Ocean Engineering 273 (2023) 113998

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Cyclic p-y curve model of unidirectional loaded pile in sand considering the
characteristic of friction in the strain wedge model
Fanhuan Zeng , Chong Jiang *, Pan Liu
School of Resources and Safety Engineering, Central South University, Changsha, 410083, China

A R T I C L E I N F O A B S T R A C T

Handling Editor: Prof. A.I. Incecik Pile foundations are always subjected to cyclic loads due to the action of wind and waves. A cyclic p-y curve
model considering the characteristic of friction force is proposed to study the lateral response of pile foundations
Keywords: under unidirectional cyclic loads in sand. The soil resistance calculated by the improved nonlinear strain wedge
Pile model (SWM) is divided into friction resistance and compression resistance according to the different charac­
Cyclic lateral load
teristics of soil resistance. The friction resistance depends on the relative displacement between pile and soil,
p-y curve
whose direction and magnitude will constantly change in the process of cyclic load, while the compression
Strain wedge model
Soil degradation resistance is analyzed by combining the extended Masing rule and soil degradation model. The validity of the
model proposed in saturated sand is verified by centrifuge test results. Furthermore, the influence of cyclic loads
on pile deflection, bending moment and soil resistance is analyzed through the results calculated by the cyclic p-y
curve model.

1. Introduction permanent cumulative deformation and the degradation of strength and

stiffness of soil under cyclic loads. For these characteristics, many
Pile foundations are widely used in various engineering structures scholars have improved the load transfer curve (p-y curve) of the BNWF
which are subjected to various lateral loads caused by wind and waves. to study the lateral response of pile foundations under cyclic loads.
To study the lateral response of pile, the beam on nonlinear Winkler Naggar and Bentley (2000) added damping into the load transfer curve
foundation (BNWF) method is widely used due to its simplicity and of static elements to generate a dynamic p-y curve based on the BNWF.
practicability. The BNWF just considers the soil as a series of discrete Allotey and El Naggar (Allotey and El Naggar, 2008) improved the
nonlinear springs and the continuity of the soil is not considered. Based model by adding appropriate reloading and unloading rules and taking
on three-dimensional pile-soil interaction, the strain wedge model into account the development of slack zone and cyclic degradation.
(SWM) method (Norris, 1986; Ashour et al., 1998) considers the conti­ Gerolymos and Gazetas (2005) also analyzed soil-pile gap, degradation
nuity of soil, and the constitutive relation of soil is reflected in the and radiation damping characteristics in clay by modifying the
stress-strain relationship of the model. Xu et al. (2013) improved SWM Bouc-wen model. Memarpour et al. (2012) proposed a model of series
by combining the Duncan-Chang model, and Yang et al. (2016) solved and parallel nonlinear springs. Liang et al. (2018) developed a
the problem of constant strain along the strain wedge which is not quasi-static p-y hysteretic curve by combining the hyperbolic p-y curve
consistent with the real response of soil by proposing a nonlinear SWM. with the extended Masing rule. Yu et al. (2020) combined the boundary
The behavior of laterally loaded piles is also studied under the effect of surface elastic-plastic theory with the soil softening model to derive the
slope topographic, which is quite different from that of the pile on level theoretical cyclic p-y model of the laterally loaded pile in undrained
ground (Deendayal et al., 2016; Rathod et al., 2018, 2019). clay. Most of the above studies are based on the existing empirical p-y
The lateral loads applied to pile foundations are not always constant, curve, or cannot well reflect the mechanism of pile-soil interaction. In
which show cyclic characteristics in many cases (Rathod et al., 2020). addition, the experiments show that the unidirectional cyclic loads can
Numerous experimental studies (Matlock, 1970; Reese et al., 1975; Zhu lead to a greater reduction in the bearing capacity of pile foundations
et al., 2017; Leblanc et al., 2010; Peralta and Achmus, 2010; Abadie (Zhu et al., 2021), which the above studies rarely focus on. Heidari et al.
et al., 2019; Wang et al., 2021., Rathod et al., 2021) present the (2014) derived the p-y backbone curve based on the SWM, and

* Corresponding author.
E-mail addresses: [email protected] (F. Zeng), [email protected] (C. Jiang), [email protected] (P. Liu).

https://doi.org/10.1016/j.oceaneng.2023.113998
Received 15 November 2022; Received in revised form 16 January 2023; Accepted 14 February 2023
Available online 21 February 2023
0029-8018/© 2023 Elsevier Ltd. All rights reserved.
F. Zeng et al. Ocean Engineering 273 (2023) 113998

established the cyclic p-y curve considering soil cave-in and recom­ 2. Modified strain wedge model
pression. Kim et al. (2015) combined the accumulated strain obtained
from the cyclic triaxial test with the SWM to analyze the cyclic response The laterally loaded pile can be considered as a beam on a nonlinear
of the laterally loaded pile. However, the assumption of linear deflection Winkler foundation, and the control equation of the laterally loaded pile
of pile makes SWM less accurate in the study of flexible piles under can be expressed as Eq. (1).
lateral cyclic load. The soil resistance calculated by SWM is caused by
d4 y
two aspects: soil compression and friction. While the direction of friction EI + ky = 0 (1)
dz4
force is always opposite to the direction of relative displacement be­
tween pile and soil, and it will reverse during unload. Therefore, the where EI is the flexural rigidity of pile, y is the pile deflection, z is the
variation characteristics of friction force and compression force under depth of pile, k is the modulus of subgrade reaction which can be
cyclic load are different, which are not considered separately under expressed as p/y, p is the soil resistance calculated by SWM (Norris,
cyclic load in the above studies, leading to an inaccurate evaluation of 1986; Ashour et al., 1998) which can be divided into compression
the cyclic p-y curve. resistance pc and friction resistance pτ .
In this paper, the soil resistance generated by nonlinear SWM is
divided into friction resistance and compression resistance. The friction p = pc + pτ (2)
resistance is combined with the relative displacement between pile and
soil to analyze its change in the process of cyclic load, while the where pc is the lateral resistance from the compressive soil in front of the
compression resistance is analyzed by combining the extended Massing pile, and pτ is the lateral resistance caused by the relative displacement
rule and the degradation or hardening of soil. By considering the char­ between pile side and soil. They will be calculated separately according
acteristics of friction resistance and compression resistance separately, to their respective characteristics
the one-way cyclic p-y curve of the laterally loaded pile is obtained. An
iterative program is designed to calculate the lateral cyclic response of 2.1. Calculation of compression resistance
the pile, which is verified by centrifuge test.
The compressive soil in front of the pile is simplified as a three-

Fig. 1. Strain wedge: (a) three-dimensional diagram, (b) side view of the strain wedge, (c) plan view of the strain wedge.

2
F. Zeng et al. Ocean Engineering 273 (2023) 113998

dimensional passive wedge in SWM. model is employed in the strain wedge model.
The geometric of SWM is characterized by base angle βm , fan angle εi
φm , and depth of the wedge h (Fig. 1a). The relationship between βm , the Δσhi = σ1i − σ 3i = (10)
1
+ (σ1i − εσi 3i )
width of the wedge face BC at depth z and φm can be expressed as follow:
Ei ult

βm = 45◦ +
φm
(3) where Ei is the initial elastic modulus of the soil and (σ 1i − σ3i )ult is the
2 asymptotic value of the deviatoric stress, which can be calculated by Eq.
(11) and Eq. (12) respectively.
L(BC) = D + 2(h − z)tan βm tan φm (4)
( )n
σ3i
where D is the diameter of pile. Ei = K E P a (11)
Pa
The compression resistance pc can be expressed by Eq. (5) (Fig. 1c).
[ ( ) ]
(5) Δσhfi σ 3i tan2 45◦ + φ2 − 1
pc = S1 Δσh L(BC) (σ 1i − σ 3i )ult = = (12)
Rf Rf
where S1 is a coefficient for the pile with different shapes which equals
0.75 and 1.0 for circular pile and square pile respectively. Δσ h is the where Pa is the atmospheric pressure, taken as 101 kPa. σ3i is the
horizontal stress change in front of the pile corresponding to the effective vertical pressure. KE , Rf and n refer to elastic modulus coeffi­
deviatoric stress in the triaxial test (Fig. 1b). cient, failure ratio and elastic modulus index, respectively, related to
Different from the traditional SWM which assumes that the pile relative density Dr (Table 1).
displacement changes linearly with depth, an improved nonlinear SWM The relationship between φm and Δσhi is defined as Eq. (13) and is
is used in this study (Yang et al., 2016) (Fig. 2). expressed as a function of stress level SLi .
The soil was divided into a series of soil layers with thickness Δh. For ( φ
)
Δσhi tan2 45◦ + 2mi − 1
every horizontal slice i, the lateral deflection at the upper and lower SLi = = ( ) (13)
Δσ hfi tan2 45◦ + φ2 − 1
interfaces are yi− 1 and yi . The rotation angle δi can be expressed as Eq.
(6). In the calculation, the initial pile deflection Y0 is assumed to calcu­
yi− 1 − yi late δi by Eq. (6), and then εi , φmi and Δσ hi of each soil layer can be
tan δi = (6) obtained by Eq. (9), Eq. (10) and Eq. (13), so that LEFi and LBCi can be
Δh
calculated by Eq. (7) and Eq. (8). Finally, the compression resistance pc
The soil strain (εi ) is assumed to vary with depth and related to the
can be obtained by Eq. (5).
wedge width in the middle of layer LEFi as
yi− 1 + yi
εi = (7)
2LEFi

Then Eq. (4) is changed to Eq. (8) to calculate LBCi .

LBCi = D + 2LEFi tan φmi (8) Table 1
Duncan-Chang model parameters (Byrne et al., 1987).
The relationship between δi , εi and φmi can be expressed as Eq. (9)
Dr (%) KE n Rf
based on the Mohr’s circle for soil strain (Ashour et al., 1998).
25 250–500 0.5 0.9
γ i εi − εvi (1 + ν)εi 50 450–900 0.5 0.8
δi = = sin (2θmi ) = cos φmi (9) 75 850–1500 0.5 0.7
2 2 2
100 1500–3000 0.5 0.6
To obtain the relationship between Δσ hi and εi , the Duncan-Chang

Fig. 2. Side view of the nonlinear strain wedge.

3
F. Zeng et al. Ocean Engineering 273 (2023) 113998

2.2. Calculation of friction resistance

The friction resistance pτ can be calculated by Eq. (14) (Fig. 1c).

pτ = 2S2 τD (14)

where S2 is also a coefficient for the pile with different shapes which
equals 0.5 and 1.0 for circular pile and square pile respectively. τ is the
pile side shear stress calculated by the hyperbolic load transfer model as
Eq. (15) (Wong and Teh, 1995; Chen et al., 2007).
Δ
τ= 1
(15)
kini
+ τΔult

where Δ is the relative displacement between pile side and soil, kini and
τult are initial shear stiffness and interfacial ultimate shear stress
respectively. τult can be expressed as
τult = K0 σ 3 tan φs (16)
Fig. 3. Relationship between shear stress and relative displacement between
where K0 is the lateral soil stress coefficient and φs is the mobilized pile and soil.
friction angle obtained from Table 2.
The initial shear stiffness kini is related to τult expressed as generated by applying static lateral loads to the pile, which is just the
static pc − y curve. In this study, the backbone curve is derived based on
kini = τult / (Δu / χ ) (17)
the nonlinear SWM at each depth. The points on the pc − y curve can be
where Δu is the ultimate relative displacement between pile and soil determined by calculating the pile displacement and the corresponding
taken as 2 mm in this study, χ is a parameter taken as 4 so that the shear compression resistance under different pile head loads.
stress approaches τult when the relative displacement reaches Δu (Chen
et al., 2007) (Fig. 3). 3.1.2. Unload and reload curve
Based on the extended Masing rule, the unload and reload curves are
similar to the backbone curves, which are scaled with a scaling factor of
3. Cyclic p–y curve modeling
2. Pyke improves the extended Masing rule by combining the scaling
factor with the force at beginning of reloading or unloading (Pyke,
In SWM, the soil resistance p is not completely caused by the
1979).
compression of soil, but also partly due to the friction between soil and
pile side. When the pile is subjected to static load, compression resis­ pur
c=1 ± (18)
tance and friction resistance can be considered together, but when the pult
pile is subjected to cyclic load, the changes in compression resistance
and friction resistance are different. The direction of compression where c is the scaling parameter, pur is the force at the unloading point
resistance always remains the same, while the friction resistance always (pu ) or reloading point (pr ), and pult is the ultimate of soil resistance.
reverses periodically. To consider the characteristic of friction resistance Assuming that the expression of pc − y curve is pc = f(y), unload and
and generate the cyclic p-y curve, the cyclic pc − y curve and the cyclic reload curves can be expressed as
pτ − y curve are analyzed separately. Additionally, there are several ±(pur − pc )
(
±(yur − y)
)
assumptions in this study for simplicity of calculation: =f (19)
c c

(1) Friction resistance does not change with the increase of cyclic where yur denotes the pile deflection of the unloading point (yu ) or
loading times, and its direction and magnitude are only related to reloading point (yr ). The plus and minus signs in the equations above
the relative displacement between pile and soil. This means that indicate that the current state is unload and reload respectively.
the shape and position of the pτ − y curve remain the same for It should be noted that the separation of pile and soil occurs during
each load or unload. unload and there is a residual displacement of soil. In the next reload,
(2) The effect of friction resistance on the residual displacement of the pile contacts the soil again at the same point. Therefore, for the pile
soil during unload is not considered. Therefore, the residual under unidirectional cyclic load, the starting point of the reload curve is
displacement is completely controlled by the pc − y curve.
(3) The pc − y curve still follows the extended Masing rule during
unload and reload.

3.1. Cyclic pc − y curve

3.1.1. Backbone curve

To generate the cyclic pc − y curve, the backbone curve is first

Table 2
Interface friction angle (Kulhawy, 1984).
Material of pile Smooth Rough

Steel 0.5 φ 0.9 φ

Concrete 0.8 φ 1.0 φ
Fig. 4. Schematic of unload and reload curves.

4
F. Zeng et al. Ocean Engineering 273 (2023) 113998

the intersection point of the previous unload curve and the horizontal prevents the relative displacement. Therefore, no matter in the process
axis without considering the collapse of the soil (Fig. 4). of load or unload, the direction of friction resistance is always opposite
to the direction of pile movement. As a result, when the pile is trans­
3.1.3. Cyclic degradation of soil formed between loading and unloading states, the direction of friction
Many cyclic loading experiments show that the stiffness and strength resistance will be reversed due to the change of relative displacement
of soil will degrade or harden under cyclic load. The degradation index direction between pile and soil. Thus, the friction resistance cannot be
model was first proposed to represent the cyclic degradation of un­ considered as part of soil resistance together with the compression
drained clay (Idriss et al., 1978) and the hardening of dry sand (Drumm resistance in the process of unload (Fig. 5).
and Desai, 1986), which defines the amount of degradation as
3.2.1. Unload and reload curve
GsN
δN = = N− t
(20) The pτ − y curve remains the same for each load and reload, which
Gs1
can be expressed as Eq. (27) according to Section 2.2.
where δN is the amount of degradation, GsN and Gs1 are the secant shear y
pτ = 2S2 D (27)
modulus at cycles N and 1, t is the fitting parameter. 1
kini
y
+ τult
Then the concept of degradation is also extended to the pile-soil load-
transfer curve (p − y curve) to represent the cyclic degradation or When the pile starts to unload, the unloading stiffness of pτ − y curve
hardening of the pile-soil system (Grashuis et al., 1990; Rajashree and is basically equal to its initial stiffness, and the friction resistance de­
Sundaravadivelu, 1996). creases linearly to zero, where the pile displacement is defined as yτr .
( )( ) When the pile continues to rebound and move in the opposite direction,
δN = 1 − 1 − δf 1 − e− Nθgs (21) the friction resistance also increases in the opposite direction as the
hyperbolic pattern. And the starting point and direction of the curve are
δN = 1 − φ(y / d)log (N) (22) changed (Eq. (28)).
yτ r − y
where δf is the ultimate of degradation or hardening which can be δft or pτ = − 2S2 D (y ≤ yτr ) (28)
δfk standing for strength or stiffness, respectively, θgs is the degradation
1
kini
+ yττrult− y
parameter, and φ(y /d) is a function of pile deflection.
When the pile starts to reload, the unloading of the friction resistance is
When the pile-soil system is subjected to cyclic load, generally, it is
ignored. The cyclic pτ − y curve can be obtained by combining the above
not accurate to determine the degradation degree only by the loading
curves as shown in Fig. 6.
times N. The stress level at a certain depth of pile-soil system may
change with each cyclic load, which cannot be considered well in the
models above. Therefore, the fatigue damage function Dk is used to 3.3. One-way cyclic p-y curve
define the amount of degradation (Allotey, 2006).
( ( )θ )1θ Based on the theories above, the cyclic p − y curve can be obtained
∑ by superimposing the pc − y curve and the pτ − y curve (Fig. 7).
N
( ) ni
Dk = Dk N, Nf (S) = 1 − 1− (23)
i=1
Nf (Si ) When reloading and unloading in the gap zone, soil resistance p is
entirely provided by friction resistance. The residual displacement yr is
where θ is the model parameter which can be θt or θk when calculating the boundary of the gap zone, which can be obtained by solving Eq. (26)
the damage of strength or stiffness, Nf (Si ) is the number of cycles to when pc = 0 during unload.
failure at the force ratio of Si (the ratio of the current force and soil ( )
c1 δt − 1 puc
strength), and ni is the number of cycles at the force ratio of Si . Nf (Si ) can yr = yu − f (29)
δk c1 δt
be obtained from the S–N curve with two parameters S1 and ksn , which
represent the stress ratio at N = 1 and negative slope of S–N curve. The where c1 = 1 + ppult
uc
, puc is the compression resistance at the beginning of
S–N curve obtained from cyclic triaxial tests may be a log-log model or a
unload. Then the reloading curve can be expressed as
semilogarithmic model.
⎧
At each half-cycle loop, the incremental damage under the current ⎪
⎪
y
⎪ 2S2 D y < yr
stress level Si is calculated as ⎨ 1/kini + y/τult
p= ( ) (30)
1 ⎪
⎪ δk (y − yr ) y
ΔDk = (24) ⎪
⎩ δt f
δt
+ 2S2 D
1/kini + y/τult
y ≥ yr
2Nf (Si )

and the fatigue damage Dk can be obtained by accumulating ΔDk and Eq. 4. Calculation procedure
(23). After the fatigue damage Dk is determined, the amount of degra­
dation can be defined as To solve Eq. (1), the subgrade reaction modulus can be obtained by
( ) ki = pi /yi according to the soil resistance calculated by SWM. However,
δ = 1 − 1 − δf Dk (25)
ki may become extremely large at the bottom of the strain wedge due to
where δ can stand for both δt and δk , indicating the degradation of the small displacement of the soil layer. Thus, the maximum horizontal
strength and stiffness. Then, the unload and reload curves represented in subgrade reaction modulus proposed by Reese et al. (1974) (Eq. (31)) is
Eq. (19) can be rewritten as used to replace the ki of this part as well as the lower soil layers where
( ) the strain wedge is assumed to be undeveloped.
±(pur − pc ) ±(yur − y) δk
=f (26) kh = nh z (31)
cδt c δt
where nh is the coefficient of soil reaction modulus obtained from Fig. 8.
3.2. Cyclic pτ − y curve Since the subgrade reaction modulus is determined, Eq. (1) can be
converted into linear equations by the finite difference method for
Different from the compression resistance, the friction resistance is solving with appropriate boundary conditions. the global solving matrix
related to the relative displacement between pile and soil, which always for the pile is

5
F. Zeng et al. Ocean Engineering 273 (2023) 113998

Fig. 5. Direction of friction resistance changes.

Fig. 7. Schematic of cyclic p − y curve when pc and pτ are consid­

ered separately.

Fig. 6. Schematic of cyclic pτ − y curve.

KY = F (32)

where K is the global stiffness matrix, Y and F are the lateral deflection
and the load vectors of pile.
The cumulative damage Dk and degradation δ at each depth are
calculated according to the resistance provided by the soil spring. To
calculate the bearing characteristics of pile at the next reload, the p− y
curve of each layer of soil spring is updated to obtain the new subgrade
reaction modulus ki . Then, the global stiffness matrix is updated to
calculate the new pile deflection. It should be noted that the accumu­
lation of damage depends on the resistance level, so the p− y curve of the
soil spring without load will not change, which is also consistent with
the actual situation. The detailed calculation process is presented as a
flow chart in Fig. 9.

Fig. 8. The coefficient of soil reaction modulus versus relative density.

6
F. Zeng et al. Ocean Engineering 273 (2023) 113998

Fig. 9. Flow chart for the proposed method.

5. Approach verification and discussions One set of static loading test (S1-1) and three sets of cyclic loading tests
(S2-2, S2-3, S2-4) are analyzed in this paper (Table 3). Parameters of pile
5.1. Approach verification and soil in tests can be found in Table 4.
According to the test parameters, the coefficient of soil reaction
A series of centrifuge model tests of single pile in saturated sand modulus nh is 22 MN/m3 , and the cyclic degradation/hardening pa­
subjected to static and cyclic loads were carried out by Zhu et al. (2016). rameters for this study are S1 = 1.0, ksn = 0.3, θt = θk = 1.1 (De Alba
et al., 1976; Popescu and Prevost, 1993). δfk = 1.6 and δft = 1.0

Table 3
Test scheme.
Table 4
Test Loading Prototype pile Loading Number of Parameters of pile and soil in tests.
number amplitude (N) load (MN) frequency (Hz) cycles
Parameters of sand in tests parameters of the corresponding prototype pile
S1-1 Static load
S2-2 290 2.02 0.5 2121 γ (kN/m3 )
′
φ (◦ ) Dr (%) L (m) Load height (m) D (m) EI (kN • m2 )
S2-3 212 1.46 0.5 1894 9.17 35 60 50 6.75 2.5 56.66 × 106
S2-4 106 0.73 0.5 3057

7
F. Zeng et al. Ocean Engineering 273 (2023) 113998

according to a verified case for sand (Heidari et al., 2014), which are displacement. Compared with increasing loading amplitude, cyclic load
assumed to simulate the stiffness hardening due to the densification of seems to contribute more to this phenomenon, especially when the
sand. loading amplitude is larger. The same phenomenon can also be found in
Fig. 10 presents the lateral deflection and bending moment along the the distribution of bending moment, which the maximum bending
pile subjected to static load, which shows that the results calculated by moment depth deepens with cyclic load (Fig. 12b).
nonlinear SWM are in good agreement with the test results. Further­ The deepening of the zero point of pile displacement and the
more, the reliability of p − y backbone curve generated by nonlinear maximum bending moment position means that deeper soil is mobilized
SWM is proved, which lays a foundation for the reliability of cyclic p-y to resist the movement of pile. Fig. 13 shows the development of damage
curve. It is worth noting that the difference between the calculated and (Dk ) when the loading amplitude is 730 kN, which represents the
measured values of bending moment increases for larger loads. The densification of sand in this study.
development model of strain wedge leads to this deviation. With the It can be seen from the figure that after a certain number of cycles,
increase of load, the growth of Δσh , φm and LBC slows down, which leads the soil damage in the deeper layer begins to accumulate. Before this
to the calculation deviation of soil resistance and affects the calculation number of cycles, the soil at this depth is not disturbed.
of bending moment (Yang et al., 2016). In order to compare the p − y curve in every cycle, two definitions of
Fig. 11a shows the characteristic of displacement accumulation of stiffness are introduced. For a soil resistance not exceeding the ultimate
piles under cyclic loads. The maximum percentage of deviation between soil resistance, there is a point on the p − y curve corresponding to it,
the measured value and the estimated value is 20% and the minimum is and the slope of the line between the point and the origin of coordinates
5% throughout the cyclic loading process when H0 is 2020 kN. When H0 is defined as the secant stiffness Esi (Eq. (33)). While the tangent stiffness
is 1460 kN, the maximum deviation is 23% and the minimum deviation Eti is defined as the slope of the tangent line to a point on the p − y curve
is 10%. The deviation is between 13% and 35% when H0 is 730 kN. (Eq. (34)).
When the number of cycles is small, the increment of displacement is
Esi = tan (αsi ) (33)
large, and after a number of cycles, the displacement increases gently.
The trend of cumulative displacement calculated by the method pro­
Eti = tan (αti ) (34)
posed is generally consistent with the test, which proves the rationality
of considering the compression resistance and friction resistance sepa­ where αsi and αti are the angles between the lines and the coordinate axis
rately during unload. The bending moments before and after cyclic load (Fig. 14). It should be noted that the values of Esi and Eti are not constant
are in good agreement, especially the calculated value of the maximum values in a cycle, which are related to the choice of the magnitude of the
bending moment is relatively accurate as shown in Fig. 11b. Compared soil resistance. Therefore, when studying the variation of Esi or Eti in
with ignoring the characteristic of friction, the position and magnitude different cycles, it must be compared under the same soil resistance. And
of the maximum bending moment can be predicted more accurately by as shown in Fig. 14, this soil resistance may not necessarily be the actual
considering the characteristic of friction. soil resistance, since the actual soil resistance exerted by soil may be
different in different cycles.
According to the improved extended Masing rule, the shape of the
5.2. Discussions
unload curve is changed by Pu /Pult , which will cause a residual
displacement during unload. Fig. 15a is the cyclic p − y curve calculated
The variation of the bearing characteristics of pile during cyclic load
at the depth Z = 1.0D. The cyclic load will lead to the decrease in the
is often noteworthy. The influence of loading amplitude and cycle
secant stiffness because of the accumulation of residual displacement,
number on pile deflection, bending moment and p − y curve, as well as
but the tangent stiffness will increase due to the densification of sand.
the changes in soil damage and resistance during cyclic load are
Due to the decrease of secant stiffness, the resistance provided by the
analyzed in this section.
soil under the same displacement will be reduced, which results in a
As shown in Fig. 12a, cyclic load cannot only lead to the increase in
weakening phenomenon of the soil resistance under cyclic load. The
pile displacement, but also deepen the position of the zero point of pile

Fig. 10. The lateral response of the pile subjected to static load (S1-1), (a) the lateral deflection along the pile, (b) the bending moment along the pile.

8
F. Zeng et al. Ocean Engineering 273 (2023) 113998

Fig. 11. The lateral response of pile subjected to cyclic load, (a) the lateral displacement of loaded position, (b) the bending moment along the pile (S2-4).

Fig. 12. The lateral response of the pile subjected to different loading amplitude, (a) the lateral deflection along the pile, (b) the bending moment along the pile.

Fig. 14. Schematic of secant stiffness and tangent stiffness.

Fig. 13. Development of damage Dk .

When the increment of pile displacement gradually decreases after a few

weakening is more obvious in the shallow layers due to the low Pult .
of cycles, the increasing soil resistance with increasing pile deflection is
While in the deep layers the Pult is relatively large so that the soil
relatively small, so that the decrease in soil resistance due to the
resistance in the deep layers decreases relatively slowly, which even
decrease in secant stiffness can be observed. The soil resistance variation
increases somewhat due to the increase in pile deflection (Fig. 15b).

9
F. Zeng et al. Ocean Engineering 273 (2023) 113998

Fig. 17. The proportion of friction resistance.

about 20% at each depth and even reaches 30% with the increase of
cyclic number. Therefore, if the friction resistance is analyzed in the
same way as the compression resistance, the friction resistance will also
be weakened, and the soil resistance may be underestimated during
cyclic load.
Since Pu /Pult controls the shape of the unloading curve, whether the
compression resistance and friction resistance are considered separately
during unload determines the magnitude of residual displacement.
Fig. 18a shows the comparison of the two methods when calculating
residual displacement. It can be seen that if the characteristic of friction
is ignored, the residual displacement will be overestimated, especially
when the number of cycles is large. The accumulation of soil damage is
also overestimated if the characteristic of friction is ignored as shown in
Fig. 18b (represents the densification of sand). This will cause the soil
state to be incorrectly estimated, and eventually lead to a deviation in
Fig. 15. The cyclic p − y curve, (a) depth Z = 1.0D, (b) depth Z = 3.0D. the evaluation of pile cyclic characteristics.

at different depths can be seen in Fig. 16. 6. Conclusions

In the method proposed, it is crucial to consider the friction resis­
tance and the compression resistance separately. As shown in Fig. 17, A cyclic p-y curve model considering the characteristic of friction
with the weakening of shallow compression resistance, the proportion of force based on the modified SWM is proposed to study the lateral cyclic
friction resistance keeps increasing. While in the deep layers, it de­ response of pile in this paper. The soil resistance calculated by the
creases with the development of compression resistance first, and then modified SWM is considered separately as the friction resistance and the
gradually increases too. No matter before or after cyclic load, the pro­ compression resistance, and appropriate unload/reload and degradation
portion of friction resistance cannot be ignored, which accounts for rules are applied to generate the cyclic p-y curve. Based on this study,
the following conclusions can be obtained.

(1) By comparing the results calculated by the model proposed with

the centrifuge tests, it can be seen that the results obtained by
considering the characteristic of friction are more accurate than
those ignoring it, which shows the rationality of the model.
(2) The friction resistance accounts for about 20% of soil resistance
and the direction of relative displacement between pile and soil
constantly changes during cyclic load. Therefore, it is necessary
to consider the compression resistance and the friction resistance
separately. So that the soil resistance will not be underestimated
under cyclic load, especially in shallow layers. In addition, the
residual displacement of soil after unload can be evaluated more
accurately by considering the compression resistance and friction
resistance separately.
(3) Cyclic load will not only increase the deflection and bending
moment of pile, but also deepen the zero point of pile deflection
and the maximum position of bending moment, which means
deeper soil layers are mobilized to resist the displacement of pile.

Fig. 16. Soil resistance variation at different depths.

10
F. Zeng et al. Ocean Engineering 273 (2023) 113998

Fig. 18. The comparison of the characteristic of friction is ignored or considered, (a) the residual displacement, (b) the damage Dk .

(4) In the process of cyclic load, the tangent stiffness of p-y curve will Gerolymos, N., Gazetas, G., 2005. Phenomenological model applied to inelastic response
of soil-pile interaction systems. Soils Found. 45 (4), 119–132.
increase due to the densification of sand, but the cumulative re­
Grashuis, A.J., Dietermann, H.A., Zorn, N.F., 1990. Calculation of cyclic response of
sidual displacement will lead to the decrease of the secant stiff­ laterally loaded piles. Comput. Geotech. 10 (4), 287–305.
ness of p-y curve. As a result, the soil resistance will be weakened Heidari, M., El Naggar, H., Jahanandish, M., Ghahramani, A., 2014. Generalized cyclic
with cyclic load, which is more obvious in the shallow layers. p–y curve modeling for analysis of laterally loaded piles. Soil Dynam. Earthq. Eng.
63, 138–149.
Idriss, I.M., Dobry, R., Singh, R.D., 1978. Nonlinear behavior of soft clays during cyclic
CRediT authorship contribution statement loading. J. Geotech. Eng. Div. 104 (12), 1427–1447.
Kim, K., Nam, B.H., Youn, H., 2015. Effect of cyclic loading on the lateral behavior of
offshore monopiles using the Strain Wedge Model. Math. Probl Eng. 1–12, 2015
Fanhuan Zeng: Conceptualization, Methodology, Validation, (PT.15).
Writing – original draft, Preparation. Chong Jiang: Resources, Formal Kulhawy, F.H., 1984. Limiting tip and side resistance: fact or fallacy. In: Analysis and
Design of Pile Foundations. ASCE, pp. 80–98.
analysis, Supervision. Pan Liu: Writing – review & editing.
LeBlanc, C., Houlsby, G.T., Byrne, B.W., 2010. Response of stiff piles in sand to long-term
cyclic lateral loading. Geotechnique 60 (2), 79–90.
Liang, F., Chen, H., Jia, Y., 2018. Quasi-static py hysteresis loop for cyclic lateral
Declaration of competing interest response of pile foundations in offshore platforms. Ocean Eng. 148, 62–74.
Matlock, H., 1970. Correlation for Design of Laterally Loaded Piles in Soft Clay. Offshore
technology conference, Houston, Texas, pp. 577–594.
The authors declare that they have no known competing financial Memarpour, M.M., Kimiaei, M., Shayanfar, M., Khanzadi, M., 2012. Cyclic lateral
interests or personal relationships that could have appeared to influence response of pile foundations in offshore platforms. Comput. Geotech. 42, 180–192.
the work reported in this paper. Naggar, M.H.E., Bentley, K.J., 2000. Dynamic analysis for laterally loaded piles and
dynamic p-y curves. Can. Geotech. J. 37 (6), 1166–1183.
Norris, G.M., 1986. Theoretically based BEF laterally loaded pile analysis. In:
Data availability Proceedings of the 3rd International Conference on Numerical Methods in Offshore
Piling, pp. 361–386. Navtes.
Peralta, P., Achmus, M., 2010. An experimental investigation of piles in sand subjected to
Data will be made available on request.
lateral cyclic loads. In: 7th International Conference on Physical Modeling in
Geotechnics.
Acknowledgments Popescu, R., Prevost, J.H., 1993. Centrifuge validation of a numerical model for dynamic
soil liquefaction. Soil Dynam. Earthq. Eng. 12 (2), 73–90.
Pyke, R.M., 1979. Nonlinear soil models for irregular cyclic loadings. J. Geotech. Eng.
This work is supported by the National Natural Science Foundation Div. 105 (6), 715–726.
of China (Grant No.51978665). Rajashree, S.S., Sundaravadivelu, R., 1996. Degradation model for one-way cyclic lateral
load on piles in soft clay. Comput. Geotech. 19 (4), 289–300.
Rathod, D., Muthukkumaran, K., Sitharam, T.G., 2018. Effect of slope on py curves for
References laterally loaded piles in soft clay. Geotech. Geol. Eng. 36 (3), 1509–1524.
Rathod, D., Muthukkumaran, K., Thallak, S.G., 2019. Experimental investigation on
behavior of a laterally loaded single pile located on sloping ground. Int. J. GeoMech.
Abadie, C.N., Byrne, B.W., Houlsby, G.T., 2019. Rigid pile response to cyclic lateral
19 (5), 04019021.
loading: laboratory tests. Geotechnique 69 (10), 863–876.
Rathod, D., Krishnanunni, K.T., Nigitha, D., 2020. A review on conventional and
Allotey, N., 2006. Nonlinear Soil-Structure Interaction in Performance-Based Design. The
innovative pile system for offshore wind turbines. Geotech. Geol. Eng. 38 (4),
University of Western Ontario (Canada).
3385–3402.
Allotey, N., El Naggar, M.H., 2008. Generalized dynamic Winkler model for nonlinear
Rathod, D., Nigitha, D., Krishnanunni, K.T., 2021. Experimental investigation of the
soil–structure interaction analysis. Can. Geotech. J. 45 (4), 560–573.
behavior of monopile under asymmetric two-way cyclic lateral loads. Int. J.
Ashour, M., Norris, G., Pilling, P., 1998. Lateral loading of a pile in layered soil using the
GeoMech. 21 (3), 06021001.
strain wedge model. J. Geotech. Geoenviron. Eng. 124 (4), 303–315.
Reese, L.C., Cox, W.R., Koop, F.D., 1974. Analysis of laterally loaded piles in sand. In:
Byrne, P.M., Cheung, H., Yan, L., 1987. Soil parameters for deformation analysis of sand
Offshore Technology Conference, pp. 473–483. Houston, TX.
masses. Can. Geotech. J. 24 (3), 366–376.
Reese, L.C., Cox, W.R., Koop, F.D., 1975. Field testing and analysis of laterally loaded
Chen, R.P., Zhou, W.H., Cao, W.P., Chen, Y.M., 2007. Improved hyperbolic model of
piles om stiff clay. In: Offshore Technology Conference. OnePetro.
load-transfer for pile-soil interface and its application in study of negative friction of
Wang, H., Wang, L., Hong, Y., Mašín, D., Li, W., He, B., Pan, H., 2021. Centrifuge testing
single piles considering time effect. Chin. J. Geotech. Eng. 29 (6), 824–830.
on monotonic and cyclic lateral behavior of large-diameter slender piles in sand.
De Alba, P.A., Chan, C.K., Seed, H.B., 1976. Sand liquefaction in large-scale simple shear
Ocean Eng. 226, 108299.
tests. J. Geotech. Eng. Div. 102 (9), 909–927.
Wong, K.S., Teh, C.I., 1995. Negative skin friction on piles in layered soil deposits.
Deendayal, R., Muthukkumaran, K., Sitharam, T.G., 2016. Response of laterally loaded
Journal of Geotechnical Engineering 121 (6), 457–465.
pile in soft clay on sloping ground. Int. J. Geotech. Eng. 10 (1), 10–22.
Xu, L.Y., Cai, F., Wang, G.X., Ugai, K., 2013. Nonlinear analysis of laterally loaded single
Drumm, E.C., Desai, C.S., 1986. Determination of parameters for a model for the cyclic
piles in sand using modified strain wedge model. Comput. Geotech. 51, 60–71.
behaviour of interfaces. Earthq. Eng. Struct. Dynam. 14 (1), 1–18.

11
F. Zeng et al. Ocean Engineering 273 (2023) 113998

Yang, X.F., Zhang, C.R., Huang, M.S., Yuan, J.Y., 2016. Modification of strain wedge Zhu, B., Zhu, Z.J., Li, T., Liu, J.C., Liu, Y.F., 2017. Field tests of offshore driven piles
method for lateral soil-pile interaction in sand. Rock Soil Mech. 37 (10), 2877–2884. subjected to lateral monotonic and cyclic loads in soft clay. J. Waterw. Port, Coast.
Yu, J., Zhu, J., Shen, K., Huang, M., Leung, C.F., Tan, Q.W.J., 2020. Bounding-surface- Ocean Eng. 143 (5), 05017003.
based py model for laterally loaded piles in undrained clay. Ocean Eng. 216, 107997. Zhu, N., Cui, L., Liu, J., Wang, M., Zhao, H., Jia, N., 2021. Discrete element simulation on
Zhu, B., Li, T., Xiong, G., Liu, J.C., 2016. Centrifuge model tests on laterally loaded piles the behavior of open-ended pipe pile under cyclic lateral loading. Soil Dynam.
in sand. Int. J. Phys. Model. Geotech. 16 (4), 160–172. Earthq. Eng. 144, 106646.

12