Seismic Analysis of a Buried Operating Steel Pipeline with

Emphasis on the Equivalent-Boundary Conditions

G. Banushi, Ph.D. 1; and N. Squeglia, Ph.D. 2
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Abstract: It is of fundamental importance in engineering design practice to accurately evaluate the structural response of buried pipelines
subjected to seismic hazards, like landslides, lateral spreading due to liquefaction, and faulting, because these seriously threaten pipeline
safety. The present paper investigates the seismic performance of a buried operating steel pipeline subjected to strike-slip faulting using the
continuum finite element modeling approach and considering appropriate equivalent boundaries representing the interaction with the rest of
the soil–pipeline system. The latter are derived analytically in terms of an axial spring expressed in function of the soil–pipeline nonlinearities,
service loads like the internal pressure and temperature variation, pipeline unanchored length, cross-section area, and the different loading
paths either in tension or compression. The obtained analytical and numerical results give a better insight into the mechanical behavior of the
soil–pipeline system subjected to strike-slip faulting that induces overall tension in the pipeline under different levels of operating loads. The
methodology described in this paper, including the formulation of the equivalent boundaries, can be applied to accurately and efficiently
assess the seismic performance of operating buried pipelines subjected to strike-slip fault movement and similar permanent ground defor-
mations (PGDs). DOI: 10.1061/(ASCE)PS.1949-1204.0000316. © 2018 American Society of Civil Engineers.
Author keywords: Buried pipelines; Seismic performance; Equivalent-boundary spring; Service loads; Finite elements.

Introduction shell-spring, and the more complex continuum models. The beam
on Winkler foundation model is unable to simulate realistic soil
Postearthquake investigations have shown that almost all seismic behavior for large-scale deformations, as well as capturing the pipe-
damage to modern continuous buried steel pipelines was because line’s local instabilities at the section level, such as buckling and
of permanent ground deformation (PGD) such as landslides, ovalization (Odina and Tan 2009; Trifonov 2015). On the other
lateral spreading due to liquefaction, and faulting (Ariman and hand, the shell-spring model realistically simulates the large
Muleski 1981; Liang and Sun 2000; O’Rourke and Ayala 1990; cross-section deformations in the pipeline but inherits the same lim-
O’Rourke and Liu 2012; Vazouras et al. 2010). Therefore, it is itations as the Winkler model concerning the soil–pipe interaction
of fundamental importance in engineering design practice to accu-
(O’Rourke et al. 2016; Saiyar et al. 2016; Xie et al. 2013). Although
rately evaluate the structural response of buried pipelines subjected
the continuum model has addressed many of the aforementioned
to these seismic hazards that seriously threaten pipeline safety.
deficiencies, it nevertheless presents significant disadvantages in
Specifically, the problem of assessing the performance of a
terms of elevated computational demands and the required expertise
straight buried pipeline under PGD such as strike-slip faulting
of the operator to analyze the models for use in routine engineering
has been investigated by many researchers during the last 50 years
by using simple analytical or more complex numerical approaches. applications (C-Core et al. 2009; Wijewickreme 2012; Banushi
The former consider the pipeline crossing the fault trace at an angle 2017; Banushi et al. 2017).
β < 90° so that the ground movement induces a combined tensile Clearly, the advanced numerical models of buried pipelines
and bending behavior in it. Clearly, this response is preferred in crossing active faults are based on the representation of the pipeline
design practice because the tensile strain capacity of the steel pipe as a cylinder of shell or solid finite elements and the surrounding
is larger than its compressive strain capacity (O’Rourke and Liu soil as an inelastic continuum discretized with solid elements. The
2012). Herein, the pipeline is modeled simplistically either as a ca- pipeline–soil interaction is modeled as a contact problem allowing
ble or beam, accounting for the soil–pipe interaction (Newmark and for sliding and separation between the pipeline and the surrounding
Hall 1975; Kennedy et al. 1977; Wang and Yeh 1985; Karamitros soil. The fault is modeled by assuming the soil volume surrounding
et al. 2007; Trifonov and Cherniy 2010; Vazouras et al. 2010, 2012, the pipeline is either a continuous medium, like in Vazouras et al.
2015). On the other hand, this problem has been addressed numeri- (2010, 2012), or a discontinuity at the fault plane where the two soil
cally within the finite element methodology using three main mod- blocks interact through contact (Paolucci et al. 2010).
eling approaches, the simplistic beam on Winkler foundation, the Recently, several researchers have tried to reduce computational
costs by using alternative modeling approaches such as replacing
1 complex finite elements like shell, solid, or surface contact with
Research Associate, Dept. of Civil and Industrial Engineering, Univ. of
Pisa, Largo Lucio Lazzarino 1, 56122 Pisa, Italy (corresponding author). simpler and more efficient elements like equivalent axial springs.
E-mail: gersena.b@[Link] Specifically, Trifonov (2015) analyzed a steel pipeline buried in
2
Assistant Professor, Dept. of Civil and Industrial Engineering, Univ. a shallow trench under strike-slip fault movement by using nonlin-
of Pisa, Largo Lucio Lazzarino 1, 56122 Pisa, Italy.
ear spring elements connected at the opposite nodes of the soil
Note. This manuscript was submitted on March 23, 2017; approved on
October 30, 2017; published online on April 9, 2018. Discussion period blocks to specify an idealized frictionless interaction along the fault
open until September 9, 2018; separate discussions must be submitted plane. Comparisons with the continuum approach described in
for individual papers. This paper is part of the Journal of Pipeline Systems Vazouras et al. (2010, 2012) were performed, concluding that
Engineering and Practice, © ASCE, ISSN 1949-1190. the fault representation influences the stress–strain evolution in

© ASCE 04018005-1 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 the pipeline; nevertheless, this effect was not critical in terms of tensile and bending behavior in the pipeline. A similar situation
limit state predictions. is commonly preferred in engineering design practice because
In the last decades, different researchers have modeled part of the tensile strain capacity of the steel pipe is larger than its com-
the soil–pipe system beyond the large deformation zone either as a pressive strain capacity.
beam on Winkler foundation or more efficiently as an equivalent- The novelty of the present work consists of the detailed
boundary spring suitably connected to the pipe ends. Therefore, the modeling of the physical discontinuity at the fault as a contact in-
length of the pipe–soil model requiring computationally expensive teraction between the two soil blocks and the accurate specification
elements needs to be only a little longer than the curved pipeline of the loads, as well as boundary conditions in the pipeline and
length undergoing large deformations because of the imposed fault soil domain. Moreover, each pipe end is suitably attached to an
movement (Liu et al. 2004a, b, 2014; Erami et al. 2011; Vazouras equivalent-boundary spring that represents the interaction with
et al. 2014, 2015; Kaya et al. 2017). the rest of the soil–pipeline system, behaving like an axial pullout
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Takada et al. (2001) used a beam–shell hybrid model to evaluate test. Compared to the current state-of-the-art, this paper presents the
the bending angle and critical axial strain in the pipeline at fault exact analytical formulation of the equivalent-boundary spring ob-
crossings. Karamitros et al. (2007) used a hybrid shell-beam model tained in function of service loads such as internal pressure and
with nonlinear springs simulating the soil to validate their analytical temperature variation, pipeline inelasticity, as well as different
methodology for assessing the pipeline response subjected to strike- behavior in compression. The proposed methodology allows accu-
slip fault movement. Liu et al. (2008) used a hybrid beam-shell rate and efficient assessment of the seismic performance of the
model to analyze the seismic response in buried steel pipeline at pipeline for different operating conditions and can be used to ana-
fault crossings in terms of axial strains, accounting for the effect lyze any configuration of the soil–pipe system under strike-slip
of pipe internal pressure as well as the fault intersection angle. fault movement.
Gantes and Bouckovalas (2013) adopted a hybrid shell-beam model This paper first describes the methodology adopted to evaluate
to evaluate the effects of seismic induced fault movement on a part the performance of a buried steel pipeline subjected to strike-slip
of the Turkey–Greece–Italy gas interconnector (TGI). faulting at different levels of internal pressure. Second, it presents
Liu et al. (2004a, b, 2014) introduced an equivalent-boundary the analytical formulation of the equivalent boundaries in terms of
method by analytically evaluating the force-displacement relation- an axial spring expressed in function of the system nonlinearities,
ship of the soil–pipe system away from the fault, which was sub- pipeline unanchored length, cross-section area, service loads, and
sequently implemented within the numerical analysis as a nonlinear the loading path either in tension or compression. Subsequently, the
spring applied to the pipe ends. Vazouras et al. (2014, 2015) used a developed equivalent-boundary is implemented within the numeri-
refined continuum model, extending their previous work (Vazouras cal simulations in terms of an axial spring attached to the pipe ends.
et al. 2010, 2012) and accounting for appropriate end effects, to Next, the obtained numerical results are carefully discussed and
investigate the mechanical behavior of buried steel pipelines cross- compared with recent research literature data, giving a better in-
ing oblique strike–slip faults. They presented a closed-form solu- sight into the structural response of the operating pipeline under
tion for the force–displacement relationship of a buried pipeline the imposed strike-slip fault movement. Finally, the concluding
subjected to tension, accounting for a finite and infinite pipeline section highlights the contributions of the present paper to the
length, the elastic deformation of the soil and pipe, and the develop- state-of-the-art research of seismic analysis of buried steel pipe-
ment of sliding at the pipe–soil interface. Kaya et al. (2017) ana- lines, suggesting further important outlooks in relation to the ad-
lyzed the failure behavior of the Thames water pipe subjected to a dressed issues.
right-lateral strike-slip fault offset during the 1999 Izmit earthquake
in Turkey, where the pressurized steel pipeline was subjected to net
compression. They used a similar continuum model as in Vazouras Methodology
et al. (2014, 2015), using equivalent-boundary springs at the pipe
ends that were calculated through a simplified model, assuming an This paper analyzes the seismic response of a straight X65 steel
elastic pipeline and a constant axial soil resistance along its length. grade pipeline with an external diameter and thickness of 36 in.
Unfortunately, the state-of-the-art formulation of the equivalent and 12.7 mm, respectively, buried in sandy soil at a burial depth
boundaries considers only the case of an empty (i.e., nonoperating) measured to the pipe centerline equal to 2.5 m. The pipeline is
pipeline subjected to overall tension. Clearly, this assumption ne- subjected quasistatically to a strike-slip fault displacement at its
glects the effect of the service loads in the pipeline, as well as the midsection, with a crossing angle β ¼ 25°, and the numerical
different behavior in compression and tension that is particularly simulations are performed using the finite element software
important when the pipeline undergoes contraction. This may be ABAQUS (2014), as illustrated in Fig. 1.
problematic when evaluating the seismic performance of buried To optimize computational costs, only a limited region of the
pipelines in operation because the associated structural response soil–pipeline system close to the fault is modeled using the con-
is significantly influenced by the assumed boundary conditions that tinuum approach, whereas the interaction with the rest of the sys-
represent the interaction of the analyzed model with the rest of the tem is suitably modeled with an equivalent-boundary spring, as
system. explained further in a subsequent section.
Moreover, the numerical implementation of the equivalent- For comparison purposes, the assumed pipeline geometry, burial
boundary spring in the finite element model requires particular con- depth, system materials, and soil–pipeline interaction are equal to
sideration, especially in the case of an operating pipeline, that needs those adopted in Vazouras et al. (2010), as described subsequently.
to be taken into account when performing such simulations. The longitudinal and transverse dimensions of the rectangular
To address this problem, this paper analyzes the structural parallelepiped delimiting the soil system are chosen equal to 60
behavior of a buried operating steel pipeline subjected to strike-slip and 20 m, respectively, so that the pipeline response to the imposed
faulting, proposing a rigorous numerical approach that properly fault displacement is not affected by these values (Banushi 2017).
takes into account the geometric and mechanical properties of The soil parallelepiped surrounding the pipeline is divided in two
the pipe-soil system. To illustrate the proposed methodology, a fault antisymmetric parts by the vertical plane containing the fault trace,
inclination angle (β ¼ 25°) was chosen that induces a combined oriented at an angle β ¼ 25° with respect to the pipe normal.

© ASCE 04018005-2 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Continuum finite-element model of the soil–pipeline system subjected to strike-slip faulting

This fault discontinuity is suitably modeled as a surface-to-surface pressure Pmax is inversely proportional to the slenderness of the pipe-
contact interaction characterized by a purely frictional behavior line represented by the diameter-to-thickness ratio (D/t).
with a friction coefficient μ ¼ tanðϕÞ, with ϕ the friction angle The pipeline is modeled as a cylindrical shell, using four-node
of sand. The interaction between the pipeline and the soil is mod- shell elements (S4) available in ABAQUS (2014) that account for
eled as a frictional contact allowing for finite sliding and separation finite membrane strains and arbitrarily large rotations, making them
between the two interfaces. The soil–pipeline friction coefficient suitable for large strain analysis. The soil is discretized using eight-
is assumed equal to 0.3, as also adopted in Vazouras et al. node linear brick continuum elements (C3D8) in the region near the
(2010, 2012). fault and surrounding the pipeline that undergoes large deforma-
Moreover, the steel pipe material model is defined within the tions, whereas the rest of the soil block is modeled with reduced
von Mises plasticity theory with nonlinear hardening. For compari- integration continuum elements (C3D8R). Each end of the shell
son purposes, the adopted engineering stress–strain curve is the pipeline is connected through appropriate constraints to an axial
same as the one used in Vazouras et al. (2010, 2012), which is ob- connector element CONN3D2 used to model the equivalent-
tained from a uniaxial tensile test. Herein, the Young’s modulus boundary spring that represents the interaction with the rest of
of the steel material is assumed to be equal to 210 GPa, and the the system beyond the shell pipe, as described in detail in the next
Poisson’s ratio is 0.3, whereas the nominal yield stress is section.
448.5 MPa, followed by a plastic plateau up to 1.48% strain and The mesh of both the soil and pipeline components is refined in
then by a strain-hardening regime. The soil material consists of the central region, close to the fault trace, in order to better capture
sand and is described within the Mohr–Coulomb constitutive the large deformation behavior of the system. To ensure the robust-
model, characterized by a friction angle equal to 30°, an elastic ness of the contact algorithm and therefore the accuracy of the re-
modulus of 8 MPa, and a Poisson’s ratio equal to 0.3. Additionally, sulting interaction forces, both soil and pipe contact surfaces are
a small artificial cohesion of 5 kPa is assumed to prevent numerical assigned with a similar mesh size that guarantees solution conver-
difficulties, as also adopted in recent research literature (Vazouras gence. Specifically, the 60-m-long pipeline segment at the fault
et al. 2010, 2012; Xie et al. 2013). Moreover, the soil density and crossing is discretized with 24 shell elements around the circum-
earth pressure coefficient at rest are assumed to be equal to ference, and approximately the same mesh size is adopted in the
20 kN=m3 and 0.5, respectively. This choice of soil material is longitudinal direction close to the fault trace where large deforma-
in agreement with the sand properties presented in detailed techni- tions occur. Instead, the surrounding soil is discretized with 32
cal references, including NCHRP (2007) and NRCS (2007). elements around the circumference, whereas the size of the soil
Clearly, any other combination of soil and pipeline material proper- elements in the longitudinal direction is 0.15 m close to the fault
ties could be analyzed using the proposed methodology. trace. Moreover, the longitudinal discretization of the pipeline and
The effect of internal pressure is taken into account by consid- confining soil is coarser toward the external boundaries in order to
ering the case of no pressure, maximum allowed operating pressure reduce computational costs. This procedure permits correct evalu-
Pmax , and an intermediate internal pressure level equal to 0.56Pmax , ation of the large pipe wall deformations, such as cross-section ov-
where Pmax is calculated through the following expression [ASME alization, when the pipeline is subjected to excessive tension forces
B31.4 (ASME 2002); ASME B31.8 (ASME 2007)]: (Banushi 2017; Banushi et al. 2018).

 The numerical simulations for assessing the pipeline perfor-
t mance subjected to strike-slip fault movement are conducted in
Pmax ¼ 0.72 · 2σy ð1Þ
D two steps. First, a geostatic analysis is performed to establish
the initial stress and strain state in the soil–pipeline system, which
where σy = steel pipe material yield stress; t and D = thickness and equilibrates the gravity loading and the pipeline internal pressure
diameter of the pipe, respectively. Clearly, the maximum operating and satisfies the boundary conditions. Specifically, the lateral

© ASCE 04018005-3 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Schematic representation of the pipeline subjected to strike-slip faulting with indication of the “straight length” region delimited by an anchor
point and the “curved length” region

external faces of both soil parts are restrained in the horizontal di- Analytical Evaluation of the Soil–Pipe System
rection, whereas the bottom faces are vertically restrained to pre- Response Away from the Fault
vent rigid body motion. Moreover, the exposed portion of each
fault surface following the imposed ground displacement is suitably The present section first describes the mechanical behavior of the
restrained in the direction perpendicular to the fault plane. This en- soil–pipeline system, focusing on the region away from the fault
sures the continuity of normal interaction along the fault interface, that undergoes axial loading, behaving like an axial pullout test.
preventing any soil instabilities because of gravity that may lead to Second, it presents the definition of the constitutive behavior of
numerical difficulties. Furthermore, to guarantee the force equilib- the pipeline and the pipe–soil interaction. Third, the distribution
rium in the pipeline because of the service loads in geostatic con- of the axial forces in the pipeline is evaluated using the beam
ditions, both ends of the shell pipeline are restrained in the axial on elastic foundation theory for the static friction length and
direction and each free end of the equivalent-boundary springs the force equilibrium for the frictional sliding length. Then, the pipe
is fixed. elongation is assessed by integrating the axial strains along the
Finally, in the second quasistatic step, the lateral external faces pipeline from the anchor point to the considered section. The latter
of the fixed soil part remain restrained in the horizontal direction, are obtained through the pipe axial forces and its constitutive
whereas the lateral external faces of the moving soil part are dis- behavior. Finally, the influence of the service loads on the pipeline
placed horizontally, parallel to the fault plane. Simultaneously, the response under axial loading is investigated and subsequently the
axial restrain at each end of the shell pipeline is removed. The free developed formulations are applied to calibrate the equivalent-
end of the equivalent-boundary spring in the fixed side of the fault boundary spring for numerical analysis purposes.
remains constrained, whereas the free end of the other equivalent-
boundary spring is assigned a displacement vector equal to the axial
Mechanical Behavior of the Soil–Pipe System Away
component of the fault movement.
from the Fault
The total fault displacement of Δf ¼ 4 m is applied quasistati-
cally with an offset rate equal to 0.01 m=s. On each loading step, In order to accommodate the dislocation of its axis in the horizontal
the global equilibrium equations are solved iteratively by the plane, the pipe is bent and stretched within a relatively short curved
Newton-Raphson method, allowing assessment of the pipe and soil length, characterized by a localization of large deformations in
deformation state at every increment. the pipeline and surrounding soil. As schematically illustrated in
With the pipeline subjected to overall tension during the im- Fig. 2, the transverse component of the imposed fault movement
posed fault movement, it is not susceptible to local buckling but is absorbed within the curved length, whereas beyond it, the rela-
only to large tensile strains and cross-section ovalization. tive transverse soil–pipeline interaction is definitely negligible
Therefore, the local buckling limit is not considered in this paper (Karamitros et al. 2011; Liu et al. 2004b). Herein, the pipeline
and the pipeline performance is assessed in terms of the critical remains straight and the soil–pipeline interaction is only axial be-
fault displacement where the following main limit states are cause of the friction reaction. The latter assumes its limit value over
reached: a large length of the pipeline, where the soil slides relative to the
• Longitudinal tensile strain limit of 3%, as recommended within pipe, as schematically illustrated in Fig. 3.
the Eurocode 8, Part 4 (CEN 2006). In fact, large plastic strains Beyond this frictional sliding zone, the axial soil–pipe interac-
make the pipeline more susceptible to material failure with con- tion is static and the pipeline behaves like a beam on elastic foun-
sequent rupture of the pipe wall and pressure boundary loss. dation, with the soil friction reaction decreasing to zero at the
• Excessive ovalization of the pipeline cross-section, characterized anchor point. The latter is defined as the point where the pipe ex-
by the critical ovalization parameter, intended as the ratio of the pansion stops because of the achieved equilibrium with either the
minimum pipe diameter to its initial diameter and assumed to be soil friction resistance or the reaction of local constraints such as
equal to 15% (Gresnigt 1986; Vazouras et al. 2012). pumping stations or flanges. The region between the two anchor

© ASCE 04018005-4 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Schematic representation of the (a) buried pipeline subjected to axial loading; and (b) force equilibrium in the axial direction of the pipe
element of infinitesimal length dx, where the soil’s resistance to the pipe axial movement is modeled by a linear spring with stiffness k

points is defined as the “unanchored length” and is significantly Clearly, the axial force in the pipeline F can be calculated by
larger than the curved length, on the order of 100 pipe diameters multiplying both members of Eq. (2) by the initial transverse area
at both sides of the fault trace. A of the pipeline
Given that the pipeline behavior in the “straight length” region is
solely axial, it is possible to evaluate the correspondent force– F ¼ Fi−1 þ AEi ðε − εi−1 Þ; jεi−1 j ≤ jεj ≤ jεi j; i ¼ 1; 2; : : :
displacement relationship for any pipe section in this zone. ð3Þ

where Fi−1 ¼ A · σi−1 the axial force in the pipe corresponding to a

Constitutive Behavior of the Pipeline and the Pipe–Soil nominal axial stress and strain equal to σi−1 and εi−1 , respectively.
Interaction under Axial Loading The axial soil–pipeline interaction is determined by the fric-
First, the axial constitutive behavior of the pipeline is defined by the tional reaction of the soil f (per unit length of the pipeline) to
engineering strain–stress values ðεi ; σi Þ, either in tension (ε > 0) or the pipeline displacement in the axial direction u, assumed elastic-
in compression (ε < 0), describing a piecewise linear curve repre- perfectly plastic, as schematically illustrated in Fig. 4(b), and
sented analytically by the following formula: expressed by the following formula:

k · uðxÞ; juj ≤ u0 ; x ≥ L
σ ¼ σi−1 þ Ei ðε − εi−1 Þ; jεi−1 j ≤ jεj ≤ jεi j; i ¼ 1; 2; : : : f¼ ð4Þ
 fS ; juj > u0 ; x < L
ð2Þ
where fs = maximum soil friction force per unit length of the pipe-
where Ei ¼ ðσi − σi−1 Þ=ðεi − εi−1 Þ is the slope of the ith segment line; u0 = relative displacement between the soil and the pipeline
constituting the pipe multilinear stress–strain relationship, as sche- when sliding occurs; k ¼ f s =u0 the rigidity of friction interaction
matically illustrated in Fig. 4(a). To effectively present the analyti- at the soil pipeline interface; uðxÞ = pipeline axial displacement
cal procedure, this paper refers to the Appendix for the detailed with respect to the anchor point (x ¼ La ); and L = distance of
calculation of the axial stress–strain response of the pipeline in the point where frictional sliding starts to occur with respect to
function of service loads such as internal pressure and temperature the considered pipe section (x ¼ 0). Evidently, the soil reaction
variation. per unit length of the pipeline f is considered positive if the

Fig. 4. Schematic representation of the axial constitutive behavior of the (a) steel pipeline; (b) pipe–soil interaction

© ASCE 04018005-5 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Evaluation of the axial soil–pipeline interaction: (a) finite-element model; (b) comparison between the axial soil reaction per unit length of the
pipeline assessed through the continuum model and the ALA guidelines (ALA 2001)


pipeline undergoes overall tension; otherwise, it is assumed to be La ; for jΔLj ≤ u0
negative, as indicated by the  sign in Eq. (4). Le ¼ ð5Þ
In the present analysis, the axial pipe–soil interaction is deter- La − L; for jΔLj > u0 ; L ≤ x ≤ La
mined by evaluating the reaction to the pipe displacement in the where La = distance of the anchor point from the reference pipe
axial direction using the continuum model as a prototype, as sche- section (x ¼ 0), as illustrated in Fig. 3.
matically illustrated in Fig. 5(a). The resulting maximum friction First, the pipeline displacement uðxÞ in the static friction zone is
force per unit length of pipe is f s ¼ 31.379 kN=m, and the corre- determined by solving the differential equation of a beam on elastic
sponding displacement at yield is u0 ¼ 0.0095 m. This numerical foundation, taking into account the boundary conditions
response is in good agreement with the prediction of the axial soil
8
reaction per unit length of the pipeline according to the ALA (2001) > e−αðLa −xÞ − eαðLa −xÞ
guidelines, as shown in Fig. 5(b). < ΔL · ; for jΔLj ≤ u0
uðxÞ ¼ e−αLe − eαLe
This observation is consistent with the study of Sarvanis et al. > −αðL −xÞ αðL
−e a −xÞ
: u · e a
; for jΔLj > u0 ; L ≤ x ≤ La
(2016), showing good agreement between the finite-element model, 0
e−αLe − eαLe
the formula for assessing the axial soil reaction per unit length of ð6Þ
the pipeline, and the experimental data of Karimian (2006).
p
where α ¼ ðk=E1 AÞ, with A the cross-section area of the pipe,
E1 = elastic stiffness of the pipe material [Fig. 4(a)], and k = elastic
Evaluation of the Axial Forces along the Pipeline
rigidity of the soil–pipe friction interaction [Fig. 4(b)].
Depending on if the applied pipe displacement at the free end The axial deformation εðxÞ in the static friction zone can be ob-
(x ¼ 0) is less than the relative soil–pipe displacement when sliding tained by differentiating the pipe axial displacement in the Eq. (6)
starts (jΔLj < u0 ) or not (jΔLj > u0 ), the length of the static fric- with respect to the distance x from reference pipe section (x ¼ 0),
tion zone Le results equal as follows:

8
>
> e−αðLa −xÞ þ eαðLa −xÞ
< αΔL · ; for jΔLj ≤ u0
duðxÞ e−αLe − eαLe
εðxÞ ¼ ¼
dx >
> −αðLa −xÞ
þ eαðLa −xÞ
: αu · e ; for jΔLj > u0 ; L ≤ x ≤ La
0 −αLe
e − eαLe
ð7Þ

Instead, the pipe axial force in the static friction zone (juj < u0 ), assuming the pipeline in the elastic range (jFj < jF1 j ¼ Ajσ1 j) in order to
avoid excessive deformation, can be calculated as
8
>
> e−αðLa −xÞ þ eαðLa −xÞ
< Aσ0 þ AE1 αΔL · ; for jΔLj ≤ u0
e−αLe − eαLe
FðxÞ ¼ Aσ0 þ AE1 · εðxÞ ¼ ð8Þ
>
> −αðLa −xÞ þ eαðLa −xÞ
: Aσ  AE αu · e ; for jΔLj > u0 ; L ≤ x ≤ La
0 1 0
e−αLe − eαLe

where σ0 = axial stress at zero deformation (ε ¼ 0) because of the effect of internal pressure and temperature variation, as indicated in
the expression (43b) in the Appendix.

© ASCE 04018005-6 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 When the applied pipe displacement at the free end is less than F − Fi−1
the relative soil–pipe displacement when sliding starts (jΔLj < u0 ), ε¼ þ εi−1 ; jεi−1 j ≤ jεj ≤ jεi j; i ¼ 1; 2; : : :
AEi
both the friction interaction and the pipe material are in the elastic
range and thus the force-displacement relationship at the pipe free ð16Þ
end is linear
The pipe elongation ΔL is straightforward in the case of
AE1 α jΔLj < u0 , where only static friction occurs at the pipe–soil inter-
F ¼ Fðx ¼ 0Þ ¼ Aσ0 þ ΔL; for jΔLj ≤ u0 ð9Þ face, and can be easily obtained from Eq. (9)
tanhðαLa Þ
tanhðαLa Þ tanhðαL Þ
On the other hand, when the applied pipe displacement at ΔL ¼ ðF − F0 Þ ¼ pffiffiffiffiffiffiffiffiffiffiffia ðF − F0 Þ;
the free end exceeds the relative soil–pipe displacement when AE1 α kAE1
sliding starts (jΔLj > u0 ), the axial force at the pipe anchor point for jΔLj < u0 ð17aÞ
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

N a and the point where frictional sliding starts N 0 are given,

respectively, by where F0 ¼ A · σ0 represents the initial axial force in the pipeline
in operation induced by the temperature variation and internal pres-
AE1 αu0
N a ¼ FðLa Þ ¼ Aσ0  ; for jΔLj > u0 ; jN 0 j < jF1 j sure before the application of the fault offset, where the pipeline is
sinhðαLe Þ assumed to be fully constrained at its ends.
ð10Þ Eq. (17a) can be suitably rewritten as

AE1 αu0 ΔL ¼ tanhðαLa Þ · ΔL∞ ; for jΔLj < u0 ð17bÞ

N 0 ¼ FðLÞ ¼ Aσ0  ; for jΔLj > u0 ; jN 0 j < jF1 j
tanhðαLe Þ where L∞ = elongation of the pipeline with an unanchored length
ð11Þ sufficiently long to be considered infinite

Consequently, the axial force along the pipeline in the region 1 1

ΔL∞ ¼ ðF − F0 Þ ¼ pffiffiffiffiffiffiffiffiffiffiffi ðF − F0 Þ; for jΔLj ≤ u0
where friction sliding occurs at the soil–pipeline interface (x ≤ L), AE1 α kAE1
is given by ð17cÞ


AE1 αu0
FðxÞ ¼ N 0  fS · ðL − xÞ ¼ Aσ0   f S · ðL − xÞ; Instead, in the case where friction sliding occurs at the soil–
tanhðαLe Þ pipeline interface (jΔLj > u0 ) and the pipeline is still in the elastic
for jΔLj > u0 ; x<L ð12Þ range (jFj < jF1 j), the elongation of the pipeline ΔL is given by
Z F−N0

Clearly, the axial force F at the pipeline free end when the ap- fS F − f S x − F0
ΔL ¼ u0  dx; for jΔLj > u0 ;
plied displacement ΔL exceeds the relative soil–pipe displacement 0 AE1
at the onset of sliding (jΔLj > u0 ) can be calculated through
jFj < jF1 j ð18Þ
Eq. (12), assuming x ¼ 0
AE1 αu0 where N 0 is given by Eq. (11).
F ¼ Aσ0   f S · ðLa − Le Þ; for jΔLj > u0 ð13Þ Solving the integral in Eq. (18), the elongation of the pipeline
tanhðαLe Þ
ΔL at its free end can be expressed in function of the associated
From Eq. (13), there exists evidently a biunivocal relation axial force F
between the applied force F and the developed static sliding length

1 ðF − F0 Þ2
Le for a given pipeline with unanchored length La subjected to an ΔL ¼  1 − 2
· u0  ;
axial load F at its free end. As the static friction length Le ap- 2tanh ðαLe Þ 2AE1 fS
proaches infinity (Le → ∞), Eq. (13) for calculating the pipeline for jΔLj > u0 ; jFj < jF1 j ð19aÞ
axial force at its free end becomes a linear function of the developed
friction sliding length L
Eq. (19a) can be suitably rewritten as
F∞ ¼ Aσ0  AE1 αu0  f S · L; for jΔLj > u0 ð14Þ
Le →∞

∞ u0 1
ΔL ¼ ΔL ∓ − 1 ; for jΔLj > u0 ð19bÞ
More generally, in the case of an anchor point situated at a finite 2 tanh2 ðαLe Þ
distance La from the pipeline free end subjected to an axial force F,
the developed static friction length Le can be obtained by numeri- where L∞ = elongation of the pipeline with an unanchored length
cally solving Eq. (13). La and associated static sliding length Le sufficiently long to be
considered infinite

Evaluation of the Elongations along the Pipeline u0 ðF − F0 Þ2

ΔL∞ ¼   ; for jΔLj > u0 ; jFj < jF1 j ð19cÞ
The pipeline elongation ΔL at the reference pipe section (x ¼ 0) 2 2AE1 fS
can be evaluated analytically by integrating the axial deformations
along the pipeline from zero (x ¼ 0) to the anchor point (x ¼ La ) Clearly, the force–displacement relationship given in Eq. (19c)
is parabolic.
Z L
a More generally, when the axial force at the considered pipe
ΔL ¼ εðxÞdx ð15Þ section exceeds the (i − 1)th yielding point and approaches the
0
ith (jFi−1 j ≤ jFj ≤ jFi j), the pipeline elongation in the case of in-
where the axial strain in the pipeline ε results from Eq. (3) finite unanchored length (La → ∞) can be calculated as follows:

© ASCE 04018005-7 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 Z
 Clearly, the evaluated pipeline elongation in function of the ax-
F − f S x − Fi−1
F−Fi−1
∞
ΔL∞
fS
ΔL ¼ i−1  þ εi−1 dx ial force is conceptually similar to the formulation developed by
0 AEi
Kennedy et al. (1977) for assessing the total available elongation
1 in the pipeline in function of the axial stress. Kennedy et al. (1977)
¼ ΔL∞
i−1  ½F2 þ 2ðAEi εi−1 − Fi−1 Þ
2fS AEi used the inelastic Ramberg-Osgood stress–strain relationship to re-
· F þ ðF2i−1 − 2AEi εi−1 Fi−1 Þ for jFj ≥ jF1 j ð20Þ present the pipe material. Evidently, the equivalent-boundary
spring formulation developed in this paper has the additional ad-
where ΔL∞ i−1 = elongation of the pipeline when the axial force is
vantage of considering any type of material stress–strain relation-
equal to Fi−1 in the case of infinite unanchored length (La → ∞) ship, defined as a piecewise linear curve, unlike Kennedy et al.
(1977), which is limited to the Ramberg-Osgood relation.
u0 ðF1 − F0 Þ2
ΔL∞
1 ¼   ; for i ¼ 2; F ¼ F1 ð21aÞ
2 2AE1 fS
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Influence of the Service Loads on the Pipeline

Response under Axial Loading
1
ΔL∞ ¼ ΔL∞
i−2  ½F2 þ 2ðAEi−1 εi−2 − Fi−2 Þ
i−1
2fS AEi−1 i−1 Pipeline elongation corresponding to the achievement of the first
yielding can be evaluated using Eqs. (22b) and (24b) in the pre-
· Fi−1 þ ðF2i−2 − 2AEi−1 εi−2 Fi−2 Þ; i>2 ð21bÞ
vious paragraph


u0 1
Instead, in the case of finite unanchored length, the pipeline ΔL1 ðLa Þ ¼ ΔL∞
1 ∓ − 1 ð25Þ
elongation is evaluated as in Eq. (19b), which is practically valid 2 tanh2 ½αLe1 ðLa Þ
for any jΔLj > u0 :
Summarizing, the force–displacement relationship of the with
equivalent-boundary spring is given by expressions in Eqs. (17b)
and (19b) u0 AE1 2
ΔL∞
1 ¼   ε ðP Þ ð26Þ
2 2f S 1 i
ΔL ¼ tanhðαLa Þ · ΔL∞ ; for jΔLj ≤ u0 ð22aÞ

 where Le1 ðLa Þ represents the static friction length when the first
u0 ∞ 1
ΔL ¼ ΔL ∓ −1 ; for jΔLj > u0 ð22bÞ yielding is reached in the pipeline with unanchored length La ;
2 tanh2 ðαLe Þ ΔL∞ 1 = pipe elongation at first yield in the case of infinite unanch-
ored length (La → ∞); and ε1 ðPi Þ = first nominal yielding strain in
where ΔL∞ is given by Eqs. (17c), (19c), and (20)
function of the service loads, such as the internal pressure Pi , as
ðF − F Þ indicated in Eq. (43e) of the Appendix:
ΔL∞ ¼  pffiffiffiffiffiffiffiffiffiffiffi0 ; for jΔLj ≤ u0 ð23aÞ
kAE1 0 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
1−2ν D 3 D 2 2
@ð 4 · t ÞPi  ð1 þ ε1 Þ2 σ21 − 16 ð t Þ Pi A
ε1 ðPi Þ ¼ exp −1
u0 ðF − F0 Þ2
ΔL∞ ¼   ; for jΔLj > u0 ; jFj ≤ jF1 j E
2 2AE1 fS
ð27Þ
ð23bÞ

1 where ε1 and σ1 = nominal strain and stress, respectively, at first

ΔL∞ ¼ ΔL∞
i−1  ½F2 þ 2ðAEi εi−1 − Fi−1 Þ yield of the pipeline material in the case of no internal pressure; ν =
2fS AEi
Poisson’s ratio of the pipe material; and D and t = internal diameter
· F þ ðF2i−1 − 2AEi εi−1 Fi−1 Þ; jFj > jF1 j; i>1 and thickness of the pipeline, respectively. The pipeline elongation
ð23cÞ capacity is maximum in the case of unanchored length La suffi-
ciently long to be considered infinite, as evident from Eq. (25).
where ΔL∞ The value of internal pressure Pi that maximizes the axial yield-
i−1 is given by Eq. (21).
Clearly, Eq. (23) can be reformulated in function of the axial ing strain ε1 ðPi Þ and consequently the pipeline elongation at first
strains, taking into account Eq. (3), as demonstrated thoroughly yield ΔL∞ 1 in the case of infinite unanchored length can be evalu-
in Banushi (2017) ated by solving the derivative of Eq. (27) set equal to zero. On the
one hand, if the applied displacement is in contraction (ΔL < 0),

ΔL∞ ¼  αε the magnitude of the axial strain evaluated through Eq. (27) is a
; for jΔLj ≤ u0 ð24aÞ decreasing function of the internal pressure Pi , like the pipe elon-
F ¼ F0 þ AE1 ðε − ε0 Þ
gation at first yield ΔL∞ 1 . On the other hand, if the applied dis-
8 placement is in elongation (ΔL > 0), the axial strain at first
< ΔL∞ ¼  u0  AE1 ε2 yield and the associated elongation ΔL∞ 1 increase monotonically
2 2f S ; for jΔLj > u0 ; jFj ≤ jF1 j ð24bÞ up to a level of internal pressure Pi given by
:
F ¼ F0 þ AE1 ðε − ε0 Þ


8 2ð1 − 2νÞ 2t
Pi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1 þ ε1 Þ · σ1 ≈ 0.36 · Pmax ð28Þ
>
< ΔL∞ ¼ ΔL∞  AEi ðε2 − ε2 Þ 9 þ 3ð1 − 2νÞ D 2
i−1
2f S i−1
;
>
: F ¼ F þ AE ðε − ε Þ
i−1 i i−1 where Pmax = maximum allowable internal pressure given by
Eq. (1). Beyond this point, the axial strain at first yield and the
for jΔLj > u0 ; jFj > jF1 j; i>1 ð24cÞ corresponding elongation capacity of the pipeline ΔL∞ 1 evaluated

© ASCE 04018005-8 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Variation of the pipe elongation at first yielding in function of the internal pressure for different values of the unanchored length La and
elongation paths, either positive (ΔL > 0) or negative (ΔL < 0)

through Eqs. (26) and (27), respectively, decrease monotonically Aσ1 0.036 m2 · 448.5 MPa
with increasing levels of internal pressure. La ¼ ¼ ¼ 514.2 m ð29Þ
fs 31.379 kN=m
This latter observation is consistent with the results of Dadfar
et al. (2015), showing that the bending capacity of buried pipelines
depends on the level of internal pressure, increasing under low to Calibration of the Equivalent-Boundary Spring for
moderate internal pressures and drastically decreasing under high Numerical Analysis Purposes
internal pressure. This paper considers a pipeline unanchored length sufficiently long
The relationship between the pipe elongation at first yield ΔL1 to be considered infinite; therefore, the force–displacement relation-
and the level of internal pressure Pi =Pmax for different values of the ship of the equivalent-axial spring is calculated using Eq. (23)
unanchored length La , in the situation where the pipeline is sub- for each level of internal pressure, as illustrated graphically in Fig. 7.
jected to elongation (ΔL > 0) as well as contraction (ΔL < 0), is In the plastic range, the axial stiffness of the equivalent-
illustrated in Fig. 6. boundary spring decreases significantly; therefore, its force-
Herein, the pipeline cross-section area is A ¼ 0.036 m2 , displacement relationship is mainly characterized by its elastic
whereas the friction force per unit length of pipe is fs ¼ behavior (jFj < jF1 j). The latter is defined by the value of the axial
31.379 kN=m and the corresponding displacement at yield is force F0 at zero elongation and the values of the axial force F1 and
u0 ¼ 0.0095 m, as described previously. corresponding elongation ΔL1 at the first yielding in function
Clearly, the pipeline elongation capacity at first yield increases of the internal pressure in tension and compression. Within this in-
with the increase of the unanchored length La , remaining practi- terval, the force–displacement curve is linear for values of pipe
cally constant for unanchored length values larger than 500 m. elongation less than the relative soil–pipe displacement at the onset
Therefore, in order to optimize the pipeline deformation capacity of friction sliding u0 and thereafter parabolic, as results from
at the fault crossing, a minimum unanchored length of approxi- Eqs. (23a) and (23b). These two equations are sufficient for prop-
mately 500 m should be used. This conclusion agrees with the sim- erly calculating the equivalent-boundary spring, provided that the
plified formulation often adopted in engineering design practice pipe segment beyond the region discretized with the finite element
(IITK-GSDMA 2007) remains in the elastic range, as also assumed in recent research

© ASCE 04018005-9 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Force–displacement relationship of the equivalent-boundary springs for different levels of internal pressure Pi and loading path, either in
tension (ΔL > 0) or compression (ΔL < 0)

publications (Liu et al. 2004a, b, 2014; Vazouras et al. 2014, 2015; The calculated equivalent-boundary spring is implemented in
Kaya et al. 2017). In this case, the evaluation of the inelastic part of ABAQUS (2014) finite-element software for numerical analysis
the equivalent-boundary spring given by Eq. (23b) could be purposes. This modeling procedure permits largely reducing the
avoided, further simplifying the analytical calculations. needed memory and computation time of the calculator compared
Clearly, the pipeline elongation capacity is inversely proportional to the one where the entire length of the pipeline is modeled with
to the axial soil–pipe interaction force at sliding f S . Therefore, in nonlinear shell elements and the surrounding soil with solid ele-
order to optimize the pipeline structural performance, friction with ments (e.g., Özcebe et al. 2015).
the surrounding soil should be minimized, for example, by using
loose and lightweight backfill or adopting appropriate coating or
wrapping, as recommended within current guidelines (CEN 2006; Structural Response of the Pipeline for Different
IITK-GSDMA 2007). Furthermore, the pipeline elongation capacity Levels of Internal Pressure
is directly proportional to its cross-sectional area A, suggesting adop-
tion of a thicker pipe along the region susceptible to large ground This paragraph describes the mechanical behavior of the buried
deformations, as prescribed in Eurocode 8, Part 4 (CEN 2006). pipeline for different levels of internal pressure Pi . Herein, the

Fig. 8. Deformation of the pipeline–soil system at the spring-line level for a fault displacement Δf ¼ 2.70 m

© ASCE 04018005-10 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 Table 1. Fault Displacement Capacity (in Meters) Corresponding to internal pressure Pi considered. Additionally, for comparison pur-
Various Performance Criteria for Different Levels of Internal Pressure Pi poses, the results reported in Vazouras et al. (2012) and Karamanos
Local Ovalization Strain et al. (2015) that basically consider analogous mechanical charac-
Finite element model Pi (Pmax ) buckling 15% 3% teristics of the soil–pipe system are indicated.
Continuum 0 None 2.70 2.90 The peak compressive strains occur for a fault offset of approx-
Vazouras et al. (2012) 0 None 0.84 0.62 imately 1.0 m, subsequently decreasing until there are only tensile
Continuum 0.56 None None 2.94 strains in the pipe for fault displacements larger than 2.0 m, as shown
Karamanos et al. (2015) 0.56 None None 2.15 in Fig. 9(a). Instead, the tensile longitudinal strains continuously in-
Continuum 1.0 None None 2.62 crease with the fault offset and the 3% performance limit is reached
for a fault displacement of 2.90 and 2.94 m, respectively, for the
nonoperating and operating pipelines with internal pressure equal
to 0.56Pmax . This limit state is reached earlier for a fault displace-
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

developed compressive strains result mainly in the elastic range,

thus avoiding the risk of local buckling. Hence, the continuum ment equal to 2.62 m in the pipeline operating with maximum in-
model described in the “Methodology” section is suitably used ternal pressure Pmax because of the detrimental effect of the high
to analyze the pipeline structural response, permitting accurate internal pressure. As observed previously, high levels of internal
and efficient evaluation of the seismic performance in terms of pressure anticipate yielding in the pipeline, diminishing its perfor-
the 3% tensile strain and 15% ovalization limit. mance in terms of critical fault offsets corresponding to the achieve-
In order to accommodate the lateral offset of its axis, the pipe- ment of the 3% tensile strain. Clearly, all models exhibit a similar
line is bent within the curved length and stretched longitudinally in pattern of tensile longitudinal strains in function of the fault offset,
tension because of the arch-length effect, following the applied which increase significantly for fault displacements larger than
fault displacement, as illustrated in Fig. 8. 2.50 m as the plasticity spreads to the whole pipeline’s curved length.
Table 1 presents the fault displacement capacities of the pipeline As shown in Table 1, the pipeline performance in terms of criti-
corresponding to the different performance criteria for each level of cal fault displacement corresponding to the achievement of a

Fig. 9. Evolution of the strains in the pipeline in function of the fault offset for different levels of Pi : (a) peak tension and compression longitudinal
strains; (b) peak axial strains; (c) peak bending strains

© ASCE 04018005-11 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 specific limit state is significantly lower in Vazouras et al. (2012), increasing the bending strains in the pipeline, as also noted in
that considered fixed boundaries at the two pipeline ends 30 m Kennedy et al. (1977).
away from the fault. This is straightforward, considering that a Evidently, for large fault offsets, the maximum axial strain is
shorter unanchored length induces additional axial forces in the larger than the maximum bending strain because of the relatively
buried pipeline, diminishing its capacity to sustain fault movement. large positive fault inclination angle (β ¼ 25°); thus, the peak
Although more conservative, the results obtained in Karamanos longitudinal strain follows the trend of the peak axial strain.
et al. (2015) using a refined model for infinitely long pipelines bet- This observation is consistent with the centrifuge testing data of
ter approximate the performance of the pipeline with internal pres- Ha et al. (2008).
sure equal to 0.56Pmax evaluated in the present paper. Moreover, The variation of the longitudinal strain at the most stressed gen-
the representation of the discontinuity at the fault interface as well erator of the pipeline as well as the axial and bending strains along
as the width of the soil prism may affect lateral soil–pipe interaction the pipe axis for different values of the fault displacement are illus-
and the associated bending moment in the pipeline, leading to a trated in Figs. 10(a–c), respectively.
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

different performance evaluation between the aforementioned Evidently, the maximum bending strains localize around two
models. inflection points, situated antisymmetrically at a distance of
Furthermore, the axial strains, calculated as the average of the approximately 5.0 m with respect to the fault, and beyond them
longitudinal strains at the opposite springlines, follow a similar decrease rapidly to zero. For the smaller value of the fault displace-
evolution pattern to the previously mentioned longitudinal strains, ment, the axial strains increase linearly with the distance from the
increasing significantly beyond a fault displacement of approxi- fault trace until two significant peaks develop at the two inflection
mately 2.50 m, as shown in Fig. 9(b). points associated with the maximum bending strain, as shown in
On the other hand, the bending strains are calculated as one-half Fig. 10(b). These local peaks are produced by the interaction
of the difference between the longitudinal strains at the opposite between bending and axial strains as a result of the elastoplastic
springlines. Generally, the bending strains increase until the response of the pipe section, as noted in Karamitros et al. (2007).
widespread advancement of plasticity for a fault displacement of In fact, once the maximum longitudinal strain exceeds the yielding
2.50 m, remaining approximately constant thereafter, as shown limit, axial strains increase locally so that the integral of the
in Fig. 9(c). Evidently, the internal pressure has the effect of corresponding longitudinal stresses is equal to the continuously

Fig. 10. Variation of the strains along the pipeline axis for different values of fault displacement Δf and different levels of internal pressure Pi :
(a) longitudinal strains at the most stressed generator of the pipeline wall; (b) axial strains; (c) bending strains

© ASCE 04018005-12 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. (a) Variation of the ovalization parameter along the unpressurized pipeline for different values of fault displacement; (b) evolution of the pipe
ovalization parameter in function of the fault offset

increasing axial force due to the fault displacement. Instead, for pressure exerted on the pipe by the lateral component of the soil
larger fault offsets, an increasing length of the pipeline around movement. Beyond the curved length, where the pipeline is not
the two inflection points undergoes extensive plastic deformation, subjected to large lateral passive soil pressures, the axial pipeline
and the inelastic axial strains increase progressively toward the interaction is constantly equal to the longitudinal soil resistance in
intersection with the fault trace where their maximum value is geostatic conditions. The latter slightly increases with the level of
reached, as illustrated in Fig. 10(b). Clearly, the pipe axial and internal pressure in the pipeline, as shown in Fig. 12(a).
bending stiffness within the curved length are severely reduced be- The lateral soil–pipeline interaction occurs mainly along the
cause of widespread plasticity, and the pipe behaves like a cable. curved length of the pipeline within the two inflection points, be-
Therefore, the soil lateral pressure is resisted mainly by the pipe yond which it decreases rapidly to zero. Within the large deforma-
tension force rather than by the developing bending moment, as tion zone, the lateral force per unit length in the pipeline increases
noted in Kennedy et al. (1977). with the level of internal pressure as the latter exerts an additional
The ovalization of the unpressurized pipeline decreases progres- outward force in the pipeline because of its curvature, as originally
sively from the fault trace where it assumes the maximum value start- observed in Kennedy et al. (1977).
ing from small fault offsets, as shown in Fig. 11(a). Before the The evolution of the maximum pipe axial force, shear force, and
inelasticity spreads at the fault trace, the ovalization increases locally bending moment in function of the fault offset for different levels of
at the inflection points because of the Brazier effect (Revie 2015) pipe internal pressure is illustrated, respectively, in Figs 13(a–c).
associated with the significant bending strains and the reduced axial The axial force along the pipeline increases with the level of inter-
stiffness. Thereafter, the ovalization increases toward the fault trace nal pressure, reaching its maximum value in the proximity of the
because of the widespread plasticity and the predominant effect of fault. The peak axial force increases parabolically with the fault
the lateral soil pressure, reaching the 15% ovalization limit for a fault offset up to approximately 2.50 m, when the inelastic deformations
displacement equal to 2.70 m, as shown in Fig. 11(b). extend to the pipe section at the fault trace from the two inflection
The soil reaction to the pipeline movement in the local axial and points, as shown in Fig. 13(a). Hereafter, the pipe axial stiffness
lateral direction, for fault displacement Δf ¼ 1.5 m, is indicated, decreases abruptly, meaning that the whole curved length under-
respectively, in Figs. 12(a and b). Clearly, the axial soil reaction goes inelastic deformation, and consequently the pipeline tends
increases within the curved length because of the higher lateral soil to behave like a cable.

© ASCE 04018005-13 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Variation of the soil reaction the along the pipeline for a fault displacement Δf ¼ 1.5 m and different levels of internal pressure Pi : (a) in the
pipe axial direction; (b) in the pipe lateral direction

Moreover, the maximum bending moment in the pipeline efficiently represented with an equivalent-boundary spring con-
increases until plasticity in terms of axial strain starts at the nected at each pipe end. Compared to the current state-of-the-
two inflection points for a relatively low fault offset, equal to ap- art, this paper presents the exact analytical formulation of the
proximately 0.6 m. Hereafter, the bending moment decreases equivalent-boundary spring obtained in function of the service loads
progressively until the inelastic deformations extend to the whole such as the internal pressure and temperature variation, the pipeline
pipe curved length, when the fault displacement reaches approxi- inelasticity, and the different behavior in compression. The pro-
mately 2.50 m, remaining mainly constant afterwards, as shown in posed methodology allows accurate and efficient assessment of
Fig. 13(c). The bending moment backbone curve results from the the seismic performance of the pipeline for different operating
material and geometric nonlinearities of the pipe–soil system that conditions and can be suitably used to analyze any configuration
are accurately taken into account in the present analysis. Further- of the soil–pipe system under strike-slip fault movement.
more, the variation of the peak shear force in the pipeline as a func- The obtained analytical and numerical results are compared
tion of the fault displacement is consistent with the evolution of the with recent research literature data, giving a better insight into
peak bending moment, as illustrated in Fig. 13(b). the mechanical behavior of the operating pipeline subjected to
strike-slip fault movement.
Evidently, the pipeline response is significantly influenced by
Conclusions the assumed boundary conditions that need to be accurately con-
sidered in the numerical analysis. Moreover, the representation of
The present paper presents an advanced methodology for evaluat- the discontinuity at the fault interface as well as the width of the soil
ing the seismic response of buried operating steel pipelines sub- prism may affect the lateral soil–pipe interaction and associated
jected to strike-slip fault movement. Some innovative modeling bending moment in the pipeline that ultimately influences its seis-
aspects are introduced, such as the representation of the physical mic performance.
discontinuity at the fault as a contact interaction between the Clearly, the pipeline elongation capacity is directly propor-
two soil blocks, the accurate specification of the loads, and the tional to its cross-sectional area A and inversely proportional to
boundary conditions on the pipeline and soil domain. the soil–pipe friction force f S , suggesting adoption of thicker
To optimize computational costs, only the curved length pipes and minimization of friction with the surrounding soil at
of the pipeline–soil system undergoing large deformations is the fault-crossing area. Furthermore, the anchor points such as
modeled, whereas the interaction with the rest of the system is pumping stations or flanges should be positioned far away enough

© ASCE 04018005-14 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Evolution of the pipeline loads in function of the fault offset for different levels of internal pressure Pi : (a) peak axial force; (b) peak shear
force; (c) peak bending moment

Fig. 14. Representation of the pipeline response during the small, intermediate, and large fault offset stages, in terms of (a) peak bending moment M;
(b) axial force N; (c) deformations ε

from the fault trace because they exert additional axial forces on monotonically with the fault offset, as well as the associated
the buried pipeline, diminishing its capacity to sustain the fault bending and axial strains that are both important in this phase,
offset. as schematically illustrated in Fig. 14. Herein, the peak bending
Specifically, three stages can be individuated to describe the strain εb exceeds the peak axial strain εa so that there is nonzero
pipeline response for different levels of fault offset, as also reported net longitudinal compressive strain ε− . Moreover, the ovaliza-
in O’Rourke and Liu (2012) for pipeline oriented perpendicularly tion in the unpressurized pipeline decreases progressively from
to the fault trace: the fault trace where the maximum value is reached because of
1. For small fault offsets, the pipeline is mainly in the elastic the increasing lateral soil pressure from the inflection points
range. Both the peak bending moment and axial force increase toward the fault trace;

© ASCE 04018005-15 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 2. For intermediate fault offsets, inelasticity in terms of axial result of the Poisson effect. If the pipeline is sufficiently long
strains develops at the two inflection points, locally decreasing for the pipe–soil friction to fully restrain the pipe, or attached to
its axial and bending stiffness. Herein, the peak axial strain εa fixed objects that tend to anchor it to the ground, the longitudinal
tends to increase more than the peak bending strain εb , and con- contraction because of internal pressure is prevented and longitu-
sequently, the net longitudinal compressive strain ε− approaches dinal tensile stress arise in the buried pipeline. Secondly, the tem-
zero. After the onset of plasticity in the pipeline, the bending perature increase produces an expansion of the pipeline in all
moment decreases progressively, whereas the peak axial force directions. If longitudinal expansion is prevented as a result of
continues to increase parabolically with the fault offset until the axial constraint provided by the pipe–soil friction, longitudinal
the inelastic deformations extend to the whole pipe’s curved compressive stress will be induced in the pipe. Therefore, the hoop
length. In this phase, the ovalization of the unpressurized pipe- stress σy and the longitudinal stress σx in a fully restrained operat-
line may increase locally at the inflection points because of the ing pipeline can be calculated by ALA (2001) and Palmer and
King (2008)
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Brazier effect associated with significant bending strains and re-

duced axial stiffness;
σy ¼ pi D=2t ð30Þ
3. For large fault offsets, the inelasticity spreads to the whole pipe’s
curved length, causing an abrupt decrease of its axial stiffness σx ¼ νσy − EαðT 2 − T 1 Þ ð31Þ
and consequent increase of the axial strains, whereas the bend-
ing strains remain constant. Herein, the pipeline response is sig- where pi = pipe internal pressure; D = pipe diameter; t the pipe
nificantly influenced by the material strain hardening behavior thickness; ν = Poisson ratio of the steel material; E = Young’s
because a greater inelastic stiffness improves its performance in modulus of the steel material; T 2 = service temperature; T 1 =
terms of axial strains and ovalization. Moreover, the pipe bend- installation temperature; and α = coefficient of thermal
ing stiffness within the curved length is severely reduced expansion.
because of the widespread plasticity, and the pipe behaves like With the third principal stress in the radial direction generally
a cable. Therefore, the soil lateral pressure is resisted mainly neglected, the pipe wall stress-state results are biaxial, with both
by the pipe tension force rather than by developing bending circumferential and longitudinal stress components, which leads
moment, as noted in Kennedy et al. (1977). In this phase, the to plane stress conditions. In order to evaluate how near the com-
ovalization of the unpressurized pipeline increases significantly bined stress approaches yielding, the equivalent stress σeq , based
toward the fault trace because of the widespread plasticity and on the Von Mises plasticity theory, is used, which is given for plane
predominant effect of the lateral soil pressure. stress conditions by the following expression:
Moreover, the pipeline elongation capacity is different in ten- qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sion and compression and depends on the level of internal pressure, σeq ¼ σ2x þ σ2y − σx σy ð32Þ
being minimum for the pipeline operating with maximum internal
pressure. Therefore, high levels of internal pressure anticipate Modeling the plastic behavior of the pipe steel material within
yielding in the pipeline, diminishing its performance in terms of the Von Mises plasticity with isotropic hardening, the yield surface
critical fault offset corresponding to the achievement of the tensile remains the same shape but expands with increasing stress in func-
limit state. tion of the equivalent plastic strain εpeq . Hence, the yield surface for
Clearly, pipeline ovalization is accentuated by inelastic longitu- plane stress conditions and isotropic hardening is given by the fol-
dinal straining and lateral soil pressure following the strike-slip lowing equation:
fault movement. Conversely, higher levels of internal pressure im- qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
prove pipeline performance in terms of cross-section ovalization, fðσx ; σy Þ ¼ σ2x þ σ2y − σx σy − σe ðεpeq Þ ¼ σeq − σe ðεpeq Þ ¼ 0
contrasting the flattening effect of lateral soil pressure.
ð33Þ
Evidently, the relation between the pipe cross-section ovaliza-
tion, bending moment, and external soil pressure during fault where σe ðεpeq Þ = yield stress, which is a function of the equivalent
movement deserves further attention because this phenomenon plastic strain εpeq .
is not sufficiently considered within the seismic engineering design Herein, the equivalent plastic strain εpeq is related to the equiv-
practice of buried pipelines (Konuk et al. 1999). Furthermore, for alent stress σeq by the same plastic stress–strain relationship (flow
large displacements, the performed simulations exhibited severe curve) obtained from the data of a simple tensile test. The latter are
mesh distortion and associated numerical difficulties. This issue generally represented in terms of engineering strain–stress values
has also been noted in C-Core et al. (2009), suggesting the need ðεi ; σi Þ and describe a piecewise linear curve, as indicated in
to develop efficient means to reformulate the model mesh for large Eq. (2):
deformation analysis.
σ ¼ σi−1 þ Ei ðε − εi−1 Þ; jεi−1 j ≤ jεj ≤ jεi j; i ¼ 1; 2; : : :
ð34Þ
Appendix. Axial Constitutive Behavior of the
Operating Pipeline where Ei ¼ ðσi − σi−1 Þ=ðεi − εi−1 Þ is the slope of the ith
segment constituting the pipe’s multilinear nominal stress–strain
A pipeline in operation is subjected to temperature differentials and relationship.
internal pressure that induce the development of longitudinal stress In order to be used within plasticity theory, the pipe material’s
as well as hoop stress in the pipe. The longitudinal stresses are constitutive relationship has to be defined in terms of true stress σt
mainly caused by the Poisson effect following the pipeline circum- and true strain εt , which are related to the engineering stress σn and
ferential expansion because of internal pressure as well as temper- strain εn through
ature variation.
In fact, when subjected to internal pressure, hoop stresses de- σt ¼ σn ð1 þ εn Þ ð35Þ
velop in the pipeline associated with a circumferential extension,
which in turn induces a longitudinal contraction of the pipe as a εt ¼ lnð1 þ εn Þ ð36Þ

© ASCE 04018005-16 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 stress–true strain relationship, as schematically illustrated in
Fig. 16(a).
Moreover, the relationship between the tangent modulus of the
true stress–true strain curve Eoi and the plastic modulus H oi of the
true stress–true plastic strain curve (flow curve) is given by

1 1 1
¼ þ o; for i > 1 ð38Þ
Eoi E Hi

where E = Young’s modulus of the pipe material.

Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Using the Prandtl-Reuss equation (Kelly 2013), in the case of

plane stress conditions, the true axial strain increment dεx is related
to the true axial stress increment dσx by
Fig. 15. Schematic representation of the Von Mises yield surfaces un-
der isotropic hardening and the pipe stresses following axial loading in

the case of no internal pressure (pi ¼ 0) and internal pressure (pi ≠ 0) 1 dεpeq 1
dεx ¼ ðdσx − νdσy Þ þ σx − σy ð39Þ
E σeq 2

The evolution of the yield locus for the pipeline under increasing Introducing the plastic modulus Hi ¼ ∂σeq =∂εeq , Eq. (39) can
axial loading is schematically illustrated in Fig. 15. Herein, two be written as follows:
loading scenarios are represented:
1. Pipeline with no internal pressure pi ¼ 0, subjected to increas-

ing axial loading either in tension or compression; and 1 1 dσeq 1
dεx ¼ ðdσx − νdσy Þ þ o σx − σy ; σoi−1 ≤ σeq ≤ σoi
2. Pipeline with internal pressure pi ≠ 0, subjected to increasing E Hi σeq 2
axial loading either in tension or in compression. ð40Þ
The response of the pipeline subjected to axial loading, in terms
of the axial stress–strain relationship, for cases of no internal
pressure and internal pressure is schematically illustrated in substituting Eq. (32) into Eq. (40) and observing from Eq. (30) that
Figs. 16(a and b), respectively. Clearly, in the case of a nonoper- σy is constant, results in
ating pipeline (pi ¼ 0), the axial stress–strain response coincides
with the pipe material’s constitutive relationship obtained from
1 1 ðσx − 12 σy Þ2
the data of a simple tensile test, expressed in terms of true stress dεx ¼ dσx þ o · 2 dσ ; σoi−1 ≤ σeq ≤ σoi
σo and true strain εo E H i ðσx þ σ2y − σx σy Þ x
ð41Þ
σo ¼ σoi−1 þ Eoi ðεo − εoi−1 Þ; jεi−1 j ≤ jεj ≤ jεi j; i ¼ 1; 2; : : :
ð37Þ
Eq. (41) can be integrated exactly for σoi−1 ≤ σeq ≤ σoi, leading
where Eoi¼ ðσoi
− σoi−1 Þ=ðεoi
− εoi−1 Þ
the slope (or tangent modu- to the analytical expression for evaluating the true axial strain–
lus) of the ith segment constituting the pipe’s multilinear true stress values ðεx ; σx Þ for an operating pipeline

Fig. 16. Axial stress–strain response in the pipeline subjected to axial loading either in tension or compression for the case of (a) no internal pressure
(pi ¼ 0); (b) internal pressure (pi ≠ 0)

© ASCE 04018005-17 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðσ − σp Þ
εx ¼ εpi−1
þ x o i−1 σpi ¼ σy =2  ðσoi Þ2 − 3σ2y =4; for i ≥ 1 ð43dÞ
Ei
pffiffiffi 

p 
3σ y 2σx − σy 2σi−1 − σy ðσp1 − σp0 Þ
− arctan p ffiffi
ffi − arctan p ffiffi
ffi ; ε1 ¼ εp0 þ ; for σx ¼ σp1 ð43eÞ
2H oi 3σ y 3σ y E
σoi−1 ≤ σeq ≤ σoi ð42Þ ðσpi − σpi−1 Þ
εpi ¼ εpi−1 þ
Eoi
where εpi−1 and σpi−1 represent, respectively, the true strain and true pffiffiffi 
p 
p 
stress value of the (i − 1)th yielding point defining the stress–strain 3σ y 2σi − σy 2σi−1 − σy
− arctan p ffiffi
ffi − arctan p ffiffi
ffi ;
curve of the steel pipeline subjected to axial loading in operating 2Hoi 3σ y 3σ y
conditions, as schematically illustrated in Figs. 15 and 16(b).
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Finally, the response of the operating pipeline under increasing for i > 1 ð43fÞ
axial loading, either in tension or compression, can be analytically
evaluated in terms of axial true strain–true stress values using ðσx − σp0 Þ
εx ¼ εp0 þ ; for σx ≤ σp1 ð43gÞ
Eqs. (30), (33), and (42), as follows: E

σy ¼ pi D=2t ð43aÞ ðσx − σpi−1 Þ

εx ¼ εpi−1 þ
Eoi
pffiffiffi 

p 
σp0 ¼ νσy − EαðT 2 − T 1 Þ ð43bÞ 3σ y 2σx − σy 2σi−1 − σy
− arctan p ffiffi
ffi − arctan p ffiffi
ffi ;
2Hoi 3σ y 3σ y
εp0 ¼ 0 ð43cÞ for σx > σp1 ð43hÞ

Table 2. Discrete Values Defining the Response of the Operating Pipeline under Axial Loading in Terms of Nominal Stress–Nominal Strain
p¼0 p ¼ 0.56pmax p ¼ pmax
F<0 F>0 F<0 F>0 F<0 F>0
i εn σn (MPa) εn σn (MPa) εn σn (MPa) εn σn (MPa) εn σn (MPa) εn σn (MPa)
i¼0 0.0 0.0 0.0 0.0 0.0000 53.1 0.0000 53.1 0.0000 94.9 0.0000 94.9
i¼1 −0.0021 −450.4 0.0021 448.5 −0.0018 −334.6 0.0022 509.9 −0.0014 −198.6 0.0020 513.5
i¼2 −0.0146 −461.9 0.0148 448.5 −0.0136 −344.7 0.0141 509.9 −0.0114 −207.8 0.0121 515.4
i¼3 −0.0476 −521.1 0.0500 472.7 −0.0449 −401.5 0.0474 535.2 −0.0386 −266.1 0.0408 549.6
i¼4 −0.0909 −616.8 0.1000 509.7 −0.0866 −493.5 0.0951 573.3 −0.0762 −358.5 0.0831 597.7
i¼5 −0.1304 −705.3 0.1500 533.3 −0.1249 −577.4 0.1432 596.9 −0.1118 −439.8 0.1266 627.4
i¼6 −0.1667 −788.9 0.2000 547.9 −0.1603 −656.0 0.1914 610.9 −0.1453 −514.4 0.1708 645.6
i¼7 −0.2308 −946.4 0.3000 560.0 −0.2234 −802.4 0.2881 621.2 −0.2059 −650.4 0.2601 660.8

Fig. 17. Engineering axial stress-strain response of the pipeline subjected to axial loading either in tension or compression for the case of no internal
pressure (pi ¼ 0), intermediate internal pressure (pi ¼ 0.56pmax ) and maximum internal pressure (pi ¼ 1.0pmax )

© ASCE 04018005-18 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 where Eoi ¼ ðσoi − σoi−1 Þ=ðεoi − εoi−1 Þ is the slope (or tangent modu- Dadfar, B., El Naggar, M. H., and Nastev, M. (2015). “Ovalization of steel
lus) of the ith segment constituting the pipe multilinear true stress– energy pipelines buried in saturated sands during ground deformations.”
true strain relationship; and H oi = associated plastic modulus that Comput. Geotech., 69, 105–113.
can be calculated using Eq. (38) Erami, M. H., Miyajima, M., and Fallahi, A. (2011). “A state-of-the-art
review on fault crossing buried pipeline analysis.” 6th Int. Conf. on
E · Eoi Seismology and Earthquake Engineering, International Institute of
Hoi ¼ ; for i > 1 ð44Þ
E − Eoi Earthquake Engineering and Seismology, Tehran, Iran.
Gantes, C. J., and Bouckovalas, G. D. (2013). “Seismic verification of high
Last, the nominal strain–stress values ðεn ; σn Þ corresponding to pressure natural gas pipeline Komotini-Alexandropoulis–Kipi in areas
the true stress–strain curve ðεt ; σt Þ calculated through Eq. (43) can of active fault crossings.” Struct. Eng. Int., 23(2), 204–208.
be obtained from Eqs. (35) and (36), as follows: Gresnigt, A. M. (1986). “Plastic design of buried steel pipes in settlement
areas.” HERON, 31(4), 133.
εn ¼ expðεt Þ − 1 ð45Þ Ha, D., et al. (2008). “Centrifuge modeling of earthquake effects on buried
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

high-density polyethylene (HDPE) pipelines crossing fault zones.”

J. Geotech. Geoenviron. Eng., 10.1061/(ASCE)1090-0241(2008)
σn ¼ σt =ð1 þ εn Þ ð46Þ 134:10(1501), 1501–1515.
IITK-GSDMA. (2007). Guidelines for seismic design of buried pipelines,
Using Eq. (43), the discrete nominal stress–strain values Indian Institute of Technology Kanpur, Kanpur, India.
defining the response of the X65 steel grade pipeline under axial Karamanos, S. A., et al. (2015). “Safety of buried steel pipelines under
loading are evaluated in the case of no internal pressure (pi ¼ 0), ground-induced deformations.” RFCS Project Final Rep., RFSR-CT-
intermediate internal pressure (pi ¼ 0.56pmax ), and maximum in- 2011-00027, Brussels, Belgium.
ternal pressure (pi ¼ pmax ), as indicated in Table 2. Moreover, a Karamitros, D. K., Bouckovalas, G. D., and Kouretzis, G. P. (2007). “Stress
plot of the aforementioned engineering stress–strain curves is analysis of buried steel pipelines at strike-slip fault crossings.” Soil Dyn.
shown in Fig. 17. Earthquake Eng., 27(3), 200–211.
Finally, the calculated nominal stress–strain values in function Karamitros, D. K., Bouckovalas, G. D., Kouretzis, G. P., and Gkesouli, V.
(2011). “An analytical method for strength verification of buried steel
of the service loads can be used to determine the equivalent-
pipelines at normal fault crossings.” Soil Dyn. Earthquake Eng., 31(11),
boundary spring in terms of the force–displacement relationship 1452–1464.
expressed through Eq. (22) and (23). Karimian, H. (2006). “Response of buried steel pipelines subjected to
longitudinal and transverse ground movement.” Ph.D. thesis, Univ.
of British Columbia, Vancouver, BC.
Acknowledgments Kaya, E. S., et al. (2017). “Failure analysis of a welded steel pipe at Kullar
fault crossing.” Eng. Fail. Anal., 71, 43–62.
The research described herein is part of the doctoral work of the Kelly, P. (2013). “Solid mechanics lecture notes.” 〈[Link]
corresponding author within the joint International Doctorate in .[Link]/∼pkel015/SolidMechanicsBooks/Part_II
Civil and Environmental Engineering between the University of /[Link]〉 (Feb. 15, 2016).
Florence and the Technical University of Braunschweig. It was sup- Kennedy, R. P., Chow, A. W., and Williamson, R. A. (1977). “Fault
ported with the Regione Toscana Pegaso Doctoral Scholarship that movement effects on buried oil pipeline.” J. Transp. Eng. Div., 103(5),
is greatly acknowledged. 617–633.
Konuk, I., Phillips, R., Hurley, S., and Paulin, M. J. (1999). “Preliminary
ovalisation measurements of buried pipelines subjected to lateral
References loading.” Proc., OMAE99, 18th Int. Conf. on Offshore Mechanics
and Arctic Engineering, Vol. 4, ASME, New York, 113–119.
ABAQUS. (2014). User’s guide, Simulia, Providence, RI. Liang, J., and Sun, S. (2000). “Site effects on seismic behavior of pipelines:
ALA (American Lifeline Alliance). (2001). Guidelines for the design of A review.” ASME J. Pressure Vessel Technol., 122(4), 469–475.
buried steel pipe, FEMA, Washington, DC. Liu, A., Takada, S., and He, Q. (2014). “The failure performance of Thames
Ariman, T., and Muleski, G. E. (1981). “A review of the response of buried water pipeline at fault crossing in Kocaeli earthquake.” 2nd
pipelines under seismic excitations.” Earthquake Eng. Struct. Dyn., European Conf. on Earthquake, Engineering and Seismology, EAEE,
9(2), 133–152. Istanbul, Turkey.
ASME. (2002). “Pipeline transportation systems for liquid hydrocarbons Liu, A., Takada, S., and Hu, Y. (2004a). “A shell model with an equivalent
and other liquids.” ASME B31.4, New York. boundary for buried pipelines under the fault movement.” 13th World
ASME. (2007). “Gas transmission and distribution piping systems.” ASME Conf. on Earthquake Engineering, Indian Institute of Technology
B31.8, New York. Kanpur, Kanpur, India.
Banushi, G. (2017). “Seismic design of buried steel pipelines.” Ph.D. dis- Liu, A. W., Hu, Y. X., Zhao, F. X., Li, X. J., Takada, S., and Zhao, L. (2004b).
sertation 〈[Link] “An equivalent-boundary method for the shell analysis of buried
Banushi, G., Squeglia, N., Thiele, K., and Salvatore, W. (2017). “Finite pipelines under fault movement.” Acta Seismol. Sin., 17(1), 150–156.
element analysis of operating buried pipelines subjected to strike-slip Liu, M., Wang, Y.-Y., and Yu, Z. (2008). “Response of pipelines under fault
fault movement.” 12th Pipeline Technology Conf. 2017, EITEP crossing.” Proc., Int. Offshore and Polar Engineering Conf., ISOPE,
Institute, Hannover, Germany. Mountain View, CA.
Banushi, G., Squeglia, N., and Thiele, K. (2018). “Innovative analysis of NCHRP (National Cooperative Highway Research Program). (2007).
a buried operating pipeline subjected to strike-slip fault movement.” “Synthesis 368—Cone penetration testing—A synthesis of highway
Dyn. Earthquake Eng., 107, 234–249. practice.” Transportation Research Board of the National Academies,
C-CORE, D.G. Honegger Consulting, and SSD Inc. (2009). Guidelines for Washington, DC.
constructing natural gas and liquid hydrocarbon pipelines through Newmark, N. M., and Hall, W. J. (1975). “Pipeline design to resist large
areas prone to landslide and subsidence hazards, Design, Materials, fault displacements.” Proc., U.S. National Conf. on Earthquake Engi-
and Construction Committee of Pipeline Research Council neering, Earthquake Engineering Research Institute, Oakland, CA.
International, Inc., Chantilly, VA. NRCS (Natural Resources Conservation Service). (2007). “Soil properties
CEN (European Committee for Standardization). (2006). “Design of struc- and special geotechnical problems related to stream stabilization
tures for earthquake resistance. 4: Silos, tanks and pipelines.” Eurocode projects.” 〈[Link]
8 EN 1998-4, Brussels, Belgium. (Aug. 8, 2017).

© ASCE 04018005-19 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005

 Odina, L., and Tan, R. (2009). “Seismic fault displacement of buried pipe- Takada, S., Hassani, N., and Fukuda, K. (2001). “A new proposal for sim-
lines using continuum finite element methods.” Proc., ASME 2009 28th plified design of buried steel pipes crossing active faults.” Earthquake
Int. Conf. on Ocean, Offshore and Arctic Engineering, OMAE2009- Eng. Struct. Dyn., 30(8), 1243–1257.
79739, ASME, New York. Trifonov, O. V. (2015). “Numerical stress-strain analysis of buried steel
O’Rourke, M. J., and Liu, X. (2012). “Seismic design of buried and off- pipelines crossing active strike-slip faults with an emphasis on fault
shore pipelines.” 〈[Link] modeling aspects.” J. Pipeline Syst. Eng. Pract., 10.1061/(ASCE)PS
O’Rourke, M. J., and Ayala, G. (1990). “Seismic damage to pipeline: .1949-1204.0000177, 04014008.
Case study.” J. Transp. Eng., 10.1061/(ASCE)0733-947X(1990) Trifonov, O. V., and Cherniy, V. P. (2010). “A semi-analytical approach to a
116:2(123), 123–134. nonlinear stress-strain analysis of buried steel pipelines crossing active
O’Rourke, T. D., Jung, J. K., and Argyrou, C. (2016). “Underground pipe- faults.” Soil Dyn. Earthquake Eng., 30(11), 1298–1308.
line response to earthquake-induced ground deformation.” Soil Dyn. Vazouras, P., Dakoulas, P., and Karamanos, S. A. (2014). “Structural
Earthquake Eng., 91, 272–283. performance of buried steel pipelines crossing strike-slip faults.”
Özcebe, A. G., Paolucci, R., Mariani, S., and Santoro, D. (2015). “A 10th Int. Pipeline Conf., ASME, New York.
Downloaded from [Link] by MISSOURI, UNIV OF/COLUMBIA on 04/14/18. Copyright ASCE. For personal use only; all rights reserved.

Vazouras, P., Karamanos, S. A., and Dakoulas, P. (2010). “Finite element

numerical study on the pressurized gas pipeline-normal fault interaction
analysis of buried steel pipelines under strike-slip fault displacements.”
problem.” 6th Int. Conf. on Earthquake Geotechnical Engineering,
Soil Dyn. Earthquake Eng., 30(11), 1361–1376.
Christchurch, New Zealand.
Vazouras, P., Karamanos, S. A., and Dakoulas, P. (2012). “Mechanical
Palmer, A. C., and King, R. A. (2008). Subsea pipeline engineering,
behavior of buried steel pipes crossing active strike-slip faults.” Soil
2nd Ed., PennWell Corporation, Tulsa, OK. Dyn. Earthquake Eng., 41, 164–180.
Paolucci, R., Griffini, S., and Mariani, S. (2010). “Simplified modeling Vazouras, P., Karamanos, S. A., and Dakoulas, P. (2015). “Pipe-soil inter-
of continuous buried pipelines subject to earthquake fault rupture.” action and pipeline performance under strike-slip fault movements.”
Earthquakes Struct., 1(3), 253–267. Soil Dyn. Earthquake Eng., 72, 48–65.
Revie, R. W. (2015). Oil and gas pipelines: Integrity and safety handbook, Wang, L. R. L., and Yeh, Y. (1985). “A refined seismic analysis and design
Wiley, Hoboken, NJ. of buried pipeline for fault movement.” J. Earthquake Eng. Struct. Dyn.,
Saiyar, M., Ni, P., Take, W. A., and Moore, I. D. (2016). “Response of 13(1), 75–96.
pipelines of differing flexural stiffness to normal faulting.” Géotechni- Wijewickreme, D. (2012). “Role of geotechnical engineering in assuring
que, 66(4), 275–286. the integrity of buried pipeline systems.” Int. Conf. on Sustainable Built
Sarvanis, G. C., et al. (2016). “Soil-pipe interaction models for simulating Environment (ICSBE), Univ. of Moratuwa Library, Moratuwa, Sri Lanka.
the mechanical response of buried steel pipelines crossing active faults.” Xie, X., et al. (2013). “Numerical modeling of buried HDPE pipelines sub-
Proc., Int. Offshore and Polar Engineering Conf., International Society jected to normal faulting: A case study.” Earthquake Spectra, 29(2),
of Offshore and Polar Engineers, Mountain View, CA 412–420. 609–632.

© ASCE 04018005-20 J. Pipeline Syst. Eng. Pract.

J. Pipeline Syst. Eng. Pract., 2018, 9(3): 04018005