Computers and Geotechnics 152 (2022) 105020

Contents lists available at ScienceDirect

Computers and Geotechnics

journal homepage: www.elsevier.com/locate/compgeo

Innovative Timoshenko soil-pile integrated element for lager diameter

laterally-loaded piles considering soil-pile interactions
Xue-You Li a, Hai-Peng Zhao a, Si-Wei Liu b, Jian-Hong Wan a, Rui Bai a, *
a
School of Civil Engineering, Sun Yat-Sen University & Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai 519082, PR China
b
Department of Civil and Environmental Engineering, Hong Kong Polytechnic University, Hong Kong, China

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

Keywords: The large diameter piles (LDP) are designed to provide strong bearing capacity, which should be carefully
Large diameter piles analysed to ensure the superstructures’ reliability. The conventional discrete spring element (DSE) method with
Timoshenko beam theory the Euler-Bernoulli beam element is inappropriate for LDPs as the pile’s shear deformations may be considerable.
Soil-pile integrated element
Besides, the nonlinear soil-pile interactions (SPI) cannot be conveniently simulated with the DSE method. To
Soil-pile interaction
Geometrically nonlinear analysis
accurately and effectively evaluate the lateral performance of LDP under extreme load, this paper develops an
innovative soil-pile integrated element based on the Timoshenko beam theory. The Green-Lagrange strain is used
to formulate the potential energy function with the SPI considered. The Gauss integration method is adopted to
simplify the element formulation and evaluate the SPI within the element end nodes. The secant and tangent
matrixes are derived for calculating the element deformations and internal forces. The transformation matrix
describing the nodal parameters’ relation between element local and global coordinate is presented for practical
numerical analysis. The soil-pile integrated element avoids using discrete soil spring elements, significantly
reducing the modelling and computing efforts. The verification examples demonstrate that the proposed soil-pile
integrated element has high accuracy and efficiency in geometrically nonlinear analysis.

1. Introduction and Poulos, 1993; Kim and Jeong, 2011; Nadeem et al., 2015). The
accuracy of the analysis results can be easily achieved. However, the
Large diameter piles (LDP) are used to provide strong bearing ca­ efficiency is considerable low due to huge pre-processing and compu­
pacity for structures, such as offshore wind turbines, long-span bridges tational efforts (Ouyang et al., 2020; Liu et al., 2020b; Wan et al., 2021).
and high-rise buildings (Ng et al., 2001; Xiao et al., 2018; Li, 2019). The Therefore, a simplified method, namely the discrete spring elements
LDP should be carefully designed to ensure the superstructure’s safety (DSE) method, is proposed. The method adopts a series of Winkler spring
and serviceability. With the increase of pile diameter in engineering and beam-column elements to model the continuum soil and pile,
construction, the primary purpose of monopile gradually changes from respectively (Bouzid et al., 2013; DNV, 2014; API, 2017; Garbellini and
carrying vertical loads to both horizontal and vertical loads at the same Laloui, 2019). The accuracy of the DSE method is greatly depended on
time, or even only horizontal loads. Based on PISA project, Byrne et al. the constitutive relation and spacing of the spring elements. The spring
(2020) and Burd et al. (2020) pointed out that vertical loads may be constitutive relations usually adopt the p-y models to describe the soil
assumed to have an insignificant influence on the performance of the resistance against lateral pile displacement (Ashour and Norris, 2000;
monopile. Therefore, an accurate calculation method evaluating LDP Kim et al., 2009). Various p-y models for different soil types have been
performances under significant lateral loads and complex soil conditions developed for practical applications. Yang and Jeremić (2002) proposed
is essential. the p-y curves of sand and clay through numerical analysis. Dash et al.
In pile analysis and design, the simulation of the soil-pile interactions (2017) proposed a method for calculating the p-y curves of liquefiable
(SPI) is most critical and challenging due to the nonlinear soil properties. soils. Li et al. (2020a) presented a p-y curve for analysis of laterally-
The finite element analysis method is the most effective approach to loaded piles in undrained clay. Some scholars noticed the p-y curve
comprehensively consider SPI. The intuitive method of modelling the varies greatly with the pile’s diameter (Yan and Byrne, 1992; Lee et al.,
pile and surrounding soil with solid elements is first developed (Chen 2010). For analysis of larger diameter piles, Achmus et al. (2016)

* Corresponding author.
E-mail address: [email protected] (R. Bai).

https://doi.org/10.1016/j.compgeo.2022.105020
Received 15 May 2022; Received in revised form 7 September 2022; Accepted 8 September 2022
Available online 29 September 2022
0266-352X/© 2022 Elsevier Ltd. All rights reserved.
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

(a) General pile loading scenario (b) Integration soil-pile analysis model
Fig. 1. Soil-pile integrated analysis method.

distinct feature of the elements is the direct incorporation of the SPI in

the element formulation, where the Gauss integration method is adopted
to consider the variations of soil resistance within the element. The
discrete spring elements are unnecessary. Therefore, less elements and
nodes are required, which greatly increases the efficiency. However,
previous research is mainly based on the EB theory, where the shear
deformation is ignored. The EB theory is suitable for the piles with
length significantly larger than the section dimensions. For large
diameter piles, the use of the EB pile element would lead to some errors
in the analysis results of lateral deformations. (Gupta and Basu, 2018;
Zheng et al., 2020; Liu et al., 2020a).
Therefore, this paper aims to propose an advanced soil-pile inte­
grated element for geometrically nonlinear analysis of the behavior of
laterally-loaded large diameter piles. Based on the Timoshenko beam
theory (Timoshenko, 1921; Bai et al., 2020), the element shape func­
tions considering the shear deformation are presented. The effect of the
soil-pile interaction is explicitly considered in the element equilibrium
formulation. Based on the potential energy principle, the element secant
relation and the tangent stiffness matrixes will be derived for easily
calculating the element resisting forces and deformations in geometri­
cally nonlinear analysis. Verification examples are presented to illus­
trate the accuracy and efficiency of the proposed pile integrated
element.

2. Integrated soil-pile model and assumptions

2.1. Soil-pile integrated analysis model

Fig. 2. Element deformations and nodal forces.
A general loading scenario of a pile under compression, horizontal
force and moment with different soil layers, L1 to Lj, is plotted in Fig. 1
evaluated the effect of pile diameter on p-y curves. Zhang et al. (2020) (a). Due to the lateral deformation, the soil would provide distributed
proposed a p-y curve construction method for large diameter monopile pressures continuously along the pile height. Based on numerical inte­
considering the roughness of the soil-pile interface. Liu et al. (2021) gration theory, the influences of the total SPI forces on the pile can be
proposed p-y curve models for large diameter piles using piezocone efficiently evaluated according to the pressure at the Gauss points
penetration test. However, as the spring element can only be located at (Bathe, 2006; Chen et al., 2021). Therefore, the soil-pile system can be
the nodes of beam-column elements, sufficiently small spring spacing represented as the analytical model shown in Fig. 1(b). Different p-y
should be used to precisely capture the distributed SPI along a pile. curves at controlling points can be adopted to represent different soil
Consequently, a large number of beam-column and spring elements are layers. Following the idea, a soil-pile integrated element is developed in
needed to simulate a single pile. This drawback restricts the DSE this paper.
application in larger-scale and highly nonlinear problems.
To remedy the shortages of the DSE method, researchers developed
advanced pile elements. Liu et al. (2020b) and Li et al. (2020b) proposed 2.2. Assumptions
a 2D pile element for analysing piles under lateral and axial loads based
on the Euler-Bernoulli (EB) beam theory. Subsequently, the elements The following assumptions are adopted in the derivation of the soil-
have been improved for three-dimensional analysis (Li et al., 2021). The pile integrated element for LDP:

2
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Fig. 3. Element deformation based on Euler-Bernoulli beam theory and Timoshenko beam theory.

(1) Shear deformations of pile are considered;

dv(x)
(2) The material of pile is homogeneous, linear elastic, and isotropic; = θ(x) + γ(x) (5)
dx
(3) Strains are small, but the deformations can be moderately large;
(4) There is no slippage and separation between the soil layers. where θ(x) is the cross-section rotation caused by bending defor­
mation; γ(x) is the shear strain along the element length.
3. Soil-pile integrated element formulation Generally, the shear stress distribution on the element cross-section
is uneven, so it is necessary to introduce the shear correction coeffi­
To simplify the element expressions, the formulation is derived in the cient k to correct the cross-section shear area. Therefore, the shear force
element local coordinate. The element nodal forces and deformations acting on the cross-section is given by the following equation:
are shown in Fig. 2. The origin of the coordinate is located at the top of A
the element. The element nodal force matrix {f}node has six components Q(x) = Gγ(x) (6)
k
as.
where G is the shear modulus; A is the cross-section area of pile. k is
{f }node = { P1 , H1 , M1 , P2 , H2 , M2 } (1) the shear correction coefficient, which adopts 2 and 1.11 for the hollow
where P, H are the nodal forces along the x-axis and y-axis, respec­ circular and solid circular sections, respectively, in this paper (Hutch­
tively; M is the moment at the element nodes; and subscripts 1 and 2 inson, 2001).
represent two nodes of the element, respectively. The bending moment acting on the cross-section is given by the
The corresponding nodal deformation matrix {u}node is. following equation:

(2) dθ(x)
{u}node = { u1 , v1 , θ1 , u2 , v2 , θ2 } M(x) = EI (7)
dx
where u, v are the nodal displacements along the x-axis and y-axis,
Based on Eq. (5), (6) and (7), an equilibrium equation associated
respectively; and θ is the rotation at the element nodes.
with the shear strain can be derived:
( )
3.1. Element shape functions dM(x) d2 θ(x) GA dv(x)
− Q(x) = EI 2
− − θ(x) = 0 (8)
dx dx k dx
The linear and the Hermite interpolation functions are adopted to By submitting Eq. (4) into Eq. (8), the interpolation function of
describe the element’s axial and lateral deformation, respectively. rotation is deduced:
u(x) = a0 + a1 x (3) (
bL2
)
θ(x) = c1 + 2c2 x + 3x2 + c3 (9)
2
v(x) = c0 + c1 x + c2 x2 + c3 x3 (4)
where b = 12EIk/GAL2 is the dimensionless factor; E is the Young’s
As shown in Fig. 3(b), the traditional beam element and pile element
modulus of pile; I is the moment of inertia.
are based on Euler-Bernoulli beams theory, whose cross-section rotation
It should be noted that Timoshenko beam theory can be degraded to
of the element is caused by the bending deformation and derived by dv
Euler-Bernoulli beam theory when b = 0. Based on the boundary con­
(x)/dx = θ(x). However, for Timoshenko beam theory, the cross-section
ditions of the element, the element displacement functions can be
rotation of the element is caused by both the bending deformation and
derived and written in a matrix form as follows:
shear strain:

3
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

{u} = [N]{u}Tnode (10)

(b − 2)ζ + 3ζ2
N10 = (22)
where {u} is the vector of the displacement functions in the local (1 + b)
coordinate system; {N} is the shape functions matrix. {u} and [N] are
where L is the length of element; ζ = x/L is the normalized
given by the followings:
coordinate.
{u}T = { u(x), v(x), θ(x) } (11)
⎡ ⎤ 3.2. Element total potential energy
N1 0 0 N2 0 0
[N] = ⎣ 0 N3 N4 0 N5 N6 ⎦ (12)
The pile-soil system total potential energy in the element local co­
0 N7 N8 0 N9 N10
ordinate can be written as:
where u(x) is the vertical displacement functions along the x-axis; v
Π = UE + US − W (23)
(x) is the horizontal displacement functions along the y-axis; θ(x) is the
rotation functions along the element length; and N1 to N10 are shape where UE is the pile strain energy; US is the energy caused by the soil
functions, which are expressed as: pressure; W is the work done by element nodal loads.
The Green-Lagrange strain is adopted to evaluate the element po­
N1 = 1 − ζ (13)
tential energy, where the Green-Lagrange strain components can be
N2 = ζ (14) simplified by ignoring the high-order terms as:
∂u(x) ∂θ(x)
bζ + 3ζ2 − 2ζ3 εlxx = − y (24)
N3 = 1 − (15) ∂x ∂x
1+b [( )2 ( )2 ]
1 ∂u(x) ∂θ(x)
(2 + b)ζ − (4 + b)ζ + 2ζ2 3 εnxx = + (25)
N4 = L (16) 2 ∂x ∂x
2(1 + b)
[ ]
1 ∂v(x)
bζ + 3ζ2 − 2ζ3 γ lxy = − θ(x) (26)
N5 = (17) 2 ∂x
1+b
[ ]
1 ∂θ(x) ∂u(x)
− bζ + (b − 2)ζ2 + 2ζ3 γ nxy = yθ(x) − θ(x) (27)
N6 = L (18) 2 ∂x ∂x
2(1 + b)
where εxx is the normal strain; γxy is the shear strain; and the su­
6( − ζ + ζ2 ) perscripts l and n denote the linear and nonlinear parts of the strain,
N7 = (19)
L(1 + b) respectively.
Based on Hooke’s law, the normal and shear stresses can be obtained
− (4 + b)ζ + 3ζ2 as:
N8 = 1 + (20)
(1 + b) ( )
σ xx = Eεxx = E εlxx + εnxx (28)
6(ζ − ζ2 ) ( )
N9 = (21) τxy = Gγxy = E γlxy + γnxy (29)
L(1 + b)

The sectional stresses induced by the nodal forces along the element
length can be expressed as:

∫
1
σ xx εxx + τxy γxy dV
UE =
2 V
∫
1 ( )2
= E(εxx )2 + G γ xy dV
2 V
∫ ( )2
1 ( )2
≈ E εlxx + 2σ xx εnxx + G γlxy + 2τxy γ nxy dV
2 V
∫ [ ( )2 ] ∫ ( )
1 ( )2
= E εlxx + G γlxy dV + σ xx εnxx + τxy γnxy dV
2 V V
∫ [( )2 ( )2 ] [ ( ) ]
1 ∂u(x) ∂θ(x) ∂u(x) ∂θ(x) ∂v(x) 2 ∂v(x) (32)
= E + y2 − 2y + G (θ(x))2 + − 2θ(x) dV
2 V ∂x ∂x ∂x ∂x ∂x ∂x
∫ [ ][( ) ( ) ] ( )
1 P ( ( x) x )y ∂u(x) 2 ∂v(x) 2 V ∂θ(x) ∂u(x)
+ + M1 1 − + M2 + + 2 yθ(x) − θ(x) dV
2 V A L L I ∂x ∂x A ∂x ∂x
∫ ( ) ( ) [ ( ) ]
1 L ∂u(x) 2 ∂θ(x) 2 GA ∂v(x) 2 ∂v(x)
= EA + EI + (θ(x))2 + − 2θ(x) dx
2 0 ∂x ∂x k ∂x ∂x
∫ [( ) ( ) ]
1 L ∂u(x) 2 ∂v(x) 2 ∂u(x)
+ P + − 2Vθ(x) dx
2 0 ∂x ∂x ∂x

4
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Table 1
Integral coefficients α1i to α10i.
Integral 1 2 3 4 5 6 7
coefficients

α1i 0.064496 0.127339 0.118449 0.052245 0.008607 0.000293 0

α2i 0.001562 0.013077 0.022074 0.013061 0.002515 0.000093 0
α3i 0.000123 0.006110 0.031930 0.052245 0.031930 0.006110 0.000123
α4i − 0.000041 − 0.001941 − 0.009329 − 0.013061 − 0.005950 − 0.000627 − 0.000003
α5i 0.000038 0.001343 0.004113 0.003265 0.000735 0.000030 0
α6i 0.000003 0.000627 0.005950 0.013061 0.009329 0.001941 0.000041
α7i − 0.000001 − 0.000199 − 0.001738 − 0.003265 − 0.001738 − 0.000199 − 0.000001
α8i 0 0.000293 0.008607 0.052245 0.118449 0.127339 0.064496
α9i 0 − 0.000093 − 0.002515 − 0.013061 − 0.022074 − 0.013077 − 0.001562
α10i 0 0.000030 0.000735 0.003265 0.004113 0.001343 0.000038

Table 2
Integral coefficients in the secant relationship.
Integral coefficients 1 2 3 4 5 6 7

μi 0.064619 0.133449 0.150379 0.10449 0.040536 0.006404 0.000124

ϖi 0.000124 0.006404 0.040536 0.10449 0.150379 0.133449 0.064619
χi 0.001565 0.013704 0.028024 0.026122 0.011844 0.002034 0.000041
ϑi 0.000041 0.002034 0.011844 0.026122 0.028024 0.013704 0.001565

Fig. 4. Single pile in homogeneous soil under horizontal load and bending moment.

5
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Fig. 5. Lateral responses of the pile with fixed end condition using different analysis methods.

P y
σ xx = + [M1 (1 − ζ) + M2 ζ ] (30) ∫ L∫ y
1
∫ L
A I US = p(y)dydx = β(y)y2 dx
0 0 2 0
M1 + M2 (33)
τxy = (31) 1∑ n
LA ≈ Hi β(yi )y2i
2 i=1
where P is the axial force.
By ignoring the high-order terms, the strain energy UE can be written where p(y) is horizontal soil resistance obtained from the p-y curve;
as: β(y) is the tangent value of the p-y curve; n is the number of Gaussian
integration points, which is 7 in this study; Hi is the weight coefficients; y
is the horizontal displacement at Gaussian point; the subscript i stands
Based on the Gauss integration method, the strain energy done by SPI for the ith Gaussian integration point.
forces, US, can be calculated as: Based on Eqs. (1) and (2), the work W is given by:

W = {f }node {u}Tnode (34)

6
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Fig. 6. Lateral responses of the pile with pinned end condition using different analysis methods.

7
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Fig. 7. Single pile in soil under horizontal load and bending moment.

3.3. Tangent stiffness matrix ⎡ ⎤

EA EA
⎢ 0 0 − 0 0 ⎥
The tangent stiffness matrix [k]E can be deduced based on the po­ ⎢ L
⎢
L ⎥
⎥
tential energy principle as δ2П=0, where δ is the sign of variation. The ⎢
⎢ 12EI 6EI 12EI 6EI ⎥
⎥
⎢ 0 − ⎥
tangent stiffness matrix is written in three parts: ⎢ (1 + b)L3 (1 + b)L2 (1 + b)L3 (1 + b)L2 ⎥
⎢ ⎥
⎢ ⎥
[k]E = [k]L + [k]G + [k]S (35) ⎢
⎢
(4 + b)EI
0 −
6EI (2 − b)EI ⎥
⎥
⎢ (1 + b)L (1 + b)L2 (1 + b)L ⎥
where [k]L and [k]G are linear and geometric stiffness matrices; [k]S [k]L = ⎢
⎢
⎥
⎥
⎢ EA ⎥
is the soil stiffness matrix. The matrixes are presented in Eqs. (36) to ⎢ 0 0 ⎥
⎢ L ⎥
(38). ⎢ ⎥
⎢ 12EI 6EI ⎥
⎢ sym. − ⎥
⎢ (1 + b)L3 (1 + b)L2 ⎥
⎢ ⎥
⎢ ⎥
⎢ (4 + b)EI ⎥
⎣ ⎦
(1 + b)L
(36)

Fig. 8. Lateral displacement of the pile with fixed end condition using different methods.

8
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Fig. 9. Bending moment of the pile with fixed end condition using different methods.

Fig. 10. Lateral displacement of the pile with pinned end condition using different methods.

9
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Fig. 11. Bending moment of the pile with pinned end condition using different methods.

Fig. 12. Schematic diagram of the centrifuge models at the prototype scale.

10
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Fig. 13. p-y curves for the steel pipe pile.

Fig. 14. Comparison of the lateral displacements of the X5.5 for the pile with fixed end condition.

⎡ ⎤
P M1 + M2 b(M1 + M2 ) P M1 + M2 b(M1 + M2 )
⎢L − −
⎢ (1 + b)L2 2(1 + b)L L (1 + b)L2 2(1 + b)L ⎥ ⎥
⎢ ⎥
⎢ ξP P M1 + M2 ξP P ⎥
⎢ − ⎥
⎢ ⎥
⎢ 5(1 + b)2 L 10(1 + b)2 (1 + b)L2 5(1 + b)2 L 2
10(1 + b) ⎥
⎢ ⎥
⎢ ⎥
⎢ (ξ + 2)P b(M1 + M2 ) P (ξ − 4)PL ⎥
⎢ − − − ⎥
⎢ 60(1 + b)2 L 2(1 + b)L 10(1 + b)2 60(1 + b) ⎥2 ⎥
⎢
[k]G = ⎢ ⎥ (37)
⎢ P M1 + M2 b(M1 + M2 ) ⎥
⎢ − − ⎥
⎢ L (1 + b)L2 2(1 + b)L ⎥
⎢ ⎥
⎢ ⎥
⎢ ξP P ⎥
⎢ sym. − ⎥
⎢ 2 ⎥
⎢ 5(1 + b)2 L 10(1 + b) ⎥
⎢ ⎥
⎢ ⎥
⎣ (ξ + 2)P ⎦
2
60(1 + b) L

11
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

Fig. 15. Bending moment of the pile with pinned end condition using different methods.

⎡ ⎤
along the x-axis and y-axis, respectively; Ms is the bending moments
⎢
⎢
⎥
⎥ induced by soil springs pressures.
⎢
⎢0 0 0 0 0 0
⎥
⎥ The element resisting forces {r}e is calculated by:
⎢ ∑n ∑n ∑n ∑n ⎥
⎢ 2 ⎥
⎢ i=1
α1i βi L i=1
α2i βi L2 0 i=1
α3i βi L i=1
α4i βi L ⎥ {r}e = [S]{u}Tnode (41)
⎢ ∑n ∑n ∑n ⎥
⎢ ⎥
where [S] is the secant stiffness matrix given in Appendix I, which
⎢ 3 2 3
⎢ α5i βi L 0 α6i βi L α7i βi L ⎥ ⎥
i=1 i=1 i=1
[k]S = ⎢
⎢ 0 0 0
⎥
⎥ can be deduced by taking the first derivative of the total potential en­
⎢ ⎥
⎢ ∑n ∑n
2 ⎥ ergy, i.e., δП = 0.
⎢ sym. α8i βi L α9i βi L ⎥
⎢ i=1
∑n
i=1 ⎥ The soil resisting forces {r}s is obtained through the Gauss-Legendre
⎢ ⎥
⎢
⎢ α 3⎥
10i βi L ⎥ integration method, which is calculated by:
i=1
⎢ ⎥ ∫ ( )
⎣ ⎦ 3x L − x ( x )2
H1s = P(x) + 1− dx
L L L L
(38) (42)
∑n

where ξ = 6 + 5b (b + 2); the integral coefficients α1i to α10i are

≈L μi P(ζi L)
i=1
tabulated in Table 1.
∫ ( )
x 3(L − x) ( x )2
H2s = − P(x) + dx
3.4. Element resisting forces L L L L
(43)
∑
n

The secant relations are used to calculate the element resisting forces ≈− L ϖ i P(ζi L)
during the incremental iterations. The element resisting forces come
i=1

from element deformations and soil springs pressures, as follows: ∫ ( x )2

M1s = xP(x) 1 − dx
{r}e = { P1e , H1e , M1e , P2e , H2e , M2e } (39) L L
(44)
∑
n

{r}s = { 0, H1s , M1s , 0, H2s , M2s } (40) ≈ L2 χ i P(ζi L)

i=1
where {r}e is the force vector caused by the element deformations;
{r}s is the force vector due to the soil springs pressures; Pe and He are the
forces induced by element deformations along the x-axis and y-axis,
respectively; Me is the bending moments induced by element de­
formations; Ps and Hs are the forces induced by soil springs pressures

12
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

∫ ( x )2
M2s = − (L − x)P(x) dx v(x) = C1 eRx cos(Qx) + C2 eRx sin(Qx) + C3 e− Rx
cos(Qx) + C4 e− Rx
sin(Qx) + q/ks
L L
(45) (52)
∑
n
≈ − L2 ϑi P(ζi L)
i=1 φ(x) = C5 eRx cos(Qx) + C6 eRx sin(Qx) + C7 e− Rx
cos(Qx) + C8 e− Rx
sin(Qx)
(53)
where P (x) is the horizontal resisting forces; μi, ϖ i, χ i, ϑi are the
integral coefficients, as shown in Table 2: where e is the Euler’s number; q is the uniform foundation reaction;
C1 to C8 are the constants determined by the boundary conditions; R and
3.5. Numerical implementation of the soil-pile integrated element Q can be calculated as:
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√̅̅̅̅̅̅̅̅̅̅ ( ̅
)
The element formulation is presented in element local coordinate. In R=
ks
+
ks
+
kg
(54)
a global analysis, the transformation matrix describing the relations 4EIθ 4Φθ 4EIθ
between the element degrees of freedom in local and global coordinates ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√√ ̅
̅̅̅̅̅̅̅̅̅̅ ( )
should be provided. According to the angles between the local and ks ks kg
global coordinate’s axes φn in nth iteration, the transformation matrix Q= − + (55)
4EIθ 4Φθ 4EIθ
for 2D problem can be written as:
[ ] AG
[t]n 0 Φ = 1+ (56)
[T]n = (46) k
0 [t]n

where [t]n as follows: θ = 1+

kg k
(57)
⎡ ⎤ AG
x2,n − x1,n y2,n − y1,n
⎢ L
−
Ln
0
⎥ ks and kg are the first and second elastic foundation parameters
n
⎢
⎢
⎥
⎥ (Tanahashi, 2004; Jian et al., 2013), respectively, which can be calcu­
[t]n = ⎢ y2,n − y1,n x2,n − x1,n 0⎥ (47)
⎢ Ln Ln ⎥ lated by:
⎣ ⎦
0 0 1 ( )1
0.65Es Es D4 12
ks = 2
(58)
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ D(1 − vp ) Ep Ip
( )2 ( )2
Ln = x2,n − x1,n + y2,n − y1,n (48)
Es t
xm,n and ym,n are the × and y global coordinates of the element node 1 kg = (59)
6(1 + vs )
and node 2 in nth load step, respectively.
The element nodal forces and deformations in the global coordinate where t is the thickness of the shear layer and t = 2.5D, as suggested
can be calculated as: by Xu (2005).
Based on the differential equilibrium equations and the boundary
{F}global = [T]{f }node (49) conditions, namely (1) the pile with free-fixed boundary condition: H
(0) = H and M (0) = M at pile top, φ (L) = 0 and v (L) = 0 at pile bottom,
{u}node = [T]T {u}global (50) (2) the pile with free-pinned boundary condition: H (0) = H and M (0) =
M at pile top, M (L) = 0 and v (L) = 0 at pile bottom, the lateral per­
Therefore, the global element stiffness is. formance of the pile in homogeneous soil can be obtained. The results
∑
N ELE
( ) from the proposed and the analytical method are compared in Fig. 5 and
[K] = [T]t [k]E [T]Tt (51) Fig. 6.
t=1
As shown in Fig. 5 and Fig. 6, the lateral displacement, flexural
where NELE is the number of the proposed soil-pile integrated rotation, bending moment and shear force obtained from the proposed
elements. method match well with the analytical method. The comparisons prove
the accuracy of the proposed linear and soil stiffness matrixes, [k]L and
4. Verification example [k]S. The example demonstrates the accuracy of the soil-pile integrated
element for predicting the lateral behaviours of piles with different
4.1. Example 1 - Single pile in homogeneous soil with constant stiffness boundary conditions in the linear analysis.
under horizontal load and moment
4.2. Example 2 - Single pile in homogeneous soil with varying stiffness
In this example, the behavior of a laterally-loaded large diameter pile under horizontal load and moment
in single-layer soil with constant stiffness is analyzed to illustrate the
accuracy of the proposed element. The pile dimension and the soil This example investigates the behavior of a laterally-loaded large
property are shown in Fig. 4. The diameter D, cross-section area A, diameter pile in the soil with varying stiffness under horizontal loads.
bending stiffness EpIp, and shear modulus G of the pile are 4 m, 0.7427 The soil lateral stiffness is assumed to vary linearly following the rate m1
m2, 2.969 × 108 kN⋅m2, and 7.923 × 107 kN/m2, respectively. The shear = 2500 kN/m4 (From 0 m to − 2.4 m) and m2 = 25000 kN/m4 (From
force correction factor, k, adopts 2. In order to validate the proposed − 2.4 m to − 12 m). The pile uses a hollow circular section with a
model for different boundary conditions, the pile tip is assumed to be diameter of 3 m and a thickness of 0.06 m. The bending stiffness and
fixed or pinned. For the fixed end condition, two loading scenarios, H = shear modulus are 1.234 × 107 kN⋅m2 and 7.923 × 107 kN/m2,
5000 kN, M = 500 kN⋅m and H = 10000 kN, M = 1000 kN⋅m, are respectively. The pile dimension and the soil property are shown in
calculated, while for the pinned end condition, another two loading Fig. 7. The pile boundary conditions consist of (1) free at the top, fixed at
scenarios, H = 500 kN, M = 50 kN⋅m and H = 1000 kN, M = 100 kN⋅m, the tip, and (2) free at the top, pinned at the tip. Two loading scenarios
are also calculated. for the pile with fixed end condition, H1 = 2000 kN, M1 = 500 kN⋅m and
The closed-form solution of a Timoshenko beam resting on an elastic H2 = 5000 kN, M2 = 1000 kN⋅m, are adopted, while another two loading
foundation can be analytically calculated as (Cheng and Pantelides, scenarios for the pile with pinned end condition, H1 = 3000 kN, M1 =
1988; Morfidis and Avramidis, 2002): 600 kN⋅m and H2 = 8000 kN, M2 = 1600 kN⋅m, are adopted. The DSE

13
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

method adopts the beam element based on Euler-Bernoulli beam theory in a desktop computer with 16 GB RAM and Intel (R) Core (TM) i5 CPUs
and Timoshenko beam theory for calculation, and the proposed method clocked at 3.0 GHz, respectively, which validates the accuracy and ef­
adopts the soil-pile integrated element based on Timoshenko beam ficiency of the proposed method in simulating nonlinear pile behaviours
theory for calculation. The lateral displacement and bending moment for the large diameter pile.
along the length of pile with fixed end condition obtained from the Fig. 15 shows the bending moment of the pile with pinned end
proposed method and the conventional DSE method in geometrically condition using different methods. The calculated bending moments by
nonlinear analysis are shown in Fig. 8 and Fig. 9. For the pile with the two beam theory are almost the same, and they both fit well with the
pinned end condition, the lateral responses are shown in Fig. 10 and the centrifuge test results. With the increase of the number of Timo­
Fig. 11. shenko beam elements, the calculation results of DSE method are
As shown in Fig. 8 and Fig. 10, the beam element based on Euler- gradually close to that of the proposed method. The proposed method is
Bernoulli beam theory underestimates the lateral displacement of piles more efficient with using smaller number of element.
and differs by about 20 % from the calculation results of soil-pile inte­
grated element (or beam element) based on Timoshenko beam theory. 5. Conclusion
The main reason is that Euler-Bernoulli beam theory does not consider
the shear deformation of pile cross-section, which proves that the soil- The paper proposes an advanced soil-pile integrated element for
pile integrated element (or beam element) based on Timoshenko beam geometrically nonlinear analysis of laterally-loaded large diameter piles.
theory has higher accuracy. However, as shown in Fig. 9 and Fig. 11, the The element is developed based on the Timoshenko beam theory. The
calculated bending moment of the two beam theory is almost the same. shape functions describing the element axial and lateral displacements
It can be concluded that the consideration of shear deformation hardly including the shear deformation are provided. The soil-pile interaction is
affects the bending moment of the pile, which is consistent with the directly incorporated in the element formulation by using the Gauss
results given by Gupta and Basu (2018). With the increase of the number integration method. Based on the protentional energy principle, the
of Timoshenko beam elements, the calculation results of DSE method are element stiffness matrixes, the linear, geometric and soil stiffness ma­
gradually close to that of the proposed method, and the accuracy of the trixes, are derived. The transformation matrix is given to describe the
former method can reach that of the latter method when 30 beam ele­ relations of the element nodal parameters in local and global coordinate.
ments are used. It should be noted that a larger number of elements The distinct feature of the element is the nonlinear lateral SPI along
dramatically increase the size of the stiffness matrix, which further arbitrary directions can be easily simulated based on the soil p-y curves.
resulting in more computational efforts. The computational consuming The discrete spring elements are avoided that greatly reduces the efforts
time for the proposed method with 5 soil-pile integrated elements and in modelling and data pre-processing. The verification examples
the DSE method with 30 beam elements are about 0.33 s and 7.2 s in a demonstrate the proposed method can accurately predict the lateral
desktop computer with 16 GB RAM and Intel (R) Core (TM) i5 CPUs performance of large diameter piles in different loading and boundary
clocked at 3.0 GHz, respectively. In reality, hundreds of piles are often scenarios. The proposed model requires less number of elements, which
used in large construction engineering. Therefore, it is proved that the greatly reduces the computational consuming time. The soil-pile inte­
proposed soil-pile integrated element has a higher efficiency in ana­ grated element can provide an efficient and feasible solution for soil-
lysing the horizontal deformation characteristics of large diameter pile. structure integrated analysis.

4.3. Example 3 – Laterally loaded pile with nonlinear soil stiffness CRediT authorship contribution statement

This example evaluates the behavior of a laterally-loaded large Xue-You Li: Writing – review & editing, Funding acquisition, Su­
diameter pile in the soil with nonlinear stiffness. Centrifugal tests of pervision. Hai-Peng Zhao: Software, Formal analysis, Validation,
laterally loaded steel piles conducted by Choo and Kim (2016) were used Visualization, Writing – review & editing. Si-Wei Liu: Conceptualiza­
to test the proposed model. As shown in Fig. 12, the geometry of the steel tion, Methodology. Jian-Hong Wan: Writing – review & editing. Rui
pipe pile are as follows: length L = 64 m, diameter Dp = 6.1 m, and wall Bai: Writing – original draft, Writing – review & editing, Methodology,
thickness tp = 0.12 m. Young’s modulus and Poisson’s ratio of the pile Software.
are 199 GPa and 0.3, respectively. Lateral loads are applied at the pile
head. Two boundary conditions were considered. For the pile with fixed Declaration of Competing Interest
end condition, the soils are sand, weathered rock, and hard rock from
top to bottom with corresponding thicknesses of 20, 5, and 6 m. The The authors declare that they have no known competing financial
lateral displacement of the pile at the location 5.5 m above the ground interests or personal relationships that could have appeared to influence
surface, X5.5, is recorded. For the pile with pinned end condition, the pile the work reported in this paper.
is embedded into a sand layer with a thicknesses of 31 m, and the
bending moment of the pile is recorded. The p-y curves are experimen­ Data availability
tally derived by Choo and Kim (2016) and plotted in Fig. 13. The sand
layer is thick, so several p-y curves at different depths are obtained. Data will be made available on request.
As shown in Fig. 14, the predictions of the lateral displacement by
the proposed soil-pile integrated element match well with the centrifuge Acknowledgements
test than the conventional Euler-Bernoulli beam elements. The proposed
method uses 12 soil-pile integrated elements can obtain the accuracy of This work was supported by the National Natural Science Foundation
the DSE method with 48 conventional Timoshenko beam elements, and of China (Nos. 51909288 and 52025094) and the Guangdong Provincial
the corresponding computational consuming time is 0.54 s and 13.49 s Department of Science and Technology (2019ZT08G090).

Appendix I. – Secant stiffness matrix

The secant stiffness matrix [S] is given in the following:

14
X.-Y. Li et al. Computers and Geotechnics 152 (2022) 105020

⎡ ⎤
η1,1 η1,2 η1,3 η1,4 η1,5 η1,6
⎢ η2,2 η2,3 η2,4 η2,5 η2,6 ⎥
⎢ ⎥
⎢ η3,3 η3,4 η3,5 η3,6 ⎥
[S] = ⎢
⎢
⎥ (59)
⎢ η4,4 η4,5 η4,6 ⎥
⎥
⎣ sym. η5,5 η5,6 ⎦
η6,6
where:η1,1 = η4,4 = EA+P
L
, η1,2 = η4,5 = − M1 +M2
η
,
(1+b)L2 1,3
= b(M 1 +M2 )
2(1+b)L
, η1,4 = − EA+P
L
( 2 )
M1 + M2 b(M1 + M2 ) 12EI P 5b + 10b + 6
η1,5 = ,η = η = η5,5 = +
(1 + b)L2 1,6 2(1 + b)L 2,2 (1 + b)L3 5(1 + b)2 L
( )
6EI P M1 + M2 12EI P 5b2 + 10b + 6
η2,3 = η2,6 = + , η = , η = − −
(1 + b)L2 10(1 + b)2 2,4 (1 + b)L2 2,5 (1 + b)L3 5(1 + b)2 L

EI(b + 4) PL(5b2 + 10b + 8) b(M1 + M2 )

η3,3 = η6,6 = + , η3,4 = η4,6 = −
(1 + b)L 60(1 + b)2 2(1 + b)L

6EI P EI(2 − b) PL(5b2 + 10b + 2)

η3,5 = η5,6 = − − ,η = −
(1 + b)L2 10(1 + b)2 3,6 (1 + b)L 60(1 + b)2

References Lee, J., Kim, M., Kyung, D., 2010. Estimation of lateral load capacity of rigid short piles
in sands using CPT results. J. Geotech. Geoenviron. 136 (1), 48–56.
Li, Z., 2019. Torsional vibration of a large-diameter pipe pile embedded in
Achmus, M., Terceros, M., Thieken, K., 2016. Evaluation of py approaches for large
inhomogeneous soil. Ocean Eng. 172, 737–758.
diameter monopiles in soft clay. The 26th International Ocean and Polar Engineering
Li, X.Y., Wan, J.H., Liu, S.W., Zhang, L.M., 2020b. Numerical formulation and
Conference. OnePetro.
implementation of Euler-Bernoulli pile elements considering soil-structure-
American Petroleum Institute, 2017. Recommended practice for planning, designing, and
interaction responses. Int. J. Numer. Anal. Met. 44, 1903–1925.
constructing fixed offshore platforms.
Li, X.Y., Wan, J.H., Zhao, H.P., Liu, S.W., 2021. Three-dimensional analysis of nonlinear
Ashour, M., Norris, G., 2000. Modeling lateral soil-pile response based on soil-pile
pile-soil interaction responses using 3D pile element model. Int. J. Geomech. 21 (8),
interaction. J. Geotech. Geoenviron. 126 (5), 420–428.
04021129.
Bai, R., Hajjar, J.F., Liu, S.W., Chan, S.L., 2020. A mixed-field Timoshenko beam-column
Li, S., Yu, J., Huang, M., Leung, C.F., 2020a. Application of T-EMSD based py curves in
element for direct analysis of tapered I-sections members. J. Constr. Steel. Res. 172,
the three-dimensional analysis of laterally loaded pile in undrained clay. Ocean Eng.
106157.
206, 107256.
Bathe, K.J., 2006. Finite element procedures.
Liu, X., Cai, G., Liu, L., Liu, S., Duan, W., Puppala, A., 2021. Improved py curve models
Bouzid, D.J., Bhattacharya, S., Dash, S.R., 2013. Winkler Springs (py curves) for pile
for large diameter and super-long cast-in-place piles using piezocone penetration test
design from stress-strain of soils: FE assessment of scaling coefficients using the
data. Comput. Geotech. 130, 103911.
Mobilized Strength Design concept. Geomech. Geoeng. 5 (5), 379–399.
Liu, J., Shi, C., Cao, C., Lei, M., Wang, Z., 2020a. Improved analytical method for pile
Burd, H.J., Taborda, D.M.G., Zdravković, L., Abadie, C.N., Byrne, B.W., Houlsby, G.T.,
response due to foundation pit excavation. Comput. Geotech. 123, 103609.
Gavin, K.G., Igoe, D.J.P., Jardine, R.J., Martin, C.M., McAdam, R.A., Pedro, A.M.G.,
Liu, S.W., Wan, J.H., Zhou, C.Y., Liu, Z., Yang, X., 2020b. Efficient beam–column finite-
Potts, D.M., 2020. PISA design model for monopiles for offshore wind turbines:
element method for stability design of slender single pile in soft ground mediums.
application to a marine sand. Géotechnique. 70 (11), 1048–1066.
Int. J. Geomech. 20 (1), 04019148.
Byrne, B.W., Houlsby, G.T., Burd, H.J., Gavin, K.G., Igoe, D.J.P., Jardine, R.J., Martin, C.
Morfidis, K., Avramidis, I.E., 2002. Formulation of a generalized beam element on a two-
M., McAdam, R.A., Potts, D.M., Taborda, D.M.G., Zdravković, L., 2020. PISA design
parameter elastic foundation with semi-rigid connections and rigid offsets. Comput.
model for monopiles for offshore wind turbines: application to a stiff glacial clay till.
Struct. 80 (25), 1919–1934.
Géotechnique. 70 (11), 1030–1047.
Nadeem, M., Chakraborty, T., Matsagar, V., 2015. Nonlinear buckling analysis of slender
Chen, L., Poulos, H.G., 1993. Analysis of pile-soil interaction under lateral loading using
piles with geometric imperfections. J. Geotech. Geoenviron. 141 (1), 06014014.
infinite and finite elements. Comput. Geotech. 15 (4), 189–220.
Ng, C.W.W., Zhang, L.M., Nip, D.C.N., 2001. Response of laterally loaded large-diameter
Chen, L., Abdelrahman, A.H.A., Liu, S.W., Ziemian, R.D., Chan, S.L., 2021. Gaussian
bored pile groups. J. Geotech. Geoenviron. 127 (8), 658–669.
Beam-Column Element Formulation for Large-Deflection Analysis of Steel Members
Ouyang, W.H., Yang, Y., Wan, J.H., Liu, S.W., 2020. Second-order analysis of steel sheet
with Open Sections Subjected to Torsion. J. Struct. Eng. 147 (12), 04021206.
piles by pile element considering nonlinear soil-structure interactions. Adv. Steel
Cheng, F.Y., Pantelides, C.P., 1988. Static Timoshenko beam-columns on elastic media.
Constr. 16 (4), 354–362.
J. Struct. Eng. 114 (5), 1152–1172.
Tanahashi, H., 2004. Formulas for an infinitely long Bernoulli-Euler beam on the
Choo, Y.W., Kim, D., 2016. Experimental development of the p-y Relationship for large-
Pasternak model. Soils. Found. 44 (5), 109–118.
diameter offshore monopiles in sands: centrifuge tests. J. Geotech. Geoenviron. 142
Timoshenko, S.P., 1921. On the correction for shear of the differential equation for
(1), 04015058.
transverse vibration of prismatic bars. Philo. Mag 41, 744.
Dash, S., Rouholamin, M., Lombardi, D., Bhattacharya, S., 2017. A practical method for
Wan, J.H., Liu, S.W., Li, X.Y., Zhang, L.M., Zhao, H.P., 2021. Buckling analysis of tapered
construction of py curves for liquefiable soils. Soil. Dyn. Earthq. Eng. 97, 478–481.
piles using non-prismatic beam-column element model. Comput. Geotech. 139,
Det Norske Veritas, 2014. Offshore Standard DNV-OS-J101 Design of Offshore Wind
104370.
Turbine Structures.
Xiao, S., Wang, K., Gao, L., Wu, J., 2018. Dynamic characteristics of a large-diameter pile
Garbellini, C., Laloui, L., 2019. Three-dimensional finite element analysis of piled rafts
in saturated soil and its application. Int. J. Numer. Anal. Met. 42 (11), 1255–1269.
with energy piles. Comput. Geotech. 114, 103115.
Xu, L., 2005. Study on the longitudinal settlement of shield tunnel in soft soil. Tongji
Gupta, B.K., Basu, D., 2018. Applicability of Timoshenko, Euler-Bernoulli and rigid beam
University (in Chinese). Ph.D. Thesis.
theories in analysis of laterally loaded monopiles and piles. Géotechnique 68 (9),
Yan, L.I., Byrne, P.M., 1992. Lateral pile response to monotonie pile head loading. Can.
772–785.
Geotech. J. 29 (6), 955–970.
Hutchinson, J.R., 2001. Shear coefficients for Timoshenko beam theory. J. Appl. Mech.
Yang, Z., Jeremić, B., 2002. Numerical analysis of pile behaviour under lateral loads in
68 (1), 87–92.
layered elastic–plastic soils. Int. J. Numer. Anal. Met. 26 (14), 1385–1406.
Jian, Y., Zhang, C., Huang, M., 2013. Soil-pipe interaction due to tunnelling: Assessment
Zhang, H.Y., Liu, R., Yuan, Y., Liang, C., 2020. A modified py curve method for offshore
of Winkler modulus for underground pipelines. Comput. Geotech. 50 (may), 17–28.
large-diameter monopile foundations. J. Hydrol. Eng. 51 (2), 201–211.
Kim, Y., Jeong, S., Won, J., 2009. Effect of lateral rigidity of offshore piles using proposed
Zheng, C., Luan, L., Qin, H., Zhou, H., 2020. Horizontal dynamic response of a combined
py curves in marine clay. Mar. Georesour. Geotec. 27 (1), 53–77.
loaded large-diameter pipe pile simulated by the Timoshenko beam theory. Int. J.
Kim, Y., Jeong, S., 2011. Analysis of soil resistance on laterally loaded piles based on 3D
Struct. Stab. Dy. 20 (02), 2071003.
soil–pile interaction. Comput. Geotech. 38 (2), 248–257.

15