Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–

]]]

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering
journal homepage: www.elsevier.com/locate/soildyn

Analysis of buried pipelines subjected to reverse fault motion
Shantanu Joshi a, Amit Prashant a, Arghya Deb b, Sudhir K. Jain c,n
a b c

Department of Civil Engineering, Indian Institute of Technology Kanpur, UP 208016, India Department of Civil Engineering, Indian Institute of Technology Kharagpur, WB 721 302, India Department of Civil Engineering, Indian Institute of Technology Gandhinagar, Ahmedabad 382424, India

a r t i c l e i n f o
Article history: Received 5 July 2010 Received in revised form 7 February 2011 Accepted 8 February 2011 Keywords: Buried pipeline Finite element analysis Nonlinear quasi-static analysis Structural design of pipelines Oil and gas pipelines Earthquake induced ground movements Pipe–soil interaction Pipeline buckling Reverse fault motion

abstract
Presently available simplified analytical methods and semi-empirical methods for the analysis of buried pipelines subjected to fault motion are suitable only for the strike-slip and the normal-slip type fault motions, and cannot be used for the reverse fault crossing case. A simple finite element model, which uses beam elements for the pipeline and discrete nonlinear springs for the soil, has been proposed to analyse buried pipeline subjected to reverse fault motion. The material nonlinearities associated with pipe-material and soil, and geometric nonlinearity associated with large deformations were incorporated in the analysis. Complex reverse fault motion was simulated using suitable constraints between pipe-nodes and ground ends of the soil spring. Results of the parametric study suggest that the pipeline’s capacity to accommodate reverse fault offset can be increased significantly by choosing a near-parallel orientation in plan with respect to the fault line. Further improvement in the response of the pipeline is possible by adopting loose backfill, smooth and hard surface coating, and shallow burial depth in the fault crossing region. For normal or near normal orientations, pipeline is expected to fail due to beam buckling at very small fault offsets. & 2011 Elsevier Ltd. All rights reserved.

1. Introduction Buried steel pipelines with continuous joints are commonly used for transporting oil, gas and water over long distances. Such a pipeline crossing an active fault zone may be subjected to large, abrupt differential ground movement due to the fault rupture. Several major pipeline systems have been identified with the pipelines passing through active fault regions [1]. Reverse faults result from compressional plate tectonic environment and are abundantly present throughout the world. In India, major active faults are of reverse or thrust type and mainly distributed in Kachchh (Western India) and Himalayan frontal (North-western India) regions [2,3]. Some of these reverse faults can potentially produce large fault offset, as high as several metres. Many cases of pipeline damage due to fault rupture have been recorded during recent major earthquakes [4–6]. For example, a case of severe pipeline damage was reported due to the rupture of Chelungpu fault during 1999, Chi-Chi (Taiwan) earthquake [4]. The fault was steep reverse type (total length of about 105 km), and fault offsets of 4–10 m were observed along its length during the earthquake. The damaged portion of this pipeline went through local buckling

Corresponding author. Tel.: +91 79 2397 2574; fax: + 91 792397 2586. E-mail addresses: amit_prashant@yahoo.com (A. Prashant), skjain@iitk.ac.in (S.K. Jain). 0267-7261/$ - see front matter & 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.soildyn.2011.02.003

n

and large section deformations near the fault crossing point. Hence, it is necessary to design the pipeline which can safely accommodate large fault offsets without being ruptured or buckled. Faults are most commonly classified based on the direction of relative slip. Portion of the ground, which remains stationary during the slip is referred to as foot wall, and the other portion that slips over the foot wall is referred to as hanging wall. The hanging wall in normal-slip faults moves downward and in reverse-slip faults upward with respect to the foot wall. A low dip angle (less than 451) reverse fault is called a thrust fault. In strike-slip fault, the slip takes place in the horizontal direction. Response of buried pipeline is significantly influenced by the type of fault motion and orientation of the pipeline with respect to the fault line [7]. In general, a steel pipeline strained in direct tension due to fault rupture can safely accommodate a larger fault offset value compared to when it is strained in direct compression [8,9]. Pioneering work in the analysis of pipeline subjected to fault motion was done by Newmark and Hall [10]. They developed a simplified method for analysis of pipeline subjected to fault motion. This method assumed the pipeline to be subjected to direct tension due to the fault motion and ignored lateral resistance of the soil. Hence, the analysis of the pipeline was performed by assuming it to be a cable deforming in straight line. Kennedy et al. [8] revised the Newmark–Hall method by incorporating bending of the pipeline near the fault crossing point and considering the soil lateral forces. However, the formulae for

Please cite this article as: Joshi S, et al. Analysis of buried pipelines subjected to reverse fault motion. Soil Dyn Earthquake Eng (2011), doi:10.1016/j.soildyn.2011.02.003

as none of those methods can analyse a pipeline subjected to bi-axial bending along with axial compression or predict its subsequent buckling. b and c. the semi-analytical method proposed by Trifonov et al. On the other hand.soildyn. The analysis procedure was based on the assumption of uni-axial bending of pipeline.003 . Besides. the pipeline bending in this case cannot be assumed to take place in a predetermined plane as can be assumed for the case of strike-slip fault. The assumption of uni-axial pipeline bending in the case of strike-slip fault motion is justified by the fact that the problem involves a single crossing angle (b) between the fault-line and the pipeline. if L2 is too large β Anchor point L1. In this study. Based on the Kennedy’s Method. Compressive stresses may cause buckling of the buried pipeline either in the beam mode or in the shell mode. for the case of buried pipeline crossing a reverse fault. or Lo. Consequently. Anchor point ψ Δ Ground surface Deformed profile of pipeline Z Δz Δ Δ Cos (ψ) Original profile of pipeline X’ Anchor point X L2. the pipeline may be subjected to bi-axial bending. et al.2 S. Analysis of buried pipelines subjected to reverse fault motion. Soil Dyn Earthquake Eng (2011). doi:10. The possibility of axial elongation or shortening depends on the orientation of the pipeline with respect to the fault line in plan. one can expect a significant amount of axial shortening in the pipeline. The fault motion in this case depends on both the crossing angles. Similarly. Please cite this article as: Joshi S. the various simplified methods mentioned earlier prove to be inadequate for the analysis of pipeline crossing a reverse fault. [14] is also suitable for only the cases of strike-slip fault and normal-slip fault crossings. or Lo. the present research is focused on analysing the response of buried pipelines subjected reverse fault motion with due considerations to the three dimensional nature of reverse fault motion. Hence. Joshi et al. 1. [12] and simplified analytical method by Karamitros et al. bi-axial pipeline bending and probable pipe buckling. which are present in the horizontal and the vertical plane. The simplified methods for determination of total strain in a pipeline subjected to strike-slip fault motion are based on the assumption that the fault motion would cause uni-axial bending of the pipeline near the fault crossing point accompanied by either axial elongation or axial shortening [11.2011. Wang and Yeh [11] developed a simplified analysis method. the fault dip angle (c) and the pipeline crossing angle (b) are involved (Fig. if L1 is too large Anchor point Y Fault crossing point Fault line Δy Δx Fig.02. two different crossing angles.13].1016/j. Literature review suggests that previous research in the analysis of pipeline subjected to fault motion has been mainly focused on the case of strike-slip fault [8. Thus. respectively. an attempt was made to analyse the response of buried pipelines subjected to reverse fault motion by developing a simple finite element model using 3-D beam elements. due to hanging wall moving towards the foot wall during reverse fault movement. / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] radius of curvature proposed in this method were suitable only case of pipeline being in tension. which could analyse the case of pipeline subjected to direct tension as well as direct compression. the analysis would require simultaneous application of fault offset components in all three major pipeline-directions. Deformation of buried pipeline subjected to reverse fault motion: (a) section and (b) plan. and it was only applicable to the pipeline subjected to strike-slip fault motion. In other words. Recently proposed semi-empirical method by Takada et al. Soil surrounding the pipeline was modelled using nonlinear springs. Hence. [13] are also based on the assumption of uni-axial bending of pipeline and are only applicable to the case of strike-slip fault. 1).11–16].

c are the angles of intersection of the pipeline with the fault.soildyn. Complex 3-D reverse fault motion was simulated by applying non-zero displacements to the soil spring ends through suitable constraints between pipe-nodes and corresponding soil spring ends. backfill properties. The fault offset (D) along the fault plane can be resolved in components along three pipeline directions as follows [17]: Dx ¼ DcosðcÞsinðbÞ Dy ¼ DcosðcÞcosðbÞ Dz ¼ D sinðcÞ ð1Þ ð2Þ ð3Þ 2. X. Dx is resisted by frictional force developed at the pipeline–soil interface over the unanchored length of the pipeline. 1 shows a typical deformation profile of buried pipeline. geometric nonlinearity associated with large deformations. Dx results in axial shortening of the pipeline. respectively. Using this FEM model. pipe surface property. in section and in plan. and transverse vertical directions of the pipeline.02.S. in plan and in section. Please cite this article as: Joshi S. 1) is sufficiently away from the fault crossing point. and Z axes represent longitudinal. when subjected to reverse fault motion. b and Movement of the soil mass along the fault plane results in pipeline–soil interaction in all the three pipeline directions. nonlinear Winkler spring model of the soil. Joshi et al. doi:10. For a reverse fault. Y. and postbuckling behaviour of pipeline (in case of beam buckling). / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] 3 which support the pipeline at discrete points. Geometry of proposed FEM model for buried pipeline. and buckling of the pipeline. et al. transverse horizontal. 2. crossing angle. respectively. The major features incorporated in the model were pipe-material nonlinearity. pipe material and cross-section properties on maximum compressive strain. At the endpoint of this effective Transverse spring Axial spring Fault crossing point Z X Vertical spring Y 10 m 570 m 1m Δy 30 m Δx 30 m Δz 570 m Fig.003 . burial depth. Soil Dyn Earthquake Eng (2011).1016/j. Pipeline–soil interaction Fig.2011. buried steel pipelines were analysed for reverse fault motion to study the influence of design parameters viz. If the anchor point L1 or L2 (Fig. then the pipeline can develop the effective unanchored length through unrestrained sliding in the soil. Analysis of buried pipelines subjected to reverse fault motion.

4(b)) is similar to that adopted for lateral pile springs [23. The entire 1200 m length of the model was divided into three regions for the meshing. the above described pipeline–soil interaction components in three directions were equivalently represented by discrete nonlinear springs [18–20].1016/j. Convergence studies were performed on the model to arrive at the above described mesh configuration [22]. The von-Mises isotropic hardening model available in ABAQUS was used for the pipeline material.5 m in Region-2. and Transverse Vertical. 4(c). For the section of the pipe near the fault. The bending and axial compressive deformations of this nearfault pipe section are often of the same order of magnitude as the pipe itself. commonly used for steel oil and gas pipelines. Model details FEM model for the buried pipeline was developed using the commercial finite element software ABAQUS (Version 6. Element size was kept uniform within each region. the limiting uplift resistance of the soil is generally much smaller than the ultimate bearing resistance. pipeline’s movement will be restrained at the location of physical anchor point itself. which required a smaller spacing between the soil-spring sets in this section. 3. 3. For the remaining length of the model (570 m on either side) where the pipeline was expected to remain elastic. Soil Dyn Earthquake Eng (2011). In this study.2. Pipeline material stress–strain curve. doi:10.24]. For the case where L1 or L2 is smaller than L0. closest to the fault crossing point had a total length of 30 m (15 m on either side of the fault crossing point). Dy induces lateral soil pressure on the pipeline in the horizontal direction and results in bending of the pipeline in X–Y plane. Transverse Horizontal. which generate spring forces only in the local axial direction of the element. The element size was increased to 0. Dz results in bending of the pipeline in X–Z plane. Similar nonlinear winkler foundation models to represent the soil-structure interaction have been proposed by many researchers for buried pipes [11–13] and for piles [23. This formulation can be considered appropriate for thin pipe-sections with large diameters. Proposed finite element model 3. 2 shows the geometry adopted for the proposed finite element model. A pipe node was supported by a set of three mutually perpendicular soil springs viz. The smallest element size of 0. Axial. 3. a certain length of the pipeline on either side of the fault crossing point is expected to undergo material yielding due to development of large bending moments (in X–Y and X–Z planes) and axial compressive force. Analysis of buried pipelines subjected to reverse fault motion. Similarly. the anchor points were assumed to be sufficiently away from the fault crossing point on either side of it. for a shallow buried pipeline the uplift resistance Pu provided by a limited depth of top soil cover (backfill) is generally much smaller than the bearing resistance Pb of soil below.13]. as shown in Fig.2011. which assumes that the fault Please cite this article as: Joshi S. The finite element formulation for this element is based on Timoshenko’s beam theory and takes into account transverse shear deformations [21]. geometric nonlinearity was incorporated in the analysis. Joshi et al. Hence. This expected material nonlinearity was considered in the model by modelling the pipeline as an elasto-plastic material with a bi-linear strain-hardening curve of the type shown in Fig. 1b. anchor plates. the pipeline can be considered to be effectively anchored to the ground at distance L0 from the crossing point as shown in Fig. Region-3 represented the rest of the length of the model on both sides and it had elements of size of 5 m. 4. Soil surrounding the pipeline was modelled using discrete nonlinear springs.003 . which supported the pipeline in the three directions. However due to shallow burial depth of the pipeline. which began at the ends of Region-1 on both sides and extended up to 30 m. Force–deformation relationship for transverse horizontal springs of buried pipe (Fig. / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] unanchored length (say at L0 from the crossing point). 3. The force–deformation relationship of soil springs for pile is considered the same in two lateral directions. Simulation of fault motion A nonlinear quasi-static solution is desirable for the problem of a pipeline crossing an active fault. the maximum axial force developed in the pipeline near the fault crossing point is completely removed by soil frictional force. et al.24]. In this study.4 S. length of this region was assumed to be 30 m on either side from the fault crossing point and spring sets were provided at a uniform spacing of 1 m. Hence. the soil-spring sets were spaced at 10 m.soildyn. All the connector elements were associated with axial connector sections. These expressions are based on field and laboratory experimental investigations on pipeline response. These springs simulated the components of pipe–soil interaction in the respective pipeline directions. the pipe–soil interaction could be more complex due to large deformations of the pipeline. For each soil spring. The length of the pipeline was selected based on the model lengths proposed by previous researchers [12.25 m was adopted in Region-1 since the maximum strain in the pipeline is likely to occur in this region. Schematic representation of the force–deformation relationship as adopted in this study for three springs is presented in Fig. Stress E2 σy E1 εy Strain Fig. an elastic-perfectly plastic force–deformation relationship was assumed and expressions for the same were adopted from ALA-ASCE guidelines [18]. The pipeline segment was modelled using 2-noded linear beam elements (B31 in ABAQUS) in space. which allowed for development of the effective unanchored length.02. A 1200 m long straight pipeline segment was considered for the analysis. For a pipeline subjected to large reverse fault offset (of the order of several metres). However.1. as well as general geotechnical approaches for related structures such as piles. Fig. Region-1. The soil-springs were modelled using connector elements (CONN3D2 in ABAQUS) between the pipe-nodes and the corresponding ground nodes [21]. The fault was assumed to cross the pipeline segment at the centre of its length.7). This makes the vertical spring force–deformation relationship unsymmetrical for the pipeline. In the proposed model. The long length of the model justified the assumption that the anchor points were sufficiently away from the fault crossing point.

respectively. the results of the analysis for a given fault motion were observed to be independent of the assumed length of the soil springs. Similarly. over half length of the pipeline segment. Hence. This has been explained in Fig. Fig. Dy. 4. However. This simultaneous application of fault offset components resulted in rotation of springs if the ground ends of these springs were restrained in the other two directions. Node 1 is now constrained to have the same displacements as Node 3 (pipe node) in X and Y directions. pipeline crossing angle (b). Soil Dyn Earthquake Eng (2011). performs dynamic analysis on the model and requires an analysis time to be associated with non-zero displacement boundary conditions. To simulate reverse fault motion. Node 2 (axial soil spring ground end) and the Node 3 are constrained in Y and Z directions. Quasi-static conditions were simulated in the analysis by applying fault offset components to soil-spring ends through a smooth loading function of time (Smooth Step Function in ABAQUS) which avoids any sudden load changes. Fig. ABAQUS/Explicit. and vertical stationary points. 5(a) also suggests that the magnitude of the rotation of any spring will change if the original length of the spring is changed. / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] 5 Force Tu Pl Force Force Pu Xu Xu Disp. 5(b) shows the same portion of the pipeline segment as represented by Fig. 5(a). transverse loading points. To avoid this problem the springs were prevented from rotating by constraining the ground node and pipe node of each spring to move identically in the directions in which the faultoffset components of the spring were zero.003 . Force–deformation relationships for the soil-springs: (a) for axial spring. the ground ends of the soil-springs on the left side of the fault crossing point associated with axial.02. fault dip angle (c). This in turn will affect the results of the analysis. a loading duration of 20 s was chosen for the analysis. even if the magnitude of fault offset component applied to this spring remains the same. transverse stationary points. 4. and Dz (Eq. (1)–(3)) were assigned simultaneously to ground ends of the axial. A small duration of loading would generate significant dynamic effects in the simulation. Analysis of buried pipelines subjected to reverse fault motion. Node 1 (ground end of vertical soil-spring) is subjected to Dz and restrained in X and Y directions to avoid any free body motions during the analysis. Joshi et al. and vertical springs. Each ground end was constrained to move with its pipe node in two directions and it was either assigned a fault offset component or a restraint in the third direction. Springs in this case maintain their original orientation after the application of fault offset components without any rotation about their respective ground node. an excessively large loading duration increases the computational time substantially. and (c) for transverse vertical spring. Choosing a sufficiently long duration of loading ensures that these dynamic effects are negligibly small. Spring forces would act at Node-3 in the directions V0 and A0 . Hence. Zb Zu Disp. For the sake of convenience. the dependence of the rotations on the original length of the spring (which may exhibit considerable variation) will introduce a measure of arbitrariness in the results. and the satisfaction of the quasi-static condition was checked by comparing time histories of the kinetic energy (KE) and internal energy (IE) of the model for each trial run. Similarly. the ground ends of soil-springs on the right side of the fault crossing point are referred to as axial stationary points. and by keeping a sufficiently long loading duration for the fault offset components. and vertical loading points. Similarly.soildyn. transverse. the fault offset components Dx. which shows a portion of the pipeline segment in the X–Z plane subjected to fault offset components Dx and Dz.2011. These restraints would lead to rotation of the springs due to simultaneous application of Dx and Dz. The velocity of movement on one side of the fault with respect to the opposite side is generally sufficiently low. Table 1 shows the boundary conditions assigned to ground ends of the soil-springs to simulate the reverse fault motion. numerical experiments involving several trial runs were performed on the model to arrive at an optimum loading duration that satisfies quasi-static conditions [22]. doi:10. an explicit dynamic analysis was performed using ABAQUS/Explicit. being an explicit dynamic finite element code. Hence a nonlinear static solution was achieved for this problem by converting dynamic analysis into a quasi-static mode. In this study. subjected to fault offset components Dx and Dz. For the vertical spring (V) in the figure. Similar constraints were assigned between pipe nodes and corresponding ground ends of the transverse springs to avoid rotation. Yu Yu Disp. The simulation was assumed to have run quasi-statically if the kinetic energy of the model was found to be negligibly small compared to the internal energy throughout the analysis. 5(b). transverse. respectively. Tu Pl Pb Fig. native and backfill soil type. et al. (b) for transverse horizontal spring. Based on these studies. the effects of fault rupture on the pipeline can be considered similar to those for statically applied relative movement [8]. Because of these constraints between the pipe nodes and the soil ground nodes. A constant magnitude reverse fault motion was applied. Response of buried pipeline to reverse fault motion The factors influencing response of buried pipeline at reverse fault crossing include the fault offset (D). as shown in the figure. Node 2 (ground end of axial soil-spring) is subjected to Dx and restrained in Y and Z directions. as shown in Fig. and vertical springs are referred to as axial loading points. respectively.1016/j. and solver will have to achieve an equilibrated state in the deformed configuration.S. Since the rotation of these springs can be significant during simulation of large fault offsets of the order of several metres. burial depth Please cite this article as: Joshi S. displacement components are applied at a sufficiently slow rate so as to ensure that the dynamic effects are negligibly small. 5(a) in X–Z plane. The ABAQUS/Explicit solver was chosen for the nonlinear analysis due to its capabilities of analysing post-buckling behaviour more efficiently and to avoid the convergence problems typically encountered by implicit solvers in the post-buckling domain [21].

Table 1 Boundary conditions to soil spring ends with suitable constraints. c. Pipe cross-section with outer diameter (D) 0.2011. D. These analyses were performed with a view to enable selection of design parameters that would enhance the pipeline’s capacity to accommodate the reverse fault offset. need to be considered in the analysis.003 . pipe diameter (c) and thickness (t). Influence of crossing angle A reverse fault motion was generated by choosing a particular combination of the magnitudes of D. and the type of native soil depend on geological and geotechnical characteristics of the site and their values. et al. b. Soil Dyn Earthquake Eng (2011). 4. / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] Node 3 A’ Node 2 Δx A Pipeline V V’ Δz Node 1 V = Vertical spring in original orientation V’ = Vertical spring after applying displacement A = Axial spring in original orientation A’ = Axial spring after applying displacement Δx Node 3 A’ Node 2 A Pipeline V’ V Z Δz Node 1 X Fig.91 m and wall thickness (t) 12 mm was chosen for the analysis. Influence of these design parameters on the response of the buried steel pipe to reverse fault motion was studied using the FEM model described above. and f so as to improve the pipeline’s performance and optimise the design. However. 5.02.6 S. 3). Soil reference point type Boundary conditions X-direction Axial loading points Transverse loading points Vertical loading points Axial stationary points Transverse stationary points Vertical stationary points Y-direction Constrained to pipe node to pipe node to pipe node to pipe node to pipe node Z-direction Constrained to pipe node Constrained to pipe node Dx Constrained Constrained Restrained Constrained Constrained Dy Constrained to pipe node Constrained to pipe node Restrained Constrained to pipe node Dz Constrained to pipe node Constrained to pipe node Restrained (H).soildyn. the designer has the choice to vary the factors b.1. as obtained from the site investigations. and pipe surface characteristics (f) [25]. H. Pipe material was assumed to be API5LX65 grade steel [26] with a bi-linear stress–strain curve (Fig. Schematic representation of portion of FEM model in X–Z plane: (a) rotation of soil-springs due to restrained spring ends and (b) Rotation avoided by assigning suitable constraints between pipe-node and spring ground ends. Joshi et al.1016/j. and c. the backfill soil type. Please cite this article as: Joshi S. d. Analysis of buried pipelines subjected to reverse fault motion. t. Out of the above listed factors. pipe material. doi:10. pipe material.

and dense (F ¼401). Firm Native Soil. as adopted in the previous section. ecmax value increased as the compactness of granular backfill increased. The corresponding spring yield forces and mobilising displacements were derived as per ALA-ASCE guidelines [18] and are listed in Table 2. Spring type Axial Transverse Vertical uplift Vertical bearing Yield force (kN/m) 40 318 52 1360 Yield displacement (mm) 3 11 2.00 0 1 2 3 4 Fault displacement . a consistent rise in ecmax was observed for a constant value of D. The total strain includes axial strain and bending strains at a pipe section. This was found to be true for all the crossing angles (b) studied here.9144m. The native soil (soil surrounding the trench) was assumed to be firm cohesive soil having a cohesion value of c¼50 kPa. This was true for all the considered values of D. Please cite this article as: Joshi S.003 . β = 10° β = 30° β = 45° β = 60° β = 90° ψ = 70º 0. For each b–c combination.S. and the largest at b ¼901. the smallest value of ecmax was observed at b ¼101.0 Fig. Fault motion parameters b and c were chosen to be 301 and 401. The native soil and the backfill were assumed to have the same mechanical properties. Soil Dyn Earthquake Eng (2011).1016/j. doi:10. ψ = 40° Maximum total compressive strain. εcmax 0. Δ (m) Fig.05 0. for normal or nearnormal orientation. when the results are compared at a Table 2 Soil-spring properties used in the study to evaluate effect of crossing angle.00 0.25 0. for the case of c ¼401 (Fig. was studied for three granular backfills: loose (F ¼301). ecmax increases consistently as the crossing angle increased from 10–901.12 D = 0. Joshi et al.5 1.3 100 constant magnitude of D for various b angles.06 Medium Dense Backfill Dense Backfill 0. ecmax at D ¼2 m was observed to reduce from 21% to 5% when b was changed from 901 to 101.00 0 1 2 3 4 β = 10° β = 30° β = 45° β = 60° β = 90° ψ = 40º 0. for different pipeline crossing angles (b). Coefficient of friction of 0. Such increase in the 0. 6(a) and (b). respectively. 6. as b was successively increased. Effect of the type of backfill on the maximum total compressive strain in the pipeline.2. Results of this study suggest that the pipeline compressive strain can be significantly reduced at reverse fault crossing by choosing a near-parallel orientation of the pipeline with the fault line in plan.05 0.08 Loose Backfill 0. component D Cos c would need to be predominantly accommodated by axial shortening of the pipeline. β = 30°. εcmax 0.5 m. ecmax is considerably reduced when the fault dip angle is increased from 401 to 701. Such an orientation would reduce the direct axial shortening of the pipeline and cause the horizontal component of fault offset (D Cos c) to be predominantly accommodated by transverse horizontal bending of the pipeline.10 0.0 1. Δ (m) 0.0 0. for dip angles c ¼401 and 701.5 m.10 Fault displacement . 7 shows the effect of backfill type on maximum total compressive strain in the pipeline at various reverse fault offset magnitudes.20 0.15 0.25 0. the pipeline was subjected to the maximum fault offset of 4 m in increments of 0.8 was adopted for the interaction between the pipe surface and soil. This significant strain reduction due to change in the pipeline’s crossing angle was observed to be true for both the dip angles investigated during this study. ecmax can be seen to increase as the value of D gradually increased up to 4 m. At each value of D. 7. Δ (m) 2. which shows variation of the maximum total compressive strain (ecmax) in the pipeline with the fault offset magnitude (D). This suggested that the pipeline’s capacity to safely accommodate reverse fault offset magnitudes would reduce considerably at shallow dip angles. g ¼ 18 kN/m3. 6a).30 m of medium density sand with friction angle.30 Maximum total compressive strain. F ¼351 and unit weight.04 0.soildyn. For each b of the pipeline.20 0. Effect of the crossing angle (b) on the pipeline performance: (a) for the dip angle c ¼ 401 and (b) for the dip angle c ¼701.02 0.02. et al. For a constant magnitude of D. Influence of backfill type Fig. / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] 7 The top of the pipeline was assumed to be buried under 1. The same pipeline segment (with the same burial depth and pipe surface properties). 4. For example. 6 also shows that for constant values of D and b. On the other hand. and the pipeline was subjected to maximum fault offset of 2 m in increments of 0. Figure shows that. εcmax 0. medium dense (F ¼351).10 0. The results of the analyses have been summarised in Fig. Analysis of buried pipelines subjected to reverse fault motion.2011. In general.30 Maximum total compressive strain. Fig.15 0.5 Fault displacement.

8.3 m.5 2 Fault displacement . In all these cases. Steel pipeline with a smooth and hard surface coating was assumed to have the coefficient of friction f to be 0.6.5 1 1. Two pipelines having outer diameters of 0. c ¼151).5 2 Fault displacement . ψ = 40º.91 and 1. Effect of burial depth on the maximum total compressive strain. Consequently.14 D = 0. For this study the native soil was assumed to be firm cohesive soil (c ¼50 kPa) and the backfill to be loose granular (F ¼301).02 0. εcmax D = 0.04 0.04 0. c ¼151).78 0. 0. εcmax ecmax was expected due to the fact that increase in F of the backfill would increase limiting frictional force (Tu) as well as limiting uplift force (Pu) of the soil.5 1 1. Results of this study suggested that the maximum compressive strain in the pipeline could be reduced by a fair amount by placing a very loose backfill in the fault crossing region and by avoiding its unnecessary over-compaction. For a granular backfill. 9 shows plots of ecmax against the applied D for two burial depths.8 to 0.5. pipe material with higher yield strength can reduce the compressive strain in the pipeline compared to a lower yield strength material. (b ¼901. and leads to reduction in the value of limiting soil frictional force (Tu) acting over unit length of the pipe.92 H/D = 3. which generally leads to increase in the friction coefficient. Loose backfill 0. these results are available elsewhere [22]. 4. Fig. / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] Maximum total compressive strain. Fig.00 0 0. doi:10. Results of the parametric study showed that the increase in burial depth increased the values of limiting uplift soil force and limiting pipe–soil friction force acting on unit length of the pipeline.2011. due consideration should be given to the effect of aging on properties of the surface coating. for a constant magnitude of D. but with the same wall thickness of 12 mm were analysed for the case b ¼301 and c ¼451.02 0. 8 shows response of the pipeline to given reverse fault motion for two different surface characteristics. c ¼401). (b ¼301. Loose backfill 0.08 0. β = 30°. ψ = 40°.8 0. However. The consistency of these results was also confirmed for medium dense (F ¼351) and dense (F ¼401) backfills. Considerable reduction in the value of ecmax can be seen here due to use of smooth and hard surface coating in the fault crossing zone. However. One may also consider replacing the pipeline section in the fault crossing zone with the one having larger wall thickness and the same internal diameter in order to reduce the maximum compressive strain in the pipeline. Similar effect of the burial depth was observed when the same pipeline section was analysed for medium dense and dense backfills. A study was carried out to see effect of the pipe outer diameter on the performance of pipeline.1016/j.12 0. The maximum total compressive strain in the pipeline reduced from 7. was analysed for two burial depths. Δ (m) Fig.3. 4. Analysis of buried pipelines subjected to reverse fault motion. Buckling of pipeline Buried pipeline subjected to large compressive force generally buckles either in the shell mode (local buckling) or in the beam Please cite this article as: Joshi S. 4. Smooth surface coating reduces the value of coefficient of friction between pipe and the soil. for soft and stiff cohesive native soils having c¼20 and 100 KPa. At D ¼2 m. minimum cover of D/6 (up to the top of pipe) is adequate. as above.10 Maximum total compressive strain. the value of maximum axial force developed in the pipeline near the fault crossing would reduce and this force will be dissipated over larger length of the pipeline.9144m.80% when the outer diameter was increased from 0. Soil Dyn Earthquake Eng (2011).5.10 H/D = 1.8 S.003 .3 m. For a given pipe cross-section. the same pipeline was analysed again for the three backfill types. Δ (m) Fig.5 0. the live loads and environmental factors may also have an influence on the minimum soil cover. and consequently led to higher maximum compressive strain for a constant magnitude of D. Influence of pipe surface characteristics Further analyses were performed on the same pipeline section to evaluate influence of soil–pipe interface friction on pipeline’s performance.06 f = 0. Thus.8 [19].9144m. et al.06 0.08 f = 0. respectively. suggesting that the pipeline can be expected to perform best under those conditions. considering only the dead load of soil [27]. and pipeline without any special coating was assumed to have this coefficient to be 0.00 0 0. To confirm the consistency of these results. ecmax was observed to increase as the soil cover (h) increased from 1. Influence of pipe cross-section and material properties 0.5.soildyn. 9.91 to 1. respectively [22]. 4.9% for f¼0. These analyses were then repeated for three other b–c combinations.3 to 3 m. Influence of burial depth The same pipeline section. Effect of pipe surface characteristics on the maximum total compressive strain. ecmax values were the smallest for the loose backfill (F ¼301). Joshi et al. for a fault offset of 2 m.02. burial depth as shallow as possible is preferable in the fault crossing zone. and (b ¼901. The use of high yield strength material for the pipeline could be more suitable in the fault crossing zone. β = 30°.4.35% to 5. the pipeline compressive strain can be seen to reduce from 12% for f¼0.

Results of the experimental studies carried out on steel pipes suggest that the pipes would normally begin to wrinkle at strains of ½ to ¼ of the theoretical value [19]. A sharp rise in KE can be seen at the instance corresponding to the application of a fault-offset of D ¼ 1 m. Time variation of internal energy (IE) and kinetic energy (KE) of the model for the case b ¼901. ec ¼ 0:175t=R ð5Þ the pipeline with the adjacent soil near the fault crossing point. 11. This displacement was several times higher than the actual fault offset component applied to the model. c ¼151. The corresponding expression for the wrinkling strain is Fig. This rise confirmed the generation of large velocities in the pipeline. A sharp rise observed in ecmax from 1% at D ¼0. Further confirmation of the occurrence of beam buckling is provided by the time histories of the kinetic energy (KE) and the internal energy (IE) of the model during the analysis. Hence.29]. Joshi et al.S. designers may choose the value for the allowable wrinkling strain as they consider appropriate based on experience.2011. doi:10. Theoretical wrinkling strain (onset of wrinkling) for steel pipeline as given by ASCETCLEE Guidelines [19] is given by the following expression: ec ¼ 0:6t=R ð4Þ where t is the pipeline wall thickness and R the pipeline radius. 11 shows these plots for the case discussed above. Soil Dyn Earthquake Eng (2011). a distinct change in the deformed shape was observed. It was assumed that local buckling would initiate if the maximum compressive strain occurring in the pipeline at any location exceeds this allowable value. At this fault offset value.soildyn.5 m. while the beam mode of buckling was captured in the analysis. 10(a) shows schematic representation of the original deformed profile (as observed in ABAQUS Postprocessor) of the pipeline near fault crossing point for the case b ¼901 and c ¼151 at D ¼0. A certain length of pipeline on either side of the fault crossing point showed abrupt and very large displacement in the vertical direction. Deformation profile of the pipeline suggests no loss of contact of Please cite this article as: Joshi S. c ¼151. This expression for the allowable wrinkling strain was adopted in the present research. the maximum vertical displacement of the pipeline (wmax) was nearly the same as Dz (0. et al.1016/j.395 m) was observed to be about 10 times higher than the applied Dz (0.5 m to 21% at D ¼1 m appeared to confirm this result. local buckling in the model was predicted based on the value of maximum compressive strain in the pipeline.5 m was applied to the model. 10(b) shows the schematic representation of deformation profile of the same pipeline for the same b–c combination at D ¼1 m. This suggested that the pipeline displaced with the soil in the vertical direction when D ¼0. IITK-GSDMA Guidelines [17] suggest suitable value for the allowable wrinkling strain to be the mean of these experimental wrinkling strain values. Fig.02. Displacement profile suggested that the pipeline had undergone large lateral deformations over a small stretch in the vicinity of the fault crossing point consequent to beam buckling. c ¼ 151. Fig. and D ¼ 0. In the FEM model.5 m.25 m). Deformed profile of the pipeline near the fault crossing point in X–Z plane (a) for the case b ¼ 901. as the entire pipeline segment was modelled using beam elements.13 m). / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] 9 Deformed profile of soil Anchor point Δz Anchor point wmax Deformed profile of pipeline Anchor point Pipeline profile before deformation Deformed profile of pipeline Anchor point Δz Anchor point wmax Anchor point Pipeline profile before deformation Deformed profile of soil Fig. Here. Analysis of buried pipelines subjected to reverse fault motion. and (b) for the case b ¼901.003 . Fig. section deformations due to local buckling could not be captured directly. and D ¼ 1 m. However. mode [28. The model could successfully capture the beam mode of buckling. 10. resulting from the sudden movement of a certain portion of the pipeline from one configuration to the other. wmax (2.

consequently leading to its early failure. Influence of change in burial depth from 3 to 1. On the other hand. (a) For the steel pipeline subjected to reverse fault motion. the pipeline buckled in the beammode well before reaching the 4 m fault offset for the crossing angles 451. it was inferred that the buried pipeline buckled locally due to wall wrinkling. Analysis of buried pipelines subjected to reverse fault motion.02. As a result. Out of all the investigated design parameters. This reduction can be achieved by suitably changing the backfill type. Thus. Joshi et al. (b) The capacity of buried pipeline to accommodate the reverse fault offset could be increased substantially by adopting a near-parallel orientation in plan with respect to the fault line. Debasis Roy of Indian Institute of Technology Kharagpur for his personal help during the research. doi:10.3 m without changing the thickness. 751. In another parametric study the internal diameter of the pipeline was kept constant and the wall thickness was increased by 2 mm. pipe section dimensions and pipe material properties. For the cases having c ¼701. Also. Comparison of design parameters To study the relative influence of the above described design parameters on the response of pipeline.soildyn. While no beam bucking occurred when crossing angle was 301 or lower.8 to 0. Please cite this article as: Joshi S. For c ¼401. 6(a)). Hence. burial depths and pipesurface properties. burial depth and pipe-surface properties. (d) Pipeline crossing the fault line in normal or near-normal orientation generally buckled in the beam mode at very small fault offset magnitudes. pipe-surface characteristics. Authors would like to express their deep gratitude to Dr. However. for the same pipeline.003 . This local buckling would occur at a pipe-section where ecmax exceeded the allowable value of the wrinkling strain. the results were compared at D ¼2 m. beam buckling was not observed at c ¼401 for the crossing angles 101 and 301 up to the maximum applied fault offset. and 901. the maximum compressive strain reduced by about 7% when the dense backfill was replaced with loose backfill (Fig. Local buckling of the pipeline was predicted based on the value of maximum compressive strain in the pipeline. better performance of the pipeline was observed for the near-parallel orientations of the pipeline.3 m on the strain reduction was nearly the same as influence of change in friction coefficient (f) from 0. The influence of backfill type was found to be relatively insignificant in comparison to the other design parameters. 6 were similarly studied to cheque for the occurrence of beam buckling. Summary and conclusions Acknowledgements Buried steel pipeline subjected to reverse fault motion was analysed using a simple FEM model which used 3D beam elements to model the entire pipeline segment. typically encountered by implicit solver in the post-buckling domain. At D ¼2 m. using a smooth and hard surface coating. (f) The capacity of the pipeline to safely accommodate the reverse fault offset can be further increased by choosing a loose granular backfill. On the other hand. At D ¼2 m. The present study ignored the effect of internal pipe pressure as the pipeline was modelled using beam elements. maximum total compressive strain was always found to be more critical than the maximum total tensile strain. A reduction of 21% in the maximum compressive strain was observed when the pipeline’s outer diameter was increased from 0. For these orientations pipelines can be protected against local buckling by limiting the maximum compressive strains in the pipe walls. of three mutually perpendicular nonlinear springs. Pipeline–soil interaction was modelled by connecting each pipe-node to a set Financial support from Poonam and Prabhu Goel Foundation at Indian Institute of Technology Kanpur to carry out this research is gratefully acknowledged. thereby accommodating the horizontal component (Dx) of the fault offset primarily by bending of the pipeline in the horizontal plane (X–Y) plane. These displacements were applied quasistatically. no beam buckling was observed in any of the cases. precise simulation of the local buckling and large section deformations were ignored due the use of beam elements.5 at D ¼2 m. (c) For low crossing angles. Soil Dyn Earthquake Eng (2011). while the beam buckling mode was successfully captured in the simulation. no beam buckling was observed up to the maximum fault offset of 4 m for all crossing angles. Complex 3-D reverse fault motion was simulated by applying the fault offset components to the respective free ends of soil-springs over the half length of the model. Pipeline crossing angle (b) was found to be the most influencing design parameter. 8 and 9). for all combinations of backfill soil.10 S. Geometric nonlinearity associated with the large deformations was incorporated in this analysis.2011. The following conclusions could be drawn based on the studied response of pipeline subjected to reverse fault motion using the FEM model. The pipeline with b ¼101 and 301 was further analysed for different conditions of backfill. 7). about 16% reduction in the strain was observed [22]. the maximum compressive strain was reduced by 76% when b was changed from 901 to 101 (Fig. However.1016/j. Soil material nonlinearity was simulated using an elastic-perfectly plastic force–deformation curve for each type of spring. 4. it is possible that the pipeline might fail due to local buckling in these cases. Occurrence of beam buckling at normal or near normal orientation of the pipeline with the fault line lead to failure of the pipeline at small fault offset magnitudes. it was observed that beam buckling of the pipeline would be primarily influenced by the pipeline’s crossing angle (b) and the fault dip angle (c). et al. Suitable constraints were assigned between pipe-nodes and corresponding ground ends of soil springs to avoid rotations of the springs during application of the fault motion. the pipeline with b ¼901 showed beam buckling in all the cases. the capacity of a pipeline to safely accommodate fault offset magnitude may be considerably smaller at a thrust fault than at a steeper reverse fault. (e) For a constant magnitude of D. the pipeline experienced larger maximum compressive strain for a shallower dip angle.7. at smaller fault offsets the influence of change in friction coefficient (f) was relatively insignificant in comparison to the influence of change in the burial depth. adopting a shallower burial depth. Explicit solver in ABAQUS was used for the nonlinear analysis due to its capabilities of analysing post-buckling behaviour more efficiently and to avoid the convergence problems. the pipeline crossing angle (b) was found to be the most influencing factor on the response of the buried pipeline in terms of maximum compressive strain and its buckling. 601. In both the cases.91 to 1. 5. / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] Displacement profiles of the pipeline for all the cases shown in Fig. increasing pipe-wall thickness or by using higher yield strength material. However. for the pipeline segment analysed in this study. burial depth. Pipe material nonlinearity was considered in the analysis by associating a bilinear stress–strain curve to the beam elements. the maximum compressive strain was reduced by about 20% (Figs.

ASCE 1977. Fukuda K. [23] Dash SR. Finite element analysis of buried steel pipelines under strike-slip fault crossings. [10] Newmark NM. Soil Dynamics and Earthquake Engineering 2010. [13] Karamitros DK. July 2001 [with addenda through February 2005]. p. Kouretzis GP.2011.29(2):333–46.S. 2001. Specification for Line Pipe.30(11):1298–308. [11] Wang LRL. India. Lin SH. Kyriakides S. Nakata T. Buried pipe design. Soil Dynamics and Earthquake Engineering 1990. Jou LD.29(3):415–27. A semi-analytical approach to a nonlinear stress– strain analysis of buried steel pipelines crossing active faults. Design of Foundations in Seismic Areas: Principles and Applications. In: Proceedings of the 10th world conference on earthquake engineering. Journal of the Chinese Institute of Engineers 2000. [25] Abdoun TH. Remediation for buried pipeline systems under permanent ground deformation. Kanpur (India): NICEE. Gujarat. [26] American Petroleum Institute. Ann Arbor: University of Michigan. 2009. p. Active faults and related late quaternary deformation along the Northwestern Himalayan Frontal Zone. Abdoun TH. 271–8. Yeh YA. Palmer MC. 1992. ABAQUS user’s manual version 6. Buckling and rupture failures of pipelines due to large ground deformations. Simulia. ASME 1985. et al. Hall WJ. Analysis of buried pipelines subjected to reverse fault motion. Soil Dynamics and Earthquake Engineering 2007. Madrid. 5423–26. Journal of Pressure Vessel Technology 2000. Guidelines for the design of buried steel pipe. Kincaid RH.003 . [2] Malik JN.soildyn. 2007. Soil Dynamics and Earthquake Engineering 2010. O’Rourke TD. 1990. 2nd ed. Please cite this article as: Joshi S. Bhuj–Kachchh earthquake: surface faulting and its relation with neotectonics and regional structures. Karamanos SA. Lee BJ. On the beam and shell modes of buckling of buried pipelines. Factors influencing the behavior of buried pipelines subjected to earthquake faulting. 2008. 1984. editor. 416–25. Fault movement related damage examples and fault provisioned design case histories. O’Rourke MJ. / Soil Dynamics and Earthquake Engineering ] (]]]]) ]]]–]]] 11 References [1] Johansson J. Fault crossing design for buried gas and oil pipelines. A case study of damages of the Kandla Port and customs office tower supported on a mat–pile foundation in liquefied soils under the 2001 Bhuj earthquake. Annals of Geophysics 2003 2001. Sun S. In: Bhattacharya S. Bhattacharya S. [24] Dash SR. [9] Kennedy RP. Indian Institute of Technology Kanpur (India).27(3):200–11. Soil Dyn Earthquake Eng (2011).103(5):617–33. [16] Ariman T.23(4): 395–408.9(4):179–93. In: Proceedings of the US national conference on earthquake engineering.7. San Francisco. Tension/bending behavior of buried pipelines subjected to fault movements in earthquake. Indian Institute of Technology Kanpur. Response of buried pipelines subject to earthquake effects. Soil Dynamics and Earthquake Engineering 2010. [7] Choo YW. In: Proceedings of the 8th world conference on earthquake engineering. India. 2007. O’Rourke TD. PVP. [18] American Lifelines Alliance-ASCE. Govindaraju L.46(5):937–56. Thakkar MG. PVP. Williamson RA. Ha D. 1984. 3.98-4:3–10. O’ Rourke MJ. National Information Center of Earthquake Engineering. [19] ASCE Technical Council on Lifeline Earthquake Engineering. A new proposal for simplified design of buried steel pipes crossing active faults. Damage to buried water supply pipelines in the ChiChi (Taiwan) earthquake and a preliminary evaluation of seismic resistance of pipe joints. Chow AM. [6] McCaffrey MA. p. [29] Ariman T. [20] MCEER. [15] Vazouras P. [22] Joshi S. [21] ABAQUS.S. Pipeline design to resist large fault displacement.30(11):1361–76. p. doi:10. [27] Moser AP. Fault movement effects on buried oil pipeline. Stress analysis of buried steel pipelines at strike-slip fault crossings. A refined seismic analysis and design of buried pipeline for fault movement. [14] Trifonov OV. guidelines for the seismic design of oil and gas pipeline systems. Joshi et al.30(1–2):32–9. Site effects on seismic behavior of pipelines: a review. Hassani N. Bhattacharya S.02. [17] IITK-GSDMA. Dakoulas P. Earthquake Engineering and Structural Dynamics 2001. Mtech thesis. [4] Tsai JS. MCEER Monograph no. Soil Dynamics and Earthquake Engineering 2009. Ha D. et al. Transportation Engineering Journal. Bouckovalas GD.77:151–9. [8] Kennedy RP. API specification 5L. Guidelines for seismic design of buried pipelines. Annals of Geophysics 2003. Earthquake Engineering and Structural Dynamics 1985.27(12):1043–55.30(8):1243–57. Western India. Buried pipeline response to reverse faulting during the 1971 San Fernando earthquake. Differential ground movement effects on buried pipelines. Soil Dynamics and Earthquake Engineering 2009. ASME 1983.122:469–75. Bending–buckling interaction as a failure mechanism of piles in liquefiable soils. Cherniy VP.1016/j. Soil Dynamics and Earthquake Engineering 2007.13(1):75–96.46(5):917–36. Konagai K. IIT. Symans MD. Kanpur. [12] Takada S. Analysis of buried pipelines subjected to reverse fault motion. Blakeborough A. New York: McGraw-Hill. [28] Heedo Y. 1999. 1975. 150–228. [3] McCalpin JP. [5] Liang J.

Sign up to vote on this title
UsefulNot useful