numerical analysis of pipeline response to slow landslides ......ters (si), six temperature gauges,...

Draft

Numerical Analysis of Pipeline Response to Slow Landslides: A Case Study

Journal: Canadian Geotechnical Journal

Manuscript ID cgj-2018-0457.R1

Manuscript Type: Article

Date Submitted by the Author: 04-Dec-2018

Complete List of Authors: Katebi, Mohammad; University of Manitoba, Civil EngineeringMaghoul, Pooneh; University of Manitoba, Civil EngineeringBlatz, James; University of Manitoba, Civil Engineering

Keyword: pipeline, slow landslide, longitudinal, numerical, case study

Is the invited manuscript for consideration in a Special

Issue? :Not applicable (regular submission)

https://mc06.manuscriptcentral.com/cgj-pubs

Canadian Geotechnical Journal

Draft

Numerical Analysis of Pipeline Response to SlowLandslides: A Case StudyMohammad Katebi, Pooneh Maghoul, and James Blatz

Abstract: A numerical analysis is carried out to study the behaviour of pipelines subjected to slow landslides at three at-risk landslide zones of Manitoba Pipeline Network. The pipeline’s longitudinal axis is parallel to the slow landslides at allthree research sites. The ground displacements monitored for five years are imposed on the pipe using a special purposeelement (PSI element) using ABAQUS/Standard. The stiffness of PSI elements is defined based on soil-pipe interfaceproperties according to Honegger (2017). The results of the numerical analysis are compared with the instrumentation datato draw recommendations for future monitoring programs in slow landslide zones.

Key words: pipeline, slow landslide, longitudinal, numerical, case study.

Resume : Une analyse numerique est realisee pour etudier le comportement des pipelines soumis a des glissements deterrain lents dans trois zones a risque du Reseau de Pipelines du Manitoba. L’axe longitudinal du pipeline est parallele auglissement de terrain dans les trois sites de recherche. Les mouvements du sol monitores pendant cinq ans sont appliquessur le pipeline en utilisant le logiciel de calcul ABAQUS/Standard et a l’aide d’un element special appele PSI. La rigiditedes elements PSI est definie en fonction des proprietes de l’interface sol-pipeline selon Honegger (2017). Les resultats del’analyse numerique sont compares avec les donnees d’instrumentation afin de fournir des recommandations pour de futursprogrammes de surveillance dans les zones a risque de glissements de terrain.

Key words: pipeline, glissement de terrain lent, longitudinal, axial, numerique, etude de cas.

1. IntroductionBuried pipeline infrastructure is an integral component of gas

and oil transportation networks across the country, and their in-tegrity has an essential impact on the strength of Canada’s econ-omy. In Canada, there is an estimated 242,000 km of gas andoil pipelines, which is considered the second longest networkin the world. Loss-of-containment events are often environmen-tally damaging and extremely costly to clean-up and remediate.Also, civil, criminal or regulatory penalties from a pipeline lossof containment may be very high (Oswell 2016). In Canada,according to the National Energy Board (NEB) database, about750 incidents have occurred since 2008 on major pipelines,including 454 gas and oil leaks, twenty-five cases of serious in-jury, six deaths, thirteen explosions and seven cases of adverseenvironmental effects. In the United States, the average costfrom significant pipeline damage due to geotechnical incidentssuch as landslides and earthquakes over the past ten years areevaluated at more than $400M/year (twice the damage fromother hazards).

A program started in 2010 at the University of Manitoba tomonitor the strain in the pipelines subjected to slow landslides(Ferreira 2016). For this purpose, over ten Manitoba pipelinenetwork locations visited in Fall 2009, three at-risk landslideareas at St-Lazare Assiniboine River valley, Plum River Cross-ing and Harrowby Assiniboine River valley were selected. The

Mohammad Katebi, Pooneh Maghoul, and James Blatz.1 De-partment of Civil Engineering, University of Manitoba, Winnipeg,MB, R3T2M11 Corresponding author (e-mail: [email protected],[email protected]).

pipelines at these sites are ductile steel gas transmission lineswith the minimum yield strength of 241 MPa for the oldestsections to a maximum of 345 MPA for the newest sections.The pipelines have been under operation for more than 30 yearsat Harrowby, and more than 50 years at Plum River and St-Lazare. The ground displacement, temperature of the ground atthe pipe’s burial depth, temperature of the outer wall of the pipe,groundwater conditions, and longitudinal strain in the pipe’souter wall were instrumented from 2010 to 2015 with hourlymeasurements to analyze the behaviour of pipelines subjectedto slow soil movements.

In this paper, ABAQUS/Standard, a FEM software, is used tomodel the soil-pipeline interaction under the effects of grounddisplacements. The pipeline is represented by Timoshenkobeam elements, and the soil-pipeline interaction is simulated byusing a special-purpose element (PSI element) in ABAQUS/Stan-dard. PSI elements reflect force transfer between the pipelineand soil as a result of their relative movements on the pipelinethrough their stiffness. The stiffness is defined based on soil-pipe interface properties according to Honegger (2017). Theresults of the numerical simulations are compared with fielddata of Ferreira (2016) to examine the behaviour of pipelinessubjected to slow landslide movements and to provide somerecommendations for future monitoring programs in landslidezones.

2. Field Monitoring ProgramAs noted earlier, three at-risk landslide areas of Manitoba

pipeline network are chosen for the field monitoring program:St-Lazare Assiniboine River valley, Plum River Crossing andHarrowby Assiniboine River valley. The pipelines at these sitesare parallel to the slope and the slopes are undergoing ongoing

DOI: 10.1139/Zxx-xxx Published by NRC Research Press

Page 1 of 11



Draft

soil creep. At the Plum River research site, the pipeline is alsosubjected to localized riverbank landslide activities.

These research sites were instrumented for five years, fromSeptember 2010 to August 2015, with hourly monitoring ofground movements, longitudinal pipe strain, pipe/soil tempera-tures and groundwater levels to assess the behaviour of pipelinessubjected to slow landslides. A total of seven slope inclinome-ters (SI), six temperature gauges, twenty-three strain gaugesand ten vibrating wire (VW) piezometers were installed in thethree research sites (Ferreira 2016).

Uniaxial strain gauges were installed on the pipes as theground deformations are parallel to the pipeline’s longitudinalaxis. Two uniaxial strain gauges were attached to each pipe, oneat the top and another at the bottom.

One of the challenges of an instrumentation of this kind is tomeasure the initial conditions of the pipe before the instalmentof strain gauges. It has been known that the pipelines at St-Lazare, and Plum River have been subjected to slow landslidesmany years before the instalment of strain gauges. Therefore,the assessment of initial conditions for the interpretation ofmeasured data is extremely difficult.

The pipelines were cut and replaced at Plum River and St-Lazare which created an opportunity for the research team tomeasure the strain release after the cut. It was reported that thepipeline moved instantly after the cut which is a clear sign ofthe presence of accumulated strain in the pipe. However, nostrain release was picked up by the strain gauges installed onthe pipes: the reasons are explained in detail in Section 3.5. Inaddition, a pipe push test was carried out on the abandonedportion of the pipe to estimate the undrained shear strength ofthe soil-pipe interface. The results of the instrumentation areexplained and discussed in Section 2.4. In the following section,a brief description of each research site is given.

2.1. St-Lazare Research Site2.1.1. Site Location

St-Lazare research site is located on the southern side ofAssiniboine River, which is about 1.3 Km southwest of St-Lazare town. The research site is located between STA:0+00with the UTM coordinates of E335617 m, N5589666 m andSTA:4+25 with the UTM coordinates of E335973 m, N5589898m. The slope is approximately 55 m high and 300 m wide(5.5H:1V) as shown in Figure 1.

The pipeline at the St-Lazare research site has been under op-eration for more than fifty years. The pipeline was installed withan open-cut method on October 5, 1965, with the burial depthof one metre and was backfilled with the excavated material.The pipeline burial depth in the valley varies from a minimumof 0.75 m to a maximum of 4.9 m which is illustrated in Fig-ure 1b. The pipeline is a thin-walled (D/t > 20) ductile steelpipe with 168.3 mm diameter (D) and 4.78 mm wall thickness(t). The material Poisson ratio (ν), linear expansion coefficient(αL), Elastic modulus (E), yield strength (σy) and maximumoperating pressure (Pmax) are detailed in Table 1.

2.1.2. Site Investigation and InstrumentationThree boreholes were drilled for the installation of the slope

inclinometers (SI) and site investigation to the depth of 27.1 m,11.1 m and 20.5 m in the top, middle and bottom of the slope,respectively. The inclinometers were installed in September

Table 1. Pipeline properties used for the numericalsimulation.

Parameters St-Lazare Plum River HarrowbyD (mm) 168.3 88.9 88.9t (mm) 4.78 3.17 3.18ν 0.3 0.3 0.3αL 12× 10−6 12× 10−6 12× 10−6

E (GPa) 207 207 207σy (MPa) 345 241 290Pmax (kPa) 7230 6070 3450

2010. The ground displacement rate was expected to be ex-tremely slow, and as a result, the first inclinometer reading wasscheduled for June 2011. However, the inclinometer movementexceeded expectations (i.e. the ground displacement was muchfaster.)

The thickness of the moving ground can be estimated fromthe depth at which the inclinometers are impassible with the SIprobes. These depths are approximately 7.3 m at the top of theslope and 4 m at the bottom of the slope. According to Figure1, the SI at the top and bottom of the slope are approximately80 m and 240 m away from the slope crest, respectively.

The ground displacement monitoring then relied on surfacemonitoring using 12.5-mm rebars with RTK GPS survey equip-ment. The resolution of the survey was ± 5 mm in the horizon-tal and ± 10 mm in the vertical direction. It was considered thatthe downslope ground movement would far exceed any lateralmovements associated with frost action. As a result, the pin wasinstalled simply by pushing them into the ground without anyeffort to control the effect of frost action on pin movements. Thepin monitoring results for St-Lazare research site is presentedin Table 2. The negative values represent upslope movements,which is likely due to the effect of frost heave, and freeze-thawcycles at the ground surface.

Strain gauges were installed approximately 80 m, 170 m and240 m away from the slope crest according to Figure 1. Onetemperature gauge was attached to the pipeline wall to measurethe pipe temperature, and another used in the ground at theelevation of the pipe to measure the ground temperature.

According to VW piezometer data, the water table fluctuatesbetween the depth of two metres to a maximum depth of six me-tres over the five years of monitoring at the St-Lazare researchsite.

2.2. Harrowby Research Site2.2.1. Site Location

The Harrowby research site is approximately 37 km to thenorth of the St-Lazare research site and located about 12 km tothe west of Russell, Manitoba. The plan view and cross-sectionof the research site are presented in Figure 2. The researchsite is located between the approximate UTM coordinates ofE327346 m, N5626147 m and E327820 m, N5626276 m. Thevalley wall is about 75 m deep and 500 wide with a slope of6.7H:1V.

The pipeline at Harrowby is a steel gas transmission linewith a burial depth of one metre along its right-of-way. Theburial depth varies from a minimum of 0.8 m on the slope to amaximum of 4.1 m under the river as shown in Figure 2. Theproperties of the steel pipe are detailed in Table 1.

2

Page 2 of 11



Draft

Fig. 1. St-Lazare site plan and cross section (Ferreira 2016)

450440

430420

410400

390380

3700 20 100 120

SILTY CLAY

CLAY SHALE ROCK

CLAY SHALE

140 160 180 200 220 240 260 280 300 320 340 360 380 4204008040 60

450440

430420

410400

390380

370

ASSINIBOINE RIVER

PIPELINE

ASSINIBOINERIVER

A B

A) PLAN VIEW

B) CROSS SECTION

STRAIN GAUGES

NB: STA:0+25UTM,NAD83, ZONE 14N5589898, E335973

B: STA:0+00UTM,NAD83, ZONE 14N5589666, E335617

PIN

8

PIN

7

PIN

6

PIN

5

PIN

4

PIN

3

PIN

2

PIN

1

PIN

8

PIN

7

PIN

6

PIN

5

PIN

4

PIN

3

PIN

2

PIN

1 SLOPE CREST

ELEV

ATI

ON

(M

ETR

E)

HORIZONTAL DISTANCE (METRE)

GROUND SURFACE

BOREHOLE DISPLACEMENT DIRECTION

There are many indicators of historical landslides in the area,and heavy rainfall is known to trigger movements in the area.The pipeline is parallel to the slope as can be seen from the to-pographic contours of the plan view. The pipeline was installedwith an open-cut method and backfilled with the excavated ma-terial. The pipeline has been in operation for more than thirtyyears.

2.2.2. Site Investigation and InstrumentationThree boreholes were drilled at the Harrowby research site at

the top, middle and bottom of the slope to a depth of 30 m, 11 mand 30 m, respectively (Figure 2). The top, middle, and bottomboreholes, as well as the strain gauges, are located 140 m, 290m and 410 m down the crest of the slope, respectively. Threeinclinometers installed in November 2010 became inoperablebefore the first reading in June 2011. The depth to the shearingsurface can be estimated from the depth at which the SIs areimpassable with SI probes. The depth is 3.7 m, 18.3 m and 11m at the top, middle and bottom of the slope, respectively. TheSIs are located 140 m, 290 m and 410 m away from the slopecrest, respectively as shown in Figure 2.

The ground displacement was monitored using surface pinswith RTK GPS survey equipment from June 2011 to September2015. The pin monitoring results of the Harrowby research siteis presented in Table 3. Similar to St-Lazare site, the effect offrost action on pins’ movements was not assessed. The neg-ative values in Table 3, which represents the upslope grounddisplacements, are likely due to the effect of frost action on thepins.

2.3. Plum River Research Site2.3.1. Site Location

The Plum River research site is located about 10 km south ofMorris, Manitoba and just at the southwest area of the town ofSt.Jean Baptiste. The valley is about 4 to 5 m deep with a slopeof 3.7H:1V. The plan view and cross-section of Plum Riverresearch site are presented in Figure 3.

The pipeline at Plum River is a steel gas transmission linewith the properties detailed in Table 1. This pipeline has beenunder operation for more than fifty years. The burial depth ofthe pipeline is one metre along its right-of-way. The burial depthvaries from a minimum of 0.7 m on the slope to a maximum of2.5 m under the river.

2.3.2. Site Investigation and InstrumentationA borehole to the depth of 11.8 m was drilled for the site

investigation and installation of a slope inclinometer. The SIinstalled in August 2010 became inoperable within two monthsbefore the first reading. The impassable depth of the inclinome-ters with the SI probe is 3.86 m, which shows the depth to theground movements. The ground displacement at Plum Riverresearch site was not instrumented after the inclinometer failure.However, the pipe’s longitudinal deformation was instrumentedusing two sets of strain gauges; one at X=93 m and another at110 m as shown in Figure 3. A temperature gauge was installedat X=110 m.

2.4. Instrumentation ResultsThe results of the strain gauges of all three research sites

along with an estimation of thermal stress are presented inFigure 4. It is worth mentioning that the thermal stress shown

3

Page 3 of 11



Draft

Table 2. Monitoring pin results: downslope soil movement at St-Lazare (mm) (Ferreira 2016

)

Readinginterval

PIN8 PIN7 PIN6 PIN5 PIN4 PIN3 PIN2 PIN1

2011-05-312013-06-18

547.1 268.3 246.7 146.4 54.7 41.7 23.5 25

2013-06-182013-10-10

-42.9 -19.8 -16.9 -44.4 -219.9 -25.8 -30.6 -347

2013-10-102014-05-29

26.5 32.4 25.6 37.9 - - 34.7 38.8

2014-05-292014-10-07

66.3 18.2 68.1 9.9 137.8 -11.9 -8.9 -

2014-10-072015-06-10

-14.9 -3.5 -28 -28.9 -48 17 3.9

2015-06-102015-09-02

-0.1 14.8 8.9 13.9 - -19.5 -9.4 -

Table 3. Monitoring pin results: downslope soil movement at Harrowby (mm)(Ferreira 2016).Readinginterval

PIN1 PIN2 PIN3 PIN4 PIN5 PIN6 PIN7 PIN8

2011-05-312013-06-18

0.3 67.3 75.7 82.3 107.2 199.9 149.8 135.4

2013-06-182013-10-10

-8.6 -3.2 13.5 4.9 -5.3 2.4 -3.8 0.4

2013-10-102014-05-29

15.4 7.4 10.3 12 10.9 -7.2 15.5 -2.5

2014-05-292014-10-07

-12.6 12.6 24.4 5.6 3.2 25.7 21.4 15.5

2014-10-072015-06-10

15.8 2.4 13.2 5.2 18.5 4.7 29.6 24.7

2015-06-102015-09-02

4.7 -7.8 -14.5 10.4 6.3 -3.7 -4.9 10.0

by the red curve in Figure 4 is calculated based on the data fromthe temperature gauges attached to the pipeline’s wall. Katebiet al. (2018) carried out a thermal analysis using COMSOLMultiphysics to estimate the ground temperature profile usingweather data from Brandon, MB from Sep. 20th, 2011 to Aug.24th, 2016. This established the boundary conditions at theground surface. The numerical simulation suggested that thetemperature of the pipe wall is mainly influenced by seasonalweather conditions.

According to Figure 4, the axial stress captured by the straingauges fluctuates between winter and summer and resemblesthe induced thermal stress. For example, the strain gauge atX=93 m at Plum River was installed in April 2012 so compres-sive stress is generated in the pipe each summer. As anotherexample, the strain gauge at X=293 m at Harrowby was in-stalled in September 2011. Tensile stress is induced in the pipein winter due to the contraction of the steel. These rises andfalls in the stress profile due to the seasonal temperature varia-tion is illustrated in Figure 4, however, no strain due to graduallandslide is visible from the curves. The reason of this will bediscussed in detail in Section 3.

The initial rise of the stress at the beginning of the instrumen-tation following installation for two cases, shown by the greenand purple colours in Figure 4, is due to the effect of backfillingon the pipeline. The curves then follow the same trend drivenby seasonal effects, rising in winter and falling in summer.

As mentioned earlier, the strain gauges did not show anystrain release when the pipelines were cut at Plum River andSt-Lazare research sites. However, an instant movement of thepipe after the cut was observed. The strain gauges were installedin summer, and the pipe was cut in summer. Comparing thetemperature of the pipe after the installation of the tempera-ture gauge with the temperature of the pipe just before the cutshowed that almost no thermal stress existed in the pipe at thetime of the cut. As a result, the movement of the pipe clearlyshows that there was accumulated elastic strain locked in thepipe.

3. Numerical SimulationA numerical study is carried out using ABAQUS/Standard

to analyze the stress induced in a buried pipeline due to grounddeformations in slow landslides. The pipeline is modelled usingTimoshenko beam element using PIPE21, which is a planarpipe element. PIPE21 is used for the simulation because theslopes are all parallel to the pipe and the transverse relativemovements of soil and pipe are considered negligible accordingto the instrumentation. We note that the Timoshenko beams aregood for dealing with large axial strain but the axial strain dueto torsion should be small. The soil-pipeline interaction is mod-elled using PSI elements. A PSI element is a special-purpose

4

Page 4 of 11



Draft

Fig. 2. Harrowby site plan and cross section (Ferreira 2016)

A) PLAN VIEW

B) CROSS SECTION

PIPELINESTRAIN GAUGES

400

380

440

420

480

460

500

400

380

440

420

480

460

500

0 50 100 150 200 250 300 350 400 450 500 550

488

486

484

482

480

478

476

474

472

470

468

466

464

462 458

460 456 452

450

448

446

444

442

440

438

436

434

422

420

418

416

414

412

410

432

430

428

426

424

454

N

AB

ASSINIBOINE RIVER

B: STA: 4+75UTM,NAD83, ZONE 14N5626276, E327820

B: STA:0+00UTM,NAD83, ZONE 14N5626147, E327346

PIN

1

PIN

1

PIN

2

PIN

2

PIN

3

PIN

3

PIN

4P

IN4

PIN

5P

IN5

PIN

6

PIN

6

PIN

7

PIN

7

PIN

8

PIN

8

SILTY CLAY

SAND

CLAY SHALE ROCK

CLAY SHALE

ELEV

ATI

ON

(M

ETR

E)


TOPOGRAPHY CONTOURS

SLOPE CREST

GROUND SURFACE

BOREHOLE DISPLACEMENT DIRECTION

element which has only displacement degrees of freedom atits nodes. The far-field edges of the PSI elements resemble thesurface deformation of the ground, and the close-field edges ofthe PSI elements share nodes with the pipe elements as shownin Figure 5 (ABAQUS 2017). The ground displacements, mea-sured during the instrumentation from 2010 to 2015 (Tables 2and 3), are imposed on the pipeline through pipe-soil interactionelements (PSI). The force on the pipeline due to the grounddeformation is transferred to the pipeline through the stiffnessof the PSI elements along the pipeline’s length. The interactionbetween pipe and soil can numerically be modelled in fourdifferent directions: axial (longitudinal), transverse horizontal,vertical upward, and vertical downward. The stiffness of the PSIelements in these directions can be determined using Honegger(2017). The ground deformation in the transverse horizontaldirection is considered to be negligible because the pipeline’slongitudinal axis is parallel to the slow landslides at all three re-search sites. Also, the soil uplift in the lower reach of the slopeand the soil subsidence in the upper reach of the slope, whichusually occurs in landslides, are not considered in the numericalsimulation for simplicity and given that the instrumentation didnot measure them.

The axial soil-pipeline interaction based on the total andeffective stresses is defined by Equations 1 and 2, respectively,according to Honegger (2017):

- Axial soil-pipeline interaction for sand (effective stressmodel):

[1]tu = πDγH

(1 + k0

2

)tan (kφ)

xu = 3 ∼ 5mm

in which tu is the ultimate axial force applied per unit lengthof pipeline; D is the pipeline diameter; γ is the effective unitweight of soil; H is the depth to the centre line of pipeline;k0 is the coefficient of earth pressure at rest; k is the frictionfactor; φ is the soil friction angle; and xu is the ultimate relativedisplacement.

- Axial soil-pipeline interaction for clay (total stress model):

[2]tu = πDαSu

xu = 8 ∼ 10mm

in which α is the adhesion factor; Su is the undrained shearstrength of the soil.

3.1. Effects of Ground Displacement Rate on Soil-Pipeline Interface

The excess pore water pressure can be dissipated throughthe soil-pipe interface, which is about ∼ 2 mm according toWijewickreme et al. (2009), depending on the landslide speed.The rate of landslide movement in which the interface behavesin a drained manner is important to select the appropriate modelfor the interface. In this paper, Equation 3 is used to determinethe degree of drainage condition of the interface according toPaulin (1998):

[3]Vn =

vD

cv

where Vn is a normalized displacement parameter; v is thedisplacement rate; and cv is the consolidation coefficient. The

5

Page 5 of 11



Draft

Fig. 3. Plum River site plan and cross-section (Ferreira 2016)

PIPELINE

N

PLUM RIVER

SILTY CLAY

230230.5

230.5 230

231 231.5

233.5

232

232.5

233

PLUM RIVER

226224222220218

228230

234236

232

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 100 105959080 85 130125120115110 135 140 145 150

234 233.5 233 232.5 232

231

231.5

A) PLAN VIEW

B) CROSS SECTION

226224222220218

228230

234236

232

ELEV

ATI

ON

(M

ETR

E)


TOPOGRAPHY CONTOURS DISPLACEMENT DIRECTIONGROUND SURFACE

TEMPERATURE GAUGESTRAIN GAUGEABONDONED STL BOREHOLE

loading is drained if Vn < 0.1, and undrained if Vn > 10.The values of consolidation coefficient are usually between1 m2/year to 10 m2/year. If we use the lower-bound valueof 1 m2/year for the consolidation coefficient, the maximumrate of displacement in which the interface is drained wouldbe 50 mm/month at St-Lazare and 93 mm/month at Har-rowby and Plum River. The maximum displacement rates moni-tored were 3.75 mm/month at St-Lazare research site, and 5.2mm/month at Harrowby research site. Therefore, a drainedmodel is considered appropriate for the interface. It is worthnoting that the field full-scale tests on longitudinal soil-pipe in-teraction carried out by Cappelletto et al. (1998) showed that theeffective stress model should always be used regardless of thespeed of the landslide movement due to the the narrow thicknessof the interface. According to their work, the excess pore waterpressure is easily dissipated through the thin soil-pipe interfacealong the pipe’s longitudinal axis even in case of a relativelyrapid landslide. For the aforementioned reasons, the drainedmodel (Equation 1) is used for the numerical simulation.

3.2. Domain Extent and Boundary ConditionsThe pipeline should be simulated with sufficient length to

ensure that the imposed pipe loading are balanced out by thesoil-pipe interaction outside of the landslide zone. To estimatethe necessary length of the pipe for simulation, Equation 4introduced by Honegger (2009) can be used:

[4]Lanchor =

πDtσytu

where Lanchor is the necessary length of the pipeline outsidethe landslide zone to balance out the axial soil force, tu; t is thepipe wall thickness; σy is the pipe yield strength; and tu is theultimate axial force on the pipe.

The extent of the model should be long enough such that theboundary conditions at both ends of the model do not affect theresults of the analysis. Furthermore, the length of the pipelineshould be sufficient so that the axial strain originating from thelandslide dissipates within boundaries in the model.

Using Equation 4, the minimum length of the anchor at eachside of the landslide is calculated being 585 m. The computationof the analysis is not affected much by increasing the extentof the anchor zones because the pipeline behaviour at theselocations is elastic. As a result, the pipeline length of 2000 m isused to ensure that the noted boundary conditions are satisfied.Both ends of the pipeline are fixed to prevent the pipe fromrigid motion.

3.3. Pipeline and Soil PropertiesThe pipeline and soil properties used in the numerical analy-

sis are presented in Tables 1 and 4, respectively. The Atterberglimits of soils at the three research sites are provided in Table5. In this table, LL stands for Liquid Limit and PL stands forPlastic Limit. In Table 4, γ is the bulk unit weight of soil, k0is the the coefficient of earth pressure at rest, φ

′is the drained

internal friction angle of soil, k is the friction factor of the soil-pipe interface, and xu is the ultimate relative displacement ofthe soil-pipe interface. As was presented earlier, the effectivestress model is used for the soil-pipeline interface. The bulkunit weight of soil is used in the simulation due to the groundwater table being below the pipeline, and therefore not affectingthe model.

The drained internal friction angle of 15◦ is assumed forSt-Lazare and Plum River sites in relation to the slope gradesat these two sites being 10.3◦ and 15◦, respectively. The soilcreep would generally not be expected if the internal frictionangles were higher than the slope grades. However, to be con-servative, the drained friction angle of 15◦ was assumed forthe simulation in these sites. For Harrowby, the slope gradeis 8.5◦ and the drained friction angle of 13◦ was used for the

6

Page 6 of 11



Draft

Fig. 4. Longitudinal stress captured by the strain gauges (Ferreira 2016) and the thermal stress

2011

-09-01

2011

-10-31

2011

-12-30

2012

-02-28

2012

-04-28

2012

-06-27

2012

-08-26

2012

-10-25

2012

-12-24

2013

-02-22

2013

-04-23

2013

-06-22

2013

-08-21

2013

-10-20

2013

-12-19

2014

-02-17

2014

-04-18

2014

-06-17

2014

-08-16

2014

-10-15

2014

-12-14

2015

-02-12

2015

-04-13

2015

-06-12

2015

-08-11

- 8 0- 6 0- 4 0- 2 0

02 04 06 08 0

1 0 01 2 01 4 01 6 01 8 02 0 02 2 02 4 02 6 0

P l u m R i v e r ( b e y o n d s l o p e ) P l u m R i v e r ( a l o n g s l o p e ) S t - L a z a r e ( t o p ) S t - L a z a r e ( b o t ) S t - L a z a r e ( m i d d l e ) H a r r o w b y ( b o t ) H a r r o w b y ( m i d d l e ) H a r r o w b y ( t o p ) T h e r m a l s t r e s s

b a c k f i l l i n g

Stres

s (MP

a)

D a t e

b a c k f i l l i n g

Fig. 5. Pipeline-soil interaction (PSI) element

1

2

3

4

8

76

5

H=0.5(H1+H2)

H1

H2

PSI ELEMENT=1,2,6,5

PIPELINE DISCRETIZED WITH BEAM-TYPE ELEMENT

GROUND SURFACE

Table 4. soil properties used for the numerical simulationResearch sites γ kN/m3 k0 φ

′k xu (mm)

St-Lazare 16.8 0.8 15 0.7 5Plum River 16.9 0.8 15 0.7 5Harrowby 17.9 0.8 13 0.7 5

simulation. Correlations between drained friction angles andAtterberg limits are used according to Kanji (1974) to verifythat the assumed drained friction angles are reasonable. The k0coefficient for clay was chosen accrording to Mesri and Hayat(1993).

3.4. Result of Numerical SimulationThe pipeline is modelled with 2000 beam-type one-metre-

long elements. The burial depth of the pipeline is assumed tobe one metre for all nodes outside of the landslide area; inthe valley, the varying burial depth is used for the simulationaccording to Figures 1b, 2b, and 3b.

Table 5. Atterberg Limits of soil (Ferreira 2016)Site Soil LL PLSt-Lazare Silty Clay 73 26St-Lazare Clay Shale 99 21Harrowby Silty Clay 65 19Plum River Lacustrine Clay 83 27

3.4.1. St-Lazare Research SitesThe downslope ground displacement that is used in the nu-

merical simulation is presented in Figure 6. The horizontalaxis of this figure represents the distance from the slope crest(x=0 m) and the vertical axis shows the cumulative downslopeground movements relative to 2013-10-10. As shown in Figure6, the displacements at the slope crest (x=0) and toe (x=300)are assumed to be zero. The displacements at different locationsalong the slope are linearly interpolated using the data in Table2 and are used as boundary conditions at each PSI node in thesimulation. We note that the negative values of pin displace-ments in Tables 2 and 3 represent upslope pin movements. Theupslope movement likely occurred in the pins due to the frostaction in the ground. These negative values are ignored in thenumerical simulation because there is no possibility that thesoil moves upslope at the pipe’s burial depth. We note that thefrost depth was above the pipeline during the five years of in-strumentation so there is no need to consider the effect of frostaction on the pipeline.

The displacement corresponding to the second interval, 2013-06-18 to 2013-10-10, is not used in the simulation because theyare all negative. The displacement corresponding to the first in-terval, 2011-05-31 to 2013-06-18, is also not considered for thesimulation as the period between the readings is extensive thatmakes the reading unreliable. The average ground displacementrate between 2013-10-10 and 2015-09-02 is used to estimate

7

Page 7 of 11



Draft

Fig. 6. Downslope ground displacement versus longitudinal axisof the pipe at St-Lazare

0 3 0 6 0 9 0 1 2 0 1 5 0 1 8 0 2 1 0 2 4 0 2 7 0 3 0 0

0

1 0 0

2 0 0

3 0 0

4 0 0

5 0 0

6 0 0

7 0 0

2 0 1 4 - 0 5 - 2 9 2 0 1 4 - 1 0 - 0 7 2 0 1 5 - 0 6 - 1 0 2 0 1 5 - 0 9 - 0 2 2 0 2 3 - 1 0 - 1 0 0 . 6 - m b l o c k - p a t t e r n d i s p l a c e m e n t 0 . 7 - m b l o c k - p a t t e r n d i s p l a c e m e n t

Down

slope

grou

nd di

splac

emen

t rel

ative

to 20

13-10

-10 (m

m)

X ( m )

the cumulative ground displacement in 2023. Furthermore, bytrial and error, it is found that at the minimum block-patterndisplacement of 0.6 m, the ultimate interface loading capacityhas reached in all PSI elements. Increasing the ground displace-ment from 0.6 to 0.7 m does not increase the soil loading onthe pipeline because at this displacement, all interface elementsare plastic. The results of the simulation for these two cases canbe compared in Figure 7.

The longitudinal stress (S11) is increased as the displacementaccumulates over time from 2013-10-10 to 2023 as shown inFigure 7. The result of the simulation shows that the longitudinalstress is at its maximum when the block-pattern displacementof 0.6 m is used. At this displacement, the soil springs used formodelling the soil-pipeline interaction become plastic and asa result, the increase of the displacement does not result in anincrease of the loading.

According to the analysis, the pipeline at St-Lazare can with-stand the maximum soil loading of the 300-m-moving slopewith its elastic capacity, i.e. the maximum stress in the pipe islower than the pipe’s yield stress. As a result, the pipeline goesinto an ultimate deformed condition after which the soil slidesover the pipe all along the slope and the imposed forces are notgoing to be increased.

The pipeline, installed in 1965 at St. Lazard site, has beensubjected to slow landslides for many years (∼ 50 years) be-fore the beginning of the instrumentation. This implies that theultimate condition that is shown by 0.6-m/0.7-m block-patterndisplacement has happened in the pipe before the start of theinstrumentation. That the pipeline had reached its ultimate de-formed condition at St-Lazare due to the gradual landslidebefore the start of the instrumentation justifies why the straingauges did not pick any landslide-related strain.

3.4.2. Harrowby Research SitesThe downslope ground displacement data that is gathered

from the instrumentation at Harrowby is presented in Table 3.The ground displacement during the first interval, 2011-05-31to 2013-06-18, is not used in the simulation because the readinginterval was too long. The negative values of pin displacementsat the ground surface are likely due to frost action and conse-quently are not considered for the numerical simulation. Other

Fig. 7. Longitudinal stress due to slow downslope grounddisplacement at St-Lazare

- 3 0 0 - 2 0 0 - 1 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0- 2 0 0

- 1 5 0

- 1 0 0

- 5 0

0

5 0

1 0 0

1 5 0

C o m p r e s s i v e s t r e s s

Botto

m str

ain ga

uge

Midd

le str

ain ga

uge

2 0 1 4 - 0 5 - 2 9 2 0 1 4 - 1 0 - 0 7 2 0 1 5 - 0 6 - 1 0 2 0 1 5 - 0 9 - 0 2 2 0 2 3 - 1 0 - 1 0 0 . 6 - m b l o c k - p a t t e r n d i s p l a c e m e n t 0 . 7 - m b l o c k - p a t t e r n d i s p l a c e m e n t

Long

itudin

al str

ess (

MPa)

X ( m )

Top s

train

gaug

e

T e n s i l e s t r e s s

values of pin monitoring data provided in Table 3 are used toestimate the ground displacement at each PSI nodes along theslope using a linear interpolation.

The downslope ground deformation at Harrowby that is usedin the numerical simulation is presented in Figure 8. The ver-tical axis is the downslope ground displacement after 2013-06-18 and the horizontal axis represents the distance from thecrest. The average displacement rate between 2013-06-18 and2015-09-02 is used to estimate the displacement in 2018 and2024. Furthermore, two arbitrary block-pattern and one triangle-pattern displacements are used to study the pipeline behaviourunder different loading conditions. Results of the numericalanalysis is presented in Figure 9. Each curve of this figurecorresponds to the displacement patterns shown in Figure 8.Accordingly, the longitudinal stress is increased as the displace-ment accumulates over time from 2013-06-18 to 2024. By trialand error, it is found that the maximum soil loading occurs atthe 0.3-m-block-pattern displacement. The longitudinal stressin the pipe is not increased as the block-pattern displacement isincreased from 0.3 to 0.4 m as can be seen from Figure 9. Thisis due to the fact that the interface is modelled using a linearelastic perfectly plastic behaviour.

We note that the pipeline at Harrowby experiences a plasticdeformation under the ultimate loading condition (dashed blackand yellow lines in Figure 9), i.e. the maximum stress in thepipe is beyond the yield stress of the pipe. It is worth notingthat the maximum compressive stress shown in Figure 9 is S11.The maximum Von Mises stress is slightly over the yield stressof the pipe which leads to a plastic deformation in the pipe.The excessive deformation of the pipe at Harrowby eventuallyterminates the analysis due to convergence problems. This ishappening because the linear elastic perfectly plastic behaviouris assumed for the steel and the steel hardening after the yieldpoint is ignored in the analysis. To be able to solve the problemwithout changing the elastoplastic model, the yield stress ofthe pipe is increased from 290 to 350 MPa for the last twoanalysis (0.3 and 0.4 m block-pattern displacements). Since the

8

Page 8 of 11



Draft

Fig. 8. Downslope ground displacement versus longitudinal axisof the pipe at Harrowby

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0

0

1 0 0

2 0 0

3 0 0

4 0 0

5 0 0

6 0 0

2 0 1 3 - 1 0 - 1 0 2 0 1 4 - 0 5 - 2 9 2 0 1 4 - 1 0 - 0 7 2 0 1 5 - 0 6 - 1 0 2 0 1 5 - 0 9 - 0 2 2 0 2 4 2 0 3 7 R i d g e p a t t e r n 0 . 3 - m b l o c k - p a t t e r n d i s p l a c e m e n t 0 . 4 - m b l o c k - p a t t e r n d i s p l a c e m e n t

Down

slope

grou

nd di

splac

emen

t rel

ative

to 20

13-06

-18 (m

m)

X ( m )

strain gauge data shown by green lines in Figure 4 does notshow any progressive trend over time, it can be interpreted thata plastic deformation is not occurring in the pipe. If a plasticdeformation had occurred in the pipe, the pipe would havecontinued to deform by the constant soil loading, and the straingauges would have measured that behaviour. That the numericalanalysis estimated a plastic deformation in the pipe while theinstrumentation showed differently is due to the fact that thesteel hardening after the yield point is ignored in the numericalanalysis.

According to the dashed black and yellow lines of Figure 9,which are for the ultimate loading condition, the stress decreasesoutside of the landslide zone until it reaches zero about 300 maway from the crest of the slope and 200 m away of the slopetoe. These two zones, one at the top side of the slope (x=-300 to0) and the other at the bottom side of the slope (x=500 to 700),are called anchor zones. In these zones, the pipe movementis restrained by the pipe-soil interaction forces. Anchor zonesare simulated using the same soil spring models assuming thatthe soil pipeline interaction behaviour in the landslide zoneand outside of the landslide zone is similar. At anchor zones,the far edges of the PSI elements that represent the grounddisplacement are fixed (zero displacement) while the near edgesof the PSI elements that share nodes with the pipe are beingdisplaced by the pipe deformation. The loading from the stablesoil to the deformed pipe is estimated through the properties ofsoil springs and is transferred to the pipe as a resistant load tothe pipe’s movement. At lower reach of the slope, because thepipe is buried deeper (Figure 2), the soil pipeline interaction isstronger and so 200 m length of the pipe is deformed beforethe landslide forces are balanced out. In the upper reach of theslope, where the pipeline is buried shallower (Figure 2), the soilpipeline interaction is weaker that leads to a longer length ofthe pipeline being affected by the landslide.

3.4.3. Plum River Research SiteAs mentioned in Section 2.3.2, the ground displacement

is not monitored at Plum River so the analysis is carried out

Fig. 9. Longitudinal stress due to slow downslope grounddisplacement at Harrowby

- 3 0 0 - 2 0 0 - 1 0 0 0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0

- 3 0 0

- 2 5 0

- 2 0 0

- 1 5 0

- 1 0 0

- 5 0

0

5 0

1 0 0

1 5 0

2 0 0

2 5 0


2 0 1 3 - 1 0 - 1 0 2 0 1 4 - 0 5 - 2 9 2 0 1 4 - 1 0 - 0 7 2 0 1 5 - 0 6 - 1 0 2 0 1 5 - 0 9 - 0 2 2 0 2 4 2 0 3 7 R i d g e p a t t e r n 0 . 3 - m b l o c k - p a t t e r n d i s p l a c e m e n t 0 . 4 - m b l o c k - p a t t e r n d i s p l a c e m e n t

Long

itudin

al str

ess (

MPa)

X ( m )

Top s

train

gaug

e

Midd

le str

ain ga

uge

Botto

m str

ain ga

uge


with three different arbitrary displacement patterns as shownin Figure 10. These three arbitrary displacement patterns areassumed with regards to the cross section of the site (Figure 3).In the first analysis, a ground displacement of 1 mm in areaswith a slope greater than 5% is imposed on the pipe as shownwith the green line in Figure 10. The analysis is repeated withan arbitrary displacement of 250 mm (yellow line) and also1000 mm (dashed red line) for slope greater than 5% to studythe pipeline behaviour under the ultimate soil loading.

According to Figure 11, the longitudinal stress is not in-creased as the displacement is increased from 250 to 1000 mm.This implies that at the displacement of 250 mm , the interfaceis plastic everywhere on the pipe and the soil slides over thepipe. The longitudinal stress in the pipe is very low in the ulti-mate case which implies that the slow landslide is not an issuefor the pipe.

Tensile stress is induced in the pipeline at both sides of theriver because the pipe is surrounded by the crests of valley wallsat both northern and southern sides of the river. According toFigure 3, the slope (greater than 5%) in the southern side of theriver is longer compared to the northern wall. This is why thestress in Figure 11 is not symmetric relative to the river. It isworth mentioning that the pipeline is under compressive stressat the centre of the valley (X=70 to 85 m) because the groundmoves toward the river from both walls.

The effects of a 15-m-long rapid landslide at the southernvalley wall in Plum River research site is studied using the totalstress model (Equation 2). This is done because some tracesof river bank activities have been reported from a site visitby Ferreira (2016). The clay model is used for this purposebecause the river bank activities usually occur rapidly and moreimportantly, the clay model gives conservative results becauseof the high value of the undrained shear strength of clay at thissite. We note that αSu of 45 kPa is used based on the St-Lazarepush test. The results are presented in Figure 12. Accordingly,the maximum tensile stress of 105 MPa occurs at the crest ofthe slope at X=100 m, and the maximum compressive stress

9

Page 9 of 11



Draft

Fig. 10. Downslope ground displacement versus longitudinal axisof the pipe

0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0- 1 2 0 0

- 8 0 0

- 4 0 0

0

4 0 0

8 0 0

1 2 0 0 1 m m 2 5 0 m m 1 0 0 0 m m

Arbitra

ry do

wnslo

pe gr

ound

dis

place

ment

(mm)

X ( m )

Fig. 11. Longitudinal stress due to slow downslope grounddisplacement in Plum River research site

- 4 0 - 2 0 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0 2 0 0- 1 4- 1 2- 1 0- 8- 6- 4- 202468

1 0


Botto

m str

ain ga

uge

Midd

le str

ain ga

uge 1 m m

2 5 0 m m 1 0 0 0 m m

Long

itudin

al str

ess (

MPa)

X ( m )

Top s

train

gaug

e


of -105 MPa occurs at the toe of slope at X=84 m. The anchorzones are formed from X=40 m to X=85 m and X=100 m toX=150 m according to Figure 12.

The result of the rapid landslide at Plum River is somewhatsymmetric because block-pattern displacements are used forthe landslide and the depth of the pipeline in this location is notvarying much. This analysis suggests that the pipeline at PlumRiver is able to withstand the soil loading from rapid river bankactivities with its elastic capacity.

3.5. Discussion and ComparisonAs was mentioned in Section 2, to estimate the initial condi-

tions of the pipeline, the research team used the opportunity ofa scheduled pipeline cut in St-Lazare and Plum River researchsites. However, no strain release was measured by the straingauges. This is likely due to the anchoring effect of the pipelinein the soil. At Plum River, the pipeline was cut in a valve about50 m away from the strain gauge. The location of cut was be-yond the crest of the slope. At St.Lazare, the location of cutwas about 100 m away from the bottom strain gauge close tothe river . After the cut, the pipeline tends to move to relieveits stress while the soil resists to the pipeline’s movement. Asa result, depending on the soil-pipe interaction and the appliedforce, the strain is released only in a limited length from the

Fig. 12. Longitudinal stress due to rapid landslide in Plum Riverresearch site

4 0 5 0 6 0 7 0 8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0- 1 2 0

- 9 0

- 6 0

- 3 0

0

3 0

6 0

9 0

1 2 0


Long

itudin

al Str

ess (

MPa)

X ( m )


cut. This length is calculated for both St-Lazare and Plum Riverresearch sites as follows: the pipeline was cut at x = 350m andx = 0 m at St-Lazare and Plum River, respectively. Accordingto Figures 7 and 11, the compressive stress in the pipeline inthose sections are about 130 MPa and 0.5 MPa, respectively.The equivalent forces in these sections are calculated to be329 kN and 450 N for St-Lazare and Harrowby research sites,respectively. In case of a pipeline cut, the total stress model(Equation 2) should be used for the interface because cutting apipeline is a rapid procedure. Using Equation 2, the soil resis-tance to pipeline deformation in undrained condition is equal totu = 23.75 kN/m and tu = 12.5 kN/m for St-Lazare and PlumRiver sites, respectively. We note that αSu of 45 MPa is usedaccording to the St-Lazare push test. Based on this calculation,the strain release occurs only to a distance of 13.85 m at bothsides of the cut at St-Lazare and 0.036 m at Plum River. Thestrain gauges placed 100 m away from the cut at St-Lazareand 50 m away at Plum River did not capture any strain reliefsimply because they were too far from the cut. This interpreta-tion justifies the observation regarding the instant movement ofthe pipeline while no strain relief was measured by the straingauges.

4. ConclusionThe effects of slow landslides on buried pipelines in three

Manitoba pipeline sites are studied with the use of ABAQUS/Standard.The pipe is simulated using Timoshenko beam elements witha linear elastic perfectly plastic behaviour. The soil-pipelineinterface is modelled with PSI elements. The stiffness of PSIelements is defined according to Honegger (2017). The moni-tored landslide movements are imposed on the PSI elements asdisplacement boundaries. The pipeline behaviour subjected tothe monitored ground displacement is examined. Furthermore,arbitrary block-pattern displacements are used to analyze thepipeline behaviour under different loading conditions.

It is shown that the pipelines at two of the three sites, PlumRiver and St-Lazare, withstand the landslide loading withoutexperiencing any plastic deformation. At these sites, the load-ing due to the slow landslides increases over time as the dis-placements accumulate until the maximum frictional interface

10

Page 10 of 11



Draft

capacity is reached. At this stage, the soil slides over the pipeas the interface becomes plastic and as a result, the soil loadingremains constant.

At Harrowby, the simulation does not converge under theultimate loading because the pipeline at the toe of the slope ex-periences plastic deformation. The plastic deformation contin-ues to increase over the analysis steps until the analysis abortedat some point due to the excessive deformation of the pipe.To be able to solve the problem numerically without changingthe elastoplatsic behaviour of the pipe, the yield stress of thesteel is increased from 290 MPa to 350 MPa and the resultsare compared to the strain gauge data. Because a progressivetrend can not be seen from the green lines in Figure 4, we canassume that the pipeline at Harrowby has not experienced aplastic deformation over its many years of operation. If the pipehad reached its plastic behaviour under the soil loading, theslow landslide would have induced progressive strain in thepipe and the strain gauges would have picked them up. Thisoverestimation in the pipeline behaviour was expected to occuras the steel hardening after the yield point is ignored using alinear elastic perfectly plastic pipe behaviour in the simulation.

By comparing the stress in the pipe from the instrumentationwith those of the numerical simulations, and the calculated ther-mal stress, we can realize that the effect of the slow landslide onthe pipeline is not captured by the strain gauges. The numericalsimulation suggests that the pipeline would go into an ultimatedeformed condition under which the soil slides over the pipe(the interface becomes plastic). At this stage, the soil loading re-mains constant throughout the life of the pipe given that the pipewas able to withstand the ultimate loading condition withoutdeforming plastically. Given the fact that the strain gauges wereinstalled on the pipes that have already went into the ultimatedeformed condition clearly explains why the strain gauges didnot capture any of the soil loading effects on the pipe.

It is explained that cutting the pipeline is not a practicalway to estimate the initial condition of the pipe as the pipe isanchored in the soil and the soil resistant to pipe movementprevents the release of the locked-in strain.

At Plum River, a 15-m-long riverbank slide is the largest pos-sible landslide. The numerical simulation showed that the pipebehaves elastically subjected to such ground movements. Nu-merical analysis shows that the maximum tensile stress occursat the top and the maximum compressive stress at the bottomof the slope. There is a transition from tensile to compressivestress along the slope. Anchor zones form at the top and bot-tom sides of the slope to balance out the landslide loading onthe pipe. The extent of the anchor zones is related to soil-pipeinterface properties and the length of the moving ground.

For instrumentation, it is said that the stress condition of thepipeline should be examined before the installation of straingauges. If the pipeline is scheduled to be cut to capture thestress release, the position of the cut should be close to theslope, and also, the strain gauges should be placed next to thecut. However, this practice is not recommended. It is better toinstall the strain gauges either on a newly constructed pipelineor to remove the soil cover over the entire length of the pipein the active zone and also in the passive zone to ensure thelocked-in strain is released before the instalment of the straingauges.

AcknowledgmentsThis research was supported by the NSERC Discovery Grants

Program and the University of Manitoba, Graduate Enhance-ment of Tri-Council Stipends (GETS) program.

ReferencesABAQUS. 2017. SIMULIA User Assistance. Dassault

Systemes Simulia, Corp., Johnston, RI, USA.

Cappelletto, A., Tagliaferri, R., Giurlani, G., Andrei, G., Furlani,G., and Scarpelli, G. 1998. Field full scale tests on longitu-dinal pipeline-soil interaction. In 2nd International PipelineConference, American Society of Mechanical Engineers, pp.771–778.

Ferreira, N.J. 2016. Risk to buried gas pipelines in landslideareas. Ph.D. thesis, Department of Civil Engineering, TheUniversity of Manitoba, Winnipeg, MB.

Honegger, D.G. 2009. Guidelines for constructing natural gasand liquid hydrocarbon pipelines through areas prone to land-slide and subsidence hazards. Technical report, Pipeline Re-search Council International, Inc., Virginia. Catalogue No:L52292.

Honegger, D.G. 2017. Pipeline seismic design and assessmentguideline. Technical report, Pipeline Research Council Inter-national, Inc., Virginia. Catalogue No: PR-268-134501-R01.

Kanji, M. 1974. The relationship between drained friction an-gles and atterberg limits of natural soils. Geotechnique, 24(4).

Katebi, M., Liu, H., Maghoul, P., and Blatz, J. 2018. The op-timum pipeline burial depth considering slow downslopesoil movement and seasonal temperature variation. In 12thInternational Pipeline Conference, American Society of Me-chanical Engineers, p. V002T02A015.

Mesri, G. and Hayat, T. 1993. The coefficient of earth pressureat rest. Canadian Geotechnical Journal, 30(4): 647–666.

Oswell, J.M. 2016. Soil mechanics for pipeline stress analysis.Naviq Consulting Inc.

Paulin, M.J. 1998. An investigation into pipelines subjectedto lateral soil loading. Ph.D. thesis, Department of CivilEngineering, Memorial University of Newfoundland, MountPearl, NL.

Wijewickreme, D., Karimian, H., and Honegger, D. 2009. Re-sponse of buried steel pipelines subjected to relative axial soilmovement. Canadian Geotechnical Journal, 46(7): 735–752.

11

Page 11 of 11



numerical analysis of pipeline response to slow landslides ......ters (si), six temperature gauges,...

Documents