You are on page 1of 12

Int J Civ Eng

DOI 10.1007/s40999-016-0072-8

RESEARCH PAPER

Nonlinear Analysis of Single Laterally Loaded Piles in Clays Using


Modified Strain Wedge Model
Ling-Yu Xu1 • Fei Cai2 • Guo-Xin Wang3 • Guo-Xing Chen1

Received: 30 June 2016 / Revised: 22 September 2016 / Accepted: 26 September 2016


Ó Iran University of Science and Technology 2016

Abstract This paper proposes a modified strain wedge angle than to the undrained shear strength; and (4) the
(MSW) model for nonlinear analysis of laterally loaded proposed MSW model applies to normally consolidated
single piles in clays. The MSW model is used to calculate clays and to overconsolidated clays until they reach their
the soil resistance under increasing pile deflection. The peak strength.
stress–strain behavior of clays in the MSW, which is nee-
ded to calculate the soil resistance, is described in terms of Keywords Strain wedge  p–y method  Nonlinear
both hyperbolic and bilinear stress–strain relationships. stress–strain relationship  Overconsolidated clays 
The subgrade reaction modulus of soil below the MSW is Pile–soil interaction
assumed to equal the conventional subgrade reaction
modulus and to remain constant under the lateral loading of
the pile. The applicability of the proposed model was 1 Introduction
verified by eight case histories. The results indicate that (1)
the predicted results are consistent with the measurements Piles are frequently subjected to lateral loads resulting from
for all eight full-scale tested piles; (2) the bilinear stress– winds, landslides, and seismic events [1–3]. Over the years,
strain relationship is not recommended for clays because a number of approaches have been proposed to characterize
the clays usually have a large e50 and, thus, they exhibit a the behavior of laterally loaded piles. Among these meth-
linear behavior in the MSW during loading; (3) the pre- ods, the p–y method is most commonly used. The ultimate
dicted pile response is less sensitive to the effective friction passive wedge model [1] is often used to calculate the
ultimate soil resistance of the p–y curve. This type of
model was modified to calculate the ultimate soil resistance
& Fei Cai of the hyperbolic p–y curve for clays [4, 5]. However, it is
feicai@gunma-u.ac.jp interesting to note that X-ray CT [6] and particle image
Ling-Yu Xu velocimetry method [7] observed a passive wedge different
lyxu@njtech.edu.cn from that of Reese et al. [1] for sands In addition, tradi-
Guo-Xin Wang tional p–y curves do not account for certain properties such
gxwang@dlut.edu.cn as the pile bending stiffness, cross-sectional shape, head
Guo-Xing Chen restraint, and installation method [8, 9].
gxchen@njtech.edu.cn The strain wedge model developed by Norris can
1
College of Transportation Science and Engineering, Nanjing
overcome a few of the limitations of the p–y method [10].
Tech University, No. 200, Zhongshan North Rd, The strain wedge model is based on mobilization of the
Nanjing 210009, China passive wedge in front of the pile and is used to calculate
2
Department of Environmental Engineering Science, Gunma the soil reaction from the soil strain in the three-dimen-
University, 1-5-1 Tenjincho, Kiryu 376-8515, Japan sional passive wedge. The calculation relates the one-di-
3
Faculty of Infrastructure Engineering, Dalian University of mensional beam on a nonlinear elastic foundation analysis
Technology, No. 2, Linggong Rd, Dalian 116024, China (BNEF analysis) to the three-dimensional soil–pile

123
Int J Civ Eng

interaction response. Ashour et al. extended Norris’ model 2 Proposed Method


using a so-called multisublayer technique to deal with a
single pile or pile group in layered soils [8, 11]. Xu et al. 2.1 General Description of the Modified Strain
[12] proposed a modified strain wedge model to charac- Wedge Model
terize the behavior of laterally loaded piles in sand using
both the bilinear and hyperbolic stress–strain relationships The purpose of the modified strain wedge (MSW) model is
to describe the behavior of sand in the strain wedge. to calculate the appropriate subgrade reaction modulus Es
However, the current strain wedge model cannot be applied in the following differential equation for a single pile under
to overconsolidated clays. In this study, a modified strain a lateral load:
wedge model (MSM) proposed by Xu et al. [12] is  4 
extended for both normally consolidated (NC) and over d y
EI þ Es ð xÞy ¼ 0 ð1Þ
consolidated (OC) clays. dx4
In the undrained triaxial compression test, the NC clays
where EI is the flexural stiffness of the pile, y is the lateral
generally exhibit strain-hardening behavior, whereas the
deflection of the pile, and Es(x) is the subgrade reaction
OC clays generally exhibit strain-softening behavior after
modulus at a depth of x. In this study, piles are modeled as
yield [13, 14]. In addition, some sensitive clays, although
Euler–Bernoulli beam elements using the finite-element
normally consolidated, can also exhibit strain-softening
method.
behavior after yield. However, the maximum bending
The subgrade reaction modulus Es is defined as follows:
moments and deflections of piles rather than the ultimate
p
lateral resistance of the soil often govern the design. This Es ¼ ð2Þ
focus is present because the final yield of soils may be y
reached at a relatively large strain [15] and can, conse- where p is the soil resistance, which is calculated using the
quently, induce very large deflections. Such large deflec- proposed MSW model, and y is the pile deflection obtained
tions may not be tolerated by the structures [16]. Thus, the from BNEF analysis in this study.
effect of strain softening of clays is insignificant in design The MSW has a basic configuration similar to that of the
and, consequently, is not considered in this paper. strain wedge proposed by Ashour et al. [8] for layered
The novelty of the present work is to extend the appli- soils. However, in this study, the front surface of the pro-
cation of the conventional strain wedge model [8, 12] to posed MSW is assumed to be continuous, as shown in
OC clays by modifying an effective stress analysis to cal- Fig. 1. Figure 1a shows the MSW in front of the pile,
culate the excess pore water pressure in OC clays in the which is characterized by a base angle bm of single layer or
strain wedge. The stress–strain behavior of clays in the base angles of multiple layers, a fan angle gu0 m , and a
MSW, which is necessary to calculate the soil resistance, is wedge height. Here, u0 m is the mobilized effective friction
characterized based on both hyperbolic and bilinear stress– angle; g is a coefficient. The coefficient g affects the fan
strain relationships. The height of the MSW is determined angle of the MSW, as shown in Fig. 1 and, thus, affects the
using modified Newton–Raphson iterative process. This soil resistance at the front face of MSW. Kim et al. [5]
paper focuses on flexible piles in which the pile deflection calculated the ultimate soil resistance using different fan
below the MSW is small. Thus, the subgrade reaction angles for clays in a failure wedge, and found that the
modulus of soil below the MSW is assumed to equal the predicted ultimate soil resistance agrees well with that
conventional subgrade reaction modulus and to remain measured in clays if the ultimate fan angle equals 0.2u0 .
constant with lateral loading of the pile. In addition, the Thus, a fan angle of 0.2 u0 m was adopted in this study, and
size effect of the pile diameter or width on the subgrade the calculations indicate that this value was appropriate in
reaction modulus below the MSW is taken into consider- all cases.
ation. A finite-element program, SWPILE, was developed In Fig. 1a, Drh is the horizontal stress change on the
for the BNEF analysis. Eight case histories were used to front surface of the MSW, and s is the shear stress along
verify the applicability of the proposed method. Finally, a the sides of the pile. In layered soils, each layer is assumed
discussion is presented regarding the convergence of the to behave as a uniform soil and has its own properties. In
proposed MSW model, the difference between the results the finite-element analysis, each soil layer is divided into
using the hyperbolic and bilinear stress–strain relation- sublayers, and the number of sublayers in the MSW is the
ships, the maximum stress level in OC clays, and the effect same as that of the beam elements, as shown in Fig. 2a.
of the fan angle, effective friction angle of clays, and The base angle bm and its complement Hm at any depth
undrained shear strength of clays on the pile response. are given by the equation:

123
Int J Civ Eng

Fig. 1 Modified strain wedge


(MSW) consisting of multiple
layers

(a) (b)
Fig. 2 Soil-pile interactions in a cross section in the direction of lateral loading: a sublayers in the modified strain wedge and b an array of
discrete nonlinear and linear springs

/0 m;i The width of the continuous front surface of the MSW,


bm;i ¼ 45 þ ð3Þ
2 BC, is given by the equation:
 0 
/0 m;i Bi Ci ¼ D þ 2Ei Fi tan gum;i ð5Þ
Hm;i ¼ 45  ð4Þ
2
where the subscript i denotes the number of a sublayer. Even where D is the pile diameter or width, and EF is the hor-
in a given soil layer, different sublayers can have different izontal distance between the pile and the front surface of
mobilized effective friction angles /0 m and, thus, different the MSW, as shown in Fig. 1. EF at the top of the ith
base angles bm because of the different stress levels. sublayer is given by the equation:

123
Int J Civ Eng

X
i However, a consolidated drained shear strength should be
Ei Fi ¼ Lj tan bm;j ð6Þ used for sands [12] because sands have relatively larger
j¼nPW
permeability than clays.
where nPW is total number of sublayers in the MSW, L is Drhf ¼ 2su ð12Þ
the thickness of the sublayers, and bm varies for different
sublayers. Note that the number of a sublayer, i.e., the The ultimate soil resistance in clays, as suggested by
index i, increases with depth, as shown in Fig. 2a. The Gowda [19], is given as:
index i will be omitted in the equations below for pult ¼ Dð10S1 su þ 2S2 su Þ ð13Þ
simplification.
which was also used by Ashour et al. [8] and Kim et al. [5].
An array of discrete nonlinear and linear springs repre-
To obtain an accurate mobilized effective friction angle,
senting the soil in and below the MSW, respectively, was
used to represent the soil–pile interactions, as shown in which is necessary to calculate BC and, thus, the soil
Fig. 2b. resistance p (see Eqs. 5 and 7), Ashour et al. [8] proposed a
method of effective stress analysis for calculating the
2.2 Soil Resistance at the Front Face of the Modified excess pore water pressure (EPWP) in NC clays. In this
Strain Wedge study, this method of effective stress analysis is extended to
OC clays.
For a horizontal slice of the wedge at a depth of x (see The following equation for calculating the EPWP in
Fig. 1b), the soil resistance p (force per unit length of the clays is from Skempton [20]:
pile) is obtained based on the principle of force equilibrium Du ¼ ADrh ð14Þ
[8], as follows:
where Du is the EPWP in clays. A is Skempton’s pore
p ¼ Drh BCS1 þ 2sDS2 ð7Þ pressure coefficient and is defined as a function of the
stress level SL:
where Drh is the horizontal stress change on the front
 
surface of the MSW, D is the pile diameter or wideness, s A ¼ A0 þ SL Af  A0 ð15Þ
is the shear stress on the sides of the pile, and S1 and S2
equal 0.75 and 0.5, respectively, for a circular pile section, where A0 is the initial value of A, and Af is the ultimate
whereas both values equal 1.0 for a square pile [17]. value of A at failure. Skempton [20] demonstrated that
The calculation of the horizontal stress change Drh and A = 1/3 in isotropic elastic materials. Thus, the represen-
the horizontal strain e will be discussed later. The shear tative A0 value of 1/3 is used for both NC and OC clays in
stress s is defined by the following equation [8]: this study. The value of Af is given by the equation [21]
!
s ¼ SLt sult ð8Þ 1 1 1
Af ¼ 1þ  0  0 ð16Þ
sult ¼ asu ð9Þ 2 su rv0 sin u

where SLt is the stress level of shear along the sides of the where r0 v0 is the effective vertical stress, and u0 is the
pile, sult is the ultimate shear stress along the pile sides, su effective friction angle of the soil. Note that the effective
is the undrained shear strength of the clay soils, and a is an friction angle is the drained peak angle. The effective
adhesion factor. In this study, a is assigned a value of 1 for friction angle u0 can be estimated using Eqs. (17) and (18)
NC clays and 0.5 for OC clays, following the practice of for NC and OC clays, respectively, if it was not measured
De Ruiter and Beringen [18], and SLt is assumed to be directly due to time and cost constraints [22]:
double the stress level, SL, of the soil in the MSW, but not 0
u ¼ 43  10 log Ip ð :Þ : for NC clays ð17Þ
more than 1. ( 0
SLt ¼ 2SL  1 ð10Þ u ¼ 45  14 log Ip ð :Þ : 4\Ip \50
0 : for OC clays
u ¼ 26  3 log Ip ð :Þ : 50  Ip \150
The stress level SL in the MSW is defined as follows:
ð18Þ
Drh
SL ¼ ð11Þ where Ip is the index of plasticity and log is a base-10
Drhf
logarithm.
where Drhf is the horizontal stress change at failure. After calculating the EPWP using Eqs. (14) through
In this study, the MSW is proposed for pile subjected to (18), the effective horizontal stress (r0 v0 ? Drh - Du) and
short-term and static loading. Thus, undrained shear the confining stress (r0 v0 - Du) are determined for both
strength, given by Eq. 12, should be used for clays. NC and OC clays, as shown in Fig. 3a and b, respectively.

123
Int J Civ Eng

equation representing the hyperbolic stress–strain rela-


tionship is
e
r1  r3 ¼ ð23Þ
a þ be
where e is the axial strain in a triaxial test, and r1 and r3
are the major and minor principal stresses, respectively.
According to Norris [10], the stress state before lateral
loading of piles can be assumed isotropic in the strain
wedge due to pile installation. The lateral loading of piles
results in horizontal stress changes in the direction of pile
(a) movement; however, the vertical stress in the strain wedge
is mainly induced by the self-weight of soil and is nearly
unchanged with the lateral loading of piles. Thus, Eq. (23)
can be rewritten as:
e
Drh ¼ ð24Þ
a þ be
where the constants a and b are determined as follows:
1
a¼ ð25Þ
E0
1 Rf
b¼ ¼ ð26Þ
(b) ðr1  r3 Þult ðr1  r3 Þf

Fig. 3 Relationship between effective stress and total stress condi- In the above two equations, E0 is the initial elastic modulus
tions: a NC clays and b OC clays of the soil, (r1 - r3)ult is the asymptotic value of the
deviatoric stress, (r1 - r3)f is the deviatoric stress at
The mobilized effective friction angle u0 m of NC clays is failure, and Rf is the failure ratio, which is in the range of
calculated using the equation 0.75*1 and is assigned a value of 0.9 in this study. Xu
0 Drh =2 et al. [12] reported that the predicted results are insensitive
sin um ¼ ð19Þ to Rf in this range.
r0 v0  Du þ Drh =2
The initial elastic modulus E0 is obtained using the
The mobilized effective friction angle u0 m of OC clays can equation
be determined using the equation 2E50
E0 ¼ ð27Þ
0 Drh =2 2  Rf
sin um ¼ 0 0 ð20Þ
cm cot u0m þ rv0  Du þ Drh =2
where E50 is the secant modulus of elasticity at 50 % of
c0 m ¼ SLc0 ð21Þ (r1 - r3)f. E50 is used as the elastic modulus for the
bilinear stress–strain relationship, as shown in Fig. 2a, and
where c0 m is the mobilized effective cohesion and is defined
is given in the following equation for clays:
as the stress level SL multiplied by the effective cohesion
c0 . The effective cohesion is related to the undrained shear su
E50 ¼ ð28Þ
strength su of OC clays, as suggested by Sorensen and e50
Okkels [22], as follows: where e50 is the axial strain at 50 % of (r1 - r3)f. e50 can
0 be estimated based on the range of values of su corre-
c ¼ 0:1su  30 kPa ð22Þ
sponding to NC clays and OC clays [24], as shown in
Fig. 4.
2.3 Soil Stress–Strain Relationship
2.4 Relation Between the Horizontal Strain
The hyperbolic stress–strain relationship developed by and the Pile Deflection
Duncan and Chang [23] is used to describe the Drh - e
relationship, as shown in Fig. 2a. As a comparison, the In this study, the pile deflection at each sublayer in the
bilinear stress–strain relationship is also adopted. The strain wedge can be written as:

123
Int J Civ Eng

lateral load, DPn, and an incremental moment, DMn, for the


nth load step. Then, the new strain wedge height, h1, and the
new horizontal strain, e1, are given as follows [12]:
8 8 Rx
e1 > 2 0 y dx
>
< ySW0 ¼ h1 ¼ yFEM0 >
< h1 ¼ 0 FEM
WZ s x0 yFEM0
1 ) ð33Þ
>
: h1 ySW0 ¼ yFEM dx >
> y FEM0 Ws
2 : e 1 ¼
0 h1
where yFEM is the pile deflection between the ground line
and the point of first-zero deflection, and yFEM0 is the pile
deflection at the ground level. x0 is the location of first-zero
deflection point as shown by point C in Fig. 2a.

2.6 Subgrade Reaction Modulus Below the Modified


Strain Wedge
Fig. 4 Relationship between e50 and undrained shear strength su [24]
To account for the size effect of the pile diameter or width
y ¼ ðh  xÞd ð29Þ on the subgrade reaction modulus, Carter [25] and Ling
[26] suggested that the modulus of the subgrade reaction is
where h is the height of the strain wedge, x is the depth linearly proportional to the pile diameter or width, as
from the ground level to the middle of each sublayer under expressed by the following equation:
consideration, and d is the deflection pattern, as shown in
D E0
Fig. 2. The deflection pattern d is related with the soil Es ¼ ð34Þ
Dref ð1  t2 Þ
strain e in the strain wedge, as discussed in detail by Norris
[10]. where E0 is the initial elastic modulus, calculated using
e Eqs. (27) and (28); t is Poisson’s ratio of the soil; and Dref
d¼ ð30Þ is a reference size set to a value of 1 m.
Ws
where the parameter Ws varies with Poisson’s ratio of soils 2.7 Flow Chart of the Proposed Procedure
t and the mobilized effective friction angle u0 m [8]:
2 The proposed procedure has been incorporated in the
Ws ¼ ð31Þ
ð1 þ tÞ sin 2Hm SWPILE program for analysis of laterally loaded single
piles in clays, in which the modified Newton–Raphson
The Ws value varies in a small range due to material method using an initial stiffness K0 is used to solve the
properties of clays. Because Poisson’s ratio is 0.5 under nonlinear system [27]. The flow chart of the program is
undrained condition, and Hm varies between 45° (for shown in Fig. 5, in which Dfn is the incremental load vector
um = 0° at e = 0) and 30° (for um = 30° at failure), the
of the external force vector fnþ1 in the load step n, yknþ1 is
value of the parameter Ws is between 1.33*1.54 for clays
the total incremental vector of the lateral response of the
[8]. Xu et al. [12] found that the predicted results were k
pile after the kth iteration of the load step n, fnþ1 is the
insensitive to Ws. Consequently, Ws is taken as a constant k
equal to the average value of 1.44 for all clays. internal force vector, Wn?1 is the residual force vector, and
The pile deflection at the ground level ySW0 (x = 0) is ef is the relative residual error, which is defined as the norm
given as: of the residual force vector Wkn?1 divided by that of the
external force vector fnþ1 . The iterative loading by step
e
ySW0 ¼ hd ¼ h ð32Þ n stops when the relative residual error is less than 1 %.
Ws
Details of the modified Newton–Raphson method are pre-
sented by Zienkiewicz and Taylor [27].
2.5 Height of the Modified Strain Wedge

The strain wedge height, h, is calculated based on an iterative 3 Verification Based on Case Histories
process, as suggested by Xu et al. [12]. First, the initial strain
wedge height, h0, and the initial horizontal strain, e0, are The validity of the proposed method is verified by com-
specified at first trial to calculate the subgrade reaction mod- paring the results from the present approach with those
ulus, Es, for the FEM analysis of the pile under an incremental from full-scale pile tests and those from earlier research.

123
Int J Civ Eng

pile, with an outside diameter of 1.016 m and a wall


thickness of 16 mm, was driven to a penetration of 25.6 m
below the ground surface. The soil profile and properties at
test site are determined from five boreholes and conven-
tional sampling near the piles. The ground water level is at
the ground surface.
Figure 6 shows the predicted and measured deflections
and moments along the pile under a lateral load of 200 and
600 kN. The predicted results are close to those using the
p–y curve proposed by Kim et al. [5] and are in better
agreement with the measurements compared to the results
using existing p–y curves proposed by Matlock [30]. The
comparisons generally suggest that the proposed method is
capable of giving reasonably well distributions of deflec-
tions and moments for laterally loaded pile in clays.

3.2 Case of Gill [28]

Gill [28] reported the results of eight lateral load tests that
were performed on free-head pipe piles with diameters
ranging from 0.115 to 0.405 m. Tests on piles P1, P2, P3,
and P4 were performed in a dry area where the ground
water was at a depth of 2.3 m, and tests on the other four
piles, i.e., P5, P6, P7, and P8 were performed in an adjacent
saturated area where water was allowed to pond for a
number of days prior to driving and testing. Piles P3 and P7
are not modeled in present study because they behaved as
rigid piles.
Figure 7 shows the predicted and the measured load–
deflection curves for the various piles in both dry and
saturated areas. The predicted deflections generally agree
Fig. 5 Flow chart for analysis of a single laterally loaded pile in clay
well with the measured results. The predicted deflections
using the p–y method, as reported by Meyer [31], are also
The pile and soil properties are shown in Tables 1 and 2, shown in Fig. 7. The results indicate that the proposed
respectively. All the case histories involve static or short- method is able to model the response of the piles much
term loading and are, therefore, assumed to involve better than the p–y method.
undrained conditions.
3.3 Case of Rollins et al. [29]
3.1 Case of Kim et al. [5]
A static lateral load test was performed on a full-scale steel-
Kim et al. [5] conducted a series of lateral load tests on pipe pile that was 0.305 m in diameter, 9.5 mm in wall
full-scale steel-pipe piles and drilled shafts. The LTP-1 thickness, and 9.1 m in length [29]. The soil profile along the

Table 1 Pile properties used in the SWPILE program


Case histories Pile type Diameter (m) Embedded length (m) Flexural rigidity (kNm2) Distancea (m)

Kim et al. [5] Steel pipe (LTP-1) 1.016 25.6 1260,000 1.0
Gill [28] Steel pipe (P1, P5) 0.115 5.54 623 0.509
Steel pipe (P2, P6) 0.22 6.22 5460 0.509
Steel pipe (P4, P8) 0.405 8.13 48,400 0.509
Rollins et al. [29] Steel pipe 0.305 7.62 24,856 0.4
a
Distance from line of loading to the ground surface

123
Int J Civ Eng

Table 2 Soil properties used in the SWPILE program and maximum stress levels
Case histories No. of Layer thickness Soil type Total unit weight su (kN/m2) e50 u0 SLMAX
layers (m) (kN/m3) (°)

Kim et al. [5] 1 6.3 NC 17.5 20*108a, b


0.02a 28c 0.46
a, b a c
2 10.2 NC 17.5 108*170 0.01 28
3 5.5 NC 17.8 170*200a, b
0.005d 28c
4 4.6 Residual 17.8 – – 34c
soil
Gill–dry [28] 1 1 NC 18.1 20*80e, b
0.015d 30f 0.75 (P1); 0.67 (P2);
2 1.3 NC 18.1 80*21 e, b
0.015 d
30f 0.40 (P4)
3 # NCa 17.7 21e 0.015d 30f
e, b d
Gill–saturated 1 1 NC 17.7 0*50 0.02 30f 0.64 (P5); 0.58 (P6);
[28] 2 1.3 NC 17.7 50*30e, b
0.02d 30f 0.39 (P8)
3 # NC 17.7 21e 0.02d 30f
g d
Rollins et al. 1 0.8 OC 19.8 100 0.005 28c 0.93
[29] 2 2.3 OC 19.8 50 g
0.007 d
28c
3 1.5 Dense 19.8 – – 38.6h
sand
4 3.2 OC 19.8 50g 0.007d 28c
#
Pile tip located in this layer
a
su, e50 and u0 determined from laboratory tests reported by Kim et al. [5]
b
su varies linearly from the top to the bottom of the layer
c
Assumed in this study
d
e50 determined from Fig. 4 based on su
e
su slightly modified from average vane resistance reported by Ling [26]
f
u0 determined using Eq. (17) based on a value of Ip of 29 reported by Ling [26]
g
Determined from field and laboratory investigations reported by Rollins et al. [29]
h
Determined by Xu et al. [12] based on standard penetration test reported by Rollins et al. [29]

Fig. 6 Comparison of the predicted and measured a deflections and b bending moments along the pile under a lateral load of 200 and 600 kN in
the case of Kim et al. [5]

123
Int J Civ Eng

Fig. 7 Comparison of the predicted and measured load–deflection curves for piles in the a dry area and b saturated area in the case of Gill [28]

length of the pile consisted of various clay and sand-silt soils. dy for the pile under a lateral load of 180 kN in the case of
The water table was near the natural ground surface during Rollins et al. [29]. It is found that the strain wedge height
the testing. Consolidation testing indicated that the soils and the ground line deflection converge at the same time
were overconsolidated to a depth of approximately 10 m. when the relative residual force is lower than the tolerance
Figure 8 shows the predicted and measured load–deflec- 1 %. Thus, the convergence criteria used in the proposed
tion curves for the tested pile. The predicted deflections agree modified strain wedge model are appropriate.
well with the measurements and with those predicted by
Rollins et al. [29] using the p–y method. This indicates that the 4.2 Comparison of the Results Using Hyperbolic
MSW model can be applied to OC clays as well as to NC clays. and Bilinear Stress–Strain Relationships

Figure 10 shows the predicted load–deflection curves


4 Discussion using the hyperbolic and bilinear stress–strain relation-
ships for piles P1, P2 and P4 in the case of Gill [28]. For
4.1 Convergence of the Proposed Modified Strain piles P1 and P2, the difference between the predicted
Wedge Model deflections using these two stress–strain relationships is
insignificant. However, for pile P4, the predicted deflec-
Figure 9 shows relative residual force ef and errors of the tion that results from using the bilinear stress–strain
strain wedge height dh as well as the ground line deflection

Fig. 9 Relative residual force and errors of strain wedge height and
Fig. 8 Comparison of the predicted and measured load–deflection ground line deflection vs. iterations for the pile under a lateral load of
curves in the case of Rollins et al. [29] P0 = 180 kN in the case of Rollins et al. [29]

123
Int J Civ Eng

relationship is much larger than that resulting from the


use of the hyperbolic stress–strain relationship. This is
because the initial elastic modulus (E50) of the bilinear
relationship is smaller than the secant modulus of the
hyperbolic relationship when the horizontal strain is less
than e50. The calculations indicate that the soil using the
bilinear relationship behaves linearly until the test load
reaches the maximum. This behavior is due to the large
value of e50 for clays in the case of Gill [28]. Thus, the
use of the bilinear stress–strain relationship is not rec-
ommended for clays.

4.3 Effect of Fan Angle on the Pile Response

Figure 11 shows the predicted deflections with various fan


angles gu0 m (g = 0.1, 0.2, 0.5, 1.0) in the case of Rollins
et al. [29]. With the increase in the fan angle, the width of Fig. 11 The predicted deflections with various fan angles gu0 m
the front surface of the wedge BC (see Eq. 5) increases (g = 0.1, 0.2, 0.5, 1.0) in the case of Rollins et al. [29]
and, consequently, the soil resistance increases (see Eq. 7).
As a result, the predicted deflections generally decrease 4.5 Maximum Stress Level in OC Clays
with the increase in the fan angle. The calculations indicate
that a value of 0.2 for the coefficient g was appropriate for Strain softening of OC clays is not considered in this study;
all piles. thus, the predicted results are reliable for piles in OC clays
before such clays reach their peak strength. The OC clays
4.4 Effect of u0 and su on the Pile Response in this study did not reach their peak strength under the
maximum lateral loading of the test. The maximum stress
The effects of u0 and su are studied by varying their values level SLMAX of the various layers in the case histories is
(u0 = 20°, 25°, and 30°; su = 50, 75, and 100 kPa) for listed in Table 2.
layers 1–2 in the case of Rollins et al. [29], as shown in
Table 2. Figure 12, as expected, shows that the predicted
pile response is less sensitive to the effective friction angle 5 Conclusions
than to the undrained shear strength because the coefficient
g is quite small. A modified strain wedge (MSW) model is developed for
single laterally loaded piles in clays in this study. The
MSW model is used to calculate the soil resistance under
increasing pile deflections. The stress–strain behavior of
clays in the MSW, which is necessary to calculate the soil
resistance, is described using both hyperbolic and bilinear
stress–strain relationships. The height of the MSW is
determined using modified Newton–Raphson iterative
process. The subgrade reaction modulus of the soil below
the MSW is assumed to equal the conventional subgrade
reaction modulus and to remain constant with the lateral
loading of the pile. In addition, the size effect of the pile
diameter or width on the subgrade reaction modulus below
the MSW is taken into consideration. Finally, a discussion
is presented regarding the convergence of the proposed
MSW model, the difference between the results using the
hyperbolic and bilinear stress–strain relationships, the
maximum stress level in OC clays, and the effect of the fan
Fig. 10 The predicted load–deflection curves using the hyperbolic
and bilinear stress–strain relationships for piles P1, P2 and P4 in the angle, effective friction angle of clays, and undrained shear
case of Gill [28] strength of clays on the pile response. Based on the analysis

123
Int J Civ Eng

Fig. 12 Effect of a effective friction angle u0 and b undrained shear strength su on load–deflection curves in the case of Rollins et al. [29]

of eight case histories, the following conclusions can be 4. Sato H, Ohya T, Matsushima M (2001) Study on the behavior of
drawn: full-scaled single piles with rotation-fixed head under large lateral
deformation. Proc Jpn Soc Civil Eng 714:95–109 (in Japanese)
1. The predicted results are consistent with the measure- 5. Kim Y, Jeong S, Lee S (2011) Wedge failure analysis of soil
resistance on laterally loaded piles in clay. J Geotech Geoenviron
ments for all eight full-scale tested piles.
Eng 137(7):678–694
2. The bilinear stress–strain relationship is not recom- 6. Otani J, Pham KD, Sano J (2006) Investigation of failure patterns
mended for clays because the clays usually have a in sand due to laterally loaded pile using X-ray CT. Soils Found
large e50 and, thus, they exhibit a linear behavior in the 46(4):529–535
7. Hajialilue-Bonab M, Sojoudi Y, Puppala AJ (2011) Study of
MSW during loading.
strain wedge parameters for laterally loaded piles. Int J Geomech
3. The predicted pile response is less sensitive to the 13(2):143–152
effective friction angle than to the undrained shear 8. Ashour M, Norris G, Pilling P (1998) Lateral loading of a pile in
strength. layered soil using the strain wedge model. J Geotech Geoenviron
Eng 124(4):303–315
4. Strain softening is not considered in this study; thus,
9. Hokmabadi AS, Fakher A, Fatahi B (2012) Full scale lateral
the proposed method applies to piles in both normally behaviour of monopiles in granular marine soils. Mar Struct
consolidated and overconsolidated clays until the latter 29(1):198–210
reach their peak strength. 10. Norris GM (1986) Theoretically based BEF laterally loaded pile
analysis, Proceedings of third international conference on
numerical methods in offshore piling, Paris. France. 361–386
Acknowledgments This research was supported by the Natural 11. Ashour M, Pilling P, Norris G (2004) Lateral behavior of pile
Science Foundation of Jiangsu Province of China (Grant No. groups in layered soils. J Geotech Geoenviron Eng
BK20150958), the National Natural Science Foundation of China 130(6):580–592
(Grant Nos. 51508271, 51121005), and Postdoctoral Science Foun- 12. Xu LY, Cai F, Wang GX, Ugai K (2013) Nonlinear analysis of
dation of both Jiangsu Province, China (Grant No. 1501067B) and laterally loaded single piles in sand using modified strain wedge
China (Grant No. 2015M581782), which are gratefully model. Comput Geotech 51:60–71
acknowledged. 13. Robert DH, William DK (1981) An introduction to geotechnical
engineering. Printice-Hall Inc, Englewood Cliffs, New Jersey
14. Das BM (2010) Principles of geotechnical engineering, 7th edn.
Cengage Learning, Boston
References 15. Terzaghi K, Peck RB, Mesri G (1996) Soil mechanics in engi-
neering practice, 3rd edn. John Wiley & SONS INC, Hoboken
1. Reese LC, Cox WR, Koop FD (1975) Field testing and analysis 16. Nip DCN, Ng CWW (2005) Back-analysis of laterally loaded
of laterally loaded piles on stiff clay, Proceedings of Offshore bored piles. Proc ICE Geotech Eng 158(2):63–73
Technology Conference, Paper No. OTC 2312, 671–690 17. Briaud JL, Smith T, Meyer B (1984) Laterally loaded piles and
2. Sharafi H, Sojoudi Y (2016) Experimental and numerical study of the pressuremeter: comparison of existing methods. Laterally
pile-stabilized slopes under surface load conditions. Int J Civil loaded deep foundations: analysis and performance. ASTM Int
Eng 14(4):221–232. doi:10.1007/s40999-016-0017-2 835:97–111
3. Oyejobi DO, Jameel M, Sulong NR (2016) Nonlinear response of 18. De Kuiter J, Beringen FL (1979) Pile foundations for large north
tension leg platform subjected to wave, current and wind forces. Sea structures. Marine Georesour Geotechnol 3:267–314
Int J Civil Eng. doi:10.1007/s40999-016-0030-5

123
Int J Civ Eng

19. Gowda P (1991) Laterally loaded pile analysis for layered soil 26. Ling LF (1988) Back analysis of lateral load tests on piles. Rep.
based on the strain wedge model, MS thesis, University of No. 460, Department of Civil Engineering, University of Auck-
Nevada, Reno, Nevada land, New Zealand
20. Skempton AW (1954) The pore-pressure coefficients A and B. 27. Zienkiewicz OZ, Taylor RL (2005) The finite element method for
Geotechnique 4:143–147 solid and structural mechanics. Butterworth–Heinemann,
21. Wu TH (1976) Soil mechanics. Allyn and Bacon, Boston Amsterdam
22. Sorensen KK, Okkels N (2013) Correlation between drained 28. Gill HL (1968) Soil behavior around laterally loaded piles, Naval
shear strength and plasticity index of undisturbed overconsoli- civil engineering laboratory, Report R-571. Port Hueneme,
dated clays. Proceedings of 18th international conference on soil California
mechanics and geotechnical engineering, Paris. pp. 423–428 29. Rollins KM, Peterson KT, Weaver TJ (1998) Lateral load
23. Duncan JM, Chang CY (1970) Nonlinear analysis of stress and behavior of full-scale pile group in clay. J Geotech Geoenviron
strain in soils. J Soil Mech Found Div 96(5):1629–1653 Eng 124(6):468–478
24. Evans LT, Duncan GM (1982) Simplified analysis of laterally 30. Matlock H (1970) Correlations for design of laterally loaded piles
loaded piles, Report No. UCB/GT/82-04, University of Califor- in clay. Proceedings of second annual offshore technology con-
nia, Berkeley, California ference, Paper No. OTC 1204, 577–594
25. Carter DP (1984) A non-linear soil model for predicting lateral 31. Meyer BJ (1979) Analysis of single piles under lateral loading.
pile response, Report No. 359, Department of Civil Engineering, University of Texas, Austin
University of Auckland, New Zealand

123

You might also like