You are on page 1of 10

ANALYSIS OF PILES SUBJECTED TO EMBANKMENT INDUCED LATERAL

SOIL MOVEMENTS

By A. T. C. Gob; C. I. Teb,z and K. S. Wong,3 Member, ASCE

ABSTRACT: Damage to piles supporting structures, bridge abutments, and utilities can occur as a result of the
construction of nearby embankments. This is because the lateral displacements resulting from these construction
activities can induce forces and moments in the piles. The resulting stresses can be significant particularly when
soft s~il deposits are present and the lateral soil displacements are large. This paper describes a simplified
numencal procedure based on the finite-element method for analyzing the response of single piles to lateral soil
movements. The flexural bending of the pile is modeled by beam elements. The complex phenomenon of the
pile-soil interaction is modeled by hyperbolic soil springs. A framework for determining the soil parameters for
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

use in the analysis is summarized here. Comparisons are made between the observed behavior of full-scale tests
and centrifuge model tests and those computed by the proposed numerical method. Based on parametric studies,
empirical design solutions for pile foundation systems at the base of a sloped embankment are presented.

INTRODUCTION Winkler beam and the soil is represented by a series of elastic


The construction of road embankments and fills adjacent to springs. The spring constant adopted was the coefficient of
structures, bridges, or utilities has been known to cause dam- subgrade reaction. Others have adopted similar techniques and
age to piles supporting these structures. This is because the represented the soil with a series of nonlinear springs (Bourges
lateral displacements resulting from these construction activi- et al. 1980; Bigot et al. 1982; Byrne et al. 1984).
ties can induce forces and bending moments in the piles (Fig. This paper describes a simplified numerical procedure for
I). The resulting stresses can be significant, particularly when analyzing the response of single piles to lateral soil move-
soft soil deposits are present and the lateral soil displacements ments. The flexural bending of the pile is modeled by beam
are large (Broms 1972; Huder and Bucher 1981). A number elements. The complex phenomenon of the pile-soil interac-
of instrumented case studies have been reported, such as Hey- tion is modeled by hyperbolic soil springs. A framework for
man (1965), Leussink and Wenz (1969), and Bigot et al. determining the soil spring constants from conventional soil
(1977). Recent model centrifuge and numerical studies by test data is proposed. The method gives results for pile dis-
Springman (1989) and Stewart (1992) to investigate the lateral placements and bending moments that compare well with the
loading of pile groups due to embankment construction have analytical solution by Poulos (1973). Comparisons are also
led to an increased understanding of the factors governing the made between the observed behavior of full-scale and centri-
induced pile forces. fuge model tests and those computed by the proposed numer-
Several empirical and numerical methods have been pro- ical procedure. Finally, empirical solutions for pile foundation
posed for analyzing the response of single piles and pile systems at the base of a sloped embankment computed from
groups to lateral loading from horizontal soil movements. A parametric studies are presented. The numerical procedure pro-
comprehensive review has been made by Stewart et al. (1994). posed in this paper should also be valid for a group of widely
Most of the numerical methods that have been proposed utilize spaced piles. Stewart (1992) concluded from observations of
the finite-element method (Carter 1982; Broms et al. 1987; the centrifuge model pile group tests that a pile spacing of 4.9
Springman 1989; Stewart et al. 1993) or the finite difference times the pile diameter was sufficiently large to minimize the
method (Poulos and Davies 1980). For pile groups, the plane interaction between adjacent piles. In addition, the measured
strain finite-element approach was adopted by Stewart et al. maximum bending moments between the front and rear piles
(1993), in which the piles were represented by equivalent were also similar. Further enhancement of the procedure to
sheet-pile walls. Another approach (Springman 1989) used analyze a group of closely spaced piles would entail modeling
three-dimensional finite-element analyses in which the soil the interaction between the group piles and is beyond the
was assumed to be elastic. For single piles, the methods are scope of this paper. Only the immediate short-term response
generally either based on the elastic continuum approach or of the pile to the imposed total stress from the embankment
the modulus of subgrade reaction method. For example, in the load is considered.
Poulos and Davies (1980) method, the solution is based on a
point load in an elastic half-space (Mindlin's solution) and
modified empirically to account for the presence of a rigid Fill
bearing layer. A limiting pressure was specified to account for
plastic yielding of the soil around the pile. In the approach
taken by Broms et al. (1987), the pile is represented by a
Lateral
Soft Soil
'Sr. Lec!., School of Civ. and Struc!. Engrg.• Nanyang Technol. Univ.• Pressure
Singapore 639798.
'Sr. Lec!., School of Civ. and Struc!. Engrg., Nanyang Technol. Univ., From Horizontal
Singapore 639798. Soil Movements
3 Assoc. Prof., School of Civ. and Struct. Engrg., Nanyang Technol.

Univ.• Singapore 639798.


Note. Discussion open until February I, 1998. To extend the closing Stiffer Stratum
date one month, a written request must be filed with the ASCE Manager pile
of Journals. The manuscript for this paper was submitted for review and
possible publication on November 13. 1995. This paper is part of the
Journal of Geotechnical and Geoenvironmental Engineering, Vol. 123,
No.9, September. 1997. ©ASCE, ISSN 1090-0241/97/0009-0792-0801/ FIG. 1. Lateral Pressures Induced on Pile from Embankment
$4.00 + $.50 per page. Paper No. 12017. Construction

792/ JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / SEPTEMBER 1997

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


FINITE-ELEMENT METHOD
(4)
A rigorous numerical solution of the response of a vertical
pile to laterally induced soil movements from embankment where
loading is computationally expensive because of the three-di-
mensional and nonlinear nature of the problem. This paper p
P)* -_:... L, * -- PU)
y).,
describes a simplified approach that can account for complex PU)
Y)
Yu)
Yu) = Ksj
(5a-c)
conditions such as nonhomogeneous soil strength conditions,
pile stiffness, and pile-head fixity. The problem is decomposed The stiffness of the spring at any given load level may be
into two components. First, the soil displacement induced in computed as the tangent to the hyperbolic curve and may be
the substratum due to embankment construction is determined, written as
either from measured inclinometer data or from finite-element
analysis. The resultant soil displacement is termed "free- (6)
field" soil movement in this paper. Second, the determined
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

free-field soil movement is used as input to study its effect on


an existing pile. The method is, strictly speaking, an approx- Lateral Soli Spring Stiffness
imate method. Nevertheless, it provides good approximation
at a fraction of the time of more rigorous numerical solutions. As indicated in (2), a hyperbolic relationship is used to rep-
A similar approach has been used successfully to study the resent the soil spring. This section outlines the approach
effects of vertical movements on piles (Chow et al. 1990; Lee adopted to determine the initial soil stiffness khl for cohesion-
1993; Teh and Wong 1995). The following describes the pro- less and cohesive soils.
posed methodology.
The problem of a vertical pile located in a layered soil un- Cohesionless Soil
dergoing lateral movement is depicted in Fig. 2. Beam ele-
ments are used to model the pile. The pile load-deformation At low strain levels, the response of the soil to external load
behavior can then be expressed as is often assumed to be elastic. This is the basis for the sub-
grade modulus method in which the soil response is charac-
(1) terized by the coefficient k h • Various correlations between k h
and the Young's modulus of soil have been proposed in the
where [Kp] = pile stiffness matrix; {yp} = nodal displacement study of laterally loaded piles. Vesic (1961), based on the anal-
vector; and {P} = lateral nodal load vector. In the absence of ysis of an infinite beam on an elastic foundation, suggested
any externally applied load, {P} is a consequence of the rel- the following expression
ative displacement between the pile and surrounding soil. The
hyperbolic relationship given in the following is used to rep-
resent the soil spring (7)

p - y) (2) The term Eplp = the pile flexural stiffness; d = width of the
) - (1IK,) + y/P U
)
pile; E, = soil secant (elastic) modulus; and v, = Poisson's
ratio of the soil.
where the subscript j = node number; p) = lateral nodal force;
For a square concrete pile and typical ratios of EpIE, of
K,) = initial tangent of the hyperbolic curve; yp) is the pile node
between 103 and lOS, the term in the 12th root varies between
lateral displacement; y,) is the free-field soil lateral movement;
0.5 and 0.7. Substituting a mean value of 0.6 for this term in
y) = relative pile-soil movement perpendicular to the pile axis.
(7) and replacing kh with khl the initial stiffness for a hyperbolic
P U) is the maximum allowable lateral nodal load given by model leads to
(3)
k _ 0.39E,
(8)
where py) = limiting lateral soil pressure; and A,) = projected hi - d(l - v;)
shaft area normal to the pile axis at node j. By defining two
dimensionless parameters Pt and yt, (2) may be written in a In this study, the preceding expression was adopted for co-
more compact, dimensionless form as hesionless soils.

Cohesive Soil
Eq. (8) is similar in form to the equation proposed by Broms
Layer I (1964a) for cohesive soils

k _ 1.67Eso
h- d (9)
Layer n ----+
in which E so = secant modulus at half ultimate stress in an
node] undrained test.
Both (7) and (9) were proposed on the basis that the soil
LayerID
response is elastic, and therefore E, and Eso in (7)-(9) are
understood to be the representative average Young's modulus
over the working range. Therefore it is not appropriate to use
(9) directly in a hyperbolic model in which the stiffness
changes with stress level. By retaining the relationship be-
(a) (b)
tween k h and the Young's modulus in (9) and noting that E, =
FIG. 2. (a) Pile In Layered Soli Undergoing Lateral Movement; 2Eso in a hyperbolic model, the initial spring stiffness khl may
(b) Discretization of Problem be expressed as
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / SEPTEMBER 1997/793

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


k . = 1.67EI = 3.34E~0 'TJ is in the range of 60-300 for loose to dense sand, respec-
h (10) tively.
I d d

In this study, the preceding expression was adopted for co- Poisson's Ratio
hesive soils.
For undrained analysis of saturated clays, the Poisson's ratio
SOIL PARAMETERS is assumed to be equal to 0.5. For sands, the following equa-
tion based on the theory of elasticity can be used.
The complete description of the soil spring requires the in-
put of the soil shear modulus, the Poisson's ratio, and the o K
limiting lateral soil pressure. These properties were not di-
v, =---
1+ K
(15)
o
rectly available in the documented case studies. A framework
for determining these parameters from the available test data The coefficient of earth pressure at-rest Ko can be computed
is summarized herein. using the empirical equation proposed by Mayne and Kulhawy
(1982)
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

Determination of G1 Ko = (1 - sin <I>')OCR ain 4>' (16)


From elasticity theory, the initial shear modulus is related where <1>' = effective friction angle; and OCR = overconsoli-
to the elastic modulus E I by the following expression dation ratio of the soil.
EI
G I =---'-- (11) Determination of py
2(1 + v,)

The stress-strain relationship of a saturated clay determined In p-y curve analysis, it is common practice to estimate the
from undrained compression test can be idealized as a hyper- limit pressure py for clays by the expressions given by Reese
bola. For a perfect hyperbola, EI can be related to the secant (1984)

= ( 3 + 'c.YZ + d0.5Z) c.
modulus at 50% failure level E~o as
Py or (17a)
(12)

Taking v, to be 0.5 for the undrained condition, the initial shear Py = 9c. whichever is smaller (17b)
modulus may be expressed as
where Cu = undrained shear strength of the soil (from uncon-
solidated undrained triaxial tests); 'Y = soil unit weight; z =
(13) depth; and d = pile width. Randolph and Houlsby (1984) stud-
ied the problem of a laterally loaded, infinitely long pile using
In the absence of detailed soil tests, E~o may be estimated from classical plasticity theory. They recommended the following
the correlation proposed by Duncan and Buchignani (1976) expression
and shown in Fig. 3.
For sands, Wong and Teh (1995) have proposed a relation- Py = 10.Sc. (18)
ship correlating E I with the effective vertical stress cr~. Based Chen and Poulos (1994) studied the limit pressure on a row
on their correlation a similar relationship can be derived for of piles due to lateral soil movement using the finite-element
Gi method. The limit pressure obtained ranges from 6cu to 12cu•
GI (J'~
0.7
For simplicity, in this study Py is taken to be 9cu for clay.
-=TJ (
- (14) For sand, the expression proposed by Broms (1964b) is used
po Po ) to estimate py
where pa = atmospheric pressure; and 'TJ = parameter that is
1 + sin <1>/) ,
dependent on the relative density of the sand. For most sands, Py =3 ( 1 _ sin <1>/ (J'.
(19)

1200 Correlations with In-Situ Test Parameters


1000 For two of the case histories considered, the only test data
available were from in-situ pressuremeter and cone penetration
0 \ (CPT) tests. The following expressions were adopted for pre-
800
~ ~
bored pressuremeter tests

-
to) " 600
py =PL (20)

w
51
400
200
o
'" -- ~
r--
where PL and EM = pressuremeter limit pressure and pressure-
meter modulus, respectively. As it is general practice to as-
sume a Poisson's ratio of 0.33 for all soils in the calculations
of EM (Briaud 1992), GI was determined from (11) using a
Poisson's ratio of 0.33. Typical values of PL and EM are shown
(21)

in Table 1.
o 20 40 60 80 100 For cone penetration tests, py was obtained indirectly from
correlations with the cone resistance qc using the expressions
PLASTICITY INDEX, PI proposed by Briaud (1992)
FIG. 3. Variation of ElOlcu for Clay with OCR < 3 (Duncan and
Buchlgnanl1976) clay: py = 9c. = 0.2qc (22)

794/ JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / SEPTEMBER 1997

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


TABLE 1. Typical Pre88uremeter Parameters (Brlaud 1992) 0.0..,.-------1ro

PL EM
Material Characteristic (kPa) (kPa) 0.2
(1 ) (2) (3) (4)
Clay Soft 0-200 0-2,500
Medium 200-400 2,500-5,000
Stiff 400-800 5,000-12,000
Very stiff 800-1,600 12,000-25,000
Hard >1,600 >25,000
Sand Loose 0-500 0-3,500
Medium dense 500-1,500 3,500-12,000 0.8
Dense 1,500-2,500 12,000-22,500
Very dense >2,500 >22,500 (a) (b)

0.1 0.2 0.3 0.4 0.5 0.‫סס‬OO 0.0010 0.0020


Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

sand: py = O.llqc (23) yld


Similarly the elastic modulus was detennined using the fol-
lowing expressions (Briaud 1992): LId =25 free head, pinned lJp

clay: Ei = 2.5qc (24) Ys/d = 0.15

sand: E j = 1.15qc (25)


04L
Eq. (11) was then used to detennine G i •
L
1
NUMERICAL PROCEDURE Ys/d 0.45

As the load-displacement relationship of the soil spring is


nonlinear, the equation governing the development of lateral (c)
soil pressure has to be solved incrementally. At each step, the
incremental nodal displacement is computed from the equation Assumed soil
movement profile
(26)
FIG. 4. Lateral Pile Displacement and Bending Moment due to
where ~Ys = vector of incremental free-field soil lateral move- Trapezoidal Soli Deformation Profile
ment; and ~Yp = vector of incremental pile node lateral dis-
placement. The subscript n denotes the increment number. The studies described in this paper were conducted using this pro-
soil spring stiffness matrix [K.J is computed at the beginning gram.
of the increment using the known pile node displacement. Re-
arranging (26) leads to COMPARISON WITH EXISTING METHOD OF
COMPUTATION
(27)
Computations were carried out with BCPILE on a pile of
By applying a known value of {dYsl., (27) can be solved for 25-m length (L) with a diameter d of 1 m and a corresponding .
{dYpl•. At the end of each increment, the pile nodal displace- Eplp = 702 MNm2 • The soil was assumed to be unifonn with
ment is updated as c. = 20 kPa, py = 180 kPa, Vs = 0.49, and G i = 1,200 kPa. The
(28) free-field soil defonnation profile shown at the bottom of Fig.
4 was considered. The computed results of the nonnalized pile
A reasonably accurate approximation to the hyperbolic load- displacement and bending moment profiles in Fig. 4 show
displacement curve can be obtained by using small incremen- good agreement with the analyses by Poulos (1973). The slight
tal steps. Refinement to the solution accuracy can be achieved variations in the results could be due to the different methods
by implementing an iterative scheme within each increment. of discretization and the procedures used to model the soil-
Comparative studies by the authors have indicated that such pile interaction.
improvement is insignificant for most practical cases. Hence a
simple incremental scheme coupled with small incremental CASE STUDIES
step size has been adopted for this study. The procedure in-
volved in the proposed numerical scheme is outlined as fol- There are only a limited number of reported measurements
lows: of pile behavior due to lateral soil movements in which suf-
ficient data are available to enable comparisons between the
observed and predicted behavior to be made. Four case studies
1. Compute [KpJ and the incremental soil displacement
are described here. The required input parameters are deter-
vector {dys l.. mined using the procedures described in the preceding sec-
2. Update the elements of [KsJ using (6), based on the cur- tions.
rent nodal load {Pl •.
3. Solve (27) to obtain {dypl•. Heyman (1965)
4. Compute the incremental nodal force vector: {dPl. =
[Kp]{ dYpl •. Heyman (1965) conducted field studies to monitor the be-
5. Update the nodal load vector: {Pl.+ 1 = {Pl. + {dPl•. havior of piles as a result of the construction of a nearby 4-
6. Repeat steps 1-5 for the next load increment. m-high road embankment. The subsoil was comprised of
sand, peat, and clay layers underlain by sand at about 11 m,
A computer program, BCPILE (Teh 1995), has been developed as shown in Fig. 5. Two steel box piles 12.5-m-long with
using the proposed method. The case studies and parametric cross-section dimensions of 300 X 300 mm and a 6-mm wall
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING 1 SEPTEMBER 1997/795

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


Depth (m)
movement. The soil profile is shown in Fig. 7 alongside the
o 2.0 0.22 0.88
idealized py and G1 parameters adopted in the analysis. Based
on in-situ and laboratory tests, the Cu for the clay was 15 kPa
0.6 0.1 0.47 as reported by Leussink and Wenz (1969). No details about
cley 3.4 the sand were provided. In the analysis the top and bottom
3.7
sand layers were assumed to be loose and medium dense, re-
spectively. The measured lateral soil movements are illustrated
in Fig. 8. There are some discrepancies between the measured
and predicted pile displacements as shown in Fig. 8. The mea-
8.8 sured pile movements in Fig. 8 were relative to the pile head,
as the pile head did in fact move. The discrepancy in the re-
c1ey
sults may be because the measured pile movements correspond
put 11. 1.0
11.4
0.2 to a slightly larger embankment load of 200 kPa, while the
measured soil movements are for an embankment load of 180
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

And
kPa. While no data of the measured pile bending moment pro-
~ --""' 5.4
14.2 file were provided, another possible reason for the discrepancy
~ (MP_) in the results could also be due to the plastic (nonelastic)
movements in the pile during yielding. This is likely as the
FIG. 5. 5011 Profile and Assumed 5011 Parameters for Tests by
Heyman (1965)
Depth (m)
o.-:,...------------.., ,...-----, o r- 0.25 .... 1.75
Measured
sand
BCPILE
'.'.
2 '......) 5 0.135 0.9

.'
""
""
""
" .'
,~ organic clay
& peat
"

../~~e II
1.5 4.8
,ll 20 L - _....

land

20 40 60 80 (a) (b)
Bending Moment (kNm) 30
FIG. 6. Comparison of Measured and Predicted Bending Mo-
ments for Tests by Heyman (1965)
p (MPa) ~ (MPa)
y
thickness, were propped at the pile heads. Pile I was located FIG. 7. Soil Profile and Assumed 5011 Parameters for Test by
12 m away from the embankment toe, while Pile II was lo- Leusslnk and Welnz (1969)
cated at the embankment toe. The horizontal movements re-
sulting from the construction of the embankment were not de-
scribed in detail, but it was stated that for the full embankment
height, the movements were almost constant with depth. The
free-field lateral soil movement 12 m from the embankment toe
was 15 mm, while at the embankment toe the soil moved 30
mm. The py and G1 properties used in the analysis were derived
from the measured qc values. The values are summarized in
Fig. 5 alongside the idealized qc values adopted. Reasonable
agreement between the measured and predicted bending mo-
ments were obtained, as shown in Fig. 6. No data of the mea-
sured pile-displacement profile was available in the literature.

Leusslnk and Wenz (1969) 201-----i

In their investigations into the design of foundation systems


for a storage yard for iron ore, Leussink and Wenz (1969) 25 l . -_ _LL---...._..J..---o._...l----'_-l

monitored the behavior of a pile during the placement of the o 0.2 0.4 0.8
iron ore embankment. The pile was built up from four channel Hartz. Displacement (m)
sections welded together to form a square box 0.85 m wide FIG. 8. Comparison of Measured and Predicted Pile Displace-
and 30 m long. The pile head was hinged to restrain head ments for Test by Leusslnk and Welnz (1969)

796/ JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / SEPTEMBER 1997

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


Pile DIIpIacement
corresponding computed maximum bending moment (assum-
ing the pile to be linear elastic) associated with the pile lateral Measured
displacement shown in Fig. 8 was about 4,200 kNm. It is
likely that yielding of the pile would have occurred at about BCPILE
2,500 kNm.
Depth (m) E
o
alit & clay
4
0.1 0.38
-
:;; 10

!
.!! 15
Ii:

alit & peal


20
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

(8)

15 0.58
aand & gravel 0.3 0 0.1 0.2 0.3 0.4
17
Horlz. Displacement (m)
chalk
0
L- .30 3.8
24 3.0
4
p (MPa) ~ (MPa)
y .... 8
FIG. 9. Soli Profile and Assumed Soli Parameters for Teata by .ss::.
Bigot et al. (19n)
§12
PIle Dllplecemn
0 ,..----.......-
II
........--...,.....,..----..,
Measured i5: 16
alii & clay
BCPILE
5 20

....E 10 lin & peal


~1000
.... -1000 0 1000


s::.

15
Bending Moment (kNm)
FIG. 11. Comparlaon of Measured and Predicted Pile Dla-
placement and Bending Moments (1.76 m Embankment Settle-
lIlII1d & gravel
ment) for Teata by Bigot et al. (19n)

20 Bigot et al. (1977)


ch8Ik (8)
Comprehensive measurements were carried out by Bigot et
o 0.1 0.2 0.3 al. (1977) in their investigations into the behavior of a pile
adjacent to an embankment. The steel tube pile was 24 m long,

[±]
Harlz. Displacement (m)
with an outer diameter of 926 mm and a wall thickness of 15
mm. The soil profile is shown in Fig. 9 alongside the assumed
0 , . . - - - - - - - -.........- - - - ,
M~.,':Ired Py and G i parameters inferred from pressuremeter test data. The
lateral soil displacements for two stages of embankment place-
4 BCPILE ment are shown in Figs. 10 and II. The results of the predicted
pile displacement and bending moment profiles in Figs. 10 and
.... 8 11 are in good agreement with the measured results.
.s Stewart (1992)
§=12 Centrifuge model studies were carried out by Stewart (1992)
.!! to study the effects of embankment construction on adjacent
is: 16 single piles and pile groups. Six separate tests were conducted
on pile groups and one test (Test 5) was devoted to the re-
sponse of a single pile. Comparisons between the measured
20
and predicted results were carried out for Test 5. The prototype
(b)
square pile has a width of 0.4 m and a length of 22.5 m with
Eplp = 85.6 MNm 2 • The pile head was unrestrained. The soil
~1000 -1000 0 1000 comprised 0.5 m of surficial sand underlain by 18 m of kaolin
Bending Moment (kNm) and 6 m of dense sand. The bottom 7 m of the kaolin was
FIG. 10. Comparison of Measured and Predicted Pile Dla- relatively stiffer than the top II m. Based on the measured c.
placement and Bending Momenta (1 m Embankment Settle- data reported by Stewart (1992), the idealized Py and G i pro-
ment) for Tests by Bigot et al. (19n) files adopted in the analysis are shown in Fig. 12. As the
JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / SEPTEMBER 1997/797

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


o 57 0.815
properties of the clay and sand are summarized in Table 2.
The pile dimensions considered are summarized in Table 3.
The relative pile-soil stiffness ratio K R (Poulos 1973; Stewart
1992) is defined as
KR =EplpIE,oh; (29)
E
=10 where h. = thickness of the soft clay layer. The K R values in
Table 3 correspond to h. of 18 m.
~II 153 1.02
The lateral free-field soil movement profiles were deter-
if mined from plane strain finite-element studies for a number of
embankment pressures in the range q = 25 -1 00 kPa. Analyses
were carried out using the finite-element program AFENA
(Carter and Balaam 1990). The soil was represented by a
Mohr-Coulomb constitutive relationship. Eight-noded isopar-
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

(e)
L...-_ _...J 1578 (b) 20 ametric elements were used to model the soil beneath the em-
o 0.1 0.2 bankment.
fly (kPa) 0. (MPa) SolI HorIz. D.p. (m) In finite-element analysis, the embankment is usually rep-
resented by finite elements that apply their own self-weight

o .---------.--------,[±J
8.2 mhigh .mbar*m8n1

(d) M~::
loading to the mesh, allowing the incremental construction
procedure to be modeled. For this study, to reduce computa-
tional time, the embankment loading was simulated with an
incremental vertical trapezoidal pressure distribution together
5 with a horizontal component of the load due to the lateral
active pressure on the sloping side. Finite-element parametric
E
= 10
studies by Stewart (1992) have indicated that specification of
a pressure loading in the shape of the embankment yielded
!
~
very similar results to those obtained from a self-weight Mohr-
it 15 o Coulomb embankment. Further details of the numerical mod-
eling are described in Goh et al. (1996).

20 16 m 20 m
und
~--+;----~
-400 -300 -200 -100 0 100 200 300
Bending Moment (kNm)

FIG. 12. Soli Profile, Soli Displacement, and Comparison of


Measured and Predicted Bending Moments for Test 5 by Stewart
(1992)

centrifuge tests were not instrumented to measure the lateral


soil movements at the toe of the embankment, the free-field 18m Soft Clay
soil displacement profile with depth at the location of the pile or
was inferred from finite-element analysis in which the em- 9m
bankment placement was simulated and the soil stress-strain
behavior was modeled using an elastic perfectly plastic (Mohr-
Coulomb) constitutive relationship. The computed soil dis-
placement profile is shown in Fig. 12. Details of the numerical Pile Dense Sand
modeling using the program AFENA (Carter and Balaam
1990) are described in the following section. Reasonable FIG. 13. Configuration for Parametric Studies
agreement between the measured and predicted pile bending
moments were obtained as shown in Fig. 12, although the TABLE 2. Sand and Clay Properties
results did not appear to model the restraint imposed by the Properties
stiffer clay below about 12 m depth. No data of the measured
pile-displacement profile with depth was available for com- Py G,
Material (kPa) (kPa) Poisson's ratio
parison.
(1 ) (2) (3) (4)
Clay 180 2.67 0.49
PARAMETRIC STUDIES Sand 1,739 20.7 0.30
To examine the effects of the lateral pile stiffness on the
behavior of a pile located adjacent to an embankment, a series
of idealized cases were examined. A single concrete pile em-
TABLE 3. Summary of Pile Properties (Ep =2.8 x 10 7
kPa)

bedded in a layered clay-sand stratum as depicted in Fig. 13 Pile diameter


was analyzed. 1\vo soil profiles were assumed: (1) 18 m of Pile (m)
clay overlying a deep layer of dense sand; and (2) 9 m of clay (1) (2)
overlying a deep layer of dense sand. The clay was assumed I 0.5 2.0
to be uniform with C u = 20 kPa and E,o = 2oocu ' The pile was 2 0.6 4.2
assumed to have been installed prior to the embankment con- 3 0.7 7.9
4 0.8 13.4
struction. The pile was connected to a rigid cap such that the 5 1.0 32.7
pile head was restrained from rotation but not deflection. The
798/ JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / SEPTEMBER 1997

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


q-1oo kPa
!<R M* = A exp[[3(qlc.)] (31)
0 0

~
The parameters A and 13 can be obtained either from Fig. 17
4 4
0.0~:J27
or from the following expressions:
i 8 i 8 A = 1.88(KR)o,~ (32)

1 12
1 12 [3 = 0.18(KR)-o,1 (33)
iI! 18 clay 118 Eqs. (30)-(33) can be used to obtain preliminary estimates
20 .....s 20 of the maximum bending moment induced in piles located at
the embankment toe. A more accurate assessment of the max-
~ -2000 0 2000 4000 24 0 100 200 imum bending moment at the detailed design stage would en-
BendIng Moment (lcNm) PIle O'-P'-ment (mm) tail the use of a numerical procedure similar to the one de-
<a) (b) scribed earlier.
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 14. Typical Pile Displacement and Bending Moment Pro-


flies Illustrative Example
An application of the proposed empirical equations is illus-
1000
trated by comparison with the centrifuge test data by Stewart

-j 800
(1992). The results were obtained from Test 9 and Test 11,
which consisted of pile groups with two rows of seven piles
spaced 4.9d apart. The Eplp for each pile is 85.6 MNm2 • For
--Ii
~
Test 9, the piles were embedded in 18 m of soft clay and 4 m
of dense sand. The undrained shear strength of the clay was
relatively uniform over the entire layer with an average c. =
E 800
i 17 kPa. For Test 11, the clay layer was 8 m thick with an
average c. = 11 kPa, and the sand layer was 4 m thick. The
al
C
K R for Test 9 and Test 11 were approximately 2 X 10-4 and
:s 400
Ii
...
CD 0.5
RIA'IIned

I 200 from rotation



0.00020

41' ..
0.00042
0.4 ~
0.00079
clay h.
0.00134

...
0.3 ~
q (kPa) .-ld
0.00327
~
FIG. 15. Comparison of Maximum Bending Moments for K R =
2 x 10-· with Centrifuge Test Results from Stewart (1992) 0.2

'TYpical profiles of the pile displacement and the pile bend-


0.1
ing moments are shown in Fig. 14. In all the cases considered,
the maximum bending moments in these piles occurred close
to the interface of the clay and sand layers. These results are 0
consistent with the pile group centrifuge test findings of Stew- 0 2 3 4 5 6
art (1992). qlc u
The results for pile 1 with K R = 2 X 10-4 have been plotted
FIG. 16. Nondlmenslonallzed Plot of M* versus qlc.
in Fig. 15 alongside the envelope of the data from all the
centrifuge tests conducted by Stewart (1992). The centrifuge 0.1 I'T'---------~..--., 0.44
tests consisted of six tests on pile groups and one test on a
single pile. For the pile groups, the pile heads were restrained
from rotation, while for the single pile, the pile head was un-
restrained. In the majority of the tests K R was in the range of 0.08
0.4
2 X 10-4 _3 X 10- 4 • The predicted bending moments are con-
sistent with the centrifuge test results and follow the same
trend as the finite-element results by Stewart et al. (1993), in A 0.06
which the moments are initially overestimated.
'.·····.9 0.36
APPLICATION TO DESIGN 0.04 ................... J3
The results from the parametric studies have been nondi- ......
..................
mensionalized in Fig. 16 as M* versus qlc•. The dimensionless 0.32
parameter M * is defined as follows 0.02 0 ....···...-

(30)
Mmax = maximum computed bending moment in the pile; and oOL...---""----...l.----'----....I 0.28
0.002 0.004
q = applied embankment pressure. Regression analysis shows KR
that the curves can be represented by the following best-fit
equation FIG. 17. Values for). and ~ Derived from Regression Analysis

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / SEPTEMBER 1997/799

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


bankment at the preliminary design stage. More detailed anal-
yses can be carried out using the numerical procedure de-
scribed previously to obtain a more reliable estimate of the
pile bending moments at the final design stage.
APPENDIX I. REFERENCES
Bigot, G., Bourges, E, and Frank, R. (1982). "Etude experimentale d'un
pieu soumis aux poussees laterales du sol." Revue Francaise de Geo-
technique, Paris, France, 18, 29-47 (in French).
Bigot, G., Bourges, F., Frank, R., and Geugan, Y. (1977). "Action du
deplacement lateral du sol sur un pieu." Proc.• 9th ICSMFE, Vol. I,
407-410 (in French).
Bourges, F., Frank, R., and Mieussens, C. (1980). "Calcu des efforts et
des deplacements engendres par des poussees laterals de sol sur les
pieux." Note Technique, Dept. des Sols et Fondations, LCPC, France.
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

Briaud, J. L. (1992). The pressuremeter. A. A. Balkema, Rotterdam, The


Netherlands.
Broms. B. B. (1964a). "Lateral resistance of piles in cohesive soils." J.
Soil Mech. and Found. Div., ASCE, 90(2).27-63.
Broms, B. B. (1964b). "Lateral resistance of piles in cohesionless soils."
J. Soil Mech. and Found. Div., ASCE, 90(3), 123-156.
Broms, B. B. (1972). "Stability of flexible structures (piles and pile
groups)." Proc.• 5th ECSMFE, Vol. 2. 239-269.
Broms, B. B., Pandey, P. C., and Goh, A. T. C. (1987). "The lateral
displacement of piles from embankment loads." Proc.• Japan Soc. of
Civ. Engrs., Tokyo, Japan, 338/111-8(12), I-ll.
Byrne, P. M., Anderson, D. L., and Janzen, W. (1984). "Response of
piles and casings to horizontal free-field soil displacements." Can.
Geotech J., Ottawa, Canada, 21(4), 720-725.
Carter, J. P. (1982). "A numerical method for pile deformations due to
nearby surface loads." Proc., 4th Int. Con! Numer. Methods in Geo-
mech.• Vol. 2, 811-817.
Carter, J. P., and Balaam, N. (1990). AFENA users manual. The Univ. of
Sydney, Australia.
Chen, L., and Poulos, H. G. (1994). "A method of pile-soil interaction
analysis for piles subjected to lateral soil movement." Proc., 8th Int.
Con! on Compo Methods and Adv. in Geomech.. Vol. 3,2311-2316.
Chow, Y. K., Chin, J. T., and Lee, S. L. (1990). "Negative skin friction
on pile groups." Int. J. Numer. and Anal. Methods in Geomech., 14(1),
75-91.
Duncan, 1. M., and Buchignani, A. L. (1976). "An engineering manual
for settlement studies." Geotech. Engrg. Rep., Dept. of Civ. Engrg.,
Univ. of Calif., Berkeley, Calif.
Goh, A. T. C., Teh, C. I., and Wong, K. S. (1996). "Behavior of em-
bankment piles undergoing lateral movements." Geotech. Res. Rep.
No. NTU/GT/96-I, School of Civ. and Struct. Engrg., Nanyang Tech-
nol. Univ., Singapore.
1 X 10- 2 , respectively. Because of the similarity of the mea- Heyman, L. (1965). "Measurement of the influence of lateral earth pres-
sured bending moments between the edge and center piles, sure on pile foundations." Proc.• 6th ICSMFE. Vol. 2, 257-260.
Stewart concluded that the pile spacing of 4.9d was sufficiently Huder, J., and Bucher, E (1981). "Underpinning of a pile foundation in
soft clay." Proc.• 10th ICSMFE, Vol. 2, 741-745.
large to minimize the load sharing between adjacent piles. In Lee, C. Y. (1993). "Pile groups under negative skin friction." J. Geotech.
addition, the measured maximum bending moments between Engrg., ASCE, 119(10), 1587 -1600.
the front and rear piles were also similar. The maximum bend- Leussink, H., and Wenz, K. P. (1969). "Storage yard foundations on soft
ing moments computed using (31)-(33) have been plotted in clay." Proc.• 7th ICSMFE, Vol. 2, 149-155.
Fig. 18 alongside the centrifuge test data for the front row of Mayne, P. W., and Kulhawy, E H. (1982). "Ko-OCR relationships in
soil." J. Geotech. Engrg., ASCE, 108(6),851-872.
piles. The calculated results provide a reasonable approxima- Poulos, H. G. (1973). "Analysis of piles in soil undergoing lateral move-
tion to the centrifuge test data, particularly at higher embank- ment." J. Soil Mech. and Found. Div., ASCE, 99(5), 391-406.
ment pressures. Poulos, H. G., and Davis, E. H. (1980). Pile foundation analysis and
design. John Wiley & Sons, Inc., New York, N.Y.
Randolph, M. E, and Houlsby, G. T. (1984). "The limiting pressure on
SUMMARY AND CONCLUSIONS a circular pile loaded lateral1y in cohesive soil." Geotechnique, Lon-
don, England, 34(4), 613-623.
A numerical procedure has been presented to assess the soil- Reese, L. C. (1984). Handbook on design ofpiles and drilled shafts under
movement-induced forces and bending moments caused by lateral load. U.S. Dept. of Transportation, Fed. Hwy. Admin., Va.
soils undergoing lateral movement and their effect on the pile Springman, S. M. (1989). "Lateral loading of piles due to simulated
integrity. Comparisons with documented case histories and embankment construction," PhD thesis, Univ. of Cambridge, England.
centrifuge model tests confirm that the method provides reli- Stewart, D. P. (1992). "Lateral loading of piled bridge abutments due to
embankment construction," PhD thesis, Univ. of Western Australia,
able estimates of the pile lateral displacements and bending Australia.
moments. Parametric studies were carried out for a single pile Stewart, D. P., Jewel1, R. J., and Randolph, M. E (1993). "Numerical
embedded in a uniform clay layer overlying a layer of dense modelling of piled bridge abutments on soft clay." Compo and Geo-
sand. From these studies, simple empirical equations have tech., London, England, 15, 21-46.
been proposed to provide initial estimates of the maximum Stewart, D. P., Jewel1, R. J., and Randolph, M. E (1994). "Design of
piled bridge abutments on soft clay for loading from lateral soil move-
bending moment for piles at the base of a sloped embankment ments." Geotechnique, London, England, 44(2),277-296.
and restrained from rotating at the pile head. These simple Teh, C. I. (1995). "BCPILE-A computer program for the analysis of
empirical equations allow a quick assessment of the maximum lateral pile behaviour." Geotech. Res. Rep. No. GTI95/05, School of
bending moment in a pile located at the toe of a sloped em- Civ. and Struct. Engrg., Nanyang Technol. Univ., Singapore.

800 I JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING I SEPTEMBER 1997

J. Geotech. Geoenviron. Eng. 1997.123:792-801.


Teh, C. I., and Wong, K. S. (1995). "Analysis of downdrag of pile ksj = initial stiffness of nodal soil spring per unit area;
group." Geotechnique, London, England, 45(2),191-207. Ko = coefficient of earth pressure at rest;
Vesic, A. S. (1961). "Bending of beams resting on isotropic elastic sol- L = pile length;
ids." J. Engrg. Mech. Div., ASCE, 87(2), 35-53.
Wong, K. S., and Teh, C. I. (1995). "Negative skin friction on piles in
M = pile bending moment;
MfDJI.x = pile maximum bending moment;
layered soil deposits." J. Geotech. Engrg., ASCE, 121(6),457-465.
M* = nondimensionalized maximum bending moment Mmu./
c.dh:;
APPENDIX II. NOTATION P = lateral nodal load vector;
The following symbols are used in this paper:
PL = pressuremeter limit pressure;
p" = maximum allowable lateral nodal load;
p. = atmospheric pressure;
c" = soil undrained shear strength; py = limiting lateral soil pressure;
d = pile width; q = embankment pressure;
EM = pressuremeter modulus; qc = cone resistance;
E1 = initial tangent modulus; Yl = relative soil-pile movement;
Downloaded from ascelibrary.org by UTEP LIBRARY-SERIALS on 08/14/14. Copyright ASCE. For personal use only; all rights reserved.

E, = soil secant (elastic) modulus; YP = nodal displacement vector;


Eplp = pile flexural stiffness; Y, = soil movement;
E 50 soil undrained secant modulus at half ultimate load; z = depth below surface;
G1 = soil shear modulus; 13 = bending moment coefficient from regression analysis;
h, = thickness of soft clay layer; "y = soil unit weight;
k h = subgrade modulus; 11 = shear modulus coefficient for sands;
khl initial spring stiffness; A = bending moment coefficient from regression analysis;
K R = relative soil-pile stiffness Ep l p /E50 h:; v, = Poisson's ratio of soil; and
Kp = pile stiffness factor; (J' ~ = effective vertical stress.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING / SEPTEMBER 1997/801

J. Geotech. Geoenviron. Eng. 1997.123:792-801.

You might also like