Soil Dynamics and Earthquake Engineering 132 (2020) 106090

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

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

Response of buried pipeline subjected to reverse faulting

M. Fadaee a, *, F. Farzaneganpour a, I. Anastasopoulos b
a
Department of Civil Engineering, Science and Research branch, Islamic Azad University, Tehran, Iran
b
Geotechnical Engineering, ETH Zurich, Switzerland

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

Keywords: The paper studies the interaction of buried pipelines with a reverse fault rupture, employing 3D finite element
Buried pipeline (FE) modelling. A thoroughly validated strain softening constitutive model is employed to conduct a parametric
Fault rupture study, exploring the effect of sediment thickness, pipeline burial depth and thickness. It is shown that the in­
Finite element analysis
crease of sediment thickness H leads to a reduction of the normalized fault offset h=H at which buckling is
initiated. Two contradicting mechanisms are identified: (i) the increase of H leads to a reduction of the inter­
section angle α of the fault rupture with the pipe, increasing the compressive component; and (ii) it leads to a
reduction of localized deformation close to the surface. From our results, the h=H required for initiation of
buckling is insensitive to burial depth Hp =D, but the one required for pipe rupture increases substantially with
increase of Hp =D. The increase of the D=t (pipe diameter to wall thickness) ratio leads to a substantial decrease of
the h=H at which buckling is initiated, and the same applies to the pipe rupture criteria, which are met at a
smaller h=H with the increase of D=t.

1. Introduction force and the bending moment on the pipeline, considering material and
geometric nonlinearities. This approach was later extended by Karami­
Pipelines constitute one of the key infrastructure elements of modern tros et al. [16]; further enhancements to account for internal pressure
societies. Used to transport gases and liquids over long distances, they and temperature were made by Zhang et al. [17] by employing elasto­
are either installed at the ground surface or buried in soil for protection plastic beam theory and a more refined analysis of axial force in the high
against environmental hazards. Permanent ground deformation, such as curvature zone. Apart from analytical approaches, numerical analysis
fault dislocation, slope instability, and liquefaction-induced displace­ methods have also been employed. For example, Ariman & Lee [18]
ment constitute a major threat for buried pipelines (e.g., Jennings [1], used the finite element (FE) method to assess axial and bending strains
MaCaffrey and O’Rourke [2], Desmod et al. [3], Nakata & Hasuda [4], in buried steel pipelines. Liu et al. [19] developed an equivalent
Takada et al. [5], Bray et al. [6], Liang & Sun [7], O’Rourke and Liu [8], boundary, composed of nonlinear springs that can be applied to the ends
O’Rourke et al. [9]). Significant research has been devoted in under­ of a shell model of the pipeline. Vazouras et al. [20,21] conducted full
standing the performance of buried pipelines subjected to such defor­ 3D analyses of a steel pipeline subjected to strike slip faulting, ac­
mation. Newmark & Hall [10] were one of the first to develop simplified counting for nonlinear soil response, sliding and detachment at the
analysis methods for the fault crossing problem, assuming that the pipeline-soil interface, and the distortion of the pipeline cross section
pipeline is subjected to direct stressing by the dislocation, ignoring due to local buckling. A refined analysis, combining 3D FE modeling
non-uniform lateral soil resistance. Their work was extended by Ken­ with mathematical solutions was presented by Vazouras et al. [22].
nedy et al. [11] and Kennedy & Kincaid [12], considering non-uniform Chaloulos et al. [23,24] and Kouretzis et al. [25] studied the trench
friction at the soil-pipe interface. Wang & Yeh [13] and Vougioukas effects on the soil-pipe response subject to PGD by using 2D plane-strain
et al. [14] moved one step further, accounting for the bending stiffness soil-pipe FE models with interface contact elements.
of the pipeline and considering both horizontal and vertical fault Using the large-scale split-box of Cornell University, Yoshizaki et al.
movement. [26] carried out large scale testing of buried steel gas pipelines with
More recently, Karamitros et al. [15] combined elbows subjected to strike-slip faulting, and used the experimental re­
beam-on-elastic-springs and elastic beam theory to calculate the axial sults to validate FE models for further parametric studies. An extensive

* Corresponding author.
E-mail addresses: Fadaee@srbiau.ac.ir (M. Fadaee), Farnoud.Farzaneganpour@srbiau.ac.ir (F. Farzaneganpour), ixa@ethz.ch (I. Anastasopoulos).

https://doi.org/10.1016/j.soildyn.2020.106090
Received 6 April 2019; Received in revised form 3 February 2020; Accepted 5 February 2020
Available online 13 February 2020
0267-7261/© 2020 Elsevier Ltd. All rights reserved.
M. Fadaee et al. Soil Dynamics and Earthquake Engineering 132 (2020) 106090

series of centrifuge model tests on the effects of strike-slip faults on behavior of the steel pipe. An API/5L X65 pipe is selected for the ana­
buried high-density polyethylene (HDPE) pipelines have been con­ lyses, which is commonly used by the oil and gas industry. The yield
ducted by Ha et al. [27–29], Abdoun et al. [30] and Xie et al. [31,32]. stress σ y ¼ 450 MPa is followed by a plastic plateau up to 3% Strain, and
The centrifuge tests were conducted at the Rensselaer Polytechnic a hardening regime with hardening modulus Es =300 (where Es is the
Institute (RPI), and were compared to large-scale tests conducted at Young’s modulus of steel) [20].
Cornell University. Different diameters of HDPE pipes were tested, Following the findings of previous studies, an elastoplastic consti­
exploring the effect of internal pressure, the angle of pipeline crossing tutive model with Mohr Coulomb failure criterion and isotropic strain
the strike-slip fault, the effects of burial depth, and moisture content softening is adopted [37]. The pre-yield behaviour is assumed elastic,
(more details can be found in NEESR-SG final report [33]). Jalali et al. with secant modulus Gs ¼ τy =γy linearly increasing with depth. Strain
[34] performed full-scale testing and FE analyses of buried steel gas softening is introduced by linearly reducing the mobilized friction angle
pipelines subjected to reverse faulting. Their validated FE model was ϕmob and the mobilized dilation angle ψ mob with the increase of octa­
then used to calculate soil-pipe interaction forces at specific sections, hedral plastic shear strain γ Poct :
and the results were compared to the recommendations of ALA [35],
8
confirming their suitability for design purposes. >
< ϕP
ϕP ϕres
*γPoct for 0 � γPoct < γ Pf
Despite the considerable research on the subject, there are still some ϕmob ¼
γ Pf
(1)
open questions that need to be addressed. So far, most analytical and >
:
ϕres for γPoct > γ Pf
experimental studies have focused on local soil-pipeline interaction, not
addressing the effect of the sediment thickness beneath the pipe on its 8 !
response subjected to reverse fault rupture. Considering the fact that > γ Poct
< ψP* 1
> for 0 � γ Poct < γPf
pipelines are not installed on bedrocks and there are layers of soils ψ mob ¼ γ Pf (2)
>
beneath them, it is discovered that the underlying soil beneath the pipe >
:
ψ res for γ Poct > γPf
is a factor responsible in changing the failure modes of pipelines (with
the same pipe’s mechanical characteristics and burial depth). This is one
where ϕP and ψ P are the peak mobilized friction and dilation angles; ϕres
of the key objectives of this paper, which employs 3D FE modelling to
derive insights on the performance of buried pipelines subjected to and ψ res their residual (or critical state) values; and γPf is the plastic
reverse faulting. A thoroughly validated strain softening constitutive octahedral shear strain at which softening is completed. Scale effects are
model is employed to conduct a parametric study, exploring the effect of considered through an approximate simplified scaling method [37].
pipeline thickness and burial depth, in addition to soil depths beneath An idealized dense sand deposit is considered with ϕp ¼ 42� , ϕres ¼
the pipe. 32� , ψ p ¼ 12� , ψ res ¼ 1� (purely for numerical stability), and γ Pf ¼ 0:05.
The soil density is ρ ¼ 1:8 Mg/m3 and the Poisson’s ratio v ¼ 0:3, while
2. Problem definition and analysis methodology the Young’s modulus E is linearly increasing with depth from E0 ¼ 12
MPa at the ground surface to E30 ¼ 70 MPa at 30 m depth. These pa­
The scope of the paper is to investigate the response of steel buried rameters are used for the parametric study. The depth H of the soil layer
pipelines subjected to reverse faulting. The problem is analyzed is parametrically varied from 5 m to 30 m. In order to minimize parasitic
employing the Abaqus [36] 3D FE analysis environment, with due boundary effects, based on the results of a preliminary sensitivity study,
consideration of material and geometric nonlinearities. As shown in the the length of the model is taken equal to 4H (i.e., 120 m for H ¼ 30 m),
3D representation of Fig. 1a, the FE model includes the steel pipe and the and its width equal to 10 m.
surrounding soil, connected through a frictional (Coulomb) interface The FE discretization needs to be adequately refined to allow realistic
which allows detachment and sliding. A transverse cross-section of the simulation of buckling phenomena and soil–pipe interaction. According
model is shown in Fig. 1b. The pipe is modelled with isotropic 3D shell to Vazouras et al. [20], the pipe perimeter needs to be discretized with
elements, whereas hexahedral (8-node) reduced-integration brick ele­ 54 elements. Based on the results of an initial sensitivity study, the size
ments are used for the soil. A large-strain j2 flow (von Mises) plasticity of the shell elements is taken equal to D=18; away from the area of fault
model with isotropic hardening is employed to describe the mechanical crossing, a coarser mesh is adapted. This leads to element dimension

Fig. 1. Overview of the numerical model: (a) 3D FE discretization; and (b) transverse cross-section.

2
M. Fadaee et al. Soil Dynamics and Earthquake Engineering 132 (2020) 106090

dFE ¼ 0:05 m, which cannot (and does not need to) be maintained experimental measurements, in order to demonstrate the validity of the
throughout the soil model. In terms of fault rupture propagation, it has analysis technique.
been shown by previous studies [37], and confirmed by initial sensitivity Fig. 3 compares the FE predicted longitudinal strains at the pipe
analyses, that dFE ¼ 0:5 m is adequate. Therefore, as shown in Fig. 1b, crown to the experimental measurements for imposed fault offset h ¼
the FE model is composed of two separate parts: a local interface one 0:17m and h ¼ 0:52m (equal to oblique fault offset of 0:2m and 0:6m,
around the pipe with dFE ¼ 0:05 m, suitable for soil-pipe interaction; respectively). The analysis correctly predicts the location of the buckled
and a global one with dFE ¼ 0:5 m, suitable for fault rupture propaga­ pipeline section at x ¼ 0:6 m. The comparison is equally successful
tion. The two parts are connected together through a tied interface. in terms of the peak compressive longitudinal strain, which is in satis­
A special contact interface is employed to simulate the interface factory agreement with the experimental results (less than 10%
between the outer surface of the steel pipe and the surrounding soil. The discrepancy). The location and magnitude of the peak tensile strain are
surface to surface contact algorithm accounts for interface friction, also reasonably predicted.
which is a function of the friction coefficient μ. According to Yimsiri
� �
et al. [38], μ can be assumed equal to tan 23ϕp for steel pipes. Thus, for 3. The effect of sediment thickness

the dense sand considered herein, μ is set to 0.5. The interface also al­ In this section, the effect of sediment thickness H is investigated. The
lows separation of the pipe from the surrounding soil. effect of burial depth and pipeline thickness is addressed later on. As
The bottom boundary of the FE model represents the underlying previously discussed, In the case of reverse faults, the dip angle of the
bedrock. It is divided into two parts, one (on the right) representing the fault rupture path is reduced as it approaches the ground surface [37,
footwall (stationary block); and the other one (on the left), which is 41]. When the sediment thickness increases, the pipeline burial depth
subjected to the fault offset (of vertical component) h with dip angle λ ¼ becomes smaller in relative terms. Hence, since the reduction of the dip
60� , represents the hanging wall (moving block). The left side of the soil angle is larger close to the ground surface, with the increase of the
in YZ plane (Fig. 1) also moves while the right side in the same plane is sediment thickness at the pipe burial depth the reduction is larger.
fixed. The other two sides of the soil in XZ plane (Fig. 1) are fixed in the Therefore, in this research the response of a pipeline typically buried at
normal direction. The pipe end displacement at the footwall-side shallow depths with various soil depths beneath the pipe is investigated.
boundary is fixed in all directions, while the other end of the pipe, at In practice, pipelines are installed in an artificially excavated trenches
the hanging wall, follows the fault-induced displacement. This is con­ which are filled with non-cohesive soft soils. However, since the concept
servative assumption, which ignores the flexibility of the parts of the of the study is to show the influence of fault rupture propagation path at
pipeline that are not included in the model. More accurate simulation different sediment thicknesses, a simplified uniform dense sand is
would require an equivalent spring boundary at the two ends of the pipe, considered especially since the mechanism is mainly due to deeper
representing the axial stiffness of its continuation. However, in a layers of soil and the reduction of dip angle of the fault rupture occurs in
research by Kaya et al. [39], the fixed end boundary was able to capture
an accurate results compared to field data observations. Kinematic
coupling is introduced at the end sections of the pipeline to prevent local
instabilities, as recommended by Guarracino et al. [40]. More specif­
ically, special kinematic constraints are used to connect the perimeter
nodes of the pipe at the two end sections with a reference node at the
center of pipe cross-section.

2.1. Model validation

The FE modeling technique is validated against full-scale physical

model tests of a 4-inch API-5L grade B steel pipeline using the material
property of the test, which was conducted at Sharif University of
Technology by Jalali et al. [34]. As shown in Fig. 2, the tests were
conducted using a large-scale split-box, which was built to investigate
the behavior of buried gas distribution pipelines subjected to reverse
faulting. The previously discussed 3D FE constitutive model is adapted Fig. 3. Validation of the 3D FE model against the experimental results of Jalali
et al. [34]. Comparison of numerical prediction to the experimental measure­
to the properties of the tested configurations (400 API-5L Grade B pipe
ments for imposed fault offset of 0.2 m and 0.6 m in terms of longitudinal
and soil with ϕp ¼ 33:5� , ϕres ¼ 30� , ψ p ¼ 3:5� , ψ res ¼ 1� ; E ¼ 5000 KPa;
strains at the pipe crown.
ρ ¼ 1:8 Mg/m3). The analytical predictions are compared to the

Fig. 2. Longitudinal cross-section and photo of the split-box of Jalali et al. [34], the results of which are used as benchmark for validation of the FE model.

3
M. Fadaee et al. Soil Dynamics and Earthquake Engineering 132 (2020) 106090

near ground surface. Hence, to focus on the effect of H, the remaining vertical Δh =H and horizontal Δx =H displacements along the pipe
parameters are kept constant: the burial depth to diameter ratio Hp = D ¼ (Fig. 4b). The increase of H and the associated reduction of α at the
2:5, and the pipe thickness to diameter ratio D=t ¼ 48 (D ¼ 36”, Hp ¼ intersection with the pipe leads to an increase of Δx =H. This unavoidably
2:3 m). A key focus of the study is the buckling and post-buckling leads to the generation of increased compressive strains, leading to the
behavior of the pipeline. there are some critical limit states for change of the failure mode which is discussed in Fig. 5.
compression and tension in pipelines such as those used by Wijewick­ Fig. 5 compares the response for H ¼ 15 m and 30 m, in terms of
reme et al. [42] and other criteria that are available in the literature and deformed FE mesh with superimposed longitudinal strain contours. Two
design codes [35,42–45]. snapshots are shown, one at initiation of buckling (Fig. 5a) and a second
In this paper, the main goal was to investigate the fault rupture one at rupture (Fig. 5b). A first observation is that the length of strain
propagation reduction angle effects on pipe response. The criterion concentrations is reduced with the increase of H. As shown in Fig. 4, by
which was used in this research to evaluate the buckling load of shells increase of sediment thickness, the soil movement in the shear band
was first introduced by Budiansky and Roth [46,47] and is used by region at the ground surface tend to move more horizontally rather than
various researchers [48–50]. Based on this criterion, shells are subjected vertically. Hence, the axial tensions and strains to the pipe in interaction
to different levels of loading and the load at which there is a significant with soil increases that change the failure mode from local buckling for
jump in the response for a small load increment is considered as the H ¼ 15 m to wrinkling and telescopic failure for H ¼ 30 m. Irrespective
critical buckling load. This load signifies the passing of the structure of H, the location of maximum damage is within the moving block, and
from a stable state into a critical state. By using this concept, the pipe the ultimate rupture is due to compressive strains. Taking a closer look
buckling strain might differ from one loading pattern to another one. at H ¼ 15 m, it can be seen that the pipe suffers inward folding at the
Therefore, the strain values for buckle initiations are different for each bottom, resulting to typical kinking and local buckling. No such buckling
pipe. With respect to ultimate rupture, a failure strain ε ¼ 20% is is observed at the top of the pipe, where the strains are tensile. In stark
assumed in accord with material properties [20]. It has to be mentioned contrast, for H ¼ 30 m the entire pipe cross section sustains compressive
that no internal pressure is considered in this study. strains, leading to telescopic-type buckling. This is due to the previously
Fig. 4 compares H ¼ 15 m to H ¼ 30 m ​ in terms of the fault rupture mentioned increase of the compressive component of the deformation
propagation path and its intersection angle α with the pipe for h= H ¼ due to the increase of H, which results to larger compressive strains and
1:3% (Fig. 4a), and the normalized vertical and horizontal displacement rupture of the pipe for a smaller h=H soon after buckling initiation.
along the pipe (Fig. 4b). It can be seen that the increase of H leads to a Further insights are offered in Fig. 6 which illustrates the normalized
reduction of the fault rupture propagation path angle α at the intersec­ fault offset h=H at pipe buckling and ultimate rupture, in function of
tion with the pipeline: from α ¼ 59ο for H ¼ 15 m to α ¼ 38ο for H ¼ sediment thickness H. The increase of H leads to a reduction of the h=H
30 m (Fig. 4a). This is due to the fact that the burial depth to diameter at which buckling is observed: from h=H ¼ 2:35% for H ¼ 15 m to h=H ¼
ratio is kept constant to Hp =D ¼ 2:5. By keeping Hp =D constant, the 1:7% for H ¼ 30 m. However, in absolute terms the required h for pipe
increase of H leads to a pipe that is closer to the ground surface in buckling increases with H, as it would be intuitively expected. Two
relative terms. This is confirmed by the distribution of normalized contradicting mechanisms can be identified: (a) the increase of moiety

Fig. 4. Comparison of ​ H ¼ 15 m to H ¼ 30 m ​ ðHp =D ¼ 2:5 and D=t ¼ 48Þ in terms of: (a) fault rupture propagation path and its intersection angle α with the pipe
for h=H ¼ 1.3%; and (b) normalized vertical and horizontal displacement along the pipe (the length x ​ is normalized to the total model length xo ).

4
M. Fadaee et al. Soil Dynamics and Earthquake Engineering 132 (2020) 106090

Fig. 5. Deformed mesh with superimposed longitudinal strain contours for H ¼ 15 m and 30 m ðHp =D ¼ 2:5 and D =t ¼ 48Þ: ​ (a) at the initiation of buckling; and (b)
at ultimate rupture.

angle with pipe α is almost the same. This is attributed to the fact that
fault rupture propagation angle tends to decrease in near ground surface
and for H ¼ 5 m, the reduction happens very close to the surface.
Further studies show that by change of bedrock dip angle λ to higher
or fewer values, the fault rupture dip angle still reduces in ground sur­
face but with different magnitudes. For instance, the results of bedrock
dip angle λ ¼ 45ο , shows more reduction angle compared to λ ¼ 60ο .
The same mechanism can be observed again. This time since the hori­
zontal movement of the soil are more (compared to λ ¼ 60ο ) the tele­
scopic failure is also observed at H ¼ 20 m and the length of damages in
H ¼ 15 m has reduced.

4. The effect of burial depth

The effect of burial depth is investigated by comparing the response

of the pipeline for four different ratios Hp =D ¼ 1:5; 2:5; 3:5 and 4:5
Fig. 6. Normalized fault offset h=H at which pipe buckle initiating and ultimate (which correspond to Hp ¼ 1:3; 2:3; 3:2 m and 4:1 m, respectively),
rupture is observed, as a function of sediment thickness H. (results for Hp = D ¼
keeping the remaining parameters constant: the sediment thickness H ¼
2:5 and D=t ¼ 48).
20 m and the pipe thickness to diameter ratio D=t ¼ 48. Fig. 7 depicts
the normalized fault offset h=H at which pipe buckling and ultimate
was introduced by re leads to a reduction of the rupture propagation rupture is observed, in function of burial depth Hp =D. The increase of the
path angle, increasing the compressive deformation component; and (b)
normalized burial depth Hp =D leads to a negligible increase of the fault
it leads to a reduction of localized deformation close to the surface, as
offset h=H that is required to initiate buckling. The big difference lies in
the dislocation is partially absorbed by the soil layer which acts as a
the h=H that is required for rupture, which can be seen to increase by
“cushion”. The h=H required for ultimate rupture of the pipeline also
over 60% with the increase of Hp =D from 3:5 to 4:5.
decreases with H, but more abruptly: from h=H ¼ 3:55% for H ¼ 15 m to
Fig. 8 compares the response for Hp =D ¼ 4:5 and 2:5, in terms of
h=H ¼ 1:77% for H ¼ 30 m. For H ¼ 30 m rupture takes place almost
deformed FE mesh with superimposed longitudinal strain contours.
simultaneously with buckling. The H ¼ 5 m shows somehow the similar
Three snapshots are shown, one at h=H ¼ 2%, which corresponds to the
behavior with H ¼ 15 m as the rupture propagation path intersection

5
M. Fadaee et al. Soil Dynamics and Earthquake Engineering 132 (2020) 106090

with the pipe leads to an increase of the compressive component, which

is, however, not enough to lead to the previously discussed telescopic-
type failure. The previously mentioned difference in the failure mech­
anism (Fig. 8) is attributed to the increase of the confining stresses with
the increase of burial depth. However, the reason that pipe response has
improved with increase if burial depth is that the failure mechanism is
not only a function of friction between the pipe and the soil, but also, the
reduction of dip angle (αÞ at its intersection with the fault rupture. With
the increase of burial depth, on one hand the overburden pressures and
the pipe soil frictions increase, while on the other hand, due to the less
reduction of α at the intersection with the pipe, the compressive strains
decrease.
It has to be mentioned that in shallower sediment thicknesses, the
reduction of burial depth may lead to general beam buckle failure [51,
52]. This is attributed to the previously discussed reduction of fault
Fig. 7. Normalized fault offset h=H at which pipe buckle initiating and ultimate propagation path angle α at near ground surface. Hence, in a shallower
rupture is observed, as a function of burial depth to diameter ratio Hp = D (re­
sediment, the reduction occurs in a depth that pipe is not buried.
sults for H ¼ 20 m and D=t ¼ 48Þ.
Therefore, the combination of fewer overburden pressure and axial
movement of the soil, may lead to the formation of general beam buckle
initiation of buckling for both Hp =D (Fig. 8a); one at h= H ¼ 2:23% failure mode.
where the shallower pipe reaches the rupture criteria (Fig. 8b); and one
at h=H ¼ 3:87%, where the deeper pipe finally reaches the rupture 5. The effect of pipe thickness
criteria (Fig. 8c). Evidently, the initiation of buckling is insensitive to the
burial depth Hp =D, taking place at h=H ¼ 2% in both cases examined. The effect of pipe thickness is parametrically studied by varying the
However, the fault offset h=H at which the rupture criteria are met in­ pipe thickness to diameter ratio D=t ¼ 144; 48; and 28 (which corre­
creases significantly with the burial depth, from h=H ¼ 2:23% for Hp = spond for the 36’’ D ¼ 914 mm pipe to t ¼ 6:35 mm; 19:05 mm; and
D ¼ 2:5 to h=H ¼ 3:87% for Hp =D ¼ 4:5. The deformation mechanism 32:65 mm, respectively) in the thick sediment of H ¼ 20 m. Fig. 10
also differs. It can be seen that length of damaged section has reduced shows the normalized fault offset h=H at which pipe buckling and ulti­
and consequently, sides of the pipe wall have folded in deeper pipe. mate rupture is observed in function of the D=t ratio. The increase of D=t
Fig. 9 compares the two burial ratios, Hp =D ¼ 2:5 and 4:5 in terms of leads to a substantial decrease of the fault offset h=H at which buckling is
the intersection angle α of the fault rupture propagation path with the initiated. The decrease of the pipe wall thickness t leads to a larger
pipe for h=H ¼ 2%. Although the bedrock dip angle is the same ðλ ¼ decrease of flexural stiffness compared to the axial. And as discussed
60ο Þ, the decrease of the burial depth leads to a substantial decrease of later on, this leads to a change of mechanism, which explains the
the rupture propagation path angle α at the intersection with the pipe­ observed reduction of the required h=H for buckling initiation. The same
line: from α ¼ 45ο for Hp =D ¼ 2:5 to α ¼ 59ο for Hp =D ¼ 4:5. The applies to the ultimate rupture criteria, which are also met at a smaller
decrease of Hp =D and the associated decrease of the intersection angle α h=H with the increase of D=t. It is worth observing that for D=t ¼ 144

Fig. 8. Deformed mesh with superimposed longitudinal strain contours for Hp =D ¼ 2:5 and 4:5 (results for H ¼ 20 m and D=t ¼ 48Þ: (a) at initiation of buckling; (b)
the ultimate rupture of the pipe with Hp =D ¼ 2:5; and (c) at the ultimate rupture of the pipe with Hp =D ¼ 4:5.

6
M. Fadaee et al. Soil Dynamics and Earthquake Engineering 132 (2020) 106090

Fig. 9. Comparison of Hp =D ¼ 2:5 ðHp ¼ 2:3 mÞ to Hp =D ¼ 4:5 ðHp ¼ 4:1 mÞ in terms of fault rupture propagation path and its intersection angle α with the pipe for
h=H ¼ 2% (results for H ¼ 20 m and D=t ¼ 48Þ.

the h=H required for rupture almost coincides with the one required for
buckling initiation.
The previously discussed change of mechanism is further elucidated
in Fig. 11, which compares the response for D=t ¼ 28 and 144 in terms of
deformed FE mesh with superimposed longitudinal strain contours. Two
snapshots are provided, one at initiation of buckling (Fig. 11a) and one
at ultimate rupture (Fig. 11b). While for D=t ¼ 28 the buckling mecha­
nism is observed, for D=t ¼ 144 a telescopic-type mechanism becomes
critical. The latter takes place at a smaller fault offset: h=H ¼ 1:4%
instead of h=H ¼ 2:42%. As previously discussed, the increase of D=t is
associated with a decrease of axial and flexural stiffness. The decrease of
flexural stiffness is much more pronounced than the axial one, rendering
the more compressive telescopic-type failure critical. Its earlier onset is
attributed to the decrease of the axial resistance of the pipe to buckling.
Similar trends are observed with respect to the rupture criteria, which
Fig. 10. Normalized fault offset h=H at which pipe buckle initiation and ulti­
are met much earlier with the increase of D=t: for h=H ¼ 1:45% for D=t ¼
mate rupture is observed, as a function of pipe thickness to diameter ratio D= t 144 as opposed to. h=H ¼ 4:5% ​ for ​ D=t ¼ 28:
(results for H ¼ 20 m and Hp =D ¼ 2:5Þ.

Fig. 11. Deformed mesh with superimposed longitudinal strain contours for D=t ¼ 28 and 144 ðH ¼ 20 m andHp =D ¼ 2:5Þ : (a) at initiation of buckling; and (b) at
ultimate rupture.

7
M. Fadaee et al. Soil Dynamics and Earthquake Engineering 132 (2020) 106090

6. Conclusions [5] Takada S, Nakayama M, Ueno J, Tajima C. “Report on taiwan earthquake”, RCUSS.
Earthquake Laboratory of Kobe University; 1999. p. 2–9.
[6] Bray JD, Stewart JP, Baturay MB, Durgunoglu T, Onalp A, Sancio RB, Stewart JP,
This paper has applied a thoroughly validated FE analysis method­ Ural D, Ansal A, Bardet JB, Barka A, Boulanger R, Cetin O, Erten D. Damage
ology to study the performance of pipelines subjected to reverse faulting. patterns and foundation performance in adapazari. Earthq Spectra 2000;16(S1):
The FE model was validated against full-scale experimental results of a 163–89.
[7] Liang J, Sung S. Site effects on seismic behavior of pipelines: a review. Journal of
4-inch steel pipeline subjected to reverse faulting [34]. The validated FE Pressure Vessel Technology, ASME 2000;122(4):469–75. https://doi.org/10.1115/
model was consequently employed to parametrically study the effect of 1.1285974.
sediment thickness, pipe thickness, and burial depth. [8] O’Rourke MJ, Liu X. Seismic design of buried and offshore pipelines. Buffalo:
Multidisciplinary Center for Earthquake Engineering Research; 2012. Monograph
The increase of sediment thickness H was shown to lead to a No. 4.
reduction of the normalized fault offset h=H at which buckling is initi­ [9] O’Rourke M, Filipov E, Uçkan E. Towards robust fragility relations for buried
ated. Two contradicting mechanisms were identified: (a) the increase of segmented pipe in ground strain areas. Earthq Spectra 2015;31(3):1839–58.
https://doi.org/10.1193/032311EQS076M.
H leads to a reduction of the intersection angle α of the fault rupture with [10] Newmark NM, Hall WJ. Pipeline design to resist large fault displacement. In:
the pipe, increasing the compressive deformation component; and (b) it Proceedings of U.S. national conference on earthquake engineering; 1975.
leads to a reduction of localized deformation close to the ground surface. p. 416–25.
[11] Kennedy RP, Chow AW, Williamson RA. Fault movement effects on buried oil
The normalized fault offset h=H required for ultimate rupture of the pipeline. J Transport Eng, ASCE 1977;103:617–33.
pipeline also decreases with sediment thickness H. The increase of H [12] Kennedy RP, Kincaid RH. Fault crossing design for buried gas oil pipelines. In:
leads to a change of failure mode from local buckling ðH ¼ 15 mÞ to Proceeding of the PVP conference, ASME. vol. 77; 1983. p. 1–9.
[13] Wang LRL, Yeh YA. A refined seismic analysis and design of buried pipeline for
wrinkling and telescopic-type failure ðH ¼ 30 mÞ. The location of
fault movement. Earthq Eng Struct Dynam 1985;13:75–96. https://doi.org/
maximum damage is always within the moving block, and the ultimate 10.1002/eqe.4290130109.
rupture is due to compressive strains. [14] Vougioukas EA, Theodossis C, Carydis PG. Seismic analysis of buried pipelines
The normalized fault offset h=H required for initiation of buckling subjected to vertical fault movement. J Tech Counc ASCE 1979;105(TCI):432–41.
[15] Karamitros DK, Bouckovalas GD, Kouretzis GP. Stress analysis of buried steel
was shown to be insensitive to the burial depth Hp =D for the cases pipelines at strike-slip fault crossings. Soil Dynam Earthq Eng 2007;27:200–11.
examined herein. This is not the case for the h=H that is required for https://doi.org/10.1016/j.soildyn.2006.08.001.
rupture, which increases substantially with the increase of Hp = D. The [16] Karamitros DK, Bouckovalas GD, Kouretzis GP, Gkesouli V. An analytical method
for strength verification of buried steel pipelines at normal fault crossings. Soil
latter is related to the change in the deformation mechanism, which is Dynam Earthq Eng 2011;31(11):1452–64. https://doi.org/10.1016/j.
partly attributed to the increase of the confining stresses with burial soildyn.2011.05.012.
depth Hp =D. At the same time, the decrease of Hp =D leads to a pro­ [17] Zhang L, Zhao X, Yan X, Yang X. Elastoplastic analysis of mechanical response of
buried pipelines under strike-slip faults. Int J Geomech, ASCE 2016;17(4):
nounced decrease of the intersection angle α of the fault rupture with the 04016109. https://doi.org/10.1061/(ASCE)GM.1943-5622.0000790.
pipeline, which is directly associated with an increase of the compres­ [18] Ariman T, Muleski GE. A review of seismic response of buried pipelines under
sive deformation component. seismic excitations. Earthq Eng Struct Dynam 1981;9(2):133–52. https://doi.org/
10.1002/eqe.4290090204.
The increase of the D=t (pipe diameter to wall thickness) leads to a [19] Liu M, Wang Y-Y, Yu Z. Response of pipelines under fault crossing. In: Proceedings,
substantial decrease of the fault offset h=H at which buckling is initiated. international offshore and polar engineering conference. Vancouver, BC: Canada;
The same applies to the ultimate rupture criteria, which are met at a 2008.
[20] Vazouras P, Karamanos SA, Dakoulas P. Finite element analysis of buried steel
smaller h=H with the increase of D=t. This is directly related to a pipelines under strike-slip fault displacements. Soil Dynam Earthq Eng 2010;30
mechanism change. While for D=t ¼ 28 the buckling mechanism is (11):1361–76. https://doi.org/10.1016/j.soildyn.2010.06.011.
observed, for D=t ¼ 144 a telescopic-type mechanism becomes critical. [21] 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. https://
The increase of D=t leads to a decrease of axial and flexural stiffness. And doi.org/10.1016/j.soildyn.2012.05.012.
since the latter is much more pronounced, the telescopic-type failure [22] Vazouras P, Dakoulas P, Karamanos SA. “Pipe–soil interaction and pipeline
becomes critical. performance under strike–slip fault movements”. Soil Dynam Earthq Eng 2015;72:
48–65. https://doi.org/10.1016/j.soildyn.2015.01.014.
[23] Chaloulos YK, Bouckovalas GD, Zervos SD, Zampas AL. Lateral soil-pipeline
Declaration of competing interest interaction in sand backfill: effect of trench dimensions. Comput Geotech 2015;69:
442–51. https://doi.org/10.1016/j.compgeo.2015.05.014.
[24] Chaloulos YK, Bouckovalas GD, Karamitros DK. Trench effects on lateral p-y
The authors declare that they have no known competing financial
relations for pipelines embedded in stiff soils and rocks. Comput Geotech 2017;83:
interests or personal relationships that could have appeared to influence 52–63. https://doi.org/10.1016/j.compgeo.2016.10.018.
the work reported in this paper. [25] Kouretzis GP, Sheng D, Sloan SW. Sand-pipeline-trench lateral interaction effects
for shallow buried pipelines. Comput Geotech 2013;54:53–9. https://doi.org/
10.1016/j.compgeo.2013.05.008.
CRediT authorship contribution statement [26] Yoshizaki K, O’Rourke TD, Hamada M. Large scale experiments of buried steel
pipelines with elbows subjected to permanent ground deformation. Structural
M. Fadaee: Conceptualization, Methodology, Writing - review & Engineering/Earthquake Engineering 2003;20(1):1S–11S. https://doi.org/
10.2208/jsceseee.20.1s.
editing, Supervision, Project administration. F. Farzaneganpour: [27] Ha D, Abdoun TH, O’Rourke MJ, Symans MD, O’Rourke TD, Palmer MC, et al.
Conceptualization, Methodology, Software, Validation, Formal analysis, Centrifuge modeling of earthquake effects on buried high-density polyethylene
Investigation, Resources, Data curation, Writing - original draft, Writing (HDPE) pipelines crossing fault zone. J Geotech Geoenviron Eng 2008;134(10):
1501–15. https://doi.org/10.1061/(ASCE)1090-0241(2008)134:10(1501).
- review & editing. I. Anastasopoulos: Conceptualization, Methodol­ [28] Ha D, Abdoun TH, O’Rourke MJ, Symans MD, O’Rourke TD, Palmer MC, et al.
ogy, Writing - review & editing, Visualization, Supervision. Buried high-density polyethylene pipelines subjected to normal and strike-slip
faulting: a centrifuge investigation. Can Geotech J 2008;45(12):1733–42. https://
doi.org/10.1139/T08-089.
References
[29] Ha D, Abdoun TH, O’Rourke MJ, Symans MD, O’Rourke TD, Palmer MC, et al.
Earthquake faulting effects on buried pipelines – case history and centrifuge study.
[1] Jennings PC. Engineering features of the san fernando earthquake february 9, J Earthq Eng 2010;14(5):646–69. https://doi.org/10.1080/13632460903527955.
1971. Pasadena, CA: California Institute of Technology Report; 1971. EERL 71–02. [30] Abdoun TH, Ha D, O’Rourke MJ, Symans MD, O’Rourke TD, Palmer MC, et al.
[2] MaCaffrey MA, O’Rourke TD. Buried pipeline response to reverse faulting during Factors influencing the behavior of buried pipelines subjected to earthquake
the 1971 San Fernando Earthquake. In: Proceeding of the PVP conference, ASME. faulting. Soil Dynam Earthq Eng 2009;29(3):415–27. https://doi.org/10.1016/j.
vol. 77; 1983. p. 151–9. soildyn.2008.04.006.
[3] Desmod TP, Power MS, Taylor CL, Lau RW. Behavior of large-diameter pipeline at [31] Xie X, Symans MD, O’Rourke MJ, Abdoun TH, O’Rourke TD, Palmer MC, et al.
fault crossings. In: Technical council on lifeline earthquake engineering (TCLEE), Numerical modeling of buried HDPE pipelines subjected to strike-slip faulting.
ASCE. vol. 6; 1995. p. 296–303. J Earthq Eng 2011;15(8):1273–96. https://doi.org/10.1080/
[4] Nakata T, Hasuda K. Active fault in 1995 hyogoken nanbu earthquake. Kagaku 13632469.2011.569052.
1995;65:127–42.

8
M. Fadaee et al. Soil Dynamics and Earthquake Engineering 132 (2020) 106090

[32] Xie X, Symans MD, O’Rourke MJ, Abdoun TH, O’Rourke TD, Palmer MC, et al. [42] Wijewickreme D, Honegger D, Mitchell A, Fitzell T. Seismic vulnerability
Numerical modeling of buried HDPE pipelines subjected to normal faulting: a case assessment and retrofit of a major natural gas pipeline system: a case history.
study. Earthq Spectra 2013;29(2):609–32. https://doi.org/10.1193/1.4000137. Earthq Spectra 2005;21(2):539–67.
[33] NEESR-SG final report. Prepared: Cornell University, Rensselaer Polytechnic [43] American Society of Civil Engineers (ASCE). Guidelines for the seismic design of oil
Institute and Sciencenter Discovery Center; 2008. p. 47. and gas pipeline systems, committee on gas and liquid fuel lifelines, ASCE. 1984.
[34] Jalali HH, Rofooei FR, Attari NKA, Samadian M. Experimental and finite element [44] Honegger DG, Nyman D. PRCI guidelines for the seismic design and assessment of
study of the reverse faulting effects on buried continuous steel gas pipelines. Soil natural gas and liquid hydrocarbon pipelines. Houston, Texas: Pipeline Research
Dynam Earthq Eng 2016;86:1–14. https://doi.org/10.1016/j.soildyn.2016.04.006. Council International; 2004. published by Technical Toolboxes.
[35] American Lifeline Alliance (ALA). Guidelines for the design of buried steel pipe. ” [45] American Lifeline Alliance (ALA). Design guidelines for seismic resistant water
FEMA; 2001. p. 75p. pipeline installations. FEMA; 2005. p. 255.
[36] ABAQUS 6.11. Standard user’s manual. Providence: Dassault Syst�emes Simulia [46] Budiansky B, Roth RS. Axisymmetric dynamic buckling of clamped shallow
Corp; 2011. spherical shells. 1962. p. 597–606.
[37] Anastasopoulos I, Gazetas G, Bransby MF, Davies MCR, El Nahas A. fault rupture [47] Budiansky B. Dynamic buckling of elastic structures: criteria and estimates.
propagation through sand: finite element analysis and validation through Dynamic Stability of Structures 1967:83–106.
centrifuge experiments. Journal of Geotechnical and Geoenvironmental [48] Buratti N, Tavano M. Dynamic buckling and seismic fragility of anchored steel
Engineering, ASCE 2007;133(8):943–58. https://doi.org/10.1061/(ASCE)1090- tanks by the added mass method. Earthq Eng Struct Dynam 2014;43(1):1–21.
0241(2007)133:8(943). https://doi.org/10.1002/eqe.2326.
[38] Yimsiri S, Soga K, Yoshiaki K, Dasari GR, O’Rourke TD. “Lateral and upward soil– [49] Virella JC, Godoy LA, Su� arez LE. Dynamic buckling of anchored steel tanks
pipeline interactions in sand for deep embedment conditions”. Journal of subjected to horizontal earthquake excitation. J Constr Steel Res 2006;62(6):
Geotechnical and Geoenvironmental Engineering, ASCE 2003;130(8):830–41. 521–31. https://doi.org/10.1016/j.jcsr.2005.10.001.
https://doi.org/10.1061/(ASCE)1090-0241(2004)130:8(830). [50] Miladi S, Razzaghi MS. Failure analysis of an un-anchored steel oil tank damaged
[39] Kaya ES, Uckan E, O’Rourke M, Karamanos S, Akbas B, Cakir F, et al. Failure during the Silakhor earthquake of 2006 in Iran. Eng Fail Anal 2019;96:31–43.
analysis of a welded steel pipe at kullar fault crossing. Eng Fail Anal 2016;71: https://doi.org/10.1016/j.engfailanal.2018.09.031.
43–62. https://doi.org/10.1016/j.engfailanal.2016.10.004. [51] Rojhani M, Moradi M, Galandarzadeh A, Takada S. Centrifuge modeling of buried
[40] Guarracino F, Walker AC, Giordano A. Effects of boundary conditions on testing of continuous pipelines subjected to reverse faulting. Can Geotech J 2012;49(6):
pipes and finite element modeling. Int J Pres Ves Pip 2009;86(2– 3):196–206. 659–70. https://doi.org/10.1139/t2012-022.
https://doi.org/10.1016/j.ijpvp.2008.09.009. [52] Khaksar RY, Moradi M, Galandarzadeh A. “Response of buried oil and gas pipelines
[41] Lade PV, Cole Jr DA. Multiple failure surfaces over dip-slip faults. Journal of subjected to reverse faulting: a novel centrifuge-finite element approach” Scientia
Geotechnical and Geoenvironmental Engineering, ASCE 1984;110:616–27. Iranica. Transaction A, Civil Engineering 2018;25(5):2501–16.