Soil Dynamics and Earthquake Engineering 162 (2022) 107447

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

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

Modelling and analysing failure modes of buried pipelines perpendicularly

crossing landslide boundaries
Hao-Jie Li a, Hong-Hu Zhu a, b, *, Chun-Xin Zhang a, Wei Zhang a
a
School of Earth Sciences and Engineering, Nanjing University, Nanjing, Jiangsu, 210023, China
b
Institute of Earth Exploration and Sensing, Nanjing University, Nanjing, Jiangsu, 210023, China

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

Keywords: The pipeline response to shallow translational landslides is investigated using semianalytical and numerical
Buried pipeline methods. The orthogonality of a pipeline and the landslide boundary is considered in this study. In the finite
Finite element analysis element (FE) analyses, ground movements are simplified as spatially distributed horizontal transverse defor­
Relative stiffness
mation. The pipe-soil interaction is modelled accounting for large ground deformation, inelastic material
Pipe-soil interaction
Failure prediction
behavior, and special contact conditions at the soil–pipe interface. Pipeline failure patterns are summarized
Horizontal transverse ground deformation based on a series of FE models with varying diameter thickness ratios of the pipe and soil conditions. According
to the FE analysis results and the published literature, the existing failure criteria are evaluated and extended to
cover new failure modes. The dynamic evolution in the predominant failure modes due to the pipe elongation
effects is investigated. A semianalytical solution for the sliding distance to the onset of pipe wall wrinkling is
given. In addition, a pipeline failure criterion is proposed based on the relative stiffness. This dimensionless
parameter is correlated with failure features, including failure modes, sliding distances to failure initiation, and
failure positions. The reliability of the proposed FE model and failure prediction methodology is validated using
published results from centrifuge tests and FE analyses. Simple charts for failure prediction in terms of relative
stiffness parameters are provided. The proposed approach allows efficient and reliable safety evaluation on X65
steel pipelines subjected to horizontal transverse ground deformation without conducting numerical modelling.

1. Introduction pipelines. Even when a simple translational landslide is taken into ac­
count, the pipe-soil system response depends significantly on the inter­
As a common source of permanent ground deformations (PGDs), section angle between the landslide boundary and pipeline [8]. This is
landslides can cause terrible devastations to buried pipelines [1,2]. The similar to the problem that a buried pipeline crosses the trace of
structural failure of buried pipelines due to excessive deformation may strike-slip faults [9–11]. This response ranges from the condition of pure
lead to severe economic and environmental consequences, especially for bending (the pipeline axis is normal to the soil sliding direction) to pure
gas or oil pipelines. Landslide-induced pipeline accidents have been axial compression or tension (the pipeline axis is parallel to the soil
found worldwide [3–5]. For instance, it was reported that China’s sliding direction, with tension at the top of the slide and compression at
West-East Gas Transmission Project is vulnerable to over 174 massive its toe) [12,13].
loess landslides [5]. Lee et al. [2] summarized six accidents in the Shallow translational landslides with small slope angles can be
Camisea Pipeline System, four of which were caused by landslide simplified as spatially distributed horizontal ground deformation. The
movements. Recently, an explosion accident of a gas pipeline in Guiz­ mechanical behaviours of buried pipelines to this deformation pattern
hou, China, was investigated by Cheng et al. [6]. The failure-induced have been experimentally investigated, dealing in particular with the
leakage was thought to be relevant to the ground shear deformation overbending behaviour [14–16]. The pipe materials are usually
near the landslide boundary. high-density polyethylene (HDPE) and polyvinyl chloride (PVC)
The structural response of buried pipelines is governed by the [17–19]. In contrast, theoretical and numerical analyses (e.g., Refs.
external loads, internal pressures, and stiffness of pipe-soil systems [7]. [20–24]) are two major approaches to the performance evaluation of
Large differential ground deformation causes additional loads on buried metal pipelines since failures can hardly be induced in laboratory tests.

* Corresponding author.Rm 362, Zhugongshan Bldg, Nanjing University, 163 Xianlin Ave, Nanjing, 210023, China.
E-mail addresses: [email protected] (H.-J. Li), [email protected] (H.-H. Zhu), [email protected] (C.-X. Zhang), [email protected] (W. Zhang).

https://doi.org/10.1016/j.soildyn.2022.107447
Received 1 December 2021; Received in revised form 11 July 2022; Accepted 14 July 2022
Available online 31 July 2022
0267-7261/© 2022 Elsevier Ltd. All rights reserved.
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

In FE analyses [25–27], the onset of pipeline failure is identified Vazouras et al. [25,26] and Demirci et al. [15] used the same method in
commonly according to the wall strain and cross-section distortion the numerical modelling of pipe-soil interactions.
limits suggested by codes (e.g., Refs. [28,29]). The influence of pipe The pipe diameter is 1000 mm, which is typical for low-pressure gas
stiffness and soil stiffness on pipe failure modes has been investigated by pipelines in North China and Central Asia. The embedment ratio, H/D, is
varying the material and diameter to thickness ratio of the pipe and soil set to 2.5, which is close to the statistical average of pipelines crossing
strength [25–27]. Nevertheless, the quantitative relationship between landslides along the second line of the West-East Gas Transmission
the relative stiffness and the failure mode remains to be investigated Project [5]. In addition, such a burial ratio is close to some experimental
regarding the horizontal transverse ground deformation. In addition, the and numerical studies [17,25,26]. This allows the comparison with their
transition of the predominant failure modes is rarely addressed, ac­ results. The pipeline length is 60 m (i.e., 60 times its diameter), and the
counting for the pipe elongation effect. soil prism dimensions are set to 5D and 10D in the vertical and lateral
This study focuses on the pipe-soil interaction induced by shallow directions, respectively. It is considered a rational geometry and
translational landslides with small slope angles. This kind of landslide- adequate to avoid finite-size effects according to the parametric analyses
correlated pipeline failure has caused concern in many long-distance conducted by Vazouras et al. [25,26]. Diameter-to-thickness ratios, D/t,
pipelines (e.g., Refs. [2,5]). In this paper, an orthogonal geometry of ranging from 60 to 100 are obtained by changing the wall thickness of
the pipe-soil system is considered, i.e., the pipeline crosses the landslide the pipe models.
at a right angle with respect to its boundary. A simplified semianalytical A narrow strip near the landslide boundary is emphasized in the
model is proposed to derive solutions for the peak compressive strain of modelling of the soil mass. This method was also used in similar nu­
the pipe wall and the critical landslide displacement to failure initiation. merical studies [33]. Ignoring this transition zone between the sliding
The transition of the predominant failure mode from wall wrinkling to and the stable block will lead to errors in the boundary condition
shearing-induced buckling, tensile rupture and ovalization has been assignment of two soil parts. The width w of the landslide boundary is
specified. A dimensionless parameter reflecting the relative stiffness of assumed to be 0.02 m, which was considered suitable in reported FE
the pipe-soil system is introduced. Relationships between the stiffness models with similar geometry sizes [25,26]. The results obtained from
parameter and failure characteristics, including the sliding distance, parametric studies show that the influence of width variation ranging
failure mode, and failure position, are investigated. from 0.01 to 0.05 m has little effect on the pipeline response.
Finite element meshes of the soil mass have a uniform spatial dis­
2. Numerical modelling tribution (approximately 0.3 m2 per grid). Fine meshes are used near the
pipe-soil interface. The surface of the hole in the soil prism is the slave
The FE analyses are conducted by using the commercial software surface, whereas the outer surface of the pipes is the master. An inten­
ABAQUS [30] in the present study, dealing in particular with the sive meshing approach is employed for the surrounding soil. The 20-m
nonlinearity of (1) pipe material, (2) geometry of pipe-soil system, (3) segment of the pipeline near the middle span is the region of interest
pipe-soil interaction (contact at the pipe-soil interface), and (4) inelastic in the study if not otherwise stated. To improve the computational ac­
behaviour of the surrounding soil. There are two frequently used ap­ curacy in this segment, finer meshes with each element equal to 0.15 m
proaches for simulating the soil reaction, i.e., the soil spring in multiple in the longitudinal direction are chosen. In contrast, the rest segments
directions [10,11] and surface-to-surface contact [25,26,31,32]. The far from the landslide boundary are given little attention and impose
latter is utilized in the present study since it is rigorous and suitable for relatively coarser meshes, i.e., the size of each element in the longitu­
interaction modelling. dinal direction increases evenly from 0.2 to 0.5 m, as shown in Fig. 1(b).
A prismatic soil model is established with a buried pipeline model The Mohr‒Coulomb constitutive model is used to describe the
running throughout its length. The detailed scheme for the soil and elastic-perfectly plastic behaviour of soil. In many simulations of the
pipeline is shown in Fig. 1(a) and Fig. 1(b), respectively. To improve the pipe-soil interaction induced by the large differential ground motion
computational efficiency, eight-node reduced-integration elements and [13,15,25,26], the elastic-perfectly plastic Mohr‒Coulomb constitutive
four-node reduced-integration shell elements are utilized to generate the model was selected to model the soil mechanical behaviours. These
soil and the pipe, respectively. The reduced integration elements can works show the capability of the Mohr‒Coulomb model to characterize
give a fairly similar force/displacement/stiffness response under a the soil resistance with increasing soil deformation. The values of the
refined meshing strategy. Its drawbacks mainly lie in plasticity and re­ dilation angle are determined according to the semiempirical methods
sidual stress output. The preliminary test results show that the results reported by Bolton [34]. The von Mises plasticity model is used to
based on C3D8R and C3D8 are very close for the present FE models. describe the X65 steel pipe. The Young’s modulus and yield strength of

Fig. 1. (a) Finite element model of the soil and pipe.

The width of the concentrated shear band at the
landslide boundary is 0.02 m. A 60 m-long pipe runs
throughout the length of two soil blocks. (b) Section
view of the X65 steel pipe and soil prism models. Note
that (1) the right half (corresponding to the moving
block) of the pipe-soil model is omitted in the figure
for brevity; (2) the left soil block and the left pipe end
are fixed (i.e., Ux = Uy = Uz = Uxy = Uyz = Uxz = 0),
and the right soil block and the right pipe end impose
a uniform displacement in the y-direction (i.e., Uy =
1 m, Ux = Uz = Uxy = Uyz = Uxz = 0) when performing
the ground moving step; and (3) the present coordi­
nate system is applicable to all results throughout the
paper unless emphasized.

2
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

X65 steel are 2.06 × 105 MPa and 450 MPa, respectively. The hardening Table 2
characteristics are assigned based on the uniaxial tension tests con­ Summary of experimental and numerical works.
ducted by Vazouras et al. [25,26]. The other information on the FE Research Type of study Material of pipe Max. slide disp.
model is listed in Table 1. Clay mechanical properties are chosen to a
Ha et al. [17]. Centrifuge Exp HDPE 2.5D
simulate “undrained” conditions. A small amount of artificial cohesion Sim et al. [16] Exp Acrylic & PE 1.25D
(~5–8 kPa) is considered for the sand to avoid numerical problems, such Vazouras et al. [25,26] FEM X65 steel 1D
as behaviours of purely frictional materials under low confining Demirci et al. [15] FEM & Exp HDPE 6D
pressure. a
The landslide boundary-crossing angle of the model pipe is 85◦ in the
The separation at the pipe-soil interface is permitted in the surface- centrifuge modelling conducted by Ha et al. [17].
to-surface contact algorithm. In addition, the hard contact and penalty
function rules are utilized to define the normal and tangential in­
teractions, respectively. The interface friction coefficient is summarized Table 3
in Table 1. The general static analysis algorithm is used in numerical Parameters used in the numerical model.
modelling. Two steps are carried out in order. The first is gravity Density Young’s Poisson’s Friction Dilation Yield
loading, in which the normal displacement of the external nodes of the (kg/m3) modulus ratio angle (◦ ) angle (◦ ) strength
soil prism is constrained, except for the top surface representing the (MPa) (MPa)
ground surface. Subsequently, the PGD effect is imposed through an Soil 1800 14 0.3 33 3 N/A
incremental application of landslide movements. In this step, the stable Pipe 1200 920 0.4 N/A N/A 15
soil block and the left end of the pipe are fixed, while the moving block
and the right end of the pipe are given a uniform displacement in the y-
distance from the landslide boundary were nondimensionalized against
direction. The maximum incremental ground displacement at each time
the pipe diameter D. The bending strain was plotted as a function of the
step is within 0.05D. The total sliding distance (i.e., sliding distance) is
dimensionless distance from the landslide boundary, as shown in Fig. 2.
set to ≥1D.
The FEA results show reasonable agreement with strain measurements
obtained in the experimental study. Local buckling would occur at large
3. Validation of the FE model
ground deformations, which are difficult to detect by limited-point

The validation of the proposed FE model is important. Table 2

summarizes the experimental and numerical works involving the pipe­
line response to the horizontal transverse ground deformation. In these
studies, Ha et al. [17] presented results from a series of centrifuge tests at
a gravity level of 12.2 g, in which the model pipe corresponds to a
prototype scale with D = 407.5 mm and t = 24.0 mm. Compared with
other experimental studies, he provided the most detailed information
on the experimental design and strain measurements of the pipe model.
It should be noted that the landslide boundary-crossing angle of the pipe
model is 85◦ in his tests, implying that there is a difference of 5◦ with the
present FE model.
The results obtained in the centrifuge tests of Ha et al. [17] were used
herein to verify the present FE model. The geometry of the pipe-soil
system is enlarged when performing the FEA. The size of the FE model
is consistent with the prototype scale of the pipe-soil system. The re­
sidual internal friction angle estimated according to the peak value
specified by Ha et al. [17] was used to describe the soil. Likewise, a small
artificial cohesion of 4 kPa was included to avoid numerical difficulties
and allow the observation of gaps appearing around the pipeline. The
friction coefficient at the pipe-soil interface is taken as 0.4. Other main
parameters in the simulation are listed in Table 3. Fig. 2. Comparison of bending strains of the pipe at various offsets obtained in
To facilitate the comparison, both the sliding distance and the FE analyses and centrifuge tests (data from Ha et al. [17]).

Table 1
List of tests.
Test* Pipe Soil Interface

D/t Density (kg/m3) Cohesion c (Pa) Friction angle φ (◦ ) Dilation angle (◦ ) Young’s modulus E (MPa) Poisson’s ratio υ Friction coefficient μ

M 100 100 1900 6000 31 1 30 0.3 0.4

D 100 100 2000 8000 33 3 40 0.3 0.4
C 100 100 2000 50000 0 0 25 0.46 0.3
M 80 80 1900 6000 31 1 30 0.3 0.4
D 80 80 2000 8000 33 3 40 0.3 0.4
C 80 80 2000 50000 0 0 25 0.46 0.3
M 60 60 1900 6000 31 1 30 0.3 0.4
D 60 60 2000 8000 33 3 40 0.3 0.4
C 60 60 2000 50000 0 0 25 0.46 0.3
M 40 40 1900 6000 31 1 30 0.3 0.4
D 40 40 2000 8000 33 3 40 0.3 0.4
C 40 40 2000 50000 0 0 25 0.46 0.3

*M = Medium sand; D = Dense sand; C=Clay; and 40, 60, 80, 100 are D/t values

3
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

strain gauges. Hence, the strain distribution at large sliding distances is defected locations are common damage modes. Consequently, the limit
not incorporated. The comparison indicates that the proposed FE model tension strain in the longitudinal direction is used to identify the onset of
is capable of describing the mechanical response of pipelines subjected tensile fracture in many codes. ALA [28] and EN 1998-4 [35] suggest
to spatially distributed horizontal transverse ground deformation. that 2%–4% and 3% is the critical tension strain of buried pipelines,
respectively. The value has been taken as 3% in this study.
4. Results and analyses
4.1.2. Compression-induced buckling (CIB)
4.1. Failure modes and verification criteria The wall wrinkling caused by excessive compressive strains is one of
the most common local buckling forms. With the expansion of wrinkles,
The potential pipeline failure modes due to landslide movements are the load capacity of pipes exhibits a significant decrease. A strain-based
shown in Fig. 3. Several commonly used criteria in the industry are criterion for the onset of compression-induced buckling was proposed by
discussed below to evaluate their applicability to this study [26–28,35]. Gresnigt [36], in which the critical compressive strain is calculated by
In addition, a criterion is given to identify the newly identified failure (σ )2
t
mode correlated with excessive shear strains. (1)
h
εc− c = − 0.0025 + 3000
2D E

4.1.1. Tensile rupture where εc− c is the critical compressive strain; t is the wall thickness; D is
For the stretched and bending segments, tensile fractures at weld or the diameter; E is Young’s modulus; and σh is the circumferential strain
of the pipeline, which can be determined by
⎧
⎪
⎪ pD pD
⎨ 2t , 2σ t ≤ 0.4
⎪
(2)
s
σh =
⎪
⎪ pD
⎪
⎩ 0.4σ s > 0.4
2σ s t

where p is the internal pressure of the pipeline and σ s is the yield stress of
the pipeline material.
Equations (1) and (2) have also been adopted by ALA [28] and CSA
Z662-07 [29] due to their conservatism in practical applications.
However, many explorers argued that these criteria should be further
verified. Vazouras et al. [25] suggested a new performance criterion for
wrinkling that features the “wavy” distribution of axial strain of the
pipeline. Taking the results of Medium sand 100 as an example, a
comparison between the two criteria is given in Fig. 4. According to
Equation (1) proposed by Gresnigt [36], a value of 2.5 × 10− 3 (corre­
sponding to the sliding distance u = 0.328 m) is the limit compressive
strain of the X65 steel pipeline in this study. However, Vazouras et al.
[26] suggested that a transition from a smooth configuration to a uni­
form wavy pattern of axial strain (corresponding to u = 0.337 m) is the
onset of wall wrinkling. The former is used in the following analysis due
to its conservatism.

4.1.3. Shear-induced buckling (SIB)

A new buckling mode caused by excessive shearing deformation near
the landslide boundary is addressed to distinguish it and wall wrinkling
induced by high compressive strains. This failure type is found at the
invert and crown of the pipe cross-section at the landslide boundary. It is
a specific failure of the large-diameter pipe with a smaller thickness (i.e.,
large D/t value). From the published literature, it can be concluded that
the overbending segments (wall wrinkling) are at a distance from the
concentrated shear band of the ground, such as ~3D of the HDPE pipe
and ~5D of the X65 steel pipe tests [17,25,26]. However, the
shear-induced buckling would be accurately located at the landslide
boundary. That is a major difference between SIB and CIB. In addition,
under the present pipe-soil geometry configuration, wrinkles tend to
occur at the springline of the pipe, and the compression side of the
overbending segments to be exact. However, shear failure is always
found at the crown and invert. This is another difference between SIB
and CIB.
It is important to develop a criterion to identify the onset of this
newly defined failure mode. In this study, morphological observation is
used to determine the failure initiation. Similar methods are widely used
in FEM to detect the onset of failure (e.g., Refs. [7,10,11,25,26]). As
Fig. 3. Failure modes and maximum principal stress distributions. (a)
introduced previously, Vazouras et al. [25,26] recommended that the
Shearing-induced buckling (SIB) for Dense sand 80. (b) Compression-induced
onset of local buckling can be determined when the pipe wall presents a
wrinkling (CIB) or kinking for medium sand 80. (c) Flattening for dense sand
40. Note that the coordinates are intended for the schematic presentation of shortwave pattern. Likewise, in the present study, the onset of shear
pipe deformation and are independent of Fig. 1. failure is identified according to the spatial distribution of the

4
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

Fig. 4. Onset of compression-induced wall wrinkling. (a) Distribution of longitudinal strain of the pipeline wall. (b) Comparison of critical strains to wrinkling
between the three methodologies (Medium sand 100). The landslide displacement u is in meters for all graphics in this paper unless noted otherwise.

Fig. 5. Initiation of shearing-induced wall buckling (Medium sand 100). (a) Shear strain distribution of the outer generator of the pipe; (b) the variation in the
position of the peak shear strain. x and θ are consistent with the coordinates in Fig. 1.

5
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

concentrated shear strain. Specifically, the concentrated shear band of Only three cases with a D/t value of 40 exhibit breakage processes
the pipe is distributed in a straight line before failure. It transforms from beginning with other failure modes. As is well known, the large longi­
the original configuration to an “S” shape when failure occurs (Fig. 3 tudinal wall strain is responsible for wall wrinkling. It will cost much to
(a)). This process corresponds to the change in the maximum shear obtain the distributed strain profile along an instrumented pipeline,
strain location, as illustrated in Fig. 5. especially for long-distance gas or oil pipelines. In contrast, it would be
more affordable to obtain surface ground movements. The relationship
4.1.4. Flattening between the longitudinal strain and the sliding distance of the slope can
Flattening is another potential damage for pipelines subjected to be derived using analytical methods. This is similar to the buried
significant bending moments. For the convenience of flattening evalu­ structure elongation problems, which are usually based on the elastic
ation, a parameter e calculated by the diameter variation of the cross- beam theory or the kinematic method, as reported by Karamitros et al.
section is introduced as follows: [13] and Vazouras et al. [25,26]. The kinematic method is more prac­
tical for applications due to its virtue of relative simplicity and ease of
ΔD
e= (3) computation. For this purpose, the pipelines are considered to deform in
D
an “S” shape near the landslide boundary. Its cross-section deformation
where ΔD is the maximum change in pipeline diameter. According to is neglected in the analysis. Vazouras et al. [25,26] used a variation of
Gresnigt [36], the suggested value of e to flattening is 0.15. This crite­ the cosine function to describe the shape of deforming segments (i.e., the
rion is used in this study. Meanwhile, another threshold, 5%, is intro­ “S” shaped portion). However, the circular arc configuration is consid­
duced in this part considering the current pipeline maintenance. For ered more suitable to describe pipe deformation under horizontal
instance, pigs have been increasingly used to monitor pipeline opera­ transverse deformation according to a large amount of literature [13].
tions over the last twenty years, and they would be affected when the Therefore, a kinematic method with the assumption that the “S” shaped
flattening parameter (Eq. (3)) reaches 5%. Fig. 6 illustrates the general segments deform as circular arcs is proposed here to calculate the lon­
evolution of ovalization for the “S"-shaped portion. The ellipticity peaks gitudinal strain of the pipe wall in terms of the sliding distance of the
at compression-induced buckling positions. The diameter variation in slope, length of the deforming segments, and pipe diameter.
the horizontal direction has good consistency with that in the vertical Based on the assumption that a continuous and homogeneous pipe­
direction under small deformations (u < 0.4 m). Nevertheless, this line is buried in the homogeneous surrounding soil, the kinematic model
consistency no longer holds under large deformations. of the pipe-soil system is presented in Fig. 7. Vazouras et al. [25] utilized
the trigonometric function to describe the shape of the deformed pipe­
line in terms of landslide displacement u (Pattern 1 in Fig. 7(a)). In the
4.2. Initiation of compression-induced buckling present study, however, the Arc model (Pattern 2 in Fig. 7(a)) is more
suitable to describe the pipe shape for a considerable ground movement
The failure modes and sliding distance to CIB are listed in Table 4. It associated with possible damages, as shown in Figs. 8 and 9. The validity
may be observed that wall wrinkling is almost an inescapable initial of the Arc model is premised on an accurate length of the deforming
failure type for buried pipelines crossing landslides at the right angle.

Fig. 6. (a) Flattening process. (b) Developments and distribution of e (Medium sand 40). H and V represent e values computed by horizontal and vertical deflection,
respectively. x is consistent with the coordinate in Fig. 1.

6
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

Table 4
Failure modes and critical landslide displacements corresponding to wall wrinkling initiation.
*D 100 M 100 C 100 D 80 M 80 C 80 D 60 M 60 C 60

*CIB–SIB CIB–SIB CIB-TR CIB–SIB CIB–F CIB-TR CIB–SIB CIB–F CIB

0.2358 m 0.3336 m 0.3526 m 0.3184 m 0.4708 m 0.5228 m 0.5626 m 0.9 m 1.1272 m

*D = Dense sand; M = Medium sand; C = Clay; CIB=Compression-induced buckling; SIB = Shearing-induced buckling; TR = Tensile rupture; and F = Flattening

Fig. 9. Displacement (in the y-direction) development of pipe for Dense sand
80. The data were fitted based on the Arc model (Pattern 2 in Fig. 7(a)).

Fig. 7. A kinematic method for determining the deformation and displace­ ⎧

ments of “S" shaped pipelines via the landslide displacement and sensitive area ⎪
⎪ Ls
⎪
⎪ r × sin α =
range. (a) Two deformation patterns of buried pipelines during sliding and their ⎪
⎨ 2
geometric features. Pattern 1: Cosine function fitting case. Pattern 2: Arc fitting 2α × r = Lpa (4)
⎪
case. (b) Schematic representation of the pipeline deformation and soil motion. ⎪
⎪
⎪ u
⎪
⎩ r × cos α + = r
2

where r and α are the radius and half of the radian of the arc, respec­
tively; Lpa is the length of deformed pipeline segments; Ls is the initial
length of deformed pipeline segments; and u and d are the sliding dis­
tance of slope mass and the pipeline displacement in the y-direction,
respectively. By solving Equation (4), one may have
⎧
⎪ u
⎪
⎪ α = 2 arctan
⎪
⎪ Ls
⎪
⎪
⎪
⎪
⎨ u2 + Ls 2
r=
4u (5)
⎪
⎪
⎪
⎪ ( 2 ) u
⎪
⎪ 2
u + Ls arctan
⎪
⎪ Ls
⎪
⎩ Lpa =
u
The displacement of the pipeline within stretched segments can be
readily obtained as
⎧ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅
⎪ Ls
⎪
⎨r − r 2 − x2 0≤x≤
2
d(x) = (6)
⎪
⎪ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ Ls
⎩ (r − x + Ls )(r + x − Ls ) + u − r ≤ x ≤ Ls
2
Fig. 8. Displacement distribution (in the y-direction) of the pipe for a given
landslide displacement u = 0.6 m. The data are fitted based on the Arc model According to the expression of radius r, the bending curvature κ can
(Pattern 2 in Fig. 8(a)). be readily computed as follows:
1 4u
segment. Determining the length Ls may be challenging for buried κ= = 2 (7)
r u + Ls 2
pipelines in practice. This will be further discussed in the following
section. Given a circular cross-section hypothesis for pipelines under small
According to the geometric relationships shown in Fig. 7(a), the deformation, the longitudinal strain εb due to bending is
following equations hold:

7
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

κD 2uD
εb = = 2 (8)
2 u + Ls 2
The length increase of deformed segments is
( 2 )
u + Ls 2 arctan Lus
Δ = Lpa − Ls = − Ls (9)
u
The corresponding axial strain, i.e., the tensile strain due to pipeline
stretching (referred to as εs ), can therefore be obtained:
( 2 )
Δ u + Ls 2 arctan Lus
εs = = − 1 (10)
Ls uLs
By subtracting εs from the bending strain εb , one obtains the
compressive strain of the pipe wall
( 2 )
2uD u + Ls 2 arctan LuS
ε = εb − εs = 2 − +1 (11)
u + Ls 2 uLs
If ignoring the variation of Ls and D for a given pipeline during
ground movements, the value of ε depends only on the sliding distance u. Fig. 10. Critical landslide displacement u0 corresponding to the maximum
The maximum value of ε and corresponding sliding distance can be compressive strain versus the length of deformed segments Ls and pipeline
computed by differentiating Equation (11) with respect to u: diameter D.
[( ) ]
2
dε ( 2 )( 2 ) 1 1 Ls u
2
= 2D Ls − u u + Ls 2 − 2
− + − 1 arctan (12)
du u Ls u Ls

To our knowledge, the exact analytical solution cannot be obtained

from the equation du dε
= 0. Through an iterative solution method of
nonlinear equations, the approximate solution of u0 (Ls , D), at which the
compressive strain ε peaks, can be expressed in the following form:
∑∑
u0 = pij Di Ls j (13)
i j

where u0 is termed the critical sliding distance. Considering typical

values of Ls andD, coefficients in Equation (13) are computed with a
certain confidence level, as listed in Table 5. The relationship between
u0 and Ls , D is shown in Fig. 10. Equation (11) can be used to identify
whether wall wrinkling will appear. Equation (13) permits engineers to
calculate the sliding distance to the onset of wrinkling.
At small ground movements, ε increases with the application of
ground movement, and it peaks when u reaches u0 , as determined by
Equation (13). After that, it decreases with increasing sliding distance u,
implying that the elongation effect of the pipe has a much greater impact
Fig. 11. Critical sliding distance u0 versus length of the deformed pipeline. Ls
on the longitudinal wall strain than the bending behaviour. Therefore, it
can be inferred that the pipe would exhibit a damaging process starting
with the CIB if ε(u0 , Ls , D) is greater than the ultimate strain; otherwise, compressive strains have yet to reach the threshold.
such a failure mode would not occur, regardless of the magnitude of
ground movement. For conservative considerations, the threshold for 4.3. Transition of the dominant failure mode
wrinkling initiation can be determined according to Equations (1) and
(2). Failure modes predicted according to the initial deforming trend are
Regarding the two parameters of Equation (13), common diameter always considered the governing (e.g., Refs. [15,26,27]). However, it
values range from 0.4 to 2 m for most onshore pipeline applications, should be noted that the elongation of pipelines with increasing ground
whereas Ls depends on the RS of pipe-soil systems and the frictional displacements would significantly affect the deformation behaviour of
coefficient at their interface. More importantly, this value might change pipelines. For instance, a failure process toward wall wrinkling due to
with increasing ground movement, making it difficult to determine. high compressive strain may be alleviated or stopped due to the elon­
Fig. 11 plots the relationship between u0 and Ls for the typical diameter gation of “S" shaped pipe segments. Based on the theoretical and nu­
values in pipeline practices. In the three FE models, wall wrinkling is not merical simulation results under perpendicular crossing configurations,
observed throughout the ground movement application. The develop­ three distinct forms of the transition from wall wrinkling to other modes
ment of compressive strains recorded in these models is shown in are identified in the current study.
Fig. 12. The compressive strains increase with the sliding distance before
u0 and then decrease. No wrinkling is observed since the maximum a) CIB to SIB

Table 5
Coefficient of Equation (13). (with 95% confidence bounds).
p00 p01 p02 p03 p10 p11 p12 p13 p21

− 0.0522 0.0368 − 0.0055 0.0002 1.1880 0.0767 − 0.0033 − 0.3183 0.0143

8
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

according to the shear strain at the pipe middle span (Fig. 14(b)). Fig. 14
(c) indicates that the excessive distortion of the pipe cross-section is the
final failure mode.
The tensile fracture needs special attention, as it would be magnified
with increasing ground movements. However, this transformation was
not clearly shown in any FE models listed in Table 1. Taking the case of
Clay 80, the critical displacement for the initiation of tensile rupture is
far greater than CIB. Therefore, it can be inferred that tensile fracture is
only a trend and does not occur because wrinkling causes damage at the
initial stage.

5. RS-dependent failure prediction method

5.1. Role of RS in pipeline failure

Previous studies have confirmed that the stiffness of the pipe and
surrounding soil significantly influences the structural response of
pipelines under large differential ground motion [37–40]. An available
dimensionless parameter is needed to investigate the relationship be­
Fig. 12. Maximum longitudinal compressive strain versus the sliding distance tween the pipe-soil relative stiffness and pipeline failure. Peck et al. [38]
of slope.u0 suggested that a flexibility ratio can be used to estimate the stiffness of
buried structures. To match the concept of RS, it can be rewritten as
Fig. 13 shows the strain development correlated with three failure 6Ep Ip (1 + vs )
types, i.e., CIB, SIB, and tensile rupture. According to the evaluation Sp,s = ( ) (14)
Es 1 − v p 2 R 3
criteria defined previously, the pipeline exhibited wall wrinkling when
the sliding distance reached 0.3 m (Fig. 13(a)). Subsequently, where Ep and Es are Young’s moduli of the pipe and soil, respectively, νp
compressive strains decrease after the occurrence of SIB (Fig. 13(b)). As and νs are the Poisson’s ratios of the pipe and soil, respectively, R is the
shown in Fig. 13(c), the SIB is the final failure mode of the pipeline. The radius of the pipe, and Ip is the moment of inertia of the pipe.
same results hold in tests Medium sand 100 and Dense sand 80. Eq. (14) is widely adopted for estimating buried structures that are
Another failure transition process from CIB to ovalization is found in stiff relative to the surrounding soil (e.g., tunnel linings). However, its
Medium sand 80 and Medium sand 60. As shown in Fig. 14(a), the applicability for steel pipes buried in sand remains to be completely
compressive strain experiences a reduction as the sliding distance in­ verified. Alternatively, a dimensionless parameter RS is also available by
creases from 0.5 m to 0.7 m. It can be seen that the SIB does not occur extending the stress distribution model hypothesized by Spangler [37].

Fig. 13. Evolution of the failure mode from CIB to SIB. (a) The distribution of normal axial strain of the pipeline. (b) Development of shear strain at x = 11 m. (c)
Distribution of tensile strain and developments of maximum shear strain, tensile strain, and compressive strain. (Dense sand 100). x is consistent with the coordinate
in Fig. 1. b) CIB to flattening.

9
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

Fig. 14. The transition of the failure mode from compression-induced buckling to flattening. (a) The distribution of normal axial strain of the pipeline from 11 m to
15 m. (b) The distribution of shear strain of the pipe section at x = 11 m. (c) The development of flattening parameter e at x = 10 m. (Medium sand 80). x is consistent
with the coordinate in Fig. 1. c) CIB to tensile rupture.

It had a broad impact on practical applications because it is the first overbending zones. The failure types in these positions are also sum­
analytical solution to the flexible pipe incorporating pipe-soil interac­ marized in Fig. 16. The failure type is strongly dependent on the RS. SIB
tion. It was supported by an extensive database on metal pipes. Although at the pipe crown and invert could be expected at a low RS and can be
several weaknesses have been issued, the Spangler model can still pro­ avoided by increasing the RS to ~105. The ovalization and wall wrin­
vide an approximation of pipe response. The expression of RS derived kling at the springline are typical failure modes for medium RS (~105-5
from the Iowa formula (i.e., resultant calculation from the Spangler × 105). Wrinkling or kinking is rarely found at high RS. Instead, tensile
model [37]) is introduced into the current study. Herein, the parameter cracks and hinges are the most destructive damages when RS is more
RS is defined as a ratio of the soil displacement at the pipe springline to than 5 × 105. Of note, there might be a slight error in the thresholds
the pipeline deflection and can be written as: stated above due to the limited database.
( )3 ′
Ep Rt + 12Kv E
RS = (15) 5.2. Discussion of the prediction methodology
12Es Kh

where Ep and Es are Young’s moduli of pipe and soil, respectively, Kv and In the proposed kinematic method, the axial strain is assumed to be
Kh are the bedding and lateral pressure coefficients of the Iowa formula, uniformly distributed. This assumption was also used by Vazouras et al.
and t and R are the thickness and the radius of pipe, respectively. E is the [25,26] and Karamanos et al. [7]. As a result, the sliding distance to the
′

soil reaction modulus. onset of wall wrinkling computed by Equation (13) is overestimated.
As suggested by Masada [41], the representative value of the Another problem is how to estimate the active length (termed “S” sha­
bedding angle correlated with Kv is taken as 50◦ . Hartley and Duncan ped segments in the present study) of the pipeline. It depends on the
[42] suggested that E′ can be estimated according to the function of ground displacement, the size of the pipe (diameter and thickness), and
Young’s modulus, Poisson’s ratio and Ms. the stiffness of the pipe and soil, in which the relative stiffness of the
The pipeline failures (possible transitions are included) and RS pipe-soil system plays an important role. According to the FEA per­
values are listed in Table 6. It can be inferred that wall wrinkling will not formed by Yao et al. [9], the change in Ls in the current study is not
occur under the perpendicular crossing geometry, provided that RS is obvious under a small sliding distance. This can also be proven by the
greater than 5 × 105. A highly significant linear relationship is observed FEA and experimental results shown in Fig. 2. Hence, it is treated as a
between the dimensionless sliding distance and RS, as shown in Fig. 15. constant to facilitate the calculation. However, in other cases, the active
This implies that it is feasible to predict potential failure modes and the length might vary obviously with increasing ground deformation.
sliding distance based on the RS value of the pipe-soil system. The Considering the assumptions and approximant treatments in the pro­
approach is also validated using the data reported by Vazouras et al. cedure, the predicted sliding distance might be slightly different from
[26], as shown in Fig. 15. It should be noted that the difference between the actual sliding distance. Nevertheless, this method is effective for
Vazouras et al. [26] and this study can be attributed to the estimating the safety of buried pipelines subjected to horizontal trans­
non-conservative verification criteria adopted by Vazouras et al. [26]. verse ground deformation.
A linear relationship between the position of the wrinkle and RS is For the orthogonal crossing geometry, the axial pipe-soil interaction
presented in Fig. 15(d). Unexpectedly, the soil conditions greatly affect has a limited influence on the failure modes. However, it does not hold
the location, which implies that the axial interaction should be for other crossing angles. It is reported that additional failure modes,
emphasized in the failure location. Typical failure modes of pipelines in such as beam-type buckling, could be generated for pipelines crossing
terms of RS are shown in Fig. 16. Attention should be given to three the landslide boundary at a small angle. Tensile fracture more easily
positions where failures are most likely to appear, i.e., the segment appears under large angles.
directly contacted with the landslide boundary and a pair of Determination of the soil reaction modulus value is a conundrum in
the computation of RS. Its approximate value can be obtained according

Table 6
Pipe failure modes and RS values (105).
D 100 M 100 C 100 D 80 M 80 C 80 D 60 M 60 C 60 D40 M40 C40
a a
CIB–SIB CIB–SIB CIB-TR CIB–SIB CIB–F CIB-TR CIB–SIB CIB–F CIB F N/A N/A
0.33 0.44 0.53 0.65 0.87 1.04 1.54 2.06 2.50 5.20 6.94 8.33
a
No failure was observed in tests M40 and C40 under the landslide displacement scope of 0–2 m.

10
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

Fig. 15. (a), (b), and (c) Critical landslide displacement to CIB, SIB, and tensile rupture, versus RS. (d) Distance from the buckling location to the landslide boundary
Lc/D against RS.

Fig. 16. Typical performance limits of buried X65 steel pipelines subjected to horizontal transverse ground deformation (not to scale).

to the relationship with other properties of soil, which can be measured impacts.
from laboratory tests [40,41]. In addition, the influence of backfill
conditions and trench dimensions on soil stiffness should be taken into 6. Concluding remarks
consideration. It is also important to reevaluate the relative stiffness
considering the change in the environmental factors, such as pipeline This paper investigates the failure modes of X65 steel pipelines
leakage, corrosion, and soil liquefaction. crossing shallow translational landslides at a right angle through FE
In this study, the expression of RS is developed from the interaction analyses and the semianalytical method. The structural response of
theory. A database of 16 FEA results is used to facilitate the failure an­ pipelines is examined from a series of FE models with varying diameter
alyses of X65 steel pipelines under horizontal transverse ground defor­ thickness ratios of the pipe and soil conditions. According to strain-
mation. Most of the failures fall within the plastic plateau in the stress‒ related criteria, critical states to four failure modes of the pipeline are
strain curve of X65 steel. However, the plasticity of the pipe and soil is identified, namely, tensile rupture, compression-induced buckling,
not yet taken into consideration in the expression of RS. The results and shearing-induced buckling, and flattening.
conclusions should be treated with reservation since they only apply to A kinematic method with the assumption that the “S” shaped seg­
X65 steel pipelines and are validated from a limited database. In addi­ ments deform as circular arcs is proposed to calculate the longitudinal
tion, it should be viewed with caution when predicting the true strain of the pipe. The semianalytical solution for the sliding distance of
behaviour of the surrounding soil due to the limitations of the MC model the slope to the onset of pipe wall wrinkling is given in terms of the
used in FEA. Notably, the crossing angle and the internal pressure can length of the active length and the pipe diameter. Accounting for the
affect the structural response of the pipeline. The conclusions drawn in elongation effect of deformed pipe segments, the transition of governing
this study are targeted at pipelines subjected to horizontal transverse failure modes from the CIB to SIB, tensile rupture, and flattening is
ground deformation. Further research is needed to address these identified.

11
H.-J. Li et al. Soil Dynamics and Earthquake Engineering 162 (2022) 107447

Furthermore, the RS of the pipe-soil system, defined as a ratio of soil [12] Haque Chaudhuri C, Choudhury D. Semianalytical solution for buried pipeline
subjected to horizontal transverse ground deformation. J Pipeline Syst Eng Pract
displacements to pipeline springline, is introduced to reveal its decisive
2021;12(4):04021038.
impact on pipeline failures. It is concluded that both the critical sliding [13] Karamitros DK, Bouckovalas GD, Kouretzis GP. Stress analysis of buried steel
distance and the wrinkling position (i.e., distance from wrinkles to the pipelines at strike-slip fault crossings. Soil Dynam Earthq Eng 2007;27(3):200–11.
landslide boundary) exhibit a linear relationship with RS. The semi­ [14] O’Rourke TD, Jung JK, Argyrou C. Underground pipeline response to earthquake-
induced ground deformation. Soil Dynam Earthq Eng 2016;91:272–83.
analytical solution and RS-based criterion allow engineers to predict the [15] Demirci HE, Karaman M, Bhattacharya S. Behaviour of buried continuous pipelines
failure mode, failure position, and magnitude of ground movement to crossing strike-slip faults: experimental and numerical study. J Nat Gas Sci Eng
failure initiation for buried steel pipelines subjected to horizontal 2021;92:103980.
[16] Sim WW, Towhata I, Yamada S. One-g shaking-table experiments on buried
transverse ground deformation. Further explorations are needed to pipelines crossing a strike-slip fault. Geotechnique 2012;62:1067–79. 0016-8505.
derive an improved RS parameter incorporating the internal pressure [17] Ha D, Abdoun TH, O’Rourke MJ. Buried high-density polyethylene pipelines
and the crossing angle. subjected to normal and strike-slip faulting—a centrifuge investigation. Can
Geotech J 2008;45(12):1733–42.
[18] Li HJ, Zhu HH, Wu HY, Zhu B, Shi B. Experimental investigation on pipe-soil
CRediT authorship contribution statement interaction due to ground subsidence via high-resolution fiber optic sensing. Tunn
Undergr Space Technol 2022;127:104586.
[19] Wang DY, Zhu HH, Wang BJ, Shi B. Performance evaluation of buried pipe under
Hao-Jie Li: Conceptualization, Methodology, Investigation, Writing loading using fiber Bragg grating and particle image velocimetry techniques.
– original draft, Writing – review & editing. Hong-Hu Zhu: Project Measurement 2021;186:110086.
administration, Supervision, Writing – review & editing. Chun-Xin [20] Ni P, Mangalathu S, Yi Y. Fragility analysis of continuous pipelines subjected to
transverse permanent ground deformation. Soils Found 2018;58(6):1400–13.
Zhang: Software. Wei Zhang: Formal analysis. [21] Tsatsis A, Gelagoti F, Gazetas G. Performance of a buried pipeline along the dip of a
slope experiencing accidental sliding. Geotechnique 2018;68(11):968–88.
[22] Chaudhuri CH, Choudhury D. Buried pipeline subjected to seismic landslide: a
Declaration of competing interest simplified analytical solution. Soil Dynam Earthq Eng 2020;134:106155.
[23] Zahid U, Godio A, Mauro S. An analytical procedure for modelling pipeline-
landslide interaction in gas pipelines. J Nat Gas Sci Eng 2020;81:103474.
The authors declare that they have no known competing financial
[24] Qin X, Ni P, Wang Y, Du YJ. Mechanical response estimation of jointed rigid pipes
interests or personal relationships that could have appeared to influence under normal fault rupture. Soil Dynam Earthq Eng 2021;146:106754.
the work reported in this paper. [25] Vazouras P, Karamanos SA, Dakoulas P. Mechanical behavior of buried steel pipes
crossing active strike-slip faults. Soil Dynam Earthq Eng 2012;41:164–80.
[26] Vazouras P, Dakoulas P, Karamanos SA. Pipe–soil interaction and pipeline
Acknowledgements performance under strike–slip fault movements. Soil Dynam Earthq Eng 2015;72:
48–65.
[27] Melissianos VE, Vamvatsikos D, Gantes CJ. Methodology for failure mode
This work was financially supported by the National Natural Science
prediction of onshore buried steel pipelines subjected to reverse fault rupture. Soil
Foundation of China (Grant Nos. 42077235, 41672277, and 42077232). Dynam Earthq Eng 2020;135(2020):106116.
[28] American Lifelines Alliance. Guidelines for the design of buried steel pipe. New
York, USA: American Society of Civil Engineers; 2005.
References [29] Standard Association Canadian. Oil and gas pipeline systems, CSA Z662-07.
Mississauga, Ontario: Canada; 2007.
[1] Feng W, Huang R, Liu J, Xu X, Luo M. Large-scale field trial to explore landslide [30] ABAQUS. Users’ manual. Providence RI, USA: Simulia; 2012. version 6.12.
and pipeline interaction. Soils Found 2015;55(6):1466–73. [31] Wang F, Han J, Parsons RL, Corey R. Performance of steel-reinforced high-density
[2] Lee E, Fookes P, Hart A. Landslide issues associated with oil and gas pipelines in polyethylene pipes in soil during installation: a numerical study. Acta Geotechnica
mountainous terrain. Q J Eng Geol Hydrogeol 2016;49(2):125–31. 2020;15(4):963–74.
[3] Cevik E, Topal T. GIS-based landslide susceptibility mapping for a problematic [32] Zhang CX, Zhu HH, Zhang W, Li HJ, Liu W. Modeling uplift failure of pipes buried
segment of the natural gas pipeline, Hendek (Turkey). Environ Geol 2003;44(8): in sand using material point method. Tunn Undergr Space Technol 2022;119:
949–62. 104203.
[4] Chan PD, Wong RC. Performance evaluation of a buried steel pipe in a moving [33] Anastasopoulos I, Callerio A, Bransby M, Davies M, ElNahas A, Faccioli E.
slope: a case study. Can Geotech J 2011;41(5):894–907. Numerical analyses of fault–foundation interaction. Bull Earthq Eng 2008;6(4):
[5] Xi S. Study on deformation and failure criterion and susceptible zones for buried 645–75.
pipelines across landslides [PhD Thesis]. Beijing), China: China University of [34] Bolton M. The strength and dilatancy of sands. Geotechnique 1986;36(1):65–78.
Geosciences; 2018 (in Chinese). [35] Comité Européen de Normalisation. Eurocode 8, Part 4: silos, tanks and pipelines.
[6] Cheng Z, Gong W, Tang H, Juang CH, Deng Q, Chen J, Ye X. UAV photogrammetry- In: CEN EN 1998-4, Brussels, Belgium; 2006.
based remote sensing and preliminary assessment of the behavior of a landslide in [36] Gresnigt A. Plastic design of buried steel pipelines in settlement areas. Heron 1986;
Guizhou, China. Eng Geol 2021;289:106172. 31(4):1–113.
[7] Karamanos A, Gresnigt M, Dijkstra J. Geohazards and pipelines: state-of-the-art [37] Spangler M. The structural design of flexible pipe culverts. Ames, Iowa: Iowa Eng.
design using experimental, numerical and analytical methodologies. Springer Exp. Station. Bull. No 1941;153.
Nature; 2020. [38] Peck RB, Hendron AJ, Mohraz B. State of the art of soft ground tunnelling. In:
[8] Luo X, Ma J, Zheng J, Shi J. Finite element analysis of buried polyethylene pipe Proceedings of the rapid excavation and tunnelling conference, Chicago, IL, vol. 1;
subjected to seismic landslide. J Pressure Vessel Technol 2014;136(3):31801. 1972.
[9] Yao C, He C, Takemura J, Feng K, Guo D, Huang X. Active length of a continuous [39] Klar A, Marshall AM. Linear elastic tunnel pipeline interaction: the existence and
pipe or tunnel subjected to reverse faulting. Soil Dynam Earthq Eng 2021;148: consequence of volume loss equality. Geotechnique 2015;65(9):788–92.
106825. [40] Klar A. Elastic continuum solution for tunneling effects on buried pipelines using
[10] Liu X, Zhang H, Li M, Xia M, Zheng W, Wu K. Effects of steel properties on the local Fourier expansion. J Geotech Geoenviron Eng 2018;144(9):04018062.
buckling response of high strength pipelines subjected to reverse faulting. J Nat [41] Masada T. Modified Iowa formula for vertical deflection of buried flexible pipe.
Gas Sci Eng 2016;33:378–87. J Transport Eng 2000;126(5):440–6.
[11] Liu X, Zhang H, Wu K, Xia M, Chen Y, Li M. Buckling failure mode analysis of [42] Hartley JD, Duncan JM. E′ and its variation with depth. J Transport Eng 1987;113
buried X80 steel gas pipeline under reverse fault displacement. Eng Fail Anal 2017; (5):538–53.
77:50–64.

12