Вы находитесь на странице: 1из 6

Frontiers in Offshore Geotechnics III Meyer (Ed.

)
2015 Taylor & Francis Group, London, ISBN: 978-1-138-02848-7

Evaluation of pipe-soil interaction in liquefied soil


A. Murianni, O. Zanoli & E.J. Parker
DAppolonia S.p.A, Genoa, Italy

ABSTRACT: This paper deals with the pipe-soil interaction of pipelines subjected to horizontal loading in
liquefied soil. Lateral response of piles (p-y curves) in liquefied soil has been studied by a number of authors,
but these approaches cannot be directly used for soil-pipe interaction given the different failure mechanism.
This paper analyses soil-pipe interaction using an elasto-plastic model implemented in the TOCHNOG finite
element software. Liquefied conditions are simulated changing the pore pressure field, after the geostatic stage.
Afterwards, the pipe is moved in horizontal direction with a given displacement rate allowing the evaluation
of the soil reaction forces (soil springs). The soil-pipe FEM response is compared to literature results for nonliquefied and liquefied conditions. Moreover, hysteretic loops consisting of loading, unloading and re-loading
phases were simulated to assess damping as a function of displacement.

INTRODUCTION

Safe design of a buried pipeline cannot disregard a


reliable evaluation of the forces that the soil can exert
on the pipe due to relative pipe-soil displacements.
Pipelines along unstable slopes, crossing an active
fault plane, or embedded in liquefiable soil layers are
clearly affected by this kind of overloads that can lead
to pipe failure.
Buried pipe-soil interaction is extensively studied
for static and cyclic conditions in competent soil (e. g.
Meyerof & Adams 1968, Neely et al. 1973, Audibert &
Nymann 1977, Rowe & Davis 1982, Matyas & Davis
1983,Trautman & ORourke 1983, 1985, Dickin 1994,
Aprea & Luisi 1999, Scarpelli et al. 2003, di Prisco &
Galli 2006), but almost no references are available on
pipeline response in liquefied soil, in particular for the
lateral displacement.
Several methodologies were developed for lateral pile response (p-y curves) in liquefied soil,
but these approaches cannot be directly used for
pipe-soil interaction given the different soil failure
mechanism.
In this work the pipe-soil interaction is analyzed
using a finite element model, implemented in the
TOCHNOG code. The FEM predictions were compared to literature results for non-liquefied and liquefied conditions.
Moreover, hysteretic loops consisting of loading,
unloading and re-loading phases were simulated to
obtain a damping versus displacement curve. In order
to validate the FEM results, the hysteresis loops
were also calculated from p-y curve (backbone curve)
following the modified Masing Rule.

2D FINITE ELEMENT MODEL

The soil-pipe interaction was evaluated using a 2D


finite element model. The model was implemented
using the TOCHNOG finite elements code (FEAT,
2004). A 32 (0.812 m) diameter pipe is considered.
The distance from top of model (mudline) to center of
spool is 1 m. This results in a soil cover of 0.6 m over
the pipe. The pipe was considered as an homogeneous
linear elastic material.
The soil was modeled as a homogeneous, isotropic,
elasto-plastic material with Mohr-Coulomb failure criteria and a non-associated flow rule. Elastic properties
vary with depth.
The soil-pipe interface was modeled by means of
interface elements, allowing relative slip between soil
and pipe with a reduced interface friction angle () of
2/3 of soil internal friction angle ( ).
Figure 1 shows the 2D finite element mesh used to
analyze the soil-pipe interaction problem. The model
consisted of six-node, triangular elements for both the
soil and the pipe.
The horizontal dimensions of the model were chosen to be sufficiently large to fully develop the failure
mode around the pipe. The final model is 12 m wide.
The mesh consisted of 450 elements and 941 nodes.
40 nodes were located along the pipe surface (nodal
spacing of 0.06 m). The average spacing of nodes
within a circular area having a diameter of 1 m from
the pipe center is 0.08 m (0.16 m at 1 m distance from
pipe center decreasing to 0.06 m close to the pipe).
Average node spacing outside this area is 0.45 m. The
bottom and the lateral sides of the model were fixed
in vertical and horizontal direction, respectively.

429

Figure 1. Mesh geometry.


Figure 2. Load-displacement curve for different ru values.
Table 1.

Soil parameters.

Parameter
Unit weight
Young modulus
Poisson coefficient
Friction angle
Cohesion
Dilatancy

Value
19.5
0.87.7
0.5
35
0.0
2

Unit
kN/m
MPa

4
3

kPa

The properties of the soil (unit weight ,Young modulus E, friction angle  , cohesion c , dilatancy angle
) are reported in Table 1.
3

MODELING PROCEDURE

The soil-pipe interaction was modelled in 4 stages.


First an elastic analysis was performed to establish the initial geostatic stress state in the soil. During
this step, gravity loads of the soil and pipe are applied
together with a water load on the upper surface. This
allowed pore pressures to reach a hydrostatic distribution within the soil domain and the model to be in
equilibrium with the prescribed boundary conditions.
In the second stage the soil constitutive model is
switched to elasto-plastic.
Afterwards, the pore pressure field was changed
in order to reach the desired stress field (liquefied
conditions). Specifically, the pore pressure boundary
conditions at the bottom are switched from impermeable to permeable with an imposed pore pressure
(higher than hydrostatic).
Once the stress field reached the desired conditions,
the pipe was moved in horizontal direction with a given
displacement rate.
Reactions from the nodes located along the pipesoil interface were used to obtain the reaction provided
by the soil during pipe movement. The pipe displacement and the soil reaction are used to draw
force-displacement interaction curves.

LOAD-DISPLACEMENT CURVES FOR


DIFFERENT PORE PRESSURE RATIOS

Following the methodology previously described,


load-displacement curves (p-y curves in the following) were extracted for 6 different pore pressure ratios
ru (ru = uecc /v where uecc is the excess of pore pressure
and v is the initial effective vertical stress). Results
are plotted in Figure 2.
The condition of ru = 0% represents the nonliquefied soil with a hydrostatic pore pressures regime.
An ultimate soil resistance pu of about 52 kN/m is
reached.
Increasing the pore pressure ratio decreases the
ultimate soil resistance while the shape of the loaddisplacement curve is maintained.
When the value of ru approaches 90%, the ultimate
resistance decreases significantly (about 4.5 kN/m, for
ru = 90% and 1 kN/m for ru = 95%).
An example of horizontal and vertical displacement
field, for ru = 90%, is shown in Figure 3. The failure
mechanism is characterized by a passive zone at the
right of the pipe and an active zone to the left of the
pipe.

5 VALIDATION OF RESULTS
5.1 Non-liquefied conditions
Many studies exist in literature concerning the
response of laterally loaded pipes, anchor plates or
foundation piles in sandy soil.
Pipelines buried in sand have been studied, among
others, by Audibert & Nyman (1977), Nyman (1984),
Trautmann & ORourke (1985), American Lifelines
Alliance (ALA, 2001) and Cathie et al. (2005). Following the approach of Trautmann & ORourke, the
ultimate lateral resistance can be written as:

430

Figure 4. FEM-literature p-y comparison for non-liquefied


conditions.

Figure 3. Displacement field (ru = 90%): a) horizontal


displacement; b) vertical displacement.

where Nh is the dimensionless lateral bearing capacity factor,  the effective unit weigh, H the depth to
the pipe center. These authors showed that for loose
and medium dense sands, Nh increases approximately
linearly with the embedment for H/D < 8, whereupon
Nh becomes constant, indicative of the transition from
shallow to deep soil failure mechanism. It should be
noted that the correlation was developed for a minimum H/D of 1.6, which is slightly greater than the
value of 1.24 for the spool considered.
Trautmann & ORourke also demonstrated that the
values defined for the holding capacity of anchor
plates (Rowe & Davis 1982) were in good agreement
with their results for pipes. Rowe & Davis showed that
Nh depends primarily on the friction angle and embedment ratio, and on the roughness of the embedded
structure.
Onshore pipeline design is often based on the American Lifeline Alliance (ALA) design guidelines (ALA,
2001). The ALA guidelines update the earlier work
by Audibert and Nyman (1977) and Trautmann and
ORourke (1985), incorporating results of physical
testing, numerical modeling and field performance.
ALA provides methods to estimate drag forces and
p-y curves for horizontal, axial and vertical loading
conditions for a wide range of soils including combination of cohesive and frictional behaviors. Considering
the diameter, burial depth and soil properties for
the spool, the ALA guidelines estimate an ultimate
horizontal resistance of 70.7 kN/m for non-liquefied

conditions, corresponding to Nh = 7.4. The ALA recommendations suggest the ultimate lateral resistance
for pipelines in sand is reached at a displacement of
about 0.04(H + D/2) 0.1D to 0.15D, corresponding
to 0.08 to 0.12 m for a 32 pipe.
The PRCI (2004) guidelines for seismic design
of pipelines essentially confirm the ALA recommendations for lateral pipe response. PRCI does note,
however, that there is some evidence from laboratory
tests that the ALA procedure may overestimate the
maximum soil reaction.
Cathie et al. (2005) proposed a hyperbolic relationship for lateral force-displacement curve of buried
pipes in sands:

where y* = y/yu (with yu equal to the displacement


corresponding to the maximum force) and a and b are
the model parameters.
Figure 4 compares the results of the FEM results
to the published p-y curves for pipes in non-liquefied
soils. The Base Case model uses the soil parameters
of Table 1 and pipe-soil interface friction angle equal
to 2/3 of soil friction angle. Two alternative runs are
also shown, base case soil parameters with associated
flow rule, and base case soil with associated flow and
pipe-soil friction equal to friction angle. The following
can be noted:

431

for the Base Case the predicted ultimate resistance


is 51 kN/m, which agrees very well with Trautmann & ORourke max resistance, despite the mild
extrapolation regarding H/D ratios;
the hyperbolic model by Cathie et al. (2005) predicts
a stiffer initial response compared to FEM results
but there is a good agreement in terms of maximum
force;
the ALA curve shows an excellent match regarding
initial stiffness but higher ultimate resistance;

sensitivity runs showed that considering associated


flow and pipe-soil friction equal to soil friction angle
yielded similar initial stiffness, but a higher ultimate
strength, of the order of 90% of the ALA value.

5.2

Liquefied conditions

While soil-pipe interaction is well established for nonliquefied soils, there is little published data on the
lateral behavior of pipe embedded in liquefied soil.
Generally, for piles, lateral response is modeled
using conventional load-displacement relationships
(p-y curve). The main parameters are soil stiffness
and strength. When the differential soil-pile movement
is small (i.e. soil is not pushed to its full capacity),
the soil stiffness plays an important role. In contrast,
when the differential soil-pile movement is large, the
ultimate strength of the p-y curve is governed by the
soil strength. The presence of liquefied soils dramatically affects the load-displacement response, reducing
both the initial stiffness and ultimate resistance, before
the soil strength increase due to dilation. In order to
have a confirmation of the FEM results from published data, the calculated ultimate horizontal soil-pipe
reaction was compared with the Calvetti et al. (2004)
findings. Calvetti et al. (2004) performed both experimental tests and numerical simulations of small pipes
buried in liquefied soils ( = 32 , = 18.43 kN/m3 ,
Dr = 20%) providing failure envelopes for different
horizontal and vertical loading combinations. Considering the envelopes by Calvetti et al. (2004) for
H/D = 1.35 (shown in Figure 5 for D equal to 50 mm)
and multiplying the results by a geometric factor equal
to the ratio between the actual pipe diameter (81.2
cm) and the pipe diameter considered by Calvetti et al.
(5 cm), a horizontal capacity of the order of 45 kN/m
can be estimated. This result is in very good agreement
with the FEM results which show an ultimate capacity
equal to 3.9 kN/m (see Figure 2, ru = 90%).
The horizontal pipe capacity obtained from FE
modeling could also be compared with the ultimate
soil-pipe strength reaction (pu ) computed according
to the methods proposed for piles in liquefied soils.
For example, according to Dash (2010), the ultimate
horizontal resistance is computed as pu = SrDNs,
where Sr is the residual shear strength of the liquefied sand, D is the pile outer diameter and Ns is a
bearing capacity factor. Based on FE analysis, Martin & Randolph (2006) suggested Ns values equal to
9.2 and 11.94, for smooth and rough pile-soil interface
respectively. Considering a smooth interface, the computed ultimate resistance pu in liquefied conditions
(ru = 90%) is about 5 kN/m, 10% of the corresponding
for non-liquefied conditions.

Figure 5. Pipe-soil failure envelope in liquefied soil


(Calvetti et al. 2004).

Figure 6. Hysteresis loops.

response is expected also for a buried pipe. Furthermore, since damping increases with strain and strains
associated with liquefied soils are generally high, the
hysteretic damping of the soil-pipe system can be relatively important in pipe design. In order to obtain the
damping ratio as a function of the pipe displacement,
various loops were simulated. Each loop consisted of a
first step to a specified displacement (ymax ), an un-load
step to ymax and a final step to close the loop.
Loops obtained for different ymax are shown in
Figure 6.
Based on these results, the damping ratio was
calculated as follows (Rollins et al. 2010):

where Aloop is the area within the loop, ymax is the


maximum displacement of the given loop and k is the
stiffness computed as:

DAMPING IN LIQUEFIED CONDITIONS

Soils subjected to cyclic shear loading generally show


an hysteretic stress-strain response. A similar cyclic

where pult is the ultimate load.

432

Figure 7. FEM and Masing Rule hysteresis loop


ymax = 0.10 m.

Figure 9. FEM and Masing Rule damping curve


comparison.

A comparison of percentage of critical damping


derived from the FE model and from the Masing Rule
is shown in Figure 9. The matching is quite good confirming that the damping derived from the FE model
is reliable.

Figure 8. Sensitivity to pipe displacement rate.

The damping curve, for ru = 90%, is shown


in Figure 9 (line with dots). High damping
values are observed for displacement > 2.5 cm.
Displacements > 15 cm show a value of damping close
to the maximum theoretical damping.
The high values of the damping ratios are justified
by the fully liquefied conditions assumed for the soil.
Given the lack of published results on loaddisplacement loops, in order to validate the FEM
results, the hysteresis loops were calculated from the
soil-pipe p-y curve (backbone curve) following the
modified Masing Rule (Masing 1926 cited in Hashash
et al. 2010). Results of this comparison are plotted in
Figure 7 for 10 cm of maximum displacement. The
matching can be considered satisfactory apart from
the fluctuations in FEM results. These fluctuations
(included the peak resistance for v = 0.05 m/s and
v = 0.1 m/s) are due to numerical disturbance and are
not representative of real soil behavior. In fact, if the
pipe is displaced at lower velocity these features almost
disappear as shown in Figure 8 and confirmed by Dash
et al. (2008).

CONCLUSIONS

This paper deals with the pipe-soil interaction of


pipelines subjected to horizontal loading in liquefied
soil. In the past, several methodologies were developed for lateral response of piles (p-y curves) in
liquefied soil, but these approaches cannot be directly
used for soil-pipe interaction given the different failure
mechanism. This paper analyses soil-pipe interaction
using an elasto-plastic model, implemented in the
TOCHNOG finite element software. Some assumptions were required to model numerically the soilpipe interaction in liquefied conditions. The primary
assumption/limitation is the rather simple constitutive
model adopted to describe the soil behavior. The main
drawback of advanced constitutive models is the difficult calibration of multiple parameters which must
often be based on literature values. The constitutive
model adopted, with its relative simplicity, can be
directly calibrated using measured soil properties. It
is believed adequate to capture the key aspects of soil
behavior during shearing.
As in conventional finite element analysis, the calculations were performed under the assumption that
the influence of changes in mesh geometry is neglected
in the equilibrium conditions. However, similar conventional FE approach was successfully used in literature to study other liquefaction problems (e.g. di Prisco
& Galli 2006). Furthermore, checks with well established approaches in non-liquefied conditions indicate
that the FEM results are in good agreement also for
large strains.
A comparison of percentage of critical damping
derived from the FE model and from the Masing Rule

433

is quite good confirming the damping derived from


the FE model is reliable.
Future experimental test results in similar conditions will certainly give further validation to FEM
results.
ACKNOWLEDGMENTS
The authors gratefully acknowledge Dr. Andrea Galli
of the Technical University of Milan for his support in
the TOCHNOG analyses.
REFERENCES
American Lifelines Alliance (ALA) 2001. Guidelines for
the Design of Buried Pipelines (with addenda through
February 2005), July.
Aprea, A. & Luisi, C. 1999. Problematiche geotecniche relative alla progettazione di condotte interrate lungo pendii in
frana. Degree thesis, Technical University of Milan, Dpt.
Of Structural Engineering.
Audibert, J.M.E. & Numan, K.J. 1977. Soil restraint against
horizontal motion of pipes. Journal of the Soil Mechanics
and foundations Division, ASCE 103(GT10): 11191142.
Calvetti, F., Di Prisco, C. & Nova, R. 2004. Experimental and
Numerical Analysis of Soil-Pipe Interaction. Journal of
Geotechnical and Geoenvironmental Engineering, ASCE,
130(12): 12921299.
Cathie, D.N., Jaeck, C., Ballard, J.C. & Wintgens, J.F. 2005.
Pipeline geotechnics state-of-the-art. Frontiers in Offshore Geotechnics, Proc. intern. Symp., 1921 September
2005 , Perth, Australia,
Dash, S. 2010. PhD Thesis, Lateral Pile-Soil Interaction in
Liquefiable Soils. Department of Engineering Science,
University of Oxford, UK.
Dash, S., Bhattacharya S., Blakeborough A. & Hyodo M.
2008. P-Y curve to model lateral response of pile foundations in liquefied soils. Earthquake Engineering, Proc.
14th World Conf., Beijing, 1217 October 2008, Cina.
di Prisco, C. & Galli, A. 2006. Soil-pipe interaction under
monotonic and cyclic loads: experimental and numerical
modelling. In Darve et al. (eds.), Advances in Geomaterials and Structures; Proc. intern. symp., Hammamet, 35
May 2006, Tunisia.
Dickin, E.A. 1994. Uplift resistance of buried pipelines in
sand. Soils and Foundations 34(2): 4148.

FEAT 2004. Tochnog Professional Users Manual.


Hashash, Y.M.A., Philips, C.& Groholski, D.R. 2010. Recent
Advances in Non-Linear Site Response Analysis. Recent
Advances in Geotechnical Earthquake Engineering and
Soil Dynamics and Symposium in Honor of Professor
I.M. Idriss, Proc. 5th intern. Conf., 2429 May 2010, San
Diego, California.
Martin, C.M. & Randolph, F. 2006. Upper bound analysis of
lateral pile capacity in cohesive soil. Geotecnique: 56(2):
141145.
Masing, G. 1926. Eignespannungen und Verfestigung beim
Messing Applied Mechanics, Proc. 2nd Intern. Congress,
Zurich, Switzerland.
Matyas, E.L. & Davis, J.B. 1983. Prediction of vertical
earth loads and rigid pipes. Journal of Geotechnical
Engineering, ASCE 109(GT2): 190201.
Meyerhof, G.G. & Adams, J.L. 1968. The ultimate uplift
capacity f foundations. Canadian Geotechnical Journal
5(4): 225244.
Neely, W.J., Stuart, J.G. & Graham, J. 1973. Failure load
of vertical anchor plates in sand. Journal of the Soil
Mechanics and foundations Division, ASCE 99(SM9):
669685.
Nyman, K. J. 1984. Soil Resistance against oblique motions
of pipes. Journal of Transportation Engineering, ASCE,
110(2): 190202.
Pipeline Research Council International Inc. (PRCI) 2004.
Guidelines for the Seismic Design and Assessment of
Natural Gas and Liquid Hydrocarbon Pipelines. Catalog
No. L51927, October.
Rollins, K.M., Gerber, T.M. & Kwon, K.H. 2010. Increased
Lateral Abutment Resistance from Gravel Backfills of
Limited Width. Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 136(1): 230238.
Rowe, R.K. & Davis, E.H. 1982. The behavior of anchor
plates in sand. Geotechnique 32(1): 2541.
Scarpelli, G., Sakellariadi, E. & Furlani, G. 2003. Evaluation of soil-pipeline longitudinal forces. Rivista Italiana
di Geotecnica 4(3): 2441.
Trautmann, C.H. & ORourke, T.D. 1983. Behaviour of pipe
in dry sand under lateral and uplift loading. Geotechnical Engineering Report 837, Cornell University, Itaca,
New York.
Trautmann, C.H. & ORourke, T.D. 1985. Lateral forcedisplacement response of buried pipe. Journal of Geotechnical Engineering, ASCE 111(9): 10771092.

434

Вам также может понравиться