Large-scale field trial to explore landslide and pipeline interaction
As global economic development results in greater demand for energy, many countries are facing increasing pressure to ensure the safe transport of energy, especially by pipelines. Great attention is therefore placed on the safety assessment (Hossam et al., 2010, Huang et al., 2011, Xiao et al., 2012), monitoring (Ma et al., 2011), and design of pipelines (Ma et al., 2007). In this regard, landslides are gaining much attention because of their potentially devastating effects on the integrity of oil–gas pipelines. Deng et al. (1988) simplified pipelines inside and outside a slope as beams and developed a method to assess internal stresses and deformations. Using the pipe-soil interaction module in ABAQUS/Standard, Zhang and БыковЛИ (2001) found a relation between maximum Mises stress and slide length and displacement in loess landslides. Others (Calvetti et al., 2004; Guo, 2005; Abolmaali et al., 2011) carried out investigations into the interaction between pipeline and soil body using the same pipe-soil interaction module in ABAQUS/Standard and the interface constitutive relation.
Zang (2007) established an index system to assess pipeline safety in landslide areas, which was helpful to determine pipe risks. Liu (2008) used finite element numerical simulations and mechanical theory analyzes on three ideal landslide types namely axial, lateral and deep circular slides and developed preliminary analytical theories on pipeline deformation and strain. Feng and Huang (2009) used strength theory to assess pipeline safety based on monitoring of slope surface displacements and strains in pipes. Jin and Li (2010) and Jung and Zhang (2011) analyzed performance of buried pipelines under the impact of landslides, active faulting and seismic wave-triggered disasters.
Peter (1999) considered that the force on a pipe is a function of the relative displacement between the pipeline and the sliding body. Challamel and de Buhan (2003) simplified the interaction between the pipeline and the three-dimensional slope in a landslide. Karimian (2006) described stress-strain states of pipelines influenced by axial and lateral tensile forces. Manolis et al. (1995), Datta (1999) and Lee et al. (2009) investigated the mechanical behavior and resilience of pipelines under the influence of seismic action. Vazouras et al. (2011) used finite element software to analyze the behavior of buried pipes going through active strike-slip faults.
In brief, though previous studies on pipelines have delivered results and understandings of potential failure mechanisms, most of them were limited to numerical simulations. Detailed studies are rarely found to combine an actual landslide with pipe deformation monitoring for assessing pipeline safety conditions and few studies focused on monitoring of pipelines after large deformation or total destruction. This is possibly because remedial measures are normally taken immediately after even a slight deformation is observed according to standard safety requirements. Therefore, there is no data available for analyzing large deformations and the failure process in case of an emergency. On the other hand, some researchers (Majid and William, 1998, Calvetti et al., 2004, Kinash and Iseley, 2008, Li et al., 2011, Rojhani et al., 2012), using similitude theory, investigated the interaction of pipelines and landslides by adopting physical simulation testing. However these physical simulations still require validation because of the reduced scale and the materials selected for landslide and pipeline. Large-scale landslide and pipeline model tests are the best choice to obtain complete and continuous data about pipeline and landslide interaction.
This paper presents a large-scale field model with a pipeline buried through a potential landslide. The model test follows practical pipeline operations, actual deformation and failure of the landslide.
Large-scale test model set up
Basic test model
The large-scale model was built at Chengdu University of Technology. The foundation of the model consisted of soil that belongs to the Chengdu alluvial-proluvial plain. The soil used for the construction of the landslide body was medium-hard clay soil with some rubble and breccia.
The volume of the model landslide was about 500 m3 (the length and width are about 10 m, and thickness about 5 m). The plan shape of the landslide mass was arched, while the sliding surface was prepared in advance and approached a straight plane. The tilt angle of the sliding surface and slope surface was in the range between 15° and 20° (Fig. 1, Fig. 2, Fig. 3). The base of the slope model was the actual soil at the site, which was mainly brownish yellow clay with cohesion of 13.73 kPa, internal friction angle of 4.9 degrees, elasticity modulus of 4.20 MPa, Poisson ratio of 0.3, and bulk density of 18.95 kN/m3. The soil of the landslide was artificially made with the cohesion of 5.03 kPa, internal friction angle of 3.5 degrees, elasticity modulus of 2.88 MPa, Poisson ratio of 0.33, and bulk density of 20.98 kN/m3. The strength for both soils was tested by using unconsolidated-undrained triaxial tests and triaxial compression tests. During the set up of the landslide test model and installation of the pipe and monitoring instruments (Wei and Yan, 2008, Kinash and Iseley, 2008, Ma et al., 2011), the toe of the slope was supported by an earth wall (Fig. 2) to prevent sliding. The test started by excavating the retaining earth wall step by step.
The pipe steel used for the test was of Grade L245NB. Its yield strength is 245 MPa, tensile strength 415 MPa, minimum elongation 21%, surface roughness about 0.63 μm, and elasticity modulus 210 GPa. The pipeline was 32 m long, and each end reached at least 10 m outside the landslide boundary. The diameter was 325 mm and the wall thickness was 8 mm. The pipeline, with normal internal pressure of 2.5 MPa, was buried at a depth of 1.5 m in a ditch perpendicular to the slide direction.
Monitoring scheme and test process
The main purpose of the monitoring was to get a relationship between landslide deformation and stress and strain observed in the pipe. Nine inclinometers were installed in the test model to measure internal slope deformations (Fig. 1), and 24 vibrating wire strain gauges welded on the pipe wall were used to monitor and measure pipe stresses and strains (Fig. 4). Continuous measurements were taken during excavation stage.
In order to observe and explore the interaction between the slope and the pipeline, the test was divided into 6 stages: 1) preliminary observation and measuring; 2) observation and measuring of the first excavation of the retaining wall (1st excavation) to decrease the Safety Factor; 3) complete removal of the retaining wall (2nd excavation) to create a free face for the potential landslide; 4) Infiltration of water in the back scarp to promote sliding; 5) excavation of the collapsed material (Fig. 2) (3rd excavation), which hindered the development of the landslide; and 6) complete removal of the collapsed free face material (4th excavation).
Analysis of slope deformation and failure characteristics
Fig. 5, Fig. 6 show the monitoring results of the inner slope displacements recorded by inclinometers N-3 and N-7. The horizontal displacement was consistent with the sliding direction of the landslide, and perpendicular to the direction of the pipeline. The direction of the vertical displacement was consistent with the direction of gravity. The combined displacement was the resultant of the horizontal and vertical displacement vectors (Wei and Yan, 2008).
The monitoring results show a significant increase in displacement after the 2nd excavation. The increase in displacement on the slope surface was about 60–143 mm. Inclinometers N-3 and N-7 measured 143 mm and 133 mm respectively (see Fig. 5 and Fig. 6). The surface displacement increased along the slope surface from the back scarp to the front, showing a “pull-type” slide-deformation. The inclinometer profiles N-3 and N-5 show the depth of deformation (Fig. 7), which is almost consistent with the designed depth of the slip zone (See Fig. 1 and Fig. 2).
During the infiltration of water at the back scarp, large horizontal and vertical displacements were recorded. Maximum displacements were measured by inclinometers N-3 and N-7, namely, 1280 mm and 1035 mm respectively (Fig. 5 and Fig. 6). Deep inner slope displacement increased greatly and was distributed as an inverted triangle. Deformations became shallower forwards the front edge of the landslide (Fig. 7).
The results also indicate that the pipeline was able to resist the loading imposed by the deformations. The soil body above the pipe showed shallow deformation by sliding, while the soil under the pipe showed no obvious sliding or extrusion.
After complete removal of the collapsed material (the 3rd and 4th excavations), the slope exhibited significant additional deformation, and the pipeline was subjected to larger bending deformation (Fig. 8).
The buried pipeline showed both large deformations and features associated with failure: 1) along the pipeline, the part of the soil retained by the pipeline formed a passive wedge, while the rest extruded downward (Fig. 8); 2) under favorable surface condition, a part of the soil mass moved over the pipeline, in the same direction as the landslide, and formed a shallow sliding mass.
Analysis of pipe stress characteristics
The stress-time curves of monitoring points S-3, S-10, S-17, and S-20 are representative of the deformation history of the pipe, as shown in Fig. 9, where the tension stress is positive and compression stress is negative. The changes in stress of the pipe roughly correspond with the different stages of the test.
According to the measured data at each stage of the test the stresses on the pipeline changed as follows (Fig. 9):
1) Before the 1st excavation (Oct. 8, A.M.) of the retaining wall, the stresses on the pipeline were almost zero. The maximum tension stress increment during the 1st excavation is 8.1 MPa (S-20) and the maximum compression stress increment was −12.2 MPa (S-3).
2) The 2nd excavation (Oct. 10, A.M.) of the retaining wall led to a clear stress increase. The maximum increment of the tension stress is 67.2 MPa (S-20), which was accompanied by an almost symmetrical maximum compression stress increment of −68.3 MPa (S-22).
3) The water infiltration stage of the test (Oct. 12–14) resulted in a significant increase of the stresses in the pipe. The maximum tension stress increment was 209.6 MPa (S-5) and the maximum compression stress increment was −327.8 MPa (S-15). The curves show again almost a symmetrical increase at this stage towards a nearly constant value.
4) The 3rd (Oct. 19, PM) and 4th (Oct. 20, PM) excavations of the collapsed material had also a strong influence on the stresses of the pipe. The largest stress increments occurred during these stages due to the severe deformation of the pipeline leading to a failure of the strain gauges.
The stress distribution along the pipeline for the different test stages is shown in Fig. 10.
The figure indicates that: 1) the stress distribution along the pipeline was saddle-shaped with more or less left-right symmetry corresponding to the overall pipeline deformation; 2) within the landslide, the external side and lower part of pipeline were strained. Outside the landslide boundaries, the internal sides and upper parts were strained, and the stress disappeared on both ends of the pipeline; 3) the influence of water infiltration on the pipeline׳s front side stress in the horizontal direction, caused by deformation, is larger than the stress on the pipeline׳s underside in the vertical direction; 4) The 3rd and 4th excavations of the collapsed material had a large influence on the stresses in the pipeline. As the lower part of the pipeline became suspended after the excavation, the influence of excavation on pipeline׳s underside stress in the vertical direction is larger than that of pipeline׳s front side in the horizontal direction; and 5) at the end of the test, the stress on the middle section of the pipeline is larger than the yield stress of the pipeline.
In accordance with the “Code for design of gas transmission pipeline engineering (GB50251-2003)” and the “Code for design of oil transmission pipeline engineering (GB50253-2003)” and related provisions, a pipe should resist a maximum axial stress of 80% of the minimum yield strength. The tensile strength of this pipeline appeared to be more than 500 MPa. Although the pipe has yielded, and the bending degree is around 3%, the pipeline did not break or leak. The fact that in the design only strength criteria are considered is a subject for discussion. Future research maybe focus on the formulation of a reasonable combination of deformation and strength criteria for the design of these pipelines.
Relationships between landslide deformation and pipeline deformation and stress
Relation between landslide surface deformation and the maximum axial stress on the pipeline
A relation between landslide surface displacement and tension stress in the pipeline can be obtained from displacements from different inclinometers along the longitudinal section of the landslide combined with pipe stress changes in the same section. We will ignore such factors as depth of pipeline, pipe diameter, sliding zone depth and landslide width. The analysis was carried out only on inclinometers and strain gauges which were not broken and did not record values beyond the measuring range (Feng and Huang, 2009).
Table 1 shows values of the relationship between surface displacement and the tensile stress on the pipeline, located in three profiles: A’-A’, B-B’ and C-C’ (Fig. 11, Fig. 12, Fig. 13).
The figures revealed that: 1) tension stress on the pipeline increases exponentially with slope surface displacement in the sliding direction. The coefficients and exponents of the exponential equations of the trend lines have nearly the same value for inclinometers at the same distance from the pipeline.
It can be concluded that within a certain distance from the pipeline, slope surface deformation (especially within the stronger deformation zone) is to some extent related to stresses on the pipeline. The surface displacement is the most conspicuous and easy to capture information. A robust relationship between pipe stresses and surface displacements would be a basis for a fast judgment in pipeline risk assessment for experienced experts. The relationships presented here are not unique since they are affected by many factors such as the nature of the soil, pipeline depth, diameter variation, etc. Further systematic research will be carried out to find a rapid evaluation method for establishing a relationship between surface deformation and stress on pipelines.
Interaction between landslide and pipeline deformation
Deformations will be imparted on the pipeline as the landslide displaces around and across its length. The pipeline will then bend downwards taking a symmetrical saddle-shaped profile (Fig. 14, Fig. 15). Due to the pattern of deformation, the pipeline will also be subjected to substantial torsional stresses, in addition to tensile and bending stresses, making excess torque a viable mode of failure. It is popssible that the torque failure mode of the pipeline will occur rather than tensile failure or crushing damage after bending deformation.
Pipeline resistance has an obvious influence on slope stability and the deformation and failure mode: 1) soil mass drifts and slides over a small section along the pipe; 2) downwards extrusion of the soil at the rear of the pipeline combined with the pipeline bending down are driving forces for torque failure of the pipe.
The pipeline prevents the free movement of the slope and causes both outsides of landslide border to deform more along the pipe. Moreover, the pipeline bends in the opposite direction near both sides of the landslide boundaries. This brings changes to the stress pattern of the sliding slope at the front and rear sides of the pipeline (Fig. 16).
1) The pipeline section bending in the slide direction produces lateral extrusion forces near the landslide boundaries on the front side of the pipeline. This also generates diagonal cracks near the landslide boundaries in front of the pipeline (Fig. 16).
2) The pipeline section bending backwards near the boundary also provides an arching effect, which causes the directions of stresses and displacements to turn towards the inside of landslide (Fig. 16). Simultaneously, the pipeline section outside the landslide is also moving forward through stable soil, causing the rear of the pipe to become exposed (forming cracks parallel to the pipe). With the increase of slip deformation, lateral shear cracks become more serious (Fig. 16). The impact on the pipeline sides widens the sliding area to some degree and creates a rapid expansion of the soil deformation along the pipe line. It gradually helps the pipeline to expand further. The deformation and expansion, however, is also related to the self-deformation of the pipeline (Fig. 3, Fig. 8, Fig. 14).
With a pipeline crossing the landslide, the deformation and failure mode of the landslide shifts from an original coherent slide to a shallow drift slide above the pipe and a deep extrusion deformation below the pipe. The deformation of the landslide was mirrored by deformations of the pipeline for different test stages. The stress distribution along the pipeline shows a saddle shape.
An exponential relationship was observed between surface displacement and pipeline stress. Surface locations on the landslide, at equal distance from the pipeline have nearly the same response with similar coefficients and exponents in describing the observed relationship.
The characteristics of pipe bending deformation are connected with the pipeline stress pattern and the failure mode: bending and breaking or twisting and breaking. However, the existence of the pipeline increased the stability of the slope, spread deformations towards the edges, reducing the maximum displacements in the direction of the landslide.
Finally, future research should consider design criteria in terms of a reasonable combination of deformation and strength.
Authors: Feng Wenkai, Huang Runqiu, Liu Jintao, Xu Xiangtao, Luo Min