Department of Engineering Science, Faculty of Engineering and Science, Chatham

Maritime, Kent, ME4 4TB

ADVANCED GEOTECHNICAL ENGINEERING

Topics: Computational Geotechnics; Elastic and Elastic-plastic deformation of piles and pile groups

Student ID:

Module tutor:

Disclaimer: This report is prepared and presented as the original work of the student and has not been
copied or plagiarized in any way. It is submitted as solely my work in partial fulfilment of the requirement
of the course GEOL1028: Advanced Geotechnical Engineering. I understand the university’s plagiarism
policy and accept responsibility for every plagiarism charge where existent in this document.
Contents
1. Introduction.....................................................................................................................1

2. Theoretical analysis........................................................................................................2

2.1 Problem definition.....................................................................................................2

2.2 Problem analysis......................................................................................................2

2.3 Results and Discussions...........................................................................................4

3. Numerical analysis.........................................................................................................6

3.1 Problem definition.....................................................................................................6

3.2 Problem analysis.......................................................................................................6

3.3 Results & Discussion................................................................................................9

4. PLAXIS modelling........................................................................................................11

4.1 Problem definition....................................................................................................11

4.2 Problem analysis.....................................................................................................11

4.3 Results & Discussion...............................................................................................12

5. Conclusions..................................................................................................................15

References.......................................................................................................................16

Appendix A.......................................................................................................................17

Appendix B: PLAXIS RESULTS.......................................................................................20

 ADVANCED GEOTECHNICAL ENGINEERING

Topics: Computational Geotechnics; Elastic and Elastic-plastic deformation of piles and pile groups

Abstract: Utilizing settlement investigations of made up situations and genuine contextual analyses, this review looks at direct flexible
and nonlinear heap bunch examination techniques and talks about the ramifications for deciphering the consequences of heap load tests.
Examinations between straight versatile and nonlinear methodologies legitimize the case that heap to-heap contact is overwhelmed by
direct flexibility, depicted by the little strain soil firmness. As the size of a heap bunch increments, nonlinearity in individual heap conduct
becomes overpowered by the connection impacts. In such cases, comparable evaluations will be accomplished by both direct and
nonlinear strategies assuming the dirt modulus is gotten from the underlying digression, as opposed to some secant solidness, surveyed
from the heap test information. The review explains the abilities and constraints of direct versatility in heap bunch examination and gives.

CE Database subject headings: Pile groups; Pile foundations; Soil-pile interactions.

Author keywords: Pile groups; Pile foundations; Soil-pile interactions.

Introduction duced from various moduli acquired from heap load tests. They found
that straight versatile heap bunch examinations produce comparable
Static heap load tests keep on being the most dependable re-sults to nonlinear investigations, when the flexible modulus is
strategy for assessing single endlessly heap bunch reaction under gotten from the underlying digression to the heap dislodging reaction
plan loads, notwithstanding huge hypothetical headways in the mea-sured in a solitary heap test. Obviously, the overconservatism of
examination and expectation of heap conduct in late many years. straight flexible techniques might be credited to unseemly soil firm
Heap load test information empower designers to perform back- ness translation from static heap load tests.
examinations and reason proper strength and firmness boundaries This paper endeavors to support these discoveries of Mandolini et
(e.g., heap shaft and base capaci-ties, functional shear moduli) al. (2005) and accommodate the disparities between promoters of
that further describe subsurface circumstances, legitimize/refine direct and nonlinear strategies, through cautious audit of previ-ously
introductory designing suppositions and lead to further developed distributed examinations, mathematical investigations of speculative
expectations. Despite the fact that plan codes frequently give sce-narios and reanalyses of genuine contextual analyses. The
rules on translation of extreme heap limits from load test investigations and the ensuing conversations explain the abilities and
information to guarantee "safe" plans, there are barely any recom- impediments of straight flexible techniques, and give direction on
mendations on the understanding of soil firmness and its applica- when they would be satisfactory for heap bunch examinations and
tion to heap bunch investigation techniques. There is presently an when their nonlinear partners ought to be liked. The paper likewise
abundance Error of soil solidness from load test information has looks to explain systems in regards to heap test translation for
ap-peared to show itself in the debate of straight versatile versus examination of heap bunch reaction.
nonlinear heap bunch examination techniques. With ongoing
advances in nonlinear techniques, their predominance over direct Linear-Elastic Pile Interaction and Localized
flexible strategies Plasticity

has been accentuated, and has prompted the recommendation Shortly after the development of linear-elastic pile group analysis
that "tradi-tional" flexible examinations misjudge heap methods (e.g., Poulos and Davis 1968; Poulos 1969; Randolph
cooperation (in settlement investigations), erroneously foresee and Wroth 1978, 1979), researchers have endeavored to under-
heap bunch conduct and are in many cases excessively moderate stand the role of soil nonlinearity in pile behavior and to consider
(e.g., Jardine et al. 1986; Basile 1999). Then again, Mandolini et its inclusion in pile analyses. For example, Chow (1986) states
al. (2005) aggregated a broad information base of heap bunch that nonlinearity is confined to a narrow zone of soil close to the
case narratives, and examined the legitimacy of straight and pile shaft, and that the soil response remains essentially elastic
nonlinear investigation techniques utilizing soil boundaries de- outside of this zone, where the strain level is low. Caputo and
 Fig. 1. Load-displacement curves for a loaded pile and a load-free pile (after Lee and Xiao 2001; Caputo and Viggiani 1984)

Viggiani (1984) played out a movement of pile load tests, be similar at working load levels, using either linear-elastic analy-
during which store head settlements were assessed on both the ses or highly nonlinear analyses with elastic pile interaction ef-
stacked load and enveloping weight free piles, as shown in Fig. 1 fects.
(Test 2 is essentially equivalent to Test 1 and isn't shown here).
While the pile movement lead of the stacked store apparently was
significantly nonlinear, those for the load free loads remain
fundamentally straight adaptable. With everything taken into Interpretation of Soil Stiffness Parameters
account, while the response of individual stores is nonlinear, the in Inverse Analyses
participation influences between loads remain commonly flexible.
Following the first contentions, direct versatile examinations ought to be
Poulos (1988, 2006) in like manner analyzed the different
equipped for delivering sensibly exact expectations of heap bunch
tension frameworks working close by a load, which lead to an
conduct, insofar as suitable soil solidness boundaries are utilized as
assortment of utilitarian soil moduli with distance from the pile
information. The case history depicted by McCabe and Lehane (2006a)
shaft and base. As necessary, Poulos (1988) proposed the use of
gives a helpful illustrative model. This in-volved static burden testing of
lower adaptable soil moduli close to the store shaft and higher
a gathering of five heaps and a solitary reference heap at Belfast Lough,
moduli further away from the stack. While being useful for the
Ireland. The tests were back-examined utilizing both straight flexible
ultimate objective of illustrating, the decision of the extent
plan programming and nonlinear limited component investigations.
between "close load modulus" and "soil mass modulus" (i.e.,
While the nonlinear investigations created close gauges (McCabe and
quite far from the store) is to some degree conflicting.
Lehane 2006b), the straight versatile examinations were undeniably less
Then again, different stack assessment programs have been persuading. The straight flexible examinations were acted in their work
made considering assumptions of nonlinear single store lead and utilizing a typical plan programming, PIGLET (Randolph 2003a), in
direct adaptable pile to-load cooperation. For example, RATZ which soil twisting around the heap shaft is romanticized as shearing of
(Randolph 2003b) uses figurative or misrepresented load move concentric chambers, prompting a logarithmic decrease of misshapening
twists to portray individual pile direct, while stack interac not with distance away from the heap. Heap base reaction is displayed in
altogether settled by adaptable courses of action at first made in PIGLET as an unbending punch at the base level, and the cooperation
Randolph and Wroth (1979). According to this technique, the impacts (for heap
hidden solidness in the stack move curve of a social occasion pile
is obstruct mined by the cooperation influences (which are
flexible), while the nonlinear piece of the twist stays identical to
that of a singular store. Fig. 2 shows the schematic weight move
curves of a single load, differentiated and that of a store in a
social occasion, according to the RATZ plan frameworks. It
should be seen that as the num-ber of stacks constructs, the
nonlinear piece of the pile move twist becomes overwhelmed by
the adaptable part, due to the extended effect of store
correspondence. This smothers the impor-tance of nonlinearity,
and hence, the pile settlement direct of the social occasion load
seems to be a straight adaptable absolutely plastic gauge as the
effect of stack joint effort becomes recognizable.
Toward the day's end, considering everything, the greater the
effect of load association, the closer the results will be between
nonlinear (e.g., RATZ) and essential direct adaptable
assessments, gave that comparative strength limits (e.g., shear
modulus) are used in the two game plans of assessments. Fig. 2. Modeling group effects by factoring load transfer curve (after
Obviously immediate adaptable assessments don't confine the Randolph 2003b)
constraint of a load, but in most sensible cases, piles are arranged
with a variable of safety of 2.0 or above. Therefore the stiffness
predictions of pile groups should
 the soil Poisson’s ratio is 0.3. These correspond to a rigidity index
(ratio between small-strain shear modulus and shear strength) of
500, which is a typical value compared to real soils with nonlin-
ear behavior, such as London clay (Hight et al. 2003; Gourvenec
et al. 2005) and Bangkok clay (Shibuya and Tamrakar 2003).
In the FLAC3D analyses, the mobilized shear stress of the soil
follows a hyperbolic relationship with the engineering shear
strain in the element; such a relationship leads to a higher rate of
stiff- ness degradation than often observed in soil element tests
and hence predictions may be viewed as those corresponding to a
relatively high level of nonlinearity. The soil model is incorpo-
rated in FLAC3D in a way similar to that described in Klar and
Leung (2009).
Figs. 5(b and c) summarize the results of the simulations. In
the first analysis, the pile cap is not modeled. Only the center
pile is loaded, and the displacements are tracked both at the
loaded center pile, and at the load-free corner and edge piles in
the 3 × 3 group. Fig. 5(b) shows that while the load-settlement
response of the loaded pile is highly nonlinear, those of the load-
free piles are essentially linear. This finding is consistent with
the test data reported by Caputo and Viggiani (1984). In the
group analyses, the rigid pile cap is modeled as the entire (3
× 35
× 5) pile group is loaded, and the group response is shown in
Fig. 5(c). In addition to the nonlinear FLAC3D analyses, linear-
elastic analyses were also performed with PIGLET, using the
small-strain shear modulus of 30 MPa. At working load levels
(with theoretical factor of safety of 2.0) PIGLET produces rea-
sonable estimates of single pile and pile group stiffness, with
errors of approximately 15% compared to the FLAC3D results. In
fact, the interaction factors predicted by linear-elastic PIGLET
analyses are very close to those predicted by nonlinear FLAC3D
Fig. 4. Comparisons between predicted and measured settlements of simulations.
48 case studies (Mandolini et al. 2005, with permission) Interface elements have not been included in the FLAC3D
analyses accompanying Fig. 5(c) Therefore, the model corre-
sponds to an adhesion factor of 1.0. Interface elements can be
Example Analysis with Nonlinear Finite Difference added if a different adhesion factor (normally smaller than 1.0) is
Method to be simulated. For example, Fig. 6 shows the results of
FLAC3D analyses where the same soil model is used, but with
To investigate the foregoing arguments, a series of 3-dimensional interface elements added surrounding the pile elements to limit
finite difference analyses have been performed using FLAC3D the pile shaft resistance to 30 kPa (i.e., adhesion factor= 0.5).
(Itasca Consulting Group 2005). The two example scenarios con- This allows the pile shaft elements to “slip” corresponding to the
sidered consisted of nine concrete piles (Young’s modulus of sur- rounding soil when the shaft resistance is exceeded.
the pile, Ep, is 30 GPa) arranged in a 3 × 3 pattern, and 25 piles Accordingly, the single pile and pile group bearing capacities are
in a 5 × 5 pattern, both connected with a rigid pile cap. Table 1 reduced by roughly 50%. The FLAC3D results remain
and Fig. 5(a) summarize the pile group geometries and soil pa- comparable with PIG- LET predictions at working load levels,
rameters used in the analyses. The piles are 20 m long with di- when the same small- strain shear modulus of 60 MPa is used.
ameter (d) of 1 m, located with center-to-center spacing (s) of
3 m. The soil model used in the analyses assumes a hyperbolic
shear stress-shear strain relationship, which leads to a strong Effects of Pile Spacing and Strain Level
strain dependency of shear stiffness. The small-strain shear
modu- lus of the soil is 30 MPa, the ultimate shear strength is 60 The discrepancies between linear-elastic (LE) and nonlinear (NL)
kPa and methods depend upon the spacing between the piles, and the

Table 1. Pile Group Geometries and Material Properties in FLAC3D Analyses

Pile geometry and properties Parameters of soil model
Pile diameter (m) 1.0 Initial shear modulus (MPa) 30
Pile spacing (m) 3.0 Soil Poisson’s ratio 0.30
Pile length (m) 20 Ultimate shear strength (kPa) 60
Pile Young’s modulus (MPa) 30,000 Adhesion factors:
Pile Poisson’s ratio 0.20 Without interface elements 1.0
With interface elements 0.5
 Fig. 5. (a) Pile group layout and soil model in example scenario; (b) single pile test; and (c) nine-pile and 25-pile group tests

, which correspond to higher operational G values than those

shown in Fig. 7.
These analyses again show that pile interaction effects are
dominated by the small-strain modulus. In fact, the database

Fig. 6. Comparisons between PIGLET analyses and FLAC3D analy-

ses with interface elements
Fig. 7. Engineering shear strain induced in single pile loading and
corresponding operational shear modulus

compiled by Mandolini et al. (2005), from which they created

Fig. 4, consists of pile groups with s  d ratios ranging from 1.8 to
7.1, indicating that the LE analysis method is a reasonable ap-
proach for common pile group geometries. In some cases, how-
ever, piles are designed to act separately under different columns,
and located so far from each other that the interaction effects
between them become negligible. Consequently, they behave as
individual piles and nonlinearity contributes significantly to their
response.

Comparisons of Methods in Case Study

To further illustrate the modeling of pile groups using the LE

approach and how this compares with nonlinear methods, the case
history described by O’Neill et al. (1982a,b) and studied by other
researchers (e.g., Castelli and Maugeri 2002) is now considered.
This is analyzed using three approaches: (1) linear-elastic PIG-
LET type of analysis (as described previously); (2) elastic-plastic
continuum model, where the pile response and interaction effects
are modeled by Mindlin’s solution, with limiting contact stress
between the soil and pile shaft/base interface (Poulos 1989;
Leung et al. 2009); and (3) nonlinear load transfer with elastic
pile interactions (RATZ), where individual pile response is mod-
eled by parabolic load transfer curves and interaction effects are
considered by elastic estimates originally formulated by Randolph
and Wroth (1979) (Fig. 2). The case history involved testing of a
single pile and groups of four, five, and nine tubular piles, which
were driven into stiff overconsolidated clays and connected by
rigid caps. The piles had an external diameter of 273 mm, internal
diameter of 255 mm, and penetrated 13 m into the ground. The
nine-pile group was in a 3 × 3 pattern with spacing to diameter
ratio (s  d) of 3, while the four-pile group consisted of the edge
piles in the 3 × 3 grid, and the five-pile group was formed by the
center pile and corner piles of the grid. Detailed descriptions of
the test setup and pile geometries can be found in O’Neill et al.
(1982a,b).
Fig. 8 compares the field measurements with predictions ob- Fig. 8. Comparisons between predicted and measured settlements of
tained using the three selected methods for the groups comprising case history reported by O’Neill et al. (1982a,b)
four and nine piles. The small-strain shear modulus used for each
of these methods was identical, and derived through back analy-
ses with PIGLET, matching the initial tangent of the single pile nonlinear analyses. As far as pile group stiffness is concerned, the
test. The shaft and base resistance of the pile, used in the elastic- differences are minimal at working load levels. It should be
plastic analyses and the nonlinear analyses, were also assessed noted, however, that linear-elastic and elastic-plastic analyses do
using the single pile test data. It is evident in Fig. 8 that, with a not match the single pile test data perfectly. The engineer may be
small-strain (initial) soil modulus, the linear-elastic and elastic- tempted to lower the soil modulus for calibration purposes (simi-
plastic analyses resulted in estimates very close to those from the lar to ES approach), which can then lead to underestimation of
pile group stiffness. Also, in assessing the initial tangent to the
 Fig. 10. Pile loads at center, edge, and corner piles at different load levels for (a) 3 × 3 pile group; (b) 5 × 5 pile group

ments [Figs. 3(b), 6, and 8] indicate that the foregoing methodol- illustrates that the particular analysis method employed (whether
ogy is reasonable to model both bored and driven piles. The it is linear-elastic, elastic-plastic or highly nonlinear) has a rela-
methodology may not, however, be suitable for cases involving tively minor effect on predictions, provided that the correct (i.e.,
significant disturbance during the installation process (e.g., very small-strain) elastic soil modulus is employed. The role of non-
closely spaced driven piles, deep soil mixing columns). In these linearity in pile interaction becomes more significant when pile
cases, the difference between “near-pile modulus” and “soil mass spacings are less than the usual minimum of 2.5 times the diam-
modulus” (Poulos 1988, 2006) can be significant and the initial eter, and also when a pile is used as “settlement reducer” or when
stiffness obtained from single pile tests may not be representative the group’s factor of safety is less than about 2. In cases where
for the modeling of pile interaction. nonlinearity is crucial in the pile group analysis, nonlinear meth-
The case studies analyzed in this paper involved mainly fric- ods such as the one adopted by RATZ, or the extended mobiliz-
tional piles where the pile bases made small contributions to the able strength design method which models a nonlinear continuum
overall pile capacity and stiffness. The interaction of pile bases in the pile interaction (Klar and Leung 2009), may be applied for
was modeled by the initial elastic modulus in the current study, more accurate estimates.
following the Randolph and Wroth (1979) procedure (in PIGLET
analyses), or by using the Mindlin’s solution (in elastic-plastic
analyses). In cases where the piles derive a large proportion of
their capacity from end-bearing, it is essential to incorporate the Acknowledgments
base response observed in single pile tests in the analysis. The
overall pile and pile group base stiffness will depend greatly on The writers would like to express their gratitude toward IOS Press
the level of mobilization of base capacity, which varies substan- and the writers of Mandolini et al. (2005) for their permission to
tially according to construction procedures and workmanship. use the material in the mentioned publication. The work
presented in this paper is part of the Smart Foundations project
(Grant No. EP/D040000/1) funded by the Engineering and
Conclusions Physical Sciences Research Council (EPSRC). Their financial
support is gratefully acknowledged. The first writer would also
This paper has shown, through a review of previous literature like to express his gratitude to the Cambridge Overseas Trust
together with back-analyses of reported case histories, that adop- (COT) for additional funding supporting his studies at the
tion of the small-strain soil modulus as the operational modulus University of Cambridge.
for pile interaction enables good estimates of pile group response
to be obtained. The small-strain soil modulus leads to more accu-
rate estimates of pile interaction effects, which is the dominant References
factor in pile group settlement evaluation. This modulus may be
derived from shear-wave velocity measurements, or from the ini- Basile, F. (1999). “Non-linear analysis of pile groups.” Proc. Inst. Civ.
tial tangent of load-settlement curve of static load tests on single Eng., Geotech. Eng., 137(2), 105–115.
piles. Derivation from the initial tangent modulus is preferable as Caputo, V., and Viggiani, C. (1984). “Pile foundation analysis: Simple
it incorporates variations along the shaft of shear wave velocities approach to nonlinearity effects.” Rivista Italiana di Geotecnica,
and hence the Gmax “stratigraphy.” 18(2), 32–51.
The widespread approach of deriving a secant modulus from a Castelli, F., and Maugeri, M. (2002). “Simplified nonlinear analysis for
single pile’s settlement at working load results in overestimation settlement prediction of pile groups.” J. Geotech. Geoenviron. Eng.,
of pile interaction factors and group settlements. The paper also 128(1), 76–84.
Chow, Y. K. (1986). “Analysis of vertically loaded pile groups.” Int. J.
Numer. Analyt. Meth. Geomech., 10(1), 59–72.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / DECEMBER 2010 / 1693

Coyle, H. M., and Reese, L. C. (1966). “Load transfer for axially loaded lation of pile group in overconsolidated clay.” J. Geotech. Engrg.
piles in clay.” J. Soil Mech. and Found. Div., 92(2), 1–26. Div., 108, 1369–1386.
Gourvenec, S. M., Mair, R. J., Bolton, M. D., and Soga, K. (2005).
O’Neill, M. W., Hawkins, R. A., and Mahar, L. J. (1982b). “Load transfer
“Ground conditions around an old tunnel in London Clay.” Proc. Inst.
mechanisms in piles and pile groups.” J. Geotech. Engrg. Div., 108,
Civ. Eng., Transp., 158(1), 25–33.
1605–1623.
Hight, D. W., McMillan, F., Powell, J. J. M., Jardine, R. J., and Allenou,
Poulos, H. G. (1969). “Analysis of the settlement of pile groups.” Geo-
C. P. (2003). “Some characteristics of London clay.” technique, 18(4), 449–471.
Characterisation and engineering properties of natural soils, T. S. Poulos, H. G. (1988). “Modified calculation of pile-group settlement in-
Tan, K. K. Phoon, D. teraction.” J. Geotech. Engrg., 114(6), 697–706.
W. Hight, and S. Leroueil, eds., Taylor & Francis, 851–907. Poulos, H. G. (1989). “Pile behaviour: Theory and application.” Geotech-
Itasca Consulting Group, Inc. (2005). FLAC3D—Fast Lagrangian analy- nique, 39(3), 365–415.
sis of continua in 3 dimensions,” Itasca Consulting Group, Inc., Min- Poulos, H. G. (2006). “Pile group settlement estimation: Research to
neapolis. practice.” Proc., GeoShanghai Int. Conf. 2006, ASCE Geotechnical
Jardine, R. J., Potts, D. M., Fourie, A. B., and Burland, J. B. (1986). Special Publication 153, ASCE, Reston, Va., 1–22.
“Studies of the influence of non-linear stress-strain characteristics in Poulos, H. G., and Davis, E. H. (1968). “Settlement behavior of single
soil-structure interaction.” Geotechnique, 36(3), 377–396. axially-loaded incompressible piles and piers.” Geotechnique, 18(3),
Klar, A., and Leung, Y. F. (2009). “Simple energy-based method for
351–371.
nonlinear analysis of incompressible pile groups in clays.” J.
Randolph, M. F. (1994). “Design methods for pile groups and piled rafts.”
Geotech. Geoenviron. Eng., 135(7), 960–965.
Proc., 13th Conf. on Soil Mechanics and Foundation Engineering,
CRC Press, 5, 61–82.