, with heights and diameters vary-
Finally, a pair yo - Po of the equivalent curve is deter-
ing from 11.5 to 41.0 m and 0.60 to 2.40 m, respectively. In
mined by the equations:
one place a conventional compressive test was also avail-
c Q¢p able (see Table 1).
y o = y ¢p + De × + (17)
c¢ Kr
6.1. Continuous Flight Auger (CFA) piles in Belo Hori-
Po = Al + Q¢p (18) zonte (Brazil)
A first validation of Eqs. 17 and 18 is shown in Two CFA piles E 46 and E 46A, 0.60 m in diameter,
Figs. 6-b, 7-b and 8-b: the curves of the approximate formu- were installed in Belo Horizonte, Brazil, 2.5 m apart. One
las and the mathematical model are in excellent agreement. of them was submitted to a conventional loading test and
The usual procedure to construct the equivalent cur- the other to a bidirectional test as shown in Table 1. The
ve, also shown in Figs. 6-b, 7-b and 8-b, consists in adding subsoil consisted of 1.8 m earth fill, on top of soft silty clay
the cell loads (Al + Q’p) for equal measured movements up and clayey silt layers (SPT = 3 to 5), up to 10 m depth, and
and down; no consideration is made to the correction of the below a sandy silt residual soil (SPT = 17 to 30). The water
measured movements due to elastic shortening of the pile. table was at 8 m depth.
Figure 18 - Schematic curves of a bidirectional test-upward Figure 19 - Schematic curves of a bidirectional test-upward
movements measured at the top of the O-Cell. movements measured at the pile head.
Table 1 - Case Histories -general information of the piles.
Type of test Pile Diameter (m) Height (m) Place Source

Conventional CFA (E-46) 0.60 16.0 Belo Horizonte (BR) Alonso & Silva (2000)
Bidirectional CFA (E-46A) 0.60 14.0 + 1.5 = 15.5
Bidirectional Omega (PC-02) 0.70 8.5 + 3.0 = 11.5 São Paulo (BR) Fellenius (2014-a)
Bidirectional Omega (PC-07) 0.70 7.2 + 4.3 = 11.5
Bidirectional Bored (1) 0.90 16.0 Puerto Rico (USA) Fellenius (2015)
Bidirectional Bored (2) 1.25 40.0 Mississippi River (USA) Fellenius (2014-b)
Bidirectional Bored (3) 2.40 41.0 Tucson Arizona (USA) Loadtest (2014)
Figure 20 presents the results the bidirectional test on Table 2 shows other relevant data. The value of c was
E 46A pile, performed by Arcos. The O-Cell was placed at estimated using SPT data and the Décourt-Quaresma
14.0 m depth and the upward and downward movements Method (1978) to determine the fu and the load distribution
were measured at pile head and at the base of the O-Cell, re- along depth, like shown in Fig. 1-b.
spectively. For the conventional loading test on Pile E 46, pre-
sented in Fig. 21, the Two Lines Method (Massad & Lazo,
1998 and Fonseca et al., 2007) was applied leading to an ul-
timate side friction (Alr) of 1,900 kN, y1 equals to 0.35 mm
and a toe stiffness of 3,000 kN/mm; some strain hardening
was observed at the toe, probably due to the compression of
a partially remolded soil, an outcome of pile installation.
Figure 20 - Bidirectional test on a CFA pile in Belo Horizonte. Figure 21 - Conventional test on a CFA Pile in Belo Horizonte.
Table 2 - Results of the analysis.
Type of test Pile Kr c c’ k c’.k Pile behavior

(kN/mm)
Conventional CFA (E-46) 442 0.59 0.41 12.3 5.0 Long or Compressible
Bidirectional CFA (E-46A) 505 0.57 0.43 7.6 3.3 Long or Compressible
Bidirectional Omega (PC-02) 906 0.58 0.42 0.2 0.1 Short or rigid
Bidirectional Omega (PC-07) 1069 0.58 0.42 0.2 0.1 Short or rigid
Bidirectional Bored (Puerto Rico) 795 0.50 0.50 1.6 0.8 Short or rigid
Bidirectional Bored (Mississipi) 614 var. var 0.9 var Short or rigid
Bidirectional Bored (Tucson) 4414 0.80 0.20 4.4 0.9 Intermediate
Legend: see appended list of symbols.
Massad
The toe did not reach failure. These values were confirmed
by applying the Mathematical Model based on Cambefort
Relations: the fitting between the measured and calculated
curves is remarkable. The pile behaved as a long pile, with
k ~ 8 (Table 2). It is worth pointing out that Décourt-Qua-
resma Method (1978) led to Alr = 1,860 kN.
As far as the Pile E 46A is concerned, submitted to
the bidirectional test, the maximum side friction (Alr) up to
14 m was estimated to be 1,440 kN by the Décourt-Qua-
resma Method (1978), above the maximum upward load
of 1,350 kN (see Fig. 20-a). Adjustments were made in the
upward curve, as displayed in Fig. 20-a, eliminating the
“jump” in the beginning and extrapolating at the end as-
suming a failure load of 1,440 kN. Table 3 was prepared
using Eqs. 17 and 18 that led to the equivalent curve for
Figure 22 - Comparison of the download curves - CFA Piles in
the E 46A pile, as shown in Fig. 22. The segment of pile Belo Horizonte.
below 14 m, with a length of 1.5 m, was taken as a ficti-
tious toe, with a transfer function given by Fig. 20-b. It in-
cludes the real toe and the side friction of the 1.5 m pile test, due to an unknown reason. This fact is supported by
segment. the result presented in Fig. 20-b. Finally, the application of
Next the same Mathematical Model (Cambefort) was the usual procedure as defined above led to unrealistic val-
applied to the bidirectional test on E 46A pile, using the ul- ues of settlements.
timate side friction (Alr) of 1,440 kN, y1 equals to 0.35 mm
and toe stiffness of 150 kN/mm, this last figure gotten from 6.2. Omega piles in São Paulo City (Brazil)
the initial part of the “downward” curve (Fig. 20-b). The re- Fellenius (2014-a) presented the results (see Fig. 23)
sult is also shown in Fig. 22 together with the measured of bidirectional tests performed by Arcos at a site in São
curve of the conventional test (Pile E 46). Again, the fitting Paulo, Brazil, on two Omega Piles both with diameter
is remarkable amongst the three curves up to the full mobi- 700 mm and embedment 11.5 m (see Table 1). Pile PC-02
lization of the shaft resistance in Pile E 46A. This is a sec- was provided with a bidirectional cell at 8.5 m depth and
ond validation of the approximate formulas, Eqs. 17 and 18. Pile PC-07 at 7.2 m depth. The upward and downward
Figure 22 shows moreover that the fictitious toe resistance movements were measured at pile head and at the base of
of Pile E 46A, given by Fig. 20-b, is much smaller than the the O-Cell, respectively. These and other information are
toe resistance of Pile E 46, submitted to the conventional presented in Table 2. The subsoil consisted of 2.5 m earth
fill on top of layers of silty clay and sandy silt, SPT varying
erratically from 5 to 15, and a very dense silty sand below
Table 3 - Application of the approximate formulae to E 46A CFA 9 m depth. The water table was at 2 m depth. The value of
pile. c’ = 0.42 was estimated using SPT data and the Décourt-
Quaresma Method (1978) to determine the fu and the load
y‘p (mm) Qp (kN) Al (kN) Po (kN) yo (mm) distribution along depth.
0.00 0 0 0 0.00 The use of Eqs. 17 and 18 led to the results shown in
0.09 50 836 886 1.13 Fig. 24. The agreement with the curves obtained by Fel-
lenius with a software algorithm (UniPile) is remarkable.
0.11 55 892 947 1.23
This is a third validation of the approximate formulas,
0.13 65 945 1010 1.33 Eqs. 17 and 18. Note that it was assumed fictitious toes be-
0.15 70 995 1065 1.41 low the O-Cells. As the pile is short or rigid, the usual pro-
0.17 75 1042 1117 1.50 cedure gave reasonable results.
0.21 80 1128 1208 1.64 6.3. Bored pile in Puerto Rico
0.24 93 1185 1278 1.76
Results of a bidirectional test on a 900 mm diameter
0.28 96 1255 1351 1.89 bored pile, 16 m height (see Table I), are shown in Fig. 25.
0.34 100 1345 1445 2.06 The pile was drilled in clayey saprolite and socketed a short
0.42 140 1442 1582 2.33 distance into weathered bedrock. The O-Cell was placed
0.85 200 1442 1642 2.88
near the pile toe and the movements were taken at its base
and top. The head down equivalent curves, obtained by
3.85 500 1442 1942 6.47 Fellenius (2015) and with the application of Eqs. 17 and 18,
Figure 23 - Bidirectional tests on omega piles in São Paulo. Adapted from Fellenius (2014-a).
are again in quite a good agreement. This is a fourth valida-

tion of the approximate formulas. A value of E = 20 GPa
was assumed for the pile and c’ was taken equals to 0.5 (Ta-
ble 2).
6.4. Bored pile US82 Bridge across Mississipi River
(USA)
A bidirectional test on a 1.25 m diameter 40 m deep
bored pile (see Table 1) was performed at US82 Bridge
across Mississipi River installed into dense sand. The O-
Cell was placed near the pile toe and the movements were
taken at its base and top. The results are shown in Fig. 26.
Figure 27 displays the equivalent head-down curve for 3
hypotheses with respect to the parameter c’. Also shown is
the curve obtained by the usual procedure. The differences
are relatively small, because the pile behaved like a rigid or
short pile: the value of y1 was large and k assumed a value Figure 25 - Bidirectional test - Bored pile of Puerto Rico.
close to 1 (see Table 2).
Figure 26 - Bidirectional Test: Bored pile US82 Bridge-Missis-

Figure 24 - Equivalent curves for Piles PC-02 and PC-07. sipi River (From Fellenius, 2014-b).
Massad
Figure 29 - Head-down test: Bored Pile- Tucson (AZ).
Figure 27 - Equivalent head-down test: Bored pile US82 Bridge

across Mississipi River. cannot be said about the usual procedure: the k value was
relatively large, of the order of 4.4 (Table 2). Moreover, it is
noticeable that: a) c’.k @ 0.9 and Po3/Alr @ 0.74, in agreement
6.5. Bored pile - Tucson, Arizona (USA) with the data of Fig. 9; b) yo4/y1 @ 1.9, in accordance with
Eq. 10; and c) yh = y1, as mentioned before.
A bidirectional test was carried out in an instru-
mented bored pile, 2.4 m diameter and 41 m deep in Tuc- 7. Conclusions
son, Arizona (USA) (See Tables 1 and 2). The O-Cell was
placed near the pile toe and the movements were taken at its The elastic compression of piles under O-Cell upward
base and at pile head. Figure 28 shows the results. Based on loads is generally smaller than the corresponding value for
the distribution of shear stresses given by the strain gages it the download test or conventional loading test. Its estima-
was possible to estimate c’ = 0.20. The value of y1 was as- tion may be done using de coefficient c’ = 1 - c, where c is
sumed to be 3.8 mm, y3 = y1, R’.Sp/(Kr.k) = 0.56 and the Leonards-Lovell Coefficient for axial compressive
R’/R = 10. loading applied at the pile head. Nomograms of c’ were pre-
The application of the Mathematical Model led to a sented for two patterns of shaft resistance distribution. The
good agreement with the measured up and downward value of c’ may be estimated through the shaft resistance
load-movement curves (Figs. 28-a and b). prediction by means of empirical methods based on SPT
A good fitting (Fig. 29) was also obtained amongst data. A reasonable estimation of the pile elasticity modulus
the three equivalent curves: a) one computed with the (E) and its stiffness Kr is also needed. Both parameter c’ and
Mathematical Model; b) the other with the approximate Kr can be better determined by means of instrumentation.
formulas (Eqs. 17 and 18), being this its fifth validation, The simulation of the bidirectional test using a mathe-
and c) the curve referred by Loadtest Procedure (2001), matical model showed that the coefficient c’ plays an im-
mentioned before and shown in Loadtest (2014). The same portant role. The normalized upward curve is invariant
Figure 28 - Bidirectional test, Bored Pile - Tucson, Arizona.
when the product c’.k is constant, regardless may be the

shaft resistance distribution and the separate value of k, the
relative pile-soil (shaft) stiffness.
Approximate formulas to determine the equivalent
curve were proposed, correcting the shaft elastic shortening
induced by the upward loads in the bidirectional test. The
measured or estimated displacement at pile top (y’p) in the
bidirectional test is set as the toe movement in the down-
load conventional test. The load (Q’p) at the base of the
O-Cell associated to y’p is taken as the toe load. The shaft
elastic shortening, measured or estimated in the bidirectio-
nal test, is corrected by the factor c/c’ and added to the pile
compression due to Q’p and to y’p to get the pile head move-
ment of the equivalent curve. The head load is obtained Figure A2 - Transfer functions.
adding the O-Cell shaft load (Al) to the O-Cell base load
(Q’p), both values associated do y’p. Application was made
to five case histories, revealing the potentiality and easiness
of the proposed procedure. It is shown that, when the pile is
very rigid, the usual procedure gives good results, but, for
compressible piles, the differences are relevant.
Appendix
The conclusion about the invariance with the term
c’.k still holds even when one compares the cases of
Fig. A-1, with different transfer functions, namely, Ratio
Function (RF) and Cambefort (C) Relation, as displayed in
Fig. A-2, and different resistance distribution along depth
Figure A3 - Distribution of unit skin friction with depth of cases
(Fig. A-3). In this figure RR and TR mean, respectively, of Fig. A-1.
rectangular-rectangular and triangular-rectangular shapes.
The Ratio Function has been used by Fellenius (2014-a) to
analyze the results of bidirectional tests and its general b) the value of k associated to the Ratio Function cases was
form is included in Fig. A-2. taken as:
Note that:
Alr
a) while the Mathematical Model presented in this paper k= (A-1)
was applied to the cases related to the Cambefort Rela- K r ×( y 1 / 2)
tion, the Coyle-Reese (1966) Method was used to deal
in substitution of Eq. 7. The invariance of the upward
with the Ratio Function; and
curves at the top of the O-Cell (yf of fig. A1) is remarkable.
Acknowledgments
The author acknowledges the support given by
EPUSP - Escola Politécnica of São Paulo University.
References
Alonso, U. & Silva, P.E.C.A.F. (2000). Equivalent curve of
a continuous flight auger pile submitted to a test with
the hydrodynamic expansive cell (EXPANCELL). 4th
Seminar on Special Foundations Engineering and Geo-
technics, v. 1, p. 416-425, São Paulo (In Portuguese).
Coyle, H.M. & Reese, L.C. (1966). Load transfer for axi-
ally loaded piles in clay. Journal of Soil Mechanics and
Foundation Division. Proc. of the ASCE, v. 92:2,
p.1-26, March.
Décourt, L. & Quaresma, A.R. (1978). Load capacity of
Figure A1 - Bidirectional tests for piles from SPT. In: Proc. VI Brazilian Conference on
Massad
Soil Mechanics and Foundation Engineering, Rio de Ja- Randolph, M.F., & Wroth, C.P. (1978). Analysis of defor-
neiro, ABMS, p. 45-53 (In Portuguese). mation of vertically loaded piles. Journal of the Geo-
Fellenius, B.F. (1980). The analysis of results from routine technical Engineering Division, v. 10:12, p.:
pile load tests. Ground Engineering, v. 13, September, 1465-1488, December.
p. 19-31. Silva, P.E.C.A.F. (1983). Hydrodynamic expansive cell. A
Fellenius, B.F. (2014-a). Analysis of results from routines new way to perform loading tests. Independent pub-
static loading tests with emphasis on the bidirectional lisher, Belo Horizonte, Minas Gerais State, Brazil, 106
test. XVII Brazilian Conference on Soil Mechanics and p. (In Portuguese).
Geotechnical Engineering, Goiânia, Brasil. Silva, P.E.C.A.F. (1986). Hydrodynamic expansive cell. A
Fellenius, B.F. (2014-b). Basic of design of piled founda- new way to perform loading tests.. VIII Brazilian Con-
tions: the static loading test - Performance, instrumenta- ference on Soil Mechanics and Foundation Engi-
tion and interpretation. Lecture given on September 8, neering, p. 223-241. Porto Alegre (In Portuguese).
2014, in Belo Horizonte, Brazil.
Fellenius, B.F. (2015). The red book basic foundation de- List of Symbols
sign. Eletronic edition, Jan/2014
(http://www.fellenius.net/papers.html). Al: Total lateral (shaft) load
Alr: Total lateral (shaft) load at failure
Fonseca, A.V.; Santos, J.A.; Esteves, E.C. & Massad, F.
B; Breb: Cambefort Parameters (see Fig. 4-a)
(2007). Analysis of piles in residual soil from granite
c: Leonards & Lovell Coefficient (see Eq. 4)
considering residual loads. Soils and Rocks, v. 30, p.63
c’: Correlate of c for bidirectional tests
- 80.
D: Diameter of solid pile
Kim, S.R. & Chung, S.G. (2012); Equivalent head-down E: Modulus of elasticity of the pile
load vs. movement relationships evaluated from bi- f: Unit skin friction
directional pile load tests. Journal of Civil Engineering fu: Maximum (ultimate) unit skin friction
of KSCE 16(7):1170-1177. fu1; fu2: fu of layers 1 and 2
Kwon, O.S.; Choi, Y.K.; Kwon, O.K. & Kim, M.M. (2005). fu: Residual unit skin friction
“Comparison of the bidirectional load test with the h: Pile length embedded in soil
top-down load test.” Transportation Research Record. h1; h2: Thickness of the layers of the subsoil
1936, Transportation Research Board, Washington, k: Relative stiffness of the pile-soil (shaft)
D.C., pp. 108-116. Kr: Pile stiffness, as a structural piece
Leonards, G.A . & Lovell, D. (1979). Interpretation of load Ph : Residual toe load
tests in high capacity driven piles. Behavior of deep reb
Ph : Residual toe load at the end of rebound
foundations, ASTM STP 670, Raymond Lundgren, Ed., Po: Vertical load at the pile head
ASTM, p. 388-415. Po3: Po associated to the initial shaft friction mobilization
Loadtest (2001). Construction of the equivalent top-loaded Po4: Po associated to the full shaft friction mobilization
max
load settlement curve from the results of an O-cell test, Po : Maximum value of Po
Loadtest Appendix to Reports. qp: Toe pressure
Loadtest (2014). Retrieved on January, 04th, 2014, from qu: Ultimate toe pressure
www.loadtest.com. Qp: Toe load
Massad, F. (1995). Pile analysis taking into account soil ri- Qpmax: Maximum value of Qp
gidity and residual stresses. X Pan-American Confer- R; R’;Rreb: Soil stiffness at the pile toe (see Fig. 4-b)
ence on Soil Mechanics and Foundation Engineering, S: Cross sectional area of the pile shaft
México, v. 2. p: 1199-1210. Sp: Cross sectional area of the pile toe
Massad, F. & Lazo G. (1998). Graphical method for the in- SPT: Standard Penetration Test blow count
terpretation of the load-settlement curve from vertical y: Pile movement
load tests on rigid and short piles. XI Brazilian Confer- yo; yp: Movements of the pile at head and bottom
ence on Soil Mechanics and Geotechnical Engineering. yh; y’p: Movements of the pile top (Bidirectional test)
v. 3, p. 1407-1414, Brasília, Brazil (In Portuguese). yf: Upward movement at the top of O-Cell
Massad, F. (2001). On the use of the elastic rebound to pre- y1; y1R: See Figs. 4-a
dict pile capacity. In: 15th International Conference of y3; y2; y2R: See Figs. 4-b
Soil Mechanics and Geotechnical Engineering, 2001, z: Square root of k
Istanbul. London: Balkema, v. 2: 1100-1104. De: Elastic pile shortening
De max: Maximum elastic shortening of the pile shaft
s
Osterberg, J. (1989). New load cell testing device. Proceed-
ings - 14th Annual Conference p. 17-28. Deep Founda- r: Elastic rebound measured at the pile head
tions Institute. q: See Fig. A-2 of the Appendix
Technical Notes
Soils and Rocks

v. 38, n. 3
O.M. Vilar, R.A. Rodrigues
Abstract. The paper revisits and updates two expedite methods used to identify the occurrence of collapsible soils. The
methods were proposed by Gibbs (1961) and de Mello (1973) and have in common the assumption that collapsible soils are
usually low density soils characterized through the relative compaction at natural condition, ratio between in situ
dry-density and maximum dry density from Standard Proctor compaction test. The method by Gibbs (1961) considers in
addition the soils moisture deviation, the difference between in situ moisture content and optimum moisture content. Data
from different origin soils of all over the world were used and analyzed and equations were proposed to separate collapsible
and non-collapsible soils. The updated procedures, as the original ones, were intended for use during preliminary
investigation to identify potentially collapsible soil and to serve as a basis for planning more precise methods of
investigation.
Keywords: collapsible soils; identification; unsaturated soil.
1. Introduction are of colluvial nature and reach depths of about 10 m.

Their characteristics are: specific gravity, Gs = 2.63, dry
Some soils under constant stress show volume de- density, rd = 1.44 g/cm3, moisture content, w = 7.2%, void
crease related to an increase of moisture content. This wet- ratio, e = 0.85, porosity, n = 46%, liquid limit, wL = 18%,
ting-induced or soil collapse deformation characterizes the plastic limit, wP = 11%, clay = 15%, silt = 6% and
so called collapsible soils. Collapse deformation is a typical sand = 79% and they typically classifies as SC in the
behavior of non-saturated low dry density soils and has Unified Soil Classification System. Standard Proctor maxi-
been reported as occurring in many places all over the mum dry density, rdmax, is 2.04 g/cm3, which is associated to
world in soils of different genesis such as aeolian, alluvial, optimum moisture content, wopt, of 8.6%. From Fig. 1(a)
colluvium and residual soils and even in poorly compacted one can observe the influence of stress on the magnitude of
embankments (Dudley 1970, Clemence & Finbar 1981, soil collapse and in Fig. 1(b) the differences in unsaturated
Vilar et al., 1981, Vilar and Gaioto 1994). and soaked sample, attributed to soil collapse.
Soil collapse can take place upon wetting for a wide
range of applied stress. As the applied stress is increased, Field tests also have been performed with procedures
the amount of collapse deformation experienced by an un- similar to the ones used in laboratory tests. The tests used
saturated soil reaches a maximum and then decreases to a were plate and pile load tests (Cintra 1998), cone penetra-
negligible value. The maximum value attained depends on tion tests (Ferreira et al., 1989), and load tests especially de-
the soil type, density and moisture content. Larger collapse signed to measure the phenomenon (Ferreira 1993,
deformations are associated to low densities and low de- Houston et al., 1988).
grees of saturation. (Alonso et al., 1990). Suction has long been recognized as a fundamental
The classical oedometer test with some variation has variable in the understanding of the mechanical behavior of
been extensively used to characterize and to quantify the unsaturated soils (Escario & Saez 1973, Fredlund &
soil collapse. In one option, the unsaturated specimen is Morgentern 1977, Alonso et al., 1990). Thus modern op-
loaded until a load of interest and, after equilibrium of load tions of oedometer tests with controlled suction are cur-
deformation, is wetted, thus allowing measuring the wet- rently available, allowing a detailed description of the col-
ting induced deformation. The other option is the double lapse phenomenon. For instance, Fig. 2 shows compression
oedometer test (Jennings & Knight 1957), performed with curves from oedometer tests with controlled suction using
two similar samples, one unsaturated and the other soaked the axis translating technique obtained in the same collaps-
since the beginning of load application. ible soil presented in Fig. 1. The tests were performed on
Figure 1 illustrates typical results of soil collapse us- samples with initial suction, s, of 200 kPa. After equilib-
ing both options of test. The soil tested is from the North- rium of suction, each specimen was loaded under net nor-
west Region of Sao Paulo State, Brazil and has experienced mal stress of 50, 100, 200 and 400 kPa allowing for the
collapse strains after the filling of a reservoir (Vilar & equilibrium of deformation under load. Then the suction
Rodrigues, 2011). This region is covered by sandy soils that was gradually reduced to 0 kPa and the corresponding col-
Orencio Monje Vilar, Professor, Escola de Engenharia de São Carlos, Universidade de São Paulo, São Carlos, SP, Brazil. e-mail: [email protected].
Roger Augusto Rodrigues, Assistant Professor, Faculdade de Engenharia de Bauru, Universidade Estadual Paulista “Júlio de Mesquita Filho”, Bauru, SP, Brazil. e-mail:
[email protected].
Submitted on April 4, 2015; Final Acceptance on November 11, 2015; Discussion open until April 30, 2016.
Vilar & Rodrigues
Figure 1 - Conventional oedometer test results from Pereira Barreto-SP. Brazil: (a) single oedometer tests; (b) double oedometer tests.
lapse strains were measured. In Fig. 2 (a) the variation of ear works such as canals and roads. In this context, this pa-
collapse deformation with net vertical stress, the difference per revisits and updates former criteria proposed by Gibbs
between total stress and pore air pressure, s - ua, can be ap- (Gibbs, 1961; USBR, 1998) and by de Mello (1973), ana-
preciated, indicating the influence of stress on collapse lyzing the density characteristics of soils and data from soil
strain. These are small for lower stress, tend to increase collapse occurrences registered in many parts of the world.
with load and then decrease and negligible values of strain
can be attained if larger loads are applied.
2. Some Features of Methods to Identify
Figure 2 (b) shows collapse strain (DHc/H) that arises Collapsible Soils
as suction is reduced, where DHc is the height variation of The existing methods for identifying collapsible soils
the sample after a stage of suction reduction and H is the generally rest on some basic principles: a) regional meth-
specimen height. As can be seen in Fig. 2 (b), in the initial ods developed from empirical concepts and expedite tests,
stages of suction reduction, the collapse potential is negligi- such as the methods based on consistency limits and physi-
ble and it tends to increase for the lower values of suction, cal indexes (Denisov 1951, Feda 1966, Sultan 1971, Gibbs
reaching its maximum as suction approaches zero. & Bara 1967); b) methods based on oedometer tests with
Wetting-drying cycles such as those that seasonally takes wetting (Denisov 1951, Jennings & Knight 1957, Reginatto
place in natural soil did not introduce additional deforma- & Ferrero 1973) and c) methods based on field test, as the
tion, as expected. cone penetration test (Ferreira et al., 1989) or specially de-
Figure 2(c) shows the influence of suction on con- signed plate load tests (Houston et al., 1988, Ferreira 1993).
fined compression curves. Each pair of curves of saturated Many of the empirical criteria incorporate the idea of
specimen (s = 0 kPa) and at a known suction can be consid- a low density soil, expressed in different direct and indirect
ered as a double oedometer test and allows observing the ways. Gibbs (1961) has used a criterion based on the rela-
influence of suction (or moisture content) on collapse tionship between in-place dry density and laboratory maxi-
strain, calculated from the difference in void ratio at a mum dry density and the moisture content deficiency, as
known load. shown on Fig. 3(a). If the relative compaction and moisture
The improvement of collapsible soil testing, both in content difference plot below and to the right of the limit
laboratory and in the field, has been accompanied by the de- line, significant wetting induced deformation must be ex-
velopment of analytical models, such as the Barcelona Ba- pected even under low pressures, and treatment of in-place
sic Model (Alonso et al., 1990), which allows to reproduce materials may be required. This criterion was developed on
the unsaturated soil behavior, including soil collapse, in a the basis of the Bureau of Reclamation experience, consid-
more comprehensive way. However, it is necessary to rec- ering soils classified as ML, CL, ML-CL, SM and MH and
ognize that the use of more elaborated testing techniques the loads within the range applicable for small dams
and sophisticated constitutive models demand expertise, (USBR 1998). This option comes to the heart of the prob-
are time consuming and expensive. In many applications, lem since it encompasses soil looseness and moisture defi-
the geotechnical engineer needs simple and straightforward ciency, however it has not, apparently, gained acceptance
methods to deal with some particular problem, especially in of geotechnical community and it is hardly referenced in
the preliminary parts of the project. This is the case of iden- the literature issued after 1970.
tification or detection of the occurrence of collapsible soils In the same report Gibbs (1961) has proposed an al-
in a particular site or large areas, as those traversed by lin- ternative method for fine grained soils that was based on in
Figure 2 - Suction controlled oedometer test results from Pereira Barreto-SP. Brazil: (a) tests with initial suction of 200 kPa; (b) collapse
potential during suction decrease and wetting-drying cycle; (c) compression curves for different suction (Rodrigues & Vilar 2006; Vilar
& Rodrigues 2011).
situ dry unit weight and liquid limit, as sketched in Fig. 3(b) embodied in other proposal, usually incorporating some
taken from Gibbs & Bara (1967). This method, from now local correction factors trying to better match field observa-
on called the Gibbs and Bara method, rests on the soil void tion and criteria result (Denisov 1951, Feda 1966).
ratio expressed through in-situ dry density and liquid limit.
Case I shows a soil whose volume of voids is larger than De Mello (1973) has considered for colluvium soils
that required to hold the volume of water needed to reach similar to those illustrated in Fig. 1, that the condition for
the liquid limit. Saturation will result in moisture content in collapse deformation to occur is that the relationship be-
excess of the liquid limit, consistency will be low and the tween in-place dry density and laboratory maximum dry
potential for collapse would be high. In this case, if collapse density should be lower than 80%. In fact, Gibbs & Bara
did not occur, the soil would be in a very sensitive condition (1967) method applied to the soil of Fig. 1 lead to a collaps-
(Knodel 1981). If voids volume is less than that required to ible soil as the void ratio capable of retaining the moisture
hold the moisture content at the liquid limit, as shown by content associated to the liquid limit is 0.48, which is lower
Case III, the soil at saturation will remain in the plastic state than natural void ratio (0.85). In addition, the relationship
and have greater resistance against particle shifting and between in-place dry density and laboratory maximum dry
only settle as a normal result of loading. The extensive and density (Standard Proctor) is 76%, which is lower than the
successful use of this method is shown in Gibbs & Bara value proposed by de Mello (1973), thus also, classifying
(1967) and Knodel (1981). Similar concepts are usually this as a collapsible soil.
Vilar & Rodrigues
Figure 3 - Criteria for identifying collapsible soils. (a) based on relative compaction and moisture content difference; (b) based on dry
unit weight and liquid limit (USBR, 1998).
3. Updating an Expedite Methods to Identify (rdmax) of soil as given by a Standard Proctor Test (ASTM D
698). Using these parameters it is possible to determine the
Potentially Collapsible Soils natural relative compaction (RCn) of the soil.
In spite of the successful application of the Gibbs and rd

RCn = ´ 100(%) (1)
Bara method, it is worth to recognize that it is not able to rd max
identify non-plastic soils and, in some instances, fail to
identify some collapsible soils as can be confirmed check- To define the critical condition for collapse, it was as-
ing some of the data of Table 1 presented in the Appendix, sumed that soil density was of highest importance because,
for instance soils 12, 14, 36, 63, and 67. Thus, based on the for this type of deformation to occur, soils must be suffi-
general concepts and physical characteristics of collapsible ciently loose so that they are capable of collapsing when
soils reported in the literature, it seems consistent to pursue their particle-to-particle bond is weakened by wetting or
a criterion to identify collapsible soils, considering in a di- suction decrease.
rect way the usual looseness of collapsible soil, through However, the collapse also depends on the moisture
compaction parameters. In this sense, the authors recover content of the soil or on the degree of saturation. Although
the proposals by Gibbs (1961) and de Mello (1973) that are the soils must first be in a critically low-density condition
updated with some data of collapsible soils from various for a collapse of structure to occur, the existing moisture
parts of the world. Those proposals are reanalyzed and the content or the associated suction must be also considered an
possible range of values at which a soil should be consid- important part in the analysis. Soils which were already at
ered as collapsible is enlarged. The proposed methods take an in situ high degree of saturation were expected to be
into account the relative compaction of natural soil, relating minimally affected by additional wetting or suction de-
the in situ dry unit weight (rd) with maximum dry density crease, as they will compress under load leaving small
room for collapse to takes place. Soils drier than optimum Inspection of Table 1 and Fig. 4 shows that the vast
moisture content are capable to sustain appreciable loads majority of collapsible soils are low dry density and dry
with little compression due to the rigidity introduced by soil soils, that is, they are poorly compacted materials showing
3
suction and were considered prone to collapse as indicated dry density lower than 1.50 g/cm and moisture content de-
by the many results available (Escario and Saez 1973, Vilar ficiency. Collapsible behavior was also noticed for soils
and Gaioto 1994). To take into account the influence of wetter than optimum moisture content, however always re-
moisture, the method by Gibbs (1961) also considers the lated to loose soils, with natural degree of compaction
moisture content deviation (Dw), the difference between lower than approximately 85%. In general, collapsible soils
the actual soil moisture content (w) and the Standard Proc- showed RCn lower than 90%, although at least one soil with
tor optimum moisture content (wopt). RCn = 95% has suffered collapse. In this case the moisture
content deviation was of about -8%, suggesting that even
Dw = w - wopt (2) denser soils if dry can be subjected to collapse deformation
The value of Dw is negative when the soil is drier than if under large overburden stress. However, in many of these
optimum moisture content and positive when the soil is situations related to dense and fairly dry soils, the stress
wetter than optimum moisture content. needed to induce collapse strains are very large, beyond the
As the original criteria analyzed do not consider load range of practical interest for the Geotechnical Engineer.
and the available information about load and collapse is Very few data of non-collapsible soil has been reported and
limited, this variable was not considered in this paper. they show usually RCn larger than 90% as is the case of the
To evaluate the boundary between collapsible and soil from arid climate tested by Carvalho (1994) that has
non collapsible soils different sets of test results on natural not shown wetting induced strains although they were dry
specimens of collapsible and non-collapsible soils were and wetted at stresses as large as 400 kPa or present large
used. These data are gathered in Table 1 and as many of the moisture content when RCn is lower than 90% as reported
original papers did not present all the soil characteristics by Arman & Thornton (1973).
data, some values were calculated or even assumed, when The trend shown by the data set in Fig. 4 confirms the
needed. In assuming data, especially for the Standard Com- original proposition by Gibbs, however also indicates that it
paction Tests, practical Tables based on Universal Classifi- should be slightly enlarged since some reported collapsible
cation System were used such as Table 2 in Appendix, soils plotted above the boundary line. In this sense, the
adapted from USBR (1998). In the case of calculated or as- available information has lead to the proposition of the
sumed values, the data are inserted in Table 1 in italic. In limit line drawn in Fig. 4 that separates collapsible and
the same table, soils said as non-collapsible are identified non-collapsible soils. If the natural relative compaction and
by an asterisk. moisture content deficiency plot below the limit line, sig-
Figure 4 shows the plot of natural relative compaction nificant wetting induced deformation should be expected
(RCn) and moisture content deficiency of the soil. and additional and more directed investigation may be per-
formed to adequately characterize collapsing behavior,
which necessarily includes the state of stress acting on the
soil. As it is known, the magnitude of collapse strain de-
pends on the stress and usually after reaching a maximum it
then decreases to negligible values as the load is increased
and the soil is compressed to a denser condition.
The limit lines of Fig. 4 also allows stating a relation-
ship between dry density and moisture content deviation
and the following condition should stand for a soil to be
considered potentially collapsible:
a) for dry soils (w < wopt)
rd
100 × £ 90 - 0.6Dw (3)
rd max
b) for wet soils (w > wopt)

rd
100 × £ 90 -18
. Dw (4)
rd max
Figure 4 - Relative compaction RCn versus moisture content devi- In these equations, Dw is expressed as percentage.
ation of collapsible soils. The relationships are valid up to -12% on the dry side and to
Vilar & Rodrigues
+14% on the wet side, since those were the limiting mois- during dry season, suggesting that their relative density is
ture content deviations found in the referenced papers. high. As an example, SPT performed in the dry season in a
As there are very few data above the limit line, cau- sandy soil of Petrolândia, Pernambuco, Brazil presents SPT
tion should be exerted for soils that plot in the vicinity of higher than 10 blows. After three hours of flooding the SPT
this limit line. In this case, additional investigation should was reduced to about 70% of original value in the shallower
be performed to confirm collapse behavior, considering the portions of the soil profile (Ferreira, 1995) and after the
key factors that influence collapse deformation. rainy season, some measured values of SPT were reduced
Figure 4 allows also suggesting a tentative single in- from 10 to about 5 blows, showing the influence of mois-
dex separating collapsing and non-collapsing soils. As can ture content and suction on the number of blows and on the
be seen the dry soils, that is, the soils with moisture content relative density (Souza Neto, 2004). The influence of mois-
lower that optimum moisture content show RCn lower than ture content or suction on the blow number of SPT has been
90%, figure that tends to decrease for moisture content addressed by other authors, such as Reginatto (1971) and
above the optimum water content. This index is larger than Camapum de Carvalho et al. (2001) suggesting that SPT is
the value proposed by de Mello (1973) and therefore should of limited value in identifying collapsible soils. A possible
replace it. alternative to overcome this point is to perform field tests
These expedite methods of collapsible soil identifica- with soil at natural moisture content and after flooding, ana-
tion were checked and updated to support the initial stages lyzing the differences between both values; however the
of investigation and to orientate, but not to substitute, the subject demands additional research in order to establish its
more elaborate investigation techniques, that will support efficiency.
the designer on choosing the solution for the problem under Finally, it is worth to say that that both criteria up-
analysis. Thus, by performing very simple and conven- dated in this paper indicate the collapsible nature of the
tional measurements, as the in situ unit weight and moisture sandy soil of Petrolândia, above referred to, since it shows
content and the parameters of Standard Proctor test it is natural relative compaction (RCn) lower than 80% and
possible to define whether a given soil is potentially col- moisture content deviation of about -7% in the dry season
lapsible. In situ density can be measured by conventional (Ferreira, 1995).
methods, such as the sand replacement method, among oth- 4. Conclusion
ers, while moisture content can be measured through
speedy moisture tester or other applicable method, such as Collapsible soils are typically low density non-satu-
the oven method. In some instances, and considering that rated soils that can be of different origins. These soils expe-
the proposed method was devised for preliminary investi- rience volume reduction or collapse strain when wetted at
gation, an experienced engineer can use results from other an almost constant stress, usually larger than the overbur-
sources and indirectly evaluate the soil characteristics of in- den stress. Collapse strains are known to depend on the dry
terest. For instance, soil identification following ASTM density and moisture content of the soil and on the stress
D2488 - Standard practice for description and identifica- acting on it. An expedite method first proposed by Gibbs
tion of soils (visual-manual procedure) - allows classifying (1961) to identify collapsible soils, which considers that
the soil according to the Universal Soil Classification Sys- collapsible soils are naturally poorly compacted soils, with
tem and obtain average Standard Proctor parameters from moisture content deficiency has been updated. The method
practical tables, such as Table 2 in the Appendix. takes into account both in situ density and moisture content
It must be recognized that the methods are difficult to and Standard Proctor compaction parameters of the soil, re-
apply to deeper soil horizons, unless some measurement of lated through the natural relative compaction (RCn), rela-
soil density and moisture content is available. In this case tionship between in situ dry density (rd) and maximum dry
and indeed in a more comprehensive way, a natural strategy density (rdmax) and moisture content deviation (Dw), the dif-
would be using some field test such as the common Stan- ference between the in situ soil moisture content (w) and
dard Penetration Test, searching for low density sandy and optimum moisture content (wopt). Both parameters are re-
clayey soils that are unsaturated. Some SPT profiles of col- lated through Eq. 3 for dry soils and Eq. 4 for wet soils, that
lapsible soils are available, especially from the Center- is soils that show moisture content larger that optimum
West and Southeast Brazil (Cavalcante et al., 2007 among moisture content.
others), which shows collapsible soils as being unsaturated The data gathered also allowed a simplified criterion
loose to medium compact sandy soils and unsaturated soft that takes into account only dry density. In this option, soils
clayey soils. However, the relationship between SPT and that show negative moisture deviation and natural relative
soil collapse must be seen with caution, since SPT is influ- compaction (RCn) lower than 90% are classified as collaps-
enced by moisture content or suction. For instance, for col- ible. This index tends to reduce as moisture content in-
lapsible soils developed under arid and semi arid climate, creases above optimum moisture content.
Ferreira (1995) and Souza Neto (2004) have shown that The updated methods were devised for preliminary
these highly desiccated soils can show high SPT values studies and to optimize more complete investigation analy-
sis, not to substitute them, remembering that besides dry Clemence, S.P. & Finbarr, A.O. (1981). Design consider-
density and moisture content, the collapse strains are also ations for collapsible soils. Journal Geotechnical Engi-
function of stress whose influence is not addressed in this neering Division, ASCE, 107(3):305-317.
paper. Clevenger, W.A. (1956). Experiences with loess as founda-
tion material. Journal of the Mechanics and Foundation
References Division, ASCE, 82(3):1-26.
Alawaji, H.A. (2001). Shear induced collapse settlement of Cintra, J.C.A. (1998). Foundations in Collapsible Soils.
arid Soils. Geotechnical and Geological Engineering, Rima, São Carlos (in Portuguese).
19(1):1-19. Collares, A.C.Z.B. & Vilar, O.M. (1998). Influence of satu-
th
Alonso, E.E.; Gens, A. & Josa, A. (1990). A constitutive ration liquid pH on soil collapse. In: Proc. 11 Brazilian
model for partially saturated soil. Géotechnique, Congress of Soil Mechanics and Geotechnical Engi-
40(3):405-430. neering, Brasília, pp. 123-129 (iIn Portuguese).
Arman, A. & Thornton, S.I. (1973). Identification of col- Conciani, W. (1997). Study of Soil Collapse by Plate Load
lapsible soils in Louisiana. Highway Research Record, Tests with Tensiometers and Computed Tomography.
426: 14-22. Ph.D. Thesis, University of São Paulo (in Portuguese).
Costa Jr., F.A. (2001). Infiltration Effects of Caustic Li-
ASTM D 698 (2007). Standard Test Methods for Labora-
queur on Resistance Characteristics of a Lateritic Soil.
tory Compaction Characteristics of Soil Using Standard
Ph.D. Thesis, Pontifical Catholic University of Rio de
Effort. Annual Book of ASTM Standards. ASTM Inter-
Janeiro (in Portuguese).
national, West Conshohocken.
de Mello, V.F.B. (1973). Class Notes of Soils Mechanics.
ASTM D 2488 (2009). Standard Practice for Description
EP-USP, University of São Paulo.
and Identification of Soils (Visual-Manual Procedure).
Delage, P.; Cui, Y.J. & Antoine, P. (2005). Geotechnical
Annual Book of ASTM Standards. ASTM Interna-
problems related with loess deposits in Northern
tional, West Conshohocken.
France. In: Proc. International Conference on Problem-
Beles, A.A.; Stanculescu, I.I. & Schally, V.R. (1969). Pre- atic Soils, Eastern Mediterranean University,
wetting of loess soil foundation for hydraulic structures. Famagusta, N. Cyprus, pp. 1-24.
th
In: Proc 7 International Conference on Soil Mechanics Denisov, N.Y. (1951). Mechanical Properties of Loess and
and Foundation Engineering, Mexico, v. 2, pp. 17-25. Loams. Gosstroiizdat, Moscou (in Russian).
Benvenuto, C. (1983). Theoretical-Experimental Analysis Derbyshire, E. & Mellors, T.W. (1988). Geological and
of the Behavior of Some Collapsible Soils. M.Sc. The- geotechnical characteristics of some loess and loessic
sis, University of São Paulo (in Portuguese). soils from China and Britain: A comparison. Engi-
Camapum de Carvalho, J.; Pereira, J.H.F.; Guimarães, R.C. neering Geology, (25):135-175.
& Abreu, P.S.B. (2001). Análise da influência da suc- Dudley, J.H. (1970). Review of collapsing soils. Journal of
ção nos resultados de SPT e SPT-T em solos porosos the Soil Mechanics and Foundation Division,
colapsíveis. In: Anais 4° Simp. Bras. sobre Solos Não 96(3):925-947.
Saturados, Porto Alegre, pp. 509-520. Escario, V. & Saez, J. (1973). Gradual collapse of soils
th
Cardoso, F.B.F.; Camapum de Carvalho. J. & Martins, E.S. originated by a suction decrease. In: Proc. 8 Interna-
(1998). Collapse phenomena in weathered soils with or- tional Conference on Soils Mechanics and Foundation
igins and depth different from Federal District. In: Pro- Engineering, Moscow, v. 4.2, pp. 123-124.
th
ceedings of the 11 Brazilian Congress of Soil Feda, J. (1966). Structural stability of subsident loess soil
Mechanics and Geotechnical Engineering, Brasília, pp. from Praha-Dejvice. Engineering Geology,
59-65 (In Portuguese). 1(3):201-219.
Carvalho, M.F. (1994). Study of Collapsible Soils from the Ferreira, C.V.; Lobo, A.S.; Giacheti, H.L. & Conciani, W.
Northeast of Minas Gerais State, Brazil. M.Sc. Thesis, (2004) Behavior of piles installed in collapsible soils.
th
University of São Paulo (in Portuguese). In: Proc. 5 Brazilian Symposium on Unsaturated Soils.
Cavalcante, E.H; Danziger, F.A.B.; Giacheti, H.L.; Cou- São Carlos, v. 1, pp. 345-350 (In Portuguese).
tinho, R.Q.; Souza, A.; Kormann, A.C.M.; Belicanta, Ferreira, R.C.; Monteiro, L.B.; Peres, J.E.E. & Benvenuto,
A.; Pinto, C. S.; Costa Branco C.J.M.; Ferreira, C.V; C. (1989). Some aspects on the behaviour of Brazilian
th
Carvalho, D. de; Marinho, F.A.M.; Cintra, J.C.A.; Dou- collapsible soils. In: Proc. 12 International Conference
rado, K.C. de A.; Moraes, L.S. de; Albuquerque Filho, on Soil Mechanics and Foundation Engineering, Rio de
L. H. A.; Almeida, M.S.S de; Gutierrez, N.H.M.; Albu- Janeiro, v. 1, pp. 117-120 (in Portuguese).
querque, P.J.R de; Chamecki, P.R.; Cunha, R.P. da; Ferreira, S.R.M. (1993). Volume change in expansive and
th
Teixeira, R.S.; Menezes, S.M. & Lacerda, W.A. (2007). collapsible unsaturated soils. In: Proc. 7 Brazilian
Campos experimentais brasileiros. Geotecnia, Congress of Engineering Geology, Poços de Caldas, v.
111:99-205. 1, pp. 283-296 (in Portuguese).
Vilar & Rodrigues
rd
Ferreira, S.R.M. (1995). Colapso e Expansão de Solos Campinas-SP, Brazil. In: Proc 3 Brazilian Symposium
Naturais Não Saturados devidos à Inundação. Tese de on Unsaturated Soils, Rio de Janeiro, v. 1, pp. 113-119
Doutorado, COPPE/UFRJ, 379 p. (in Portuguese).
Ferreira, S.R.M.; Fucale, S.P. & Amorim, S.F. (1998). Be- Mustafaev, A.A.; Chigniev, G.D. & Nazirova, G.R. (1974).
havior of volume change in collapsible soils studied Rheological nature of the deformations caused by
th
through field tests. In: Proc. 11 Brazilian Congress of swelling in clays and by collapse in loess soils. Soil Me-
Soil Mechanics and Geotechnical Engineering, Bra- chanics and Foundation Engineering, 11(5):325-330.
sília, pp. 171-178 (in Portuguese). Phien-wej, N.; Pientong, T. & Balasubramaniam, A.S.
Ferreira, S.R.M. (2015). Personal Communication, Federal (1992). Collapse and strength characteristics of loess in
University of Pernambuco. Thailand. Engineering Geology, 32:59-72.
Foss, I. (1973). Red soil from Kenia as a foundation mate- Reginatto, A.R. (1971). Standard penetration test in col-
th
rial. In: Proc. 8 International Conference on Soil Me- lapsible soils. . In: Proc. IV Cong. Panam. de Mec. de
chanics and Foundation Engineering, Moscow, v. 2.2, Suelos e Ing.de Fundaciones, Puerto Rico, v. 2, p.
pp. 73-80. 77-84.
Fredlund, D.G. & Morgentern, N.R. (1977). Stress State for Reginatto, A.R. & Ferrero, J.C. (1973). Collapse potential
th
Unsaturated Soils, Journal of Geotechnical Engineering of soils and soil-water chemistry. In: Proc. 8 Interna-
Division, ASCE, 103(GT5):447-466. tional Conference on Soil Mechanics and Foundation
Gibbs, H.J. (1961). Properties Which Divide Loose and Engineering, Moscow, v. 2.2, pp. 177-183.
Dense Uncemented Soils. Bureau of Reclamation Re- Reznik, Y.M. (1992). Determination of deformation prop-
port No, EM-608, Denver, CO (apud USBR, 1998). erties of collapsible soils. Geotechnical Testing Journal,
ASTM, 15(3):248-255.
Gibbs, H.J. & Bara, J.P. (1967). Stability problems of col-
Rodrigues, R.A. & Lollo, J.A. (2004). Structural, physio-
lapsing soil. Journal of the Soil Mechanics and Founda-
graphic and mechanical characteristics of two profiles
tion Division, ASCE, 93(4):577-594.
of collapsible soils from Ilha Solteira-SP. Brazil. Solos
Gutierrez, N.H.M.; Nóbrega, M.T. & Vilar, O.M. (2009).
e Rochas, 27:131-146 (in Portuguese).
Influences of the microstructure in the collapse of a re-
Rodrigues, R.A. & Vilar, O.M. (2006). Relationship be-
sidual clayey tropical soil. Bulletin of Engineering Ge-
tween collapse and soil-water retention curve of a sandy
ology and the Environment, 68:107-116. th
soil. In: Proc. 4 International Conference on Unsatu-
Houston, S.L.; Houston, W.N. & Spadola, D.J. (1988). Pre-
rated Soils, Carefree, Arizona, v. 1, pp. 1025-1036.
diction of field collapse of soils due to wetting. Journal
Souza Neto, J.B. (2004). The Behavior of a Collapsible Soil
of Geotechnical Engineering, ASCE, 114(1):40-58.
Evaluate Through in situ and Laboratory Testing and
Jennings, J.E. & Knight, K. (1957). The additional settle- Settlement Prediction Due to Wetting. Ph.D. Thesis,
ment of foundations due to a collapse of structure of Federal University of Rio de Janeiro, COOPE (in Portu-
th
sandy subsoils on wetting. In: Proc. 4 Conference on guese).
Soil Mechanics end Foundation Engineering, London, Sultan, H.A. (1971). Some Engineering Aspects of Col-
pp. 316-319. lapsing Soils. The University of Arizona, Tucson, 35 p.
Klukanová, A. & Frankovská, J. (1998). Soil collapse. In: rd
USBR (1998). Earth Manual, Part 1. 3 ed. Bureau of Rec-
th
Proc. 8 International IAEG Congress, Association for lamation, Denver.
Engineering Geology and Environment, Vancouver, Vilar, O.M.; Rodrigues, J.E. & Nogueira, J.B. (1981). Col-
pp. 3349-3355. lapsible soils: A problem for tropical soil engineering.
Knodel, P.C. (1981). Construction of large canal on col- In: Proc. Brazilian Symposium of Tropical Soils, Rio de
lapsing soils. Journal of the Geotechnical Engineering Janeiro, v. 1, pp. 209-224 (In Portuguese).
Division, ASCE, 107(GT1):79-94. Vilar, O.M. & Gaioto, N. (1994). Collapsible behavior of a
Lutenegger, A.J. & Saber, R.T. (1998). Determination of nd
compacted lateritic soil. In: Proc. 2 Brazilian Sympo-
collapse potential of soils. Geotechnical Testing Jour- sium on Unsaturated Soils, Recife, pp. 185-190 (in Por-
nal, ASTM, 11(3): 173-178. tuguese).
Mahler, C.F. & Mendonça, M.B. (1994). Micro-structural Vilar, O.M. & Rodrigues, A.R. (2011). Collapse behavior
th
aspects of the collapsible soils from Bahia. In: Proc. 10 of soil in a Brazilian region affected by a rising water ta-
Brazilian Congress of Soil Mechanics and Geotechnical ble. Canadian Geotechnical Journal, 48:226-233.
Engineering, Foz do Iguaçu, pp. 1149-1156 (in Portu- Zur, A. & Wiseman, G. (1973). A study of collapse phe-
th
guese). nomena of an undisturbed loess. In: Proc. 8 Interna-
Monacci, M.D.; Carvalho, D. & Albuquerque, P.J.R. tional Conference on Soils Mechanics and Foundation
(1997). Collapsibility analysis of a residual soil from Engineering, Moscow, v. 2.2, pp. 265-269.
Appendix
Table 1 - Data from collapsible soil.
Soil Clay Silt Sand wL wP Class. n Sr rd w rdmax wopt Dw RCn Reference

(%) (%) (%) (%) (%) USCS (%) (%) (g/cm3) (%) (g/cm )
3
(%) (%) (%)
1 32 16 52 27 13 SC 42 37 1.551 10.1 2.125 12.4 -2.3 73 Benvenuto
(1983)
2 27 7 66 23 16 SC 50 41 1.510 - 1.936 12.4 - 78 Ferreira et al.
(1989)
3 2 8 89 NL NP SW 40 13 1.590 3.4 1.980 10.8 -7.4 80 Ferreira
(1993)
4 3 19 76 18 13 SM 42 6 1.544 1.5 1.970 11.4 -9.9 78 Ferreira
(1993)
5 50 30 20 50 25 CL 66 45 0.900 32.8 1.768 16.4 16.4 51 Ferreira et al.
(1989)
6 50 37 13 80 46 MH 65 - 0.931 - 1.372 33.1 - 68 Ferreira et al.
(1989)
7 61 6 33 54 27 CH 61 42 1.030 25.3 1.531 24.8 0.5 67 Ferreira et al.
(1989)
8 42 15 43 41 19 CL 54 38 1.270 16.1 1.768 16.4 -0.3 72 Ferreira et al.
(1989)
9 45 20 35 37 26 ML 63 39 1.130 21.6 1.645 20.1 1.5 69 Ferreira et al.
(1989)
10 41 24 25 48 35 ML 65 55 1.080 33.0 1.645 20.1 12.9 66 Ferreira et al.
(1989)
11 15 35 50 25 15 SC 46 40 1.440 12.8 1.906 12.4 0.4 76 Ferreira et al.
(1989)
12 46 12 42 58 27 CH 54 60 1.240 26.2 1.531 24.8 1.4 81 Ferreira et al.
(1989)
13 45 5 50 58 34 SM 59 50 1.120 26.2 1.906 12.4 13.8 59 Ferreira et al.
(1989)
14 60 15 25 70 30 CH 61 56 1.050 32.8 1.531 24.8 8.0 69 Ferreira et al.
(1989)
15 - - - 50 10 CL 54 33 1.250 14.4 1.768 16.4 -2.0 71 Ferreira et al.
(1989)
16 32 23 44 28 18 SC 46 34 1.414 10.7 1.802 13.9 -3.2 78 Carvalho
(1994)
17 27 14 59 NL NP SP 51 3 1.305 1.1 1.827 10.5 -9.4 71 Cardoso et al.
(1998)
18 56 3 34 36 25 CL 46 41 1.388 13.2 1.620 19.1 -5.9 86 Carvalho
(1994)
19(*) 58 6 33 38 25 CL 42 45 1.469 12.6 1.574 20.2 -7.6 93 Carvalho
(1994)
20 39 27 33 32 20 CL 44 34 1.467 9.9 1.753 14.9 -5.0 84 Carvalho
(1994)
21 17 51 32 28 18 CL 45 27 1.430 8.7 1.742 15.0 -6.3 82 Carvalho
(1994)
22(*) 44 39 17 36 22 CL 34 57 1.740 10.8 1.768 16.4 -5.6 98 Carvalho
(1994)
23(*) 5 24 71 NL NP SP 31 36 1.831 6.0 1.827 10.5 -4.5 100 Carvalho
(1994)
Vilar & Rodrigues
Table 1 - Cont.

(%) (%) (%) (%) (%) USCS (%) (%) (g/cm3) (%) (g/cm )
3
(%) (%) (%)
24 15 6 79 18 11 SC 46 22 1.440 7.2 2.040 8.6 -1.4 71 Rodrigues
and Vilar
(2006)
25 16 6 78 - - SC 36 4 1.684 0.8 2.100 8.1 -7.3 80 Mahler and
Mendonça
(1994)
26 43 12 45 37 25 ML 67 30 0.930 21.5 1.645 20.1 1.4 57 Conciani
(1997)
27 39 51 10 45 32 ML 64 50 0.971 33.1 1.645 20.1 13.0 59 Conciani
(1997)
28 67 19 13 52 35 MH 64 41 1.072 24.3 1.372 33.1 -8.8 78 Monacci et
al. (1997)
29 39 48 13 66 42 MH 60 53 1.152 28.1 1.372 33.1 -5.0 84 Monacci et
al. (1997)
30 24 35 41 28 18 SC 49 42 1.370 14.8 1.715 17.6 -2.8 80 Ferreira et al.
(1998)
31 7 22 70 23 14 SC 41 19 1.620 5.2 1.965 11.1 -5.9 82 Ferreira et al.
(1998)
32 48 39 13 51 38 MH 67 43 0.953 30.4 1.650 29.6 0.8 58 Collares and
Vilar (1998)
33 13 22 65 23 14 SC 43 22 1.540 6.1 1.906 12.4 -6.3 81 Costa Jr.
(2001)
34 74 16 10 57 42 MH 66 49 1.018 37.8 1.372 33.1 4.7 74 Gutierrez et
al. (2009)
35 66 28 6 61 42 MH 62 65 1.133 35.9 1.372 33.1 2.8 83 Gutierrez et
al. (2009)
36 27 8 65 25 15 SC 46 30 1.440 9.6 1.960 11.7 -2.1 73 Rodrigues
and Lollo
(2004)
37 clayey sand (0-6 m) 41 23 SC 51 56 1.374 20.8 1.906 12.4 8.4 72 Ferreira et al.
clayey silt (6-12 m) (2004)
38a 7 2 91 NL NP SP-SM 39 4.6 1.628 1.08 1.800 11.3 -10.2 90.4 Souza Neto
(2004)
38b 9 3 88 14 NP SP-SM 39 6.9 1.618 1.68 1.880 11.0 -9.3 86.0 Souza Neto
(2004)
38c 14 2 84 16 12 SM 39 8.7 1.615 2.11 1.970 9.7 -7.6 82.0 Souza Neto
(2004)
38d 15 0 82 17 15 SM 37 9.8 1.673 2.15 2.030 8.8 -6.6 82.4 Souza Neto
(2004)
38e 16 5 79 19 13 SM/SC 31 19.1 1.833 3.18 2.000 10.2 -7.0 91.7 Souza Neto
(2004)
39 - - - - - - - - - - - - -11.6 89.0 Ferreira
(2015)
40 - - - - - - - - - - - - -1.6 87.2 Ferreira
(2015)
41 - - - - - - - - - - - - -3.9 79.5 Ferreira
(2015)
42 10-15 82-87 3 NL NP SM 48 5 1.409 1.7 1.877 12.3 -10.6 75 Lutenegger
and Saber
(1988)
Table 1 - Cont.

(%) (%) (%) (%) (%) USCS (%) (%) (g/cm3) (%) (g/cm )
3
(%) (%) (%)
43 10-15 82-87 3 33 NP SM 51 4 1.329 1.4 1.877 12.3 -10.9 71 Lutenegger
and Saber
(1988)
44 10-15 82-87 3 33 NP SM 50 7 1.358 2.5 1.877 12.3 -9.8 72 Lutenegger
and Saber
(1988)
45 11 68 - 31 24 ML 47 23 1.343 8.5 1.645 20.1 -11.6 82 Klukanová
and
Frankovská
(1998)
46 7 38 - 24 21 SM 38 25 1.738 5.0 1.877 12.3 -7.3 93 Klukanová
and
Frankovská
(1998)
47 6 26 - 25 20 SM-SC 40 39 1.656 9.1 1.877 12.3 -3.2 88 Klukanová
and
Frankovská
(1998)
48(*) 29 50 - 43 25 CL 36 45 1.738 9.5 1.768 16.4 -6.9 98 Klukanová
and
Frankovská
(1998)
49(*) 15 62 - 27 16 CL 31 37 1.775 6.8 1.768 16.4 -9.6 100 Klukanová
and
Frankovská
(1998)
50 15 38 39 29 16 CL 44 60 1.535 17.0 1.768 16.4 0.6 87 Feda (1966)
51 17 42 41 26 16 CL 42 56 1.574 14.8 1.768 16.4 -1.6 89 Feda (1966)
52(*) 30 42 28 39 21 CL 39 76 1.664 17.8 1.768 16.4 1.4 94 Feda (1966)
53 20 30-40 40-50 18 13 SM- 46 27 1.430 8.5 1.940 9.7 -1.2 74 Phien-wej at
SC al (1992)
54 16 21 63 21 17 SM- - - 1.254 8.3 1.874 12.0 -3.7 67 Alawaji
SC (2001)
55 19 35 45 23 17 SM- - - 1.339 18.8 1.835 14.0 4.8 73 Alawaji
SC (2001)
56 67% passing 200 sieve - - - 46 1.433 9 1.868 11.0 -2.0 77 Houston et al.
(1988)
57 18 72 10 28 22 CL - 48 53 1.390 18.1 1.645 20.1 -2.0 84 Delage et al.
ML (2005)
58 - - - 39 16 CL 49 29 1.367 10.5 1.768 16.4 -5.9 77 Jennings and
Knight
(1957)
59 8 72 20 30 20 - - - 1.160 10.0 1.720 16.4 -6.4 67 Clevenger
(1956)
60 12 60 28 - - SM 48 25 1.564 6.8 1.877 12.3 -5.5 83 Mustafaev et
al. (1974)
61 18 - - - - SM 46 33 1.500 10.0 1.877 12.3 -2.3 80 Reznik
(1992)
62 78 - - 64 46 MH 60 44 1.100 24.0 1.372 33.1 -9.1 80 Foss (1973)
Vilar & Rodrigues
Table 1 - Cont.

(%) (%) (%) (%) (%) USCS (%) (%) (g/cm3) (%) (g/cm )
3
(%) (%) (%)
63 < 30 > 60 10 31 16 CL 44 53 1.540 15.0 1.768 16.4 -1.4 87 Zur and
Wiseman
(1973)
64 - - - - - CL 51 30 1.275 12.0 1.768 16.4 -4.4 72 Beles et al.
(1969)
65 - - - 45 - CL-SC 45 48 1.442 15.0 1.877 12.3 2.7 77 Gibbs and
Bara (1967)
66 silt (0.06 - 0.01 mm) 32 22 CL 47 1.404 - 1.768 16.4 - 79 Derbyshire
and Mellors
(1988)
67 18 95% passing 37 19 CL 48 24 1.380 8.5 1.772 17.5 -9.0 78 Sultan (1971)
200 sieve
68 18 95% passing 43 20 CL 43 38 1.482 11.3 1.772 17.5 -6.2 84 Sultan (1971)
200 sieve
69 18 95% passing 67 28 CH 45 40 1.432 12.6 1.772 17.5 -4.9 81 Sultan (1971)
200 sieve
70 - - - 23 17 CL-M 55 35 1.280 15.0 1.645 20.1 -5.1 78 Reginatto and
L Ferrero
(1973)
71 20 17 43 25 14 SC - - 1.280 - 1.810 - - 71 Dudley
(1970)
72 12 16 72 21 12 SC - - 1.632 - 1.984 - - 82 Dudley
(1970)
73 10 19 38 20 16 GW-S 37 34 1.708 7.4 2.226 12.3 -4.9 77 Dudley
W (1970)
74 5 85 10 NL NP SM 69 28 0.881 22.0 1.049 - - 84 Dudley
(1970)
75 2 68 30 NL NP SM 58 25 1.041 14.0 1.302 - - 80 Dudley
(1970)
76 7 84 9 33 25 ML 56 36 1.186 17.0 1.516 - - 78 Dudley
(1970)
77 10 70 20 - - ML - - 1.522 22.0 1.778 - - 86 Arman and
Thornton
(1973)
78 18 70 12 - - ML - - 1.234 18.0 1.626 - - 76 Arman and
Thornton
(1973)
79 10 70 20 - - ML - - 1.200 21.3 1.606 16.6 4.7 75 Arman and
Thornton
(1973)
80 10 70 20 - - ML - - 1.216 22.6 1.606 16.6 6.0 76 Arman and
Thornton
(1973)
81 10 70 20 - - ML - - 1.312 15.3 1.606 16.6 -1.3 82 Arman and
Thornton
(1973)
82 10 70 20 - - ML - - 1.264 12.2 1.606 16.6 -4.4 79 Arman and
Thornton
(1973)
83 10 70 20 - - ML - - 1.248 20.5 1.606 16.6 3.9 78 Arman and
Thornton
(1973)
Table 1 - Cont.

(%) (%) (%) (%) (%) USCS (%) (%) (g/cm3) (%) (g/cm )
3
(%) (%) (%)
84(*) 10 70 20 - - ML - - 1.520 18.0 1.606 16.6 1.4 95 Arman and
Thornton
(1973)
85(*) 10 70 20 - - ML - - 1.616 21.0 1.606 16.6 4.4 101 Arman and
Thornton
(1973)
86(*) 10 70 20 - - ML - - 1.664 20.0 1.606 16.6 3.4 104 Arman and
Thornton
(1973)
87(*) 10 70 20 - - ML - - 1.584 16.0 1.606 16.6 -0.6 99 Arman and
Thornton
(1973)
88 10 70 20 - - ML - - 1.168 20.0 1.606 16.6 3.4 73 Arman and
Thornton
(1973)
89 10 70 20 - - ML - - 1.328 18.0 1.606 16.6 1.4 83 Arman and
Thornton
(1973)
90(*) 10 70 20 - - ML - - 1.536 22.0 1.606 16.6 5.4 96 Arman and
Thornton
(1973)
91 10 70 20 - - ML - - 1.072 18.0 1.606 16.6 1.4 67 Arman and
Thornton
(1973)
92 18 70 12 - - ML - - 1.232 18.0 1.624 16.8 1.2 76 Arman and
Thornton
(1973)
93 18 70 12 - - ML - - 1.328 14.0 1.624 16.8 -2.8 82 Arman and
Thornton
(1973)
94 18 70 12 - - ML - - 1.248 17.0 1.624 16.8 0.2 77 Arman and
Thornton
(1973)
95 18 70 12 - - ML - - 1.248 20.0 1.624 16.8 3.2 77 Arman and
Thornton
(1973)
96(*) 12 68 20 - - ML - - 1.408 23.0 1.728 13.4 9.6 81 Arman and
Thornton
(1973)
97(*) 12 68 20 - - ML - - 1.456 20.0 1.728 13.4 6.6 84 Arman and
Thornton
(1973)
98(*) 12 68 20 - - ML - - 1.440 22.0 1.728 13.4 8.6 83 Arman and
Thornton
(1973)
99(*) 12 68 20 - - ML - - 1.392 24.0 1.728 13.4 10.6 81 Arman and
Thornton
(1973)
rs is solid density, n is porosity, rd is dry density, w is moisture content, rdmax is Standard Proctor maximum dry density, wopt is Standard
Proctor optimum moisture content, Dw is moisture deviation related to the Standard Proctor optimum moisture content and RCn is rela-
tive compaction, wL is liquid limit, wP is plasticity limit. (*) soils tested as non-collapsible.
Vilar & Rodrigues
Table 2 - Average engineering properties of compacted soils (adapted from USBR 1998, Earth Manual).
Soil GROUP NAMe USCS soil type Compaction

gd max (kN/m )
3
wopt (%)
Well-graded gravel GW 19.89 11.4
Poorly graded gravel GP 19.07 12.2
Silty gravel GM 18.19 15.7
Clayey gravel GC 18.54 14.2
Well-graded sands SW 20.19 9.1
Poorly graded sands SP 18.27 10.5
Silty sands SM 18.77 12.3
Clayey sands SC 19.06 12.4
Silt ML 16.45 20.1
Lean clay CL 17.68 16.4
Elastic silt MH 13.72 33.1
Fat clay CH 15.31 24.8
SOILS and ROCKS
Publication of
SPG - Portuguese Geotechnical Society
Volume 38, N. 3, May-August 2015
Author index
Abbaszadeh, M.M. 49 Marinho, F.A.M. 81
Bandeira,A.P.N. 27 Martins, A.P.S. 9
Bicalho, K.V. 121, 149 Massad, F. 163, 349
Bu, W. 181 Mattos, R.C. 219
Candeias, M. 3 Mercadier, D. 149
Cardoso, A.S. 95 Musarra, M. 163
Costa Guerra, N.M. da 231 Pacheco, M.P. 67
Coutinho, R.Q. 27 Pretto, J.H.F. 81
Cui, Y.-J. 121, 149 Rivera-Sar, J.M. 59
Danziger, B.R. 67
Rodrigues, R.A. 265
De Leon Ferreira, L. 135
Romanel, C. 121
Delgado-Martin, J. 59
Sanchez-Tembleque, F. 59
Falcon-Suarez, I. 59
Santana, T. 3
Freitas, A.C. 67
Froumentin, M. 149 Sestrem, L.P. 81
Futai, M.M. 9 Silva, F.T. 219
Gomes, R.C. 135 Tang, A.M. 149
Hemsi, P.S. 219 Viana, L. Andrade 231
Houston, S.L. 49 Vilar, O.M. 265
Juncosa-Rivera, R. 59 Vivacqua, G.P.D. 121, 149
Kawachi, E.Y. 219 Xu, H. 181
Kormann, A.C.M. 81 Zapata, C.E. 49
Lacerda, W.A. 9 Zhao, Y. 181
Instructions for Submission of Manuscripts
Category of the Papers Conclusions: The content of this section should be based on the data
Soils and Rocks is the international scientific journal and on the discussions presented.
edited by the Brazilian Association for Soil Mechanics Acknowledgenments: If necessary, concise acknowledgements
and Geotechnical Engineering (ABMS) and the Portu- should be presented in this section.
guese Geotechnical Society (SPG). The aim of the jour- References: References to other published sources must be made in
nal is to publish (in English) original research and the text by the last name(s) of the author(s), followed by the year of publi-
technical works on all geotechnical branches. cation, similarly to one of the two possibilities below:
According to its content the accepted paper is classi- …while Silva and Pereira (1987) observed that resistance depended
fied in one of the following categories: Article paper, on soil density… or It was observed that resistance depended on soil den-
Technical Note, Case Study or Discussion. An article pa- sity (Silva and Pereira, 1987).
per is an extensive and conclusive dissertation about a In the case of three or more authors, the reduced format must be used,
geotechnical topic. A paper is considered as a technical e.g.: Silva et al. (1982) or (Silva et al., 1982). Two or more citations be-
note if it gives a short description of ongoing studies, longing to the same author(s) and published in the same year are to be dis-
comprising partial results and/or particular aspects of the tinguished with small letters, e.g.: (Silva, 1975a, b, c.). Standards must be
investigation. A case study is a report of unusual prob- cited in the text by the initials of the entity and the year of publication, e.g.:
lems found during the design, construction or the perfor- ABNT (1996), ASTM (2003).
mance of geotechnical projects. A case study is also Full references shall be listed alphabetically at the end of the text by
considered as the report of an unusual solution given to an the first author´s last name. Several references belonging to the same au-
ordinary problem. The discussions about published pa- thor shall be cited chronologically. Some examples are listed next:
pers, case studies and technical notes are made in the Dis- Papers: Bishop, A.W. & Blight, G.E. (1963). Some aspects of effective
cussions Section. stress in saturated and unsaturated soils. Geotechnique, 13(2):177-
When submitting a manuscript for review, the authors 197.
should indicate the category of the manuscript, and is also Books: Lambe, T.W & Whitman, R.V. (1979). Soil Mechanics, SI Ver-
understood that they: sion. John Wiley & Sons, New York, 553 p.
a) assume full responsibility for the contents and ac- Book with editors: Sharma, H.D.; Dukes, M.T. & Olsen, D.M. (1990).
curacy of the information in the paper; Field measurements of dynamic moduli and Poisson’s ratios of refuse
b) assure that the paper has not been previously pub- and underlying soils at a landfill site. Landva, A. & Knowles, G.D.
lished, and is not being submitted to any other periodical (eds), Geotechnics of Waste Fills - Theory and Practice, American So-
for publication. ciety for Testing and Materials - STP 1070, Philadelphia, pp. 57-70.
Manuscript Instructions Proceedings (printed matter or CD-ROM): Jamiolkowski, M.; Ladd,
C.C.; Germaine, J.T. & Lancellotta, R. (1985). New developments in
Manuscripts must be written in English. The text is to
field and laboratory testing of soils. Proc. 11th Int. Conf. on Soil
be typed in a word processor (preferably MS Word), sin-
Mech. and Found. Engn., ISSMFE, San Francisco, v. 1, pp. 57-153.
gle column, using ISO A4 page size, left, right, top, and
(specify if CD ROM).
bottom margins of 25 mm, Times New Roman 12 font,
Thesis and dissertations: Lee, K.L. (1965). Triaxial Compressive
and line spacing of 1.5. All lines and pages should be
Strength of Saturated Sands Under Seismic Loading Conditions. PhD
numbered. The text should be written in the third person.
Dissertation, Department of Civil Engineering, University of Califor-
The first page of the manuscript must include the title nia, Berkeley, 521 p.
of the paper followed by the names of the authors with the
Standards: ASTM (2003). Standard Test Method for Particle Size Analy-
abbreviation of the most relevant academic title. The af-
sis of Soils - D 422-63. ASTM International, West Conshohocken,
filiation, address and e-mail must appear below each au-
Pennsylvania, USA, 8 p.
thor´s name. An abstract of 200 words follows after the
Internet references: Soils and Rocks available at
author´s names. A list with up to six keywords at the end
http://www.abms.com.br and downloaded on August 6th 2003.
of the abstract is required.
On line first publications must also bring the digital object identifier
Although variations on the sequence and title of each
(DOI) at the end.
section are possible, it is suggested that the text contains
Figures shall be either computer generated or drawn with India ink on
the following sections: Introduction, Material and
tracing paper. Computer generated figures must be accompanied by the
Methods, Results, Discussions, Conclusions, Acknowl-
corresponding digital file (.tif, .jpg, .pcx, etc.). All figures (graphs, line
edgements, References and List of Symbols. A brief de-
drawings, photographs, etc.) shall be numbered consecutively and have a
scription of each section is given next.
caption consisting of the figure number and a brief title or description of
Introduction: This section should indicate the state the figure. This number should be used when referring to the figure in text.
of the art of the problem under evaluation, a description Photographs should be black and white, sharp, high contrasted and
of the problem and the methods undertaken. The objec- printed on glossy paper.
tive of the work should be clearly presented at the end of Tables shall be numbered consecutively in Arabic and have a caption
the section. consisting of the table number and a brief title. This number should be
Materials and Methods: This section should include used when referring to the table in the text. Units should be indicated in
all information needed to the reproduction of the pre- the first line of the table, below the title of each column. Acronyms should
sented work by other researchers. be avoided. When applicable, the units should come right below the cor-
Results: In this section the data of the investigation responding column heading. Additional comments can be placed as foot-
should be presented in a clear and concise way. Figures notes.
and tables should not repeat the same information. Equations shall appear isolated in a single line of the text. Numbers
Discussion: The analyses of the results should be de- identifying equations must be flushed with the right margin. International
scribed in this section. System (SI) units must be used. The definitions of the symbols used in the
equations must appear in the List of Symbols. It is recom- Figures, tables and equations should be numbered following the same
mended that the symbols used are in accordance with sequence of the original paper. All instructions previously mentioned for
Lexicon in 8 Languages, ISSMFE (1981) and the ISRM the preparation of article papers, case studies and technical notes also ap-
List of Symbols. ply to the preparation of discussions.
The text of the submitted manuscript (including fig- Editorial Review
ures, tables and references) intended to be published as an Each paper will be evaluated by reviewers selected by the editors ac-
article paper or a case history should not contain more cording to the subject of the paper. The authors will be informed about the
than 30 pages, formatted according to the instructions results of the review process. If the paper is accepted, the authors will be
mentioned above. Technical notes and discussions required to submit a version of the revised manuscript with the suggested
should have no more than 15 and 8 pages, respectively. modifications. If the manuscript is rejected for publication, the authors
Longer manuscripts may be exceptionally accepted if the will be informed about the reasons for rejection. In any situation compris-
authors provide proper explanation for the need of the re- ing modification of the original text, classification of the manuscript in a
quired extra space in a cover letter. category different from that proposed by the authors, or rejection, the au-
Discussion thors can reply presenting their reasons for disagreeing with the reviewer
Discussions must be written in English. The first page comments.
of a discussion should contain: Submission
The title of the paper under discussion; The author(s) must upload a digital file of the manuscript to the Soils
Name of the author(s) of the discussion, followed by and Rocks website.
their position, affiliation, address and e-mail. The au- Follow Up
thor(s) of the discussion should refer to himself (her- The online management system will provide a password to the corre-
self/themselves) as the reader(s) and to the author(s) of sponding author, which will enable him/her to follow the reviewing pro-
the paper as the author(s). cess of the submitted manuscript at the Soils and Rocks website.
SOILS and ROCKS An International Journal of Geotechnical and Geoenvironmental Engineering
ISSN 1980-9743
Soils and Rocks

100 100
95 95
75 75
25
Table of Contents 25
5 5
0 PACHECO SILVA LECTURE 0

M. Matos Fernandes 191
ARTICLES

Soils and Rocks
Laboratory Batch Tests and Mathematical Modeling An International Journal of Geotechnical
and Geoenvironmental Engineering
Volume 38, N. 3

F. Massad 249
TECHNICAL NOTE

B
100
95
MS 100
95
75 75
25
Volume 38, N. 3 25
September-December 2015
2015
5 5
0 0

Classificação Dos Solos

Uploaded by

Copyright:

Available Formats

Classificação Dos Solos

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Classificação Dos Solos

Uploaded by

Copyright:

Available Formats

SOILS and ROCKS An International Journal of Geotechnical and Geoenvironmental Engineering

Soils and Rocks

0 PACHECO SILVA LECTURE 0

New Developments in the Control and Prediction of the Movements

Use of Sugarcane Bagasse as Carbon Substrate in Permeable Reactive Barriers:

On the Interpretation of the Bidirectional Static Load Test

Revisiting Classical Methods to Identify Collapsible Soils

ABMS - Brazilian Association for Soil Mechanics and Geotechnical Engineering

SPG – Portuguese Geotechnical Society

Issue Date: December 2015

Editorial Board Members

Soils and Rocks

ISSN 1980-9743 CDU 624.131.1

SOILS and ROCKS

PACHECO SILVA LECTURE

Apparent Pressures on Multi-Propped Retaining Walls in Soils Under Drained Conditions

On the Interpretation of the Bidirectional Static Load Test

Prof. Manuel de Matos Fernandes

Soils and Rocks

Before closing this section, it should be noticed that in

2.2. General report of Peck (1969) - The founder docu-

lowed the execution of excavations in soft ground with

Figure 11 - Axial load-displacement diagram of a strut and of its

port to the wall before the excavation is made, in comple-

Soils and Rocks

are advantageous in comparison to operation-intensive and

Organic PRBs are based on a mixture of filling solid

The results of laboratory studies involving sulfate-

This paper describes the experimental results from

Table 2 - Ion concentrations and resulting pH of the synthetic

Ion Concentration (mg/L)

where [SO] is the concentration (mg/L) of sulfate (SO42-),

The bacterial activity of organic decomposition/fer-

lulose, [S], in a reactor, coming from the packed sugarcane

4.2.2. Modeling results

1. Introduction walls supported by several levels of struts. When the

Figure 3 - Schematic representation of scenarios A, B, C, D and E.

Figure 6 - Failure points after the last excavation stage.

Figure 11 - Proposal for apparent design diagrams.

Figure 12 - Proposed values of bbase as a function of the depth of

1. Introduction this curve is in the evaluation of pile bearing capacity, us-

Figure 1 - Load distribution -a) Conventional test; and b) - Bidirectional test.

Figure 2 - Nomograms for head-down loads (conventional loading test).

Figure 3 - Nomograms for upward loads (bidirectional test).

Figure 4 - Modified Cambefort Relations: a) shaft and b) toe.

(k £ 2); intermediate (2 £ k £ 8); and “long” or compressible

3.2. Simulation of the Bidirectional test: invariance of

f u1 ³ f u 2 and Alr = Alr1 + Alr 2 (8)

Figure 6 - Bidirectional test simulation for a rigid pile (k = 1, Case 5).

Figure 7 - Bidirectional test simulation for an intermediate pile (k = 5, Case 3).

long pile (k = 10), respectively. The notable points 3, M and

Figure 14 - Distribution of unit skin friction with depth: a) case 7

Table 1 - Case Histories -general information of the piles.

Type of test Pile Diameter (m) Height (m) Place Source

Table 2 - Results of the analysis.

Type of test Pile Kr c c’ k c’.k Pile behavior

are again in quite a good agreement. This is a fourth valida-

Figure 26 - Bidirectional Test: Bored pile US82 Bridge-Missis-

Figure 29 - Head-down test: Bored Pile- Tucson (AZ).