Sie sind auf Seite 1von 16

LATERAL FORCE-DISPLACEMENT RESPONSE

OF BURIED PIPE

By Charles H. Trautmann, 1 A. M. ASCE,


and Thomas D . O'Rourke, 2 M. ASCE

ABSTRACT; The results of an experimental program to assess the response of


Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

buried pipes to lateral ground movements are presented. The effects of pipe
depth, soil density, pipe diameter, and pipe roughness are considered, and test
results are compared with published analytical models and experimental data.
The results indicate the need to consider vertical equilibrium in predicting the
horizontal response of buried pipelines, and the data agree well with several
analytical models that include this effect. Pipe surface roughness was found to
have little effect on response. Soil density has a large effect on displacements
required to mobilize the maximum force but a relatively small effect on the
value of the residual force at large displacements for depths typical of trans-
mission pipelines. The study concludes with a simplified design procedure for
predicting pipeline response to lateral ground movements.

INTRODUCTION

An understanding of pipeline response to lateral ground m o v e m e n t s


is critical for design a n d risk evaluation in a variety of adverse environ-
ments. These environments include offshore slope failures (3,5,6), earth-
quake-induced faulting (12,17,19), landsliding and liquefaction (18), a n d
urban excavations a n d tunneling (15,20). Each source of ground move-
ment can cause significant pipeline strain by means of: (1) Bending; a n d
(2) increased longitudinal friction from high lateral soil stresses. This lat-
ter effect can be particularly troublesome because analytical studies of
continuous ductile pipelines have s h o w n that longitudinal strain con-
centrations near zones of ground rupture strongly influence yielding and
post-yielding pipeline performance (12,17).
The response of buried pipelines d e p e n d s on soil stresses a n d defor-
mations, which, in practice, are difficult to characterize. Audibert a n d
Nyman (4) have pointed out that existing formulations for the coefficient
of horizontal subgrade reaction are inadequate for the design of buried
conduits. In addition, various models for evaluating the m a x i m u m lat-
eral resistance of buried pipes lead to a wide range of predicted soil
forces. For example, analytical models based on work by H a n s e n (10)
and Ovesen (21) predict maximum lateral soil forces that differ by 240%
for a pipeline buried four diameters in dense sand.
In view of these uncertainties, an experimental program w a s u n d e r -
taken to compare the values given b y available models with laboratory
data obtained from carefully controlled experiments. The experiments
were performed on sections of 102 a n d 324-mm diameter pipes, which
represent sizes typical of both distribution a n d transmission pipelines.
'Research Assoc, Cornell Univ., Ithaca, N.Y. 14853.
2
Assoc. Prof., Cornell Univ., Ithaca, N.Y. 14853.
Note.-—Discussion open until February 1, 1986. Separate discussions should
be submitted for the individual papers in this symposium. To extend the closing
date one month, a written request must be filed with the ASCE Manager of lour-
nals. The manuscript for this paper was submitted for review and possible pub-
lication on September 27, 1984. This paper is part of the Journal of Geotechnical
Engineering, Vol. I l l , No. 9, September, 1985. ©ASCE, ISSN 0733-9410/85/0009-
1077/$01.00. Paper No. 19987.
1077

J. Geotech. Engrg., 1985, 111(9): 1077-1092


In this paper, the literature is reviewed with respect to the lateral load-
ing of buried linear objects. The experimental program is described, and
test results are compared with existing analytical predictions and test
data. The experimental results are summarized, and recommendations
are given in a format convenient for design.
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

PREVIOUS RESEARCH

Previous work on the behavior of laterally loaded objects can be di-


vided into studies dealing with: (1) Maximum horizontal soil forces; and
(2) force-displacement relationships. Most analytical models have fo-
cused on the maximum soil forces developed by vertical anchor plates
and rigid vertical piles, with theoretical treatments adapted to buried
pipelines principally by Audibert and Nyman (4) and Luscher, et al. (14).
Only a few studies have dealt with the force-displacement relation-
ships. Experimental findings have generally been used to express the
displacement at maximum soil force as a fraction of the burial depth.
This value is then divided into the maximum force pertaining to a given
soil and burial condition to estimate a secant force-displacement rela-
tionship based on plastic yield of the soil. A detailed review of the an-
alytical models and test data associated with horizontal soil forces and
force-displacement relationships has been provided by Audibert and
Nyman (4) and more recently by Trautmann and O'Rourke (27). Only
a brief examination of selected models and experimental results is given
in this paper.
Analytical Studies.—Theoretical formulations for buried pipelines have
evolved from existing models for piles and vertical anchors. Hansen's
model (10) for the lateral capacity of a rigid vertical pile has been rec-
ommended for evaluating the horizontal soil forces on buried pipelines
(4). The model assumes that a pipeline behaves as a retaining wall near
the ground surface and a strip footing at depth and employs an empir-
ical interpolation function to compute lateral forces at intermediate depths.
The model is not entirely compatible with soil-pipeline interaction for
two principal reasons. First, for the shallow retaining wall mechanism,
the model assumes that wall, or interface, friction is fully mobilized.
Shallow pipelines, however, are free to rise with the passive wedge of
soil that develops during large lateral displacements. The relative dis-
placement between the soil and pipe, therefore, may not be sufficiently
large to mobilize full interface friction. Second, the model is three-di-
mensional and assumes full mobilization of shearing strength along the
sides of the narrow passive wedge extending in front of a laterally loaded
pile. In contrast, laterally loaded pipelines behave under essentially plane
strain conditions. These two assumptions lead to the overprediction of
lateral soil forces.
Ovesen (21) developed a method to predict the lateral capacity of ver-
tical anchors that accounts for vertical equilibrium as was described by
Ovesen and Stromann (22). Neely, et al. (16) applied Sokolovskii's method
(25) to the problem of vertical anchors and presented design charts for
shallow anchors. Rowe and Davis (23) applied the finite element method
to strip anchors, considering depth, friction angle, mobilized interface
friction, dilatancy, and initial stress state. Dickin and Leung (9) pre-
1078

J. Geotech. Engrg., 1985, 111(9): 1077-1092


sented a comprehensive review of design methods for vertical plate an-
chors, along with extensive comparisons of various theoretical models
for the case of dense sand. Therefore theoretical models will be com-
pared subsequently with the current test results.
Experimental Studies.—Experiments have been performed on vertical
plate anchors by means of full-scale field tests, models tests, and the
centrifuge. Smith (24) tested full-scale concrete deadman anchors in moist
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

beach sand for H/D ranging fom 0.5-1.8, in which H = the depth to
the anchor or pipe center; and D = the height or diameter of the anchor
or pipe. Ovesen (21) tested model anchors in an apparatus designed to
simulate plane strain conditions. Neely, et al. (16) tested model anchors
having a length-to-width ratio of five at H/D ranging from 0.5-4.5. Rect-
angular plate anchors were tested by Das and Seeley (7). Audibert and
Nyman (4) tested pipes with diameters of 25 mm, 60 mm, and 114 mm
for H/D ranging from 1-24 in loose and dense sand. They also reported
one field test with a 230-mm diameter pipe. Akinmusuru (1,2) tested
plate anchors in loose sand with L/D = 10 and H/D ranging from 0.5-
9.5 and found a well-defined transition from shallow to deep failure
mechanism for loose sand at H/D ranging from 4-5.5.
Model anchors were tested in a centrifuge by Dicken and Leung (8).
In these tests, H/D ranged from 0.5-6.5, and the results showed that
both the maximum anchor force and failure displacement depend strongly
on the scale of the prototype anchor. Displacements at failure were also
found to be scale dependent, ranging from 0.019H for a 25 mm anchor

TABLE 1.—Summary of Characteristics of Horizontal Load Tests


Angle of Soil
Shearing
Resistence
Gradation Test Mo del
4>> in Density
Reference degrees Test (kN/m3) Dw (mm) c„ D (mm) L/D
(D (2) (3) (4) (5) (6) (7) (8)
Smith (24) 32 NR 14.4 0.2 2.0 914 8
Ovesen (21) 32 TX 15.2 NR NR 15-150 PS
36 PA
38 TX 16.7
43 PA
Neely, et al. (16) 39a TX 15.8 0.2 2.7 51 5
Das (7) 34 NR 15.8 NR 1.7 51 5
Audibert & Nyman (4) 33 E 15.7 0.3 2.7 25-114 PS
40 E 17.3
Akinmusuru (1) 35 NR 15.4 0.3 1.9 38 10
Dickin & Leung (9) 44 TX 16.4 NRb NR 25-152 5
Present study 31 DS 14.8 0.2 2.6 102, 324 PS
36 DS 16.4
44 DS 17.7
"Measured in direct shear tests at 6.9 kPa normal stress.
b
80% of grains between 0.125 mm and 0.25 mm.
Note: TX = Triaxial test; PA = estimated from earth pressure test; DS = direct shear test;
E = estimate; NR = not reported; D ]0 = effective grain size; C„ = coefficient of uniformity;
D = height or diameter of model anchor or pipe; L/D = aspect ratio of model anchor or
pipe; and PS = plane strain test conditions.

1079

J. Geotech. Engrg., 1985, 111(9): 1077-1092


at H/D = 4.5 to 0.056H for a 1,000 mm anchor at the same depth ratio.
The experimental results show considerable variation in measured soil
forces, even though H/D and the relative density of the sand used dur-
ing some of the tests were apparently similar. For example, the soil forces
measured by Akinmusuru (1) were approximately three times those
measured by Das and Seeley (7) for reported soil friction angles of 35°
and 34°, respectively.
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

Table 1 summarizes information associated with the reported tests re-


sults, including data on soil shear strength, density, and grain size char-
acteristics. For comparison, data related to the tests performed as part
of this study are also listed. Only experimental results corresponding to
L/D equal to or greater than five are considered, as Rowe and Davis (23)
have shown that this geometry compares favorably with plane strain.
The experimental conditions in published tests varied over a wide range,
particularly with respect to the angle of soil shearing resistance, soil den-
sity, and associated methods of laboratory measurement. These vari-
ations must be kept in mind when comparing the test findings and un-
derscore the need for careful control of sand density and consistent
methods of shear strength testing when performing model tests, as
pointed out by Dickin and Leung (9).

EXPERIMENTAL PROGRAM

The goals of the test program were to: (1) Measure the maximum lat-
eral soil force as a function of pipe depth and soil density; (2) evaluate
how the force-displacement behavior varies with soil density and H/D;
(3) characterize the force-displacement behavior in terms of a simple
mathematical function; (4) compare the measured results with previ-
ously published analytical and experimental studies; and (5) summarize
the results in a format suitable for design practice. The following sec-
tions outline briefly the test apparatus and procedures.
Apparatus.—The test apparatus consisted of five parts, including a
test compartment, sand spreader, storage bin and conveyor system, in-
strumentation, and data acquisition system. The test compartment mea-
sured 1.2 m X 2.3 m x 1.2 m deep and when full held about 60 kN of
dry sand. The components of the system are described in a companion
paper by Trautmann, et al. (28) and in greater detail in Trautmann and
O'Rourke (27). A side view of the test compartment is shown in Fig. 1.
For the lateral load tests, two pipe test sections were used, having
outside diameters of 102 mm and 324 mm, wall thicknesses of 6.4 mm
and 9.5 mm, and lengths of 1.20 m. The loading system included a stiff
frame, yoke, and 275-kN hydraulic jack. Loads were transmitted by 25-
mm square steel rods, which were instrumented with strain gages to
measure loads. The tie rods, located outside the test enclosure, trans-
ferred loads through ball bearings to stiff axles protruding from the cen-
ter of the pipes through slots covered in the enclosure walls.
Properties of Sand.—Cornell filter sand was used for all tests. This
material was placed at densities of 14.8, 16.4, and 17.7 kN/m 3 , corre-
sponding to direct shear friction angles of 31°, 36°, and 44°. During prep-
aration of the tests, the soil density was carefully checked at many lo-
cations throughout the test volume by means of density pans (loose sand)
1080

J. Geotech. Engrg., 1985, 111(9): 1077-1092


Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1 .—Photograph Showing Test Compartment Used for Pipe Loading Tests

or a density scoop (medium or dense sand). As a result of these mea-


surements, the average density for each test was known with a precision
of better than 1%. Additional details on both test apparatus and mea-
surements are given by Refs. 27-29.
Test Procedure.—For each test, a minimum of 200 mm of sand was
placed beneath the pipe. The sand was spread in 50-mm thick lifts, with
thin layers of flour and the ends of wood dowels placed against the
window near the pipe to permit observation of sand displacement pat-
terns during loading. The sand was emptied from the test compartment
after each test.
A total of 30 lateral loading tests was performed. Tests at three den-
sities were prepared at five depths, with H/D equal to 1.5, 3.5, 5.5, 8,
and 11. Several replicate tests were included, and the effects of pipe
surface friction were assessed by covering the 102-mm diameter pipe
with sandpaper and a plastic film coated with SAE 90 gear oil to simulate
rough and smooth surfaces, respectively.
Dimensional Analysis.—Dimensional analysis indicates that for model
studies in sand, displacements, forces, and torques scale by the first,
second, and third powers of the prototype-to-model ratio, while stresses
remain unchanged. Dimensional analysis, however, forms only a small
portion of the total modeling problem. Effects such as sidewall friction,
progressive failure, stress-dependent soil friction, the initial stress state,
dilatancy, and pipe surface roughness all influence the applicability of
model results to the field, but do not necessarily scale according to di-
mensional analysis. In this study, the model pipes were selected to be
close or equal in diameter to field scale pipes, and the results from tests
with different sized pipes showed a relatively small scale effect.

TEST RESULTS

Maximum Force.—Force-displacement data for the lateral loading tests


are shown in Figs. 2(a-c). In the figures, forces have been plotted as the
dimensionless quantity F/(yHDL), in which F = the measured force; y
= the sand density; L = the pipe length; and H and D are as previously
defined. Displacement has been plotted as the dimensionless quantity
Y/D, in which Y = the actual displacement. Arrows on the curves show
1081

J. Geotech. Engrg., 1985, 111(9): 1077-1092


* 20.0
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

"0 0.2 0.4 0.6 0.8 1.0 1.2 0 0.2 0.4 0.6 0.8 1.0 1.2
ia) Dimensionless Displacement, Y/D (tyDimensionless Displacement, Y/D

25.0

20.0 H/D=ll
Legend

| Point corresponding
15.0 - K \ V " ~ H/D = 8 to maximum force
h — H/D-5.5
P\H/D=3.5
10.0
J H/D •1.5
5.0

0 . i . 1 • '
() 0 . 2 0.4 0.6 0.8 1.0 1.2
W )imensionless Displacement, Y / D

FIG. 2.—Force-Displacement Data for Pipe Loading Tests: (a) <t> = 31°; (b) $
36°; (c) <}> = 44°

the points representing the maximum dimensionless force, Nh, which


is defined as

Nh = - (1)
yHDL
in which Fm = the maximum measured force. The corresponding di-
mensionless displacement is Yf/D. The data are tabulated in Table 2,
which lists the pipe diameter, soil density, pipe depth, and maximum
and residual values of dimensionless force.
The method used to define Yf depended on the shape of the force-
displacement curve. For dense sand, the peak force could be clearly
identified in all cases. For medium sand, the Hansen 90% criterion (10)
was used. For loose sand, the Hansen criterion gave unclear results be-
cause the force-displacement curves did not level off, even at large dis-
placements. For loose sand, therefore, Yf was defined at the break in
slope of the curves, as shown in Figs. 2(«-c). The value of Yf for H/D
= 8 and 11 was so uncertain that it was not used in the analysis of
displacements.
The displacements corresponding to maximum force varied substan-
tially as a function of soil density. For shallow pipes in loose sand, the
maximum force was generally not attained even after a lateral displace-
ment of 0.13H. In medium sand, the maximum force was approached
asymptotically, with little change in force after a displacement of about
0.08H. In dense sand, a peak force was observed at relatively small dis-
placements, which averaged about 0.03H. As the displacement in-
1082

J. Geotech. Engrg., 1985, 111(9): 1077-1092


TABLE 2.—Summary of Results for Lateral Load Tests
Test
number Pipe diameter (mm) Density (kN/m3) H/D N„ N„r
(1) (2) (3) (4) (5) (6)
6 102 14.8 1.5 6.5 6.5
8 102 14.8 1.5 6.0 6.0
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

15 102 14.8 1.5 6.5 6.5


7 102 14.8 3.5 7.5 7.5
16 102 14.8 3.5 6.9 6.9
18 102 14.8 3.5 7.3 7.3
20 102 14.8 5.5 7.9 7.9
21 102 14.8 8.0 11.2 11.2
31 102 14.8 11.0 10.7 10.7
10 324 14.8 1.5 7.2 7.2
11 324 14.8 1.5 7.3 7.3
17 324 14.8 1.5 7.1 7.1
12 324 14.8 3.5 10.6 10.6
13 324 14.8 3.5 10.7 10.7
52 324 14.8 3.5 10.9 10.9
45 102 16.4 1.5 7.9 7.9
26a 102 16.4 3.5 7.9 7.9
a
27
102 16.4 3.5 7.9 7.9
29" 102 16.4 3.5 7.2 7.2
30b 102 16.4 3.5 7.2 7.2
46 102 16.4 5.5 10.7 10.7
51 102 16.4 8.0 12.8 12.8
48 102 16.4 11.0 12.5 12.5
49 102 16.4 11.0 12.5 12.5
22 102 17.7 1.5 8.4 7.7
23 102 17.7 3.5 10.9 8.7
24 102 17.7 5.5 14.8 12.1
25 102 17.7 8.0 17.1 14.2
32 102' 17.7 11.0 22.8 18.9
14 324 17.7 3.5 11.0 9.1
"Pipe covered with coarse sandpaper.
b
Pipe covered with polyethylene plastic and gear oil.
Notes: Nh = dimensionless coefficient for maximum force; Nhr = dimensionless
coefficient for residual force; Nilr = Nh for loose and medium sand.

creased to about one diameter in dense sand, the force decreased to a


steady, residual value, defined as Nhr.
Similar data have been reported in other studies. Audibert a n d N y m a n
(4) measured values of 0.02H a n d 0.04H for dense a n d loose sand, re-
spectively. Data obtained by other investigators show substantially larger
values for displacement at failure in loose a n d m e d i u m sand. Das a n d
Seeley (7) measured of 0.12H for loose sand, a n d Neely, et al. (16) mea-
sured values of 0.15H for m e d i u m sand.
As pointed out in conjunction with Table 1, there w a s considerable
variation in the methods used to measure the soil friction angle and to
control sand densities a m o n g the tests reviewed in this study. These

1083

J. Geotech. Engrg., 1985, 111(9): 1077-1092


1 1 1 1 1

20

e— e
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

Legend:
o Loose "
E s Medium "
b * Dense •
01 1i 1i 1i 1 1 I 01—
0 2 4 6 8 10 12 0 2 4 6 8 10 12
(a) Dimensionless Depth, H/D (b) Dimensionless Depth, H/D

FIG. 3.—Plot of Nh and N,n versus H/D for Pipe Loading Tests: (a) Nfc as Function
of H/D for Loose, Medium, and Dense Sands; (to) Nh for Medium Sand and Nhr
for Dense Sand as Function of H/D

variations, together with the difficulty of defining clearly a maximum


soil force for loose sand, are responsible for the rather wide range in
values. A more clearly defined method of characterizing the force-dis-
placement relationship for loose and medium sand involves the defini-
tion of an initial secant slope at a force less than the maximum, as de-
scribed by Thomas (26). This approach will be developed further in a
subsequent section.
Fig. 3(a) summarizes values of Nh as a function of H/D for three den-
sities of sand. The figure has a number of significant aspects. First, den-
sity has only a small effect on Nh for depths less than about four di-
ameters. Second, as depth increases, Nh for dense sand significantly
exceeds that for loose and medium sand. Third, Nh for tests in both
loose and medium sand reaches a constant value at H/D of approxi-
mately eight. This constant value of Nh has been interpreted by Vesic
(30) and Akinmusuru (1) is indicating a deep soil failure mechanism.
Fourth, Nh does not approach a constant value for dense sand at the
depths tested, implying that a deep failure mechanism did not develop
for H/D less than 11.
Last, and perhaps most significant, is the result that the Nh data for
loose and medium sand are nearly equal at all depths tested. In loose
sand, lateral movement causes compaction of the sand in front of the
pipe, leading to greater density and a soil resistance consistent with that
of medium sand. Discrepancies between the data for loose and medium
sand may result, in large part, from uncertainties in estimating the max-
imum force from the data for loose sand.
Residual Force.—Fig. 3(b) shows Nh for medium sand and Nhr for dense
sand as a function of H/D, in which Nhr is defined as the residual value
of dimensionless force at large displacements. For loose and medium
sand, Nhr was equal to Nh. Fig. 3(b) shows that for pipes with H/D less
than eight, Ni„ is nearly the same for both dense and medium sand. As
H/D increases from 2-8, Nhr for dense sand is consistently larger than
that for loose and medium sand, but the difference in residual force is
never more than about 12%. As depth increases beyond H/D of about
1084

J. Geotech. Engrg., 1985, 111(9): 1077-1092


eight, Nhr for dense sand becomes increasingly larger than Nh for me-
dium sand. Rowe and Davis (23) have attributed this effect to soil di-
latancy and the development of arching at depth.
This aspect of pipe behavior can be significant in areas of ground
movement. Because soil/pipe displacements typically will vary along the
length of the pipeline, Nh does not develop simultaneously along the
pipe. For large-displacement problems where residual forces control
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

pipeline behavior and where H/D is less than about eight, residual forces
may be essentially independent of the density of soil surrounding the
pipe.
Torque Measurements.—As pipes move laterally through soil, the dis-
placement of soil around the circumference results in a net torque. This
torque was measured in the pipe loading tests and was found to be
negligible with respect to the strength of the pipe. The maximum mea-
sured torque for the 102-mm diameter pipe was about 15 N-m per meter
of pipe, corresponding to less than 0.1% of the yield stress for ASTM
Grade A-36 steel.
Effect of Surface Roughness.—The effect of surface roughness was
determined by four tests with H/D = 3.5. In tests 17 and 18, the pipe
was covered with coarse sandpaper. In tests 19 and 20, the pipe was
covered with two layers of polyethylene plastic and coated with SAE 90
gear oil immediately before burial. As shown in Table 2, Nh for the rough
pipes was 10% greater than for the relatively smooth pipes. Therefore,
for buried pipes with H/D equal to about 3.5, a smooth coating has little
influence in reducing lateral soil forces.
Effect of Size.—The effect of size on the test results was investigated
by performing several experiments on pipes with diameters of 102 and
324 mm and identical conditions of H/D. Most of the tests were per-
formed in loose sand. The data in Table 2 indicate that, on average, Nh
values were 8% higher for tests on 324-mm diameter pipe when com-
pared with 102-mm diameter pipe. For one comparative test in dense
sand, the difference was only 1%. These relatively small changes indi-
cate that the tests on 102-mm diameter pipe are representative of the
behavior of pipelines with larger diameters and can be used to extrap-
olate performance to pipelines with diameters larger than 300 mm. The

70

40

c
ay
u

I 10
l" 1

37. •-•

-J I l_
0 2 4 6 8 10 12
Dimensionless Depth, H / D

FIG. 4.—Plot of Y,/H versus H/D for Pipe Loading Tests


1085

J. Geotech. Engrg., 1985, 111(9): 1077-1092


results also indicate the relatively small influence of end effects in the
test apparatus, as significant end effects would likely be associated with
larger Nh values for larger diameter pipes at identical H/D ratios.
Displacement Associated with Nh.—The dimensionless displacement
associated with maximum force, Yf/D, is shown by arrows in Figs. 2(a-
c). These displacements are replotted in Fig. 4 as the ratio Yf/H. As shown
by the figure, representative values of Yf/H for medium and dense sand
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

are 8% and 3%, respectively. For loose sand, the ratio is substantially
larger and difficult to define because of the gradual increase in horizon-
tal resistance with larger displacements. For H/D of 2-4, the test results
indicate that Yf/H is approximately equal to 13%. For all densities, the
data are spread over a factor of about two, except in loose sand, where
the spread in data is increased because of the difficulty in defining the
point of failure with precision.

COMPARISON WITH PUBLISHED STUDIES

Figs. 5(a-b) compare the test data with experimental results from six
other studies. The present test results plot near the middle of the values
for medium sand and agree most closely with data from Ovesen (21)
and Audibert and Nyman (4). For dense sand, the present results rep-
resent a lower bound on published measurements. As mentioned in
conjunction with Table 1, there were many differences in the sand den-
sities, friction angles, and experimental procedures associated with the
experimental measurements. These variations account for the wide range
of values shown in the figures.
Figs. 6{a-b) compare the test data for medium and dense sand with
the analytical models of Hansen (9), Ovesen (21), Neely, et al. (16), and
Rowe and Davis (23). Loose sand has been omitted from the comparison
because of the difficulty in defining the maximum force for this type of
material in which large volume reductions occur during shear. The data
in Figs. 6(a-b) show similar trends for both medium and dense sand.
The Hansen (10) model overpredicts the test data, while the Ovesen (21)
model and the Rowe and Davis (23) finite element calculations agree
closely with the test data. The model of Neely, et al. (16) generally falls
slightly below that of Hansen (10).
The comparison shows the effect of assumptions regarding vertical re-
straint and mobilization of interface friction. Both the Hansen (10) and
Neely, et al. (16) models assume full vertical restraint, or horizontal mo-
tion only, and overpredict the measured forces by 150-200%, while the
Ovesen (21) and Rowe and Davis (23) models, which provide for vertical
equilibrium, agree closely with the measured data. None of the models
predicts the constant values of Nh for H/D greater than eight in loose
or medium sand that were observed in this study.

MATHEMATICAL REPRESENTATION OF FORCE-DISPLACEMENT RELATIONSHIP

The force-displacement relationship can be modeled by a rectangular


hyperbola, as reported by Audibert and Nyman (4) and Das and Seeley
(7). Normalizing the curves with respect to Nh and Yf/D, the data were
1086

J. Geotech. Engrg., 1985, 111(9): 1077-1092


• i

.,......

• . - • • • *

- -
; -^/ - "
:=£=£- - ——•
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

0 2 4 6 B 10 12 10 12
fa,) Dimcnsionlcso Dcplh, H / D f&i Dimensionless Depth, ! ! / D

Legend;
Medium sand:
Smilh (24) d> = 3 2 " , 0=914 mm
— * — Ovesen ( 2 l ) d > = 3 2 ° , 0 = I 5 - I 5 0 m m
Das 6 Seeley (7) ^ = 3 4 " , D s 5 l m m
Audibert 8 Nyman (4) d>=33", D = 2 5 mm
Audibert 8 Nymon (4) ^ . « 3 3 ° r D = 6 2 m m
Aklnmusuru I I ) d > 3 5 , D=3mm
— t — Neely, Stuart, 8 Graham (16) £ = 3 9 ° , 0 = 5 l r
— o — T h i s study, d> = 3 6 ° , D i l 0 2 m m

— — A u d i b e r t a Nyman (4) d>=»40 , D=25mm


. Audibert ft Nyman (4) d i » 4 0 , D*62mm
— o — This study, d> = 4 4 " , D M 0 2 mm

FIG. 5.—Comparison of Model Test Results with Published Experimental Models:


(a) Medium Sand; (b) Dense Sand

• 1 1
1
-
• ^o> '"" '
/ /
• V
s
/
/
/
/ .--
-£•" •<?- •'.-

0 2 4 6 8 10 12 0 2 4 6 8 10 12
(a) H/D (b) H/D

Legend:
Honsen (10)
Ovesan (2 I )
Neely, S t u a r t , 8 Graham (16)
Rowe 8 Davis ( 2 3 )
* « Data from this study

FIG. 6.—Comparison of Model Test Results with Published Analytical Data: (a) 4>
= 36°; (b) <>
| = 44°

plotted on transformed axes, as recommended by Kondner (13). The av-


erage hyperbolic relationship for selected data from each test, shown in
Fig. 7, was found to be
Y"
F" = (2)
(0.17 + 0.83Y")
in which F" = (F/yHDL)/Nh; Y" = (Y/D)/(Yf/D); and F = the force mea-
sured at each increment of displacement. Both F" and Y" range from 0 -
1. Using this relationship and appropriate values of Nh and Yf/D, a force-
displacement relationship can be constructed for any combination of pipe
diameter, depth, and soil friction angle.
Computer programs commonly use a bilinear relationship to model
1087

J. Geotech. Engrg., 1985, 111(9): 1077-1092


1.4 1 1 r i 1 1 i i i i

1.2
^ - Bilinear representation
-
'h. 1.0 r
u
ifo.a .©A
o°J
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

•a = Y"/(0.I7+0.83Y")
a> ©
.H 0.6

o 0.4
' ° /A
°7> Legend:
A Loose sand -
-PI o Medium sand _
1" © Dense sand
II
' I 1 i i I > i • • i i

0 0.2 0.4 0.6 0.8 1.0 1.2


Normalized Displacement, Y"

FIG. 7.—Comparison of Selected Model Test Results with Hyperbolic Force-Dis-


placement Model

soil behavior. In describing the choice of initial slope, Thomas (26) has
recommended that a secant slope, defined at 70% of the maximum force,
Khj0, is appropriate for pipeline analyses involving large soil displace-
ments.
As shown in Fig. 7, a bilinear representation based on Khn results in
a maximum force at a normalized displacement of 0.4. Using this ob-
servation, it is possible to estimate the displacement at maximum force
for the bilinear relationship corresponding to loose, medium, and dense
sand. This displacement is 0.4 times the V//D values shown in Fig. 4.
Defining
OAYf
CK = (3)
D
the horizontal soil stiffness, or secant slope, is then given by the follow-
ing expression:
K,m = CkNhyDL (4)
in which CK represents the reciprocal of the displacement at failure, and
the other terms represent the force at failure. Values of CK are sum-
marized in Table 3. A convenient basis for evaluating Nh is examined in
the next section.

TABLE 3.—Factors for Estimating Horizontal Soil Stiffness


Displacement at maximum Factor for estimating horizontal
Soil density horizontal force soil stiffness, C,.
(D (2) (3)
Loose 0.13H 20
Medium 0.08H 30
Dense 0.03H 80

1088

J. Geotech. Engrg., 1985, 111(9): 1077-1092


APPLICATION OF RESULTS

The test data from this study agree most closely with the analytical
models of Ovesen (21) and Rowe and Davis (23). Based on the model
of Ovesen (21), Fig. 8 shows values of Nh as a function of H/D for fric-
tion angles between 30° and 45°. The figure assumes a diameter-to-thick-
ness ratio of 50 for the pipe and soil densities of 16.0, 17.3, 18.7, and
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

20.0 kN/m 3 for friction angles of 30, 35, 40, and 45°, respectively. For
shallow pipes, the lower limit of the ruled zones represents the lateral
forces associated with gas-filled pipelines. For H/D greater than four,
the weight of the pipe and contents has a negligible effect on lateral soil
resistance.
The line for 30° is dashed to indicate the need for caution in assuming
that loose backfill will lead to low lateral soil forces. As shown by the
experimental results in Figs. 3(a-b), the densification that accompanies
large pipe displacements in loose soil results in larger forces normally
associated with initially denser soil.
It is recommended that values of Nh be chosen after careful evaluation
of soil conditions in the field and the level of conservatism governing
design. Fig. 8 is applicable for dry or saturated sands and gravels, and
for partially saturated gravels and coarse sand. Partially saturated, me-
dium to fine sands will have increased shear strength relative to their
dry or saturated states under conditions of short term loading. Neither
the Ovesen or Hansen models account for increased strength associated
with partial saturation, but the Ovesen model is most susceptible to un-
derpredicting soil forces for these conditions.
To demonstrate the application of the results, consider the analysis of
a 610-mm diameter gas transmission pipeline with 1.22 m of cover in
medium dense, coarse sand having a friction angle of 35° and a density

24
i i i i 1 i i r
22 - v/jvm j —
20

o 16
u
o
Ll_ 14

<u

u.
l0
s 8
I
I 6
4

2
0 J I I I I I I I L
0 1 2 3 4 5 6 7 8 9 10
Dimensionless Depth, H/D

FIG. 8.—Plot of Nh versus H/D for Pipeline Design


1089

J. Geotech. Engrg., 1985, 111(9): 1077-1092


of 18 kN/m 3 . The pipeline is to be analyzed by the finite element method,
using 2-m long elements. Required are the m a x i m u m force per element
and the spring constant for spring/slider elements representing the lat-
eral soil restraint.
First, compute H/D = (1.22 + 0.305)/(0.610) = 2.5. Using Fig. 8, Nh
= 5.9 for a gas-filled pipeline. From Eq. 1, the maximum lateral force is
then computed as F = NhyHDL = (5.9) x (18) x (1.53) x (0.610) x (2)
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

= 200 kN. Next, find the spring constant by m e a n s of Eq. 4. A s s u m i n g


a m e d i u m dense soil, Khm = CkNhyDL = (30) x (5.9) x (18) x (0.610) x
(2) = 3,900 k N / m . These values can b e used directly in the analysis.

CONCLUSIONS

This paper describes an experimental study of lateral forces resulting


from relative m o v e m e n t between pipes and the surrounding soil u n d e r
plane strain conditions. A review of the literature indicates a n u m b e r of
analytical a n d experimental studies dealing with m a x i m u m force values,
but relatively few associated with the initial portion of the force-dis-
placement response.
Test results indicate that the models of Ovesen (21) a n d Rowe a n d
Davis (23) agree closely with the experimental data for medium and dense
sands. Lateral soil forces in loose, contractive sands are essentially equal
to those in m e d i u m dense sands, indicating the need for caution in as-
suming that extremely loose pipe backfill will lead to low restraint forces
in the event of large g r o u n d m o v e m e n t s .
Displacements associated with m a x i m u m force were found to be about
0.13H, 0.08H, a n d 0.03H for loose, m e d i u m , a n d dense sand, respec-
tively. Using these values, a bilinear force-displacement response can be
computed using Eqs. 1 and 4.
Judgment is required in using such values in design, principally with
regard to w h e t h e r pipe restraint or stress is to be evaluated as well as
to the effects of soil moisture.

ACKNOWLEDGMENTS

This research was funded by the National Science Foundation u n d e r


grant n u m b e r CME-8022427, a n d the writers wish to thank NSF program
managers W. Hakala, C. Liu, and K. T. Thirumalai. B. Hoefler, a n d J.
Nedzel assisted with the pipe loading tests. L. Crouse typed the m a n u -
script, and A. Avcisoy drafted the figures.

APPENDIX.—REFERENCES

1. Akinmusuru, J. O., "Horizontally Loaded Vertical Plate Anchors in Sand,"


Journal of the Geotechnical Engineering Division, ASCE, Vol. 104, No. GT2, Feb.,
1978, pp. 283-286.
2. Akinmusuru, J. O., closure to "Horizontally Loaded Vertical Plate Anchors
in Sand," Journal of the Geotechnical Engineering Division, ASCE, Vol. 105, No.
GT11, Nov., 1979, pp. 1370-1371.
3. Arnold, K. E., "Soil Movements and Their Effects on Pipelines in the Mis-
sissippi Delta Region," thesis presented to Tulane University, at New Or-
1090

J. Geotech. Engrg., 1985, 111(9): 1077-1092


leans, La., in 1967, in partial fulfillment of the requirements for the degree
of Master of Science.
4. Audibert, J. M. E., and Nyman, K. J., "Soil Restraint Against Horizontal
Motion of Pipes," Journal of the Geotechnical Engineering Division, ASCE, Vol.
103, No. GT10, Oct., 1977, pp. 1119-1142.
5. Audibert, J. M. E., Lai, N. W., and Bea, R. G., "Design of Pipelines to Resist
Seafloor Instabilities and Hydrodynamic Forces," presented at the Energy
Technology Conference and Exhibition, held in Houston, Tex., Nov., 1978,
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

16 pp.
6. Bea, R. G., and Aurora, R. P., "Design of Pipelines in Mudslide Areas,"
Proceedings, 14th Offshore Technology Conference, Vol. 4, Houston, Tex.,
1982, pp. 401-414.
7. Das, B. M., and Seeley, G. R., "Load-Displacement Relationship for Vertical
Anchor Plates," Journal of the Geotechnical Engineering Division, ASCE, Vol.
107, No. GT7, July, 1975, pp. 711-715.
8. Dickin, E. A., and Leung, C. F., "Centrifugal Model Tests on Vertical Anchor
Plates," Journal of Geotechnical Engineering, ASCE, Vol. 109, No. 12, Dec, 1983,
pp. 1503-1525.
9. Dickin, E. A., and Leung, C. F., "Evaluation of Design Methods for Vertical
Anchor Plates," Journal of Geotechnical Engineering, ASCE, Vol. I l l , No. 4,
Apr., 1985, pp. 500-520.
10. Hansen, J. B., "The Ultimate Resistance of Rigid Piles Against Transversal
Forces," Bulletin 12, Danish Geotechnical Institute, Copenhagen, Denmark,
1961, pp. 1-9.
11. Hansen, J. B., discussion of "Hyperbolic Stress-Strain Response: Cohesive
Soils," by R. L. Konder, Journal of the Soil Mechanics and Foundations Division,
ASCE, Vol. 89, No. SM4, July, 1963, pp. 241-242.
12. Kennedy, R. P., Chow, A. W., and Williamson, R. A., "Fault Movement
Effects on Buried Oil Pipeline," Journal of the Transportation Engineering Di-
vision, ASCE, Vol. 103, No. TE5, Sept., 1977, pp. 617-633.
13. Kondner, R. L., "Hyperbolic Stress-Strain Response: Cohesive Soils," Journal
of the Soil Mechanics and Foundations Division, ASCE, Vol. 89, No. SMI, Feb.,
1963, pp. 115-143.
14. Luscher, U., Thomas, H. P., and Maple, J. A., "Pipe-Soil Interaction, Trans-
Alaska Pipeline," Proceedings, 2nd ASCE Specialty Conference on Pipelines
in Adverse Environments, Vol. 2, New Orleans, 1979, pp. 486-502.
15. Maynard, T. R., and O'Rourke, T. D., "Soil Movements Effect on Adjacent
Public Facilities," presented at the ASCE Annual Meeting, San Francisco,
Calif., Oct., 1977, (Preprint 3111).
16. Neely, W. J., Stuart, J. G., and Graham, J., "Failure Load of Vertical Anchor
Plates in Sand," Journal of the Soil Mechanics and Foundations Division, ASCE,
Vol. 99, No. SM9, Sept., 1973, pp. 669-685.
17. Newmark, N. M., and Hall, W. J., "Pipeline Design to Resist Large Fault
Displacement," Proceedings, 1st U.S. National Conference on Earthquake En-
gineering, Ann Arbor, Mich., June, 1975, pp. 416-425.
18. O'Rourke, T. D., and Tawfik, M. S., "Effects of Lateral Spreading on Buried
Pipelines During the 1971 San Fernando Earthquake," Proceedings, ASME
Conference on Earthquake Behavior and Safety of Oil and Gas Storage Fa-
cilities, Buried Pipelines and Equipment, PVP-Vol. 77, Portland, Oreg., June,
1983, pp. 124-132.
19. O'Rourke, T. D., and Tratumann, C. H., "Buried Pipeline Response to Per-
manent Earthquake Ground Movements," ASME, Pressure Vessels and Pip-
ing Conference, San Francisco, Calif., Aug. 12-15, 1980, (Preprint 80-C2/
PVP-78).
20. O'Rourke, T. D., and Trautmann, C. H., "Buried Pipeline Response to Tun-
neling Ground Movements," Proceedings, Europipe 82, European Exhibition
and Conference on the Construction and Maintenance of Pipelines, Basil,
Switzerland, Jan., 1982, pp. 9-16.

1091

J. Geotech. Engrg., 1985, 111(9): 1077-1092


21. Ovesen, N. K., "Anchor Slab, Calculation Methods and Model Tests," Bul-
letin 16, Danish Geotechnical Institute, Copenhagen, Denmark, 1964, 40 pp.
22. Ovesen, N. K., and Stramann, H., "Design Method for Vertical Anchor Slabs
in Sand," Proceedings, Specialty Conference on Performance of Earth and Earth-
Supported Structures, Vol. 1, Lafayette, Ind., June, 1972, pp. 1481-1500.
23. Rowe, R. K., and Davis, E. H., "The Behavior of Anchor Plates in Sand,"
Geotechnique, Vol. 32, No. 1, Mar., 1982, pp. 25-41.
24. Smith, J. E., "Deadman Anchorages in Sand," Technical Report R-199, U.S.
Downloaded from ascelibrary.org by Ryerson University on 05/04/17. Copyright ASCE. For personal use only; all rights reserved.

Naval Civil Engineering Laboratory, Port Hueneme, Calif., 1962, 77 pp.


25. Sokolovskii, V. V., Statics of Granular Media, Paragamon Press, Oxford, En-
gland, 1965, 270 pp.
26. Thomas, H. O., discussion of "Soil Restraint Against Horizontal Motion of
Pipes," by J. M. E. Audibert and K. J. Nyman, Journal of the Geotechnical
Engineering Division, ASCE, Vol. 10, No. GT9, Sept., 1978, pp. 1214-1216.
27. Trautmann, C. H., and O'Rourke, T. D., "Behavior of Pipe in Dry Sand Un-
der Lateral and Uplift Loading," Geotechnical Engineering Report 83-7, Cornell
University, Ithaca, N.Y., 1983, 306 pp.
28. Trautmann, C. H., O'Rourke, T. D., and Kulhawy, F. H., "Uplift Force-Dis-
placement Response of Buried Pipe," Journal of Geotechnical Engineering, ASCE,
Vol. I l l , No. 9, Sept., 1985, pp. 1061-1076.
29. Trautmann, C. H., Kulhawy, F. H., and O'Rourke, T. D., "Sand Density
Measurements for Laboratory Studies," Geotechnical Testing Journal, ASTM,
in press.
30. Vesic, A. J., "Breakout Resistance of Objects Embedded in Ocean Bottom,"
Journal of the Soil Mechanics and Foundations Division, ASCE, Vol. 97, No. SM9,
Sept., 1971, pp. 1183-1205.

1092

J. Geotech. Engrg., 1985, 111(9): 1077-1092

Das könnte Ihnen auch gefallen