1

2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

Buried Corrugated Thermoplastic Pipe: Simulation and Design

Schafer, B.W., McGrath, T.J.
B.W. Schafer, Ph.D., Assistant Professor, schafer@jhu.edu
Johns Hopkins University, 203 Latrobe Hall, Baltimore, MD 21218
tel. (410)516-7801, fax (410)516-7473
T.J. McGrath, Ph.D., P.E., Principal, tjmcgrath@sgh.com
Simpson Gumpertz & Heger Inc., 297 Broadway, Arlington, MA 02474
tel. (781)643-2000, fax (781)643-2009
Submitted July 2002 to:
Committee A2C06 for presentation at the 2003 Transportation Research Board Annual Meeting
Word Count: Abstract 220 Paper = 4659 + 8 figures@250, + 3 tables@250 = 7409
ABSTRACT
The objective of this paper is to demonstrate a computational method for assessing the allowable depth of fill of a
buried thermoplastic profile wall (corrugated) plastic pipe and compare the results to the recently adopted AASHTO
design method. The computational method is demonstrated for a 1500 mm (60 in.) diameter HDPE profile wall pipe,
but is applicable to all profile wall thermoplastic pipe that exhibit local buckling limit states. The computational
model compares strain demands predicted from a two-dimensional plane strain finite element model of buried pipe
in the embankment condition with strain capacity predicted from a three-dimensional finite element model of a pipesoil segment undergoing thrust and/or positive and negative bending. The strain demands indicate the dominance of
thrust strains, as opposed to bending strains, in the overall behavior, particularly for intermediate to larger fill
depths. In the profile, the ultimate strain capacity is limited by local buckling for thrust strains and/or positive
bending (crest in compression), and inward radial movement of the crest for negative bending (liner in
compression). Depth of fill predictions between the new AASHTO design method for thermoplastic pipe and the
computational method agree within 10% of one another when uniform soil distribution is considered, and within
20% of one another when a soft haunch and other soft soils are considered in the pipe-soil envelope.

INTRODUCTION
Although there have been significant advances in the structural analysis of buried pipes used in design, for example,
the SIDD method of design for concrete pipe (1) and the proposed methods for design of large-span metal and
concrete culverts (2) behavior of buried thermoplastic profile wall (corrugated) pipe has largely been investigated
through experimental means or approximated by simplified design methods. Considering the variety of
thermoplastic profile wall pipe available combined with the variety of different installed conditions, it is prohibitive
to test every scenario. Thus, there is a need for computational methods that capture the essential nonlinearities of
both the soil and the pipe profile behavior under load, but do so in a manner that remains relatively efficient. Such a
procedure is reported herein for a 1500 mm (60 in.) diameter corrugated HDPE pipe and compared to the simplified
design method recently adopted by AASHTO (3,4).
BURIED PIPE SIMULATION
The approach taken here for simulating buried profile wall plastic pipe is to separate the demand and capacity
calculation. For the demand, CANDE (5) is employed to determine the appropriate global pipe-soil interaction under
an embankment condition and to predict the applied strain demands on the pipe as depth of fill increases, using
models as shown in Figure 1. For the capacity, general purpose three-dimensional finite element models of a pipesoil segment, as shown in Figure 2, are employed to determine the ultimate strain capacity (limited by local buckling
and other phenomena) under all expected end actions: positive bending, thrust, negative bending, and combinations

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

in between. With the ultimate strain capacity determined from the pipe-soil segment model the strain demands from
CANDE can be compared as depth of fill increases to determine nominal and allowable depths of fill.
SIMULATION OF BURIED PIPE STRAIN DEMANDS
Applied strain demands are predicted using finite element models developed in CANDE. CANDE is a twodimensional plane strain finite element analysis program developed expressly for buried pipe problems (5). We
employed the nonlinear hyperbolic soil models (6, 7) with the same soil properties used to develop the concrete pipe
SIDD design method (1). The model is built by incrementally adding pipe and soil layers, mimicking the actual
construction process. Additional surcharge loads may be added on top of the fill to simulate further increases in
depth of fill. The model predicts stress state in the soil and forces in the pipe.
The applied strain demands on the pipe are determined by converting the force and the moment on the pipe,
from CANDE analysis, to strains using engineering beam theory and ignoring curvature effects. For a particular
point, 1, in the pipe, the applied strains are the sum of thrust strain and bending strain:
outside fiber applied strain:
out1= P1/(EA) + M1cout/(EI)
inside fiber applied strain:
in1= P1/(EA) + M1cin/(EI)
where:
P1 = force in the pipe at point 1
E = pipe modulus of elasticity
A = area of the pipe cross-section
M1 = moment in the pipe at point 1
cout = distance from the centroidal axis to the extreme outside fiber (top of the crest)
cin = distance from the centroidal axis to the extreme inside fiber (bottom of the valley)
I = moment of inertia of the pipe cross-section

1
2

CANDE Models
Four models of buried 1500 mm (60 in.) diameter corrugated plastic pipe were developed. Two soil distributions:

uniform ML90 (silty soil compacted to 90% of maximum density per AASHTO T99) Backfill representing an
ideal situation with all backfill of the same material and density, including backfill in the haunch zone, and
non-uniform ML90 which includes ML90 backfill around the pipe, but with a soft haunch zone(CL50, i.e. a
clay soil compacted to 85% of maximum density per AASHTO T99) and CL85 in situ soil, see Figure 1;

and two pipe moduli (E): short-term, Epipe=Ei, (from 8, Ei=1082 MPa, 157,100 psi) and long-term, Epipe=E50, i.e.,
pipe modulus of elasticity at 50 years (from 3, Ei= 138 MPa, 20,000 psi) were considered. The finite element mesh,
construction increments, depth above the top of the pipe for each of the construction increments, and the soil zones
for the non-uniform soil case used in the pipe-soil interaction studies in CANDE is shown in Figure 1. For the
uniform soil case all the soil zones of Figure 1 are modeled as ML90, which is an ASTM D 2321, Class III backfill.
Soil stresses
Examination of the developed soil stresses in the CANDE analysis for the four models indicate:
significant vertical arching of the load around the pipe, in the uniform ML90 soil case at 6 m (20 ft) of fill, the
vertical soil stress immediately above the pipe is 34.5 kPa (5 psi), while the free-field soil stress at the same
depth is 124 kPa (18 psi);
elevated horizontal soil stresses at the springline and also for approximately 15 on each side of the springline as
the pipe pushes into the soil; and
elevated stresses in the stiffer soils (e.g., ML90 vs CL85, or in situ medium stiffness soil vs. CL85) when
nonuniform soils are present in the embankment.
Pipe Strain Demands
Strain demands around the pipe are given in Figure 3 at 12.9 m (42 ft) of fill. Results indicate that:
thrust strains are dependent on pipe modulus, thus as the pipe relaxes thrust strains increase,

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

bending strains are largely independent of pipe modulus. As the pipe relaxes, bending strains are constant, and
soft haunches and other non-uniform soil support around the pipe significantly increase bending strain demands.

The increase in thrust strains for the long term condition (lower pipe axial stiffness) highlights the importance of the
thrust capacity of the profile. The relative insensitivity of bending strains to pipe stiffness indicates that control of
bending strains is achieved during the initial installation of the pipe. Results near the soft haunch indicate that poor
effort spent in the haunch region can create bending strain demands that exceed the expected primary strain demands
at other locations in the pipe.
Pipe Diameter Change
As change in diameter is often used to assess the performance of an installed pipe, Figure 4 is provided to give the
CANDE predictions for diameter change as a function of depth. Results indicate that:
vertical diameter change due to circumferential shortening is greater for non-uniform soil support than uniform
soil support,
vertical diameter change due to bending is less for non-uniform soil support than uniform soil support, and
circumferential shortening dominates behavior for large depths of fill (greater than ~7.7 m (25 ft) in this
analysis).
SIMULATION OF BURIED PIPE STRAIN CAPACITY
Applied strain capacity is determined by using the general purpose finite element package ABAQUS (9). Threedimensional models of a segment of the pipe including local soil support are analyzed under a combination of
varying thrust and end moments to determine the pipe buckling capacity at all locations (crown, shoulder, springline,
invert, etc.). The selected pipe-soil segment is a 30 degree arc (Figure 2) with one full corrugation of the pipe, and
soil support for this corrugation, modeled. Symmetry is enforced on each side of the pipe-soil segment. Radial
support is provided for the outside face of the soil. The dimensions and thickness of the pipe are representative of
one manufacturer's 1500 mm (60 in.) diameter corrugated HDPE pipe in production in 1998.
Model and Mesh Size
Sensitivity of the model to the 30 arc length and 1 element spacing is investigated by eigenvalue buckling analysis
of the pipe, without soil. Element spacing between 0.5 and 2.5 indicates that elements every 1 provide the best
compromise between efficiency and accuracy. Strain demands are relatively uniform over a given 30 arc (Figure 3).
Thus 30 is selected to insure enough length to allow several buckling waves to form under a variety of different
moments and thrusts without impeding formation of the buckling due to end effects, while not so long as to require
more precise treatment of the strain gradient around the section.
Initial Imperfections
Simulation of the parallel plate test (8) indicates that introducing small imperfections in the initial pipe geometry are
critical to the numerical success of models for buckling capacity of corrugated pipe. A superposition of a crest
buckling imperfection (similar to Figure 6(a)) and a liner buckling imperfection (Figure 6(e)) was employed.
Maximum magnitude for imperfection of the liner is 0.1tliner = 0.38 mm (0.015 in.), maximum magnitude for
imperfection of the crest is 0.4tcrest = 1.52 mm (0.06 in.). These imperfections include the same mode of
deformation for the soil and the pipe therefore the initial state for the soil and pipe is matched.
Material Model
Material models used in the pipe-soil segment model (Figure 2) are linear elastic. The pipe is modeled using an
initial modulus (Ei) and the soil is modeled with a modulus, E, of 6.9 MPa (1,000 psi), representing a lower bound
value for ML90 fill. Selection of linear elastic material models is based on a) desire to focus on buckling
phenomena, b) past success with this approach in parallel plate modeling (8), and c) computational efficiencies
needed for complex three-dimensional models. Although the material models are linear elastic, significant geometric
nonlinearities exist and are captured.
Loading
Loading of the pipe-soil segment is performed by hoop displacement and rotation of the ends of the pipe-soil
segment. Rotation is performed about the elastic centroid of the pipe profile. Force and moment that develop at the
end of the pipe due to the deformation and rotation are recorded and are converted to strains in the same manner as

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

completed in the CANDE analysis. Both the pipe and the soil are displaced and rotated at the ends of the segment,
insuring compatibility between pipe and soil.
Pipe-Soil Contact and Analysis Method
The pipe-soil interface is modeled as frictionless (unbonded) by using linear gap elements between each node of the
pipe and a corresponding node in the soil. The analysis employs an equilibrium path technique (Riks, see 9) that
traces the entire load-displacement path of the pipe. Buckling is observed by the deformation of the pipe and
eventual loss of load carrying capacity. Nonlinear geometry effects, initial imperfections, neutral axis migration,
and other phenomena that develop during the loading history of the segment prior to buckling are included the
analysis.
Results
Nonlinear buckling analysis of the pipe-soil segment results in the buckling capacity curve provided in Figure 5.
Each nonlinear analysis is evaluated with a different set of thrust and bending strains on the pipe-soil segment. The
gap between the pipe and the soil may be wide (e.g., Figure 6(a)) although compatibility at the ends of the segment,
where the displacements and rotations are applied is always maintained. Buckling of the segment does not occur
suddenly, but rather develops gradually over the duration of the loading. Under positive bending the primary
deformation of the pipe is in the crest (Figure 6(a)) and contact between the pipe and soil in the crest is largely lost.
Under increased thrust, but still significant positive bending (b), the deformations in the web of the profile increase
and contact between the pipe and the soil is lost only in the inward buckling waves. For large amounts of thrust and
small positive bending (c) the liner buckles as well as the crest. For pure thrust (d) the entire profile buckles, i.e.
crest, web, valley and liner and contact between the pipe and the soil is lost for the inward buckling waves. Under
negative bending (e,f) the liner buckles and the crest experiences large inward radial deformations. These inward
deformations of the crest result in loss of pipe to soil contact and reduce the bending capacity.
Simulation: Depth of Fill Prediction
The capacity of a pipe may be limited by buckling of the profile, or by general yielding. Experience in stub
compression tests (3) and parallel plate analysis (8) indicate that buckling, not yielding, is the mode of greatest
concern for many of the corrugated plastic pipe profiles in current use. The nonlinear buckling analysis curve
(Figure 5), combined with the strain limits discussed in (3) are selected as the basis for the limiting capacity curve.
Any combination of strain demands that fall within these limits (buckling and yielding) are sustainable by the pipe.
In Figure 7, the strain demands in a CANDE analysis with non-uniform ML90 backfill and long-term pipe
modulus are compared versus the capacity curve. The strain demands are shown as lines emanating from the origin;
each line ends with a number, in degrees, that refers to the initial location of the point on the pipe. A label of 0
refers to the crown, 90 refers to the springline, and 180 to the invert. At 1.8 m (6 ft) of fill above the crown, Figure
7(a), strain demands are well below the capacity curve. Figure 7(b) through 7(d) show the effect of increasing the
depth of fill. At 3.6 m (12 ft) of fill, the springline (i.e., 90) as well as 30s to either side of the springline (60 and
120) experience the greatest demands. Thrust plays a significant role in the behavior even at relatively low fill
depths. At 4.9 m (16 ft) of fill, demand exceeds capacity at the springline (90). Expected deformation at the
springline is given in Figure 6(d,e). Buckling of the liner and portions of the main profile may occur before 4.9 m
(16 ft) of fill, but beyond 4.9 m (16 ft) deformations are too large for the pipe to carry additional strain without loss
in load carrying capacity. As the depth of fill increases, demand exceeds capacity in additional regions (i.e., more
lines cross the capacity curve).
Based on the methodology described above, depth of fill predictions are given in Table 1. The four separate
CANDE analyses are evaluated for nominal (ultimate) depth of fill (no safety factors), and design depth of fill
(including safety factors). Results are given for the depth of fill when a section of the pipe reaches the buckling
capacity curve of Figure 5 and when the strain limit is first reached. The limiting depth of fill predictions of Table 1
are based on the first depth of fill at which a point in the pipe reaches a buckling or yielding limit state. If the strain
demands at the failed location can be withstood, redistribution of the force would occur. However, as Figure 7
indicates the demands for a full 30s on either side of the springline are nearly the same as the springline itself. Thus,
loss of carrying capacity at the springline immediately portends loss of capacity over a significant portion of the
pipe.

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

Table 1 Depth of fill predictions

Depth of fill, m, (ft) at limit of
Buckling
Strain
of Profile
Limit
NOMINAL PREDICTION (ALL SAFETY FACTORS = 1)
Backfill

Soil

Pipe Modulus

Uniform

E50

6.4 (21)

7.9 (26)

Non-Uniform

E50

4.9 (16)

5.5 (18)

Uniform

Ei

20.4 (> 67)

20.4 (> 67)

Non-Uniform
Ei
20.4 (> 67)
DESIGN PREDICTION (WITH SAFETY FACTORS*)

20.4 (> 67)

ML90

ML90

Uniform

E50

2.7 (9)

3.7 (12)

Non-Uniform

E50

2.6 (8.5)

2.9 (9.5)

Uniform

Ei

9.1 (30)

20.1 (66)

9.8 (32)
13.4 (44)
Non-Uniform
Ei
* Safety factors: 2 for buckling, 2 for the yielding in thrust strain limit, and 1.5 for the combined yielding strain limit
Pipe Deflection
CANDE results are combined with the depth of fill predictions given in Table 1 to evaluate the predicted total
vertical deflection in the pipe at the maximum depth of fill. The predicted long-term ultimate total vertical deflection
for a ML90 backfill is 7.3% in uniform soil and 5.0% in non-uniform soil. The design deflection is 3.6% in uniform
soil and 2.6% in non-uniform soil. The relatively small predictions for the bending deflection are due to the large
levels of predicted thrust strains. The thrust strains result in loss of capacity for the pipe without large vertical
deflection. Pipe analyzed with initial modulus have much higher levels of deflection at the maximum depth of fill.
BURIED PIPE DESIGN
The newly adopted design methods for profile wall thermoplastic pipe design (3,4) are used to evaluate the same
1500 mm (60 in.) diameter pipe studied using the buried pipe simulation methods. These new design methods
capture the following essential behavior of the pipe-soil system:

soil stiffness (Ms) is a function of soil type and depth of fill,

vertical arching is a function of the ratio of soil to pipe (hoop) stiffness,
decreased capacity due to local buckling is a function of buckling strain of an element of the pipe and the
applied ultimate strain on the pipe,
bending capacity of the pipe is limited by the total strain of the material less the hoop strain the pipe undergoes.

Design: Depth of Fill Prediction

For 1500 mm (60 in.) diameter pipe, we examined the allowable depth of fill for design and ultimate conditions. The
analysis is carried out for four soil types: SW95, SW90, ML90, and CL85. The design condition employs load
factors () of 1.95 on thrust, 1.5 on combined thrust and bending, and resistance factors () of 1.0 on pipe
performance, 0.9 on soil resistance. Ultimate capacity is examined by setting all load factors (s) and resistance
factors (s) to 1.0. In all cases, the depth of fill is limited by the reduced capacity in local buckling. The results are
summarized in Table 2.

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

Table 2 Depth of fill, m, (ft) for 1500 mm (60 in.) pipe as a function of soil type
Condition
Design
Ultimate
Design
Ultimate

SW95
SW90
Considering Local Buckling
6.4 (21.1)
4.2 (13.9)
16.3 (53.4)
9.7 (31.7)
Ignoring Local Buckling
8.8 (28.9)
5.7 (18.7)
21.3 (>70)
13.1 (43.0)

Soil Type
ML90

CL85

3.0(9.7)
6.3 (20.6)

2.1 (6.9)
4.4 (14.5)

3.9 (12.8)
8.3 (27.3)

2.8 (9.2)
5.9 (19.2)

The results indicate that:

even if local buckling is removed or reduced, through some improvement in design, burial depths greater than
15 ft will require SW90 or better or ML95 or better soil,
the factor of safety with respect to depth of fill (e.g., in SW95 16.3 m/6.4 m = 2.5) is greater than the design
factor of safety with respect to the local buckling limit state (1.95).
The allowable vertical deflection capacity of the pipe wall is based upon the available strain capacity which is determined from the 50-year strength and modulus of the HDPE. For the design condition the maximum
bending deflection is 4.4% change in diameter, and the total deflection is 6.5%. For the ultimate condition the
bending deflection is 4.5% and the total deflection is 8.6%. For a given soil these deflections occur at the fill depths
shown in Table 2.
The behavior of pipe using this design method is driven by three essential quantities: constrained soil
modulus (Ms), vertical arching factor (VAF), and local buckling capacity, i.e., effective area (Aeff). As depth of fill
increases Ms increases, VAF decreases, and Aeff decreases all in a nonlinear fashion. Quality fills, such as ML95
or better, or SW90 or better provide significant increases in Ms with increased depth of fill. The pipe hoop stiffness
of 1500 mm (60 in.) diameter pipe is low enough that significant changes in VAF are realized with quality fill. The
local buckling reduction also sees some benefit with higher VAF, but since effective area is always calculated at an
ultimate limit state this benefit is limited.
DISCUSSION
Design vs. Simulation
Depth of Fill Predictions: Design vs. Simulation
The depth of fill predictions for ML90 backfill provided by the simplified design method reported in Table 2 are 3.0
m (9.7 ft) for design and 6.3 m (20.6 ft) for the ultimate condition. The depth of fill predictions provided in Table 1,
for uniform ML90 backfill, are 2.7 m (9 ft) for design and 6.4 m (21 ft) for the ultimate condition. Consideration of
non-uniform ML90 backfill yields a depth of fill prediction of 2.6 (8.5 ft) for design and 4.9 m (16 ft) for ultimate.
The agreement of the simplified method with the comprehensive numerical analysis presented herein shows that the
simplified design method can provide accurate depth of fill predictions for use in design.
Strain Demands Prediction: Design vs. Simulation
The proposed design method employs the shape factor approach for calculating bending strains, via:
c
b = D f
R D
where:
Df = shape factor
c = distance from centroid to fiber of interest (usually outside or inside of pipe)
R = centroidal pipe radius
= vertical diameter change due to bending (remove circumferential component)
D = centroidal diameter

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

A table for the shape factor as a function of pipe and soil stiffness may be found in (10). A shape factor
between 5.0 and 6.0 is selected for investigation here. Change in the vertical pipe diameter due to bending, as
determined from CANDE analysis (Figure 4), is used for . The maximum compressive strain in the outside fiber
due to bending, ignoring the local peaks in the soft haunch region, is used for comparison (Figure 8). The
comparison demonstrates that the shape factor approach provides:

a reliable prediction of initial strain demands (i.e., 0 depth of fill),

a reliable prediction of strain demands up to moderate fill depths (< 6 m, 20 ft.), and
a reasonable and general methodology for hand prediction of bending strain demands.

Since the shape factor table used in this study was developed for fiberglass pipe, additional research should be
completed to investigate the applicability of the table to all types of thermoplastic pipe under additional installation
and backfill conditions.
Positive Bending - Maximum Buckling Capacity Curve vs. Strain Limits
The computational model predicts that the maximum positive bending strain is approximately 4%, compared with
6% suggested in the LRFD specifications (4). The maximum bending strain for (4) is based, in part, on a study of
the expected bending strain at 20% vertical deflection in the parallel plate test. Analysis of 450 mm to 1200 mm (18
in. to 48 in.) diameter pipe suggests strain levels between 8.9% to 5.7% strain, with the smaller diameter pipe
undergoing the higher strain levels. Strain limits in (4) recognized the high strains experienced in the smaller
diameter pipe and thus suggested a strain limit in bending 50% higher than that used in compression, i.e., 6% strain.
However, using the shape factor approach employed in (4) the predicted strain at 20% deflection for the studied
1500 mm (60 in.) diameter pipe is only 5%. Deformations in Figure 6(a) suggest that the soil provides little or no
support in the region when the pipe buckles in positive bending. In positive bending, the maximum buckling
capacity curve is providing a strain limit that recognizes a loss of soil support; the parallel plate test is providing an
upper bound approximation that includes full soil support to the crest; and the strain limits of (4) are providing an
average strain capacity for all diameter pipe. With the current simplifications used in the LRFD hand method, the
average limiting strain of 6% in current use appears reasonable.
Simulation details
Applied vs. Local Strain Demands
Applied strain demands (P/(EA) + Mc/(EI)) differ from the local strain demands due to bending of the pipe as a
plate from buckling deformations, shear lag effects that limit the effectiveness of the liner to carry load, and any
other local or three-dimensional effects (8). Comparison of the strain capacity to the demand is completed using
applied strain demands as CANDEs modeling does not incorporate these effects. However, the detailed threedimensional ABAQUS analysis calculates local strain demands at peak capacity, the total applied strain agrees
well with average strain in the corner (Table 3) of the crest and web of the profile. An exception to this conclusion is
the case of negative moment and large thrust (case (e)). In this case, strain demands based on the initial geometry are
problematic, because a large amount of inward radial deformation occurs in the crest (for the stiffened profile
shown); however, at this location the strain demands are small. Peak strain demands identified in the model are
markedly higher than applied strains in this case. In general, highly localized strain demands, at the extreme fibers
through the thickness of the plastic, and at junctures of the different elements, must be resisted by the material for
the predicted capacities to be realized. This makes the use of a material with proven ductility an important criterion
for resin selection.

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

Table 3 Strain Demands (%) from Nonlinear Buckling Analysis (at Ultimate)

condition
(a) positive moment
(b) positive moment
and moderate thrust
(c) positive moment
and large thrust
(d) thrust
(e) negative moment
and large thrust
(f) negative moment

CANDE
applied
strain
total
4.0

ABAQUS pipe-soil model

crest/web
any location
corner
min
max
average
tension
compression
3.1
-9.5
14.9

5.2

4.8

-5.5

8.5

4.7

4.4

-5.0

6.6

4.6

4.4

-9.7

10.3

0.7

-0.8

-7.3

6.9

-4.9

-6.6

-9.1

7.6

Thrust Strains and Slip in Soil

Strain demands predicted from CANDE are primarily thrust strains, as opposed to bending strains. For instance, note
in Figure 7, strain demands are essentially horizontal lines, i.e., thrust strain demands greatly exceed bending strain
demands. The CANDE analysis is performed assuming a perfect bond between pipe and soil. If slip between the
pipe and soil is allowed, thrust strains will become more uniform around the pipe. Decreases in the peak thrust strain
demand will result in increased predictions for the allowable depth of fill and percent deflection. If the mean thrust
strain in the pipe is used as the demand (to approximate soil completely free to slip) the allowable depth of fill
(without safety factors) for an ML90 backfill increases by 1.8 m (6 ft) in uniform soil and 1.2 m (4ft) in non-uniform
soil and the design (with safety factors) allowable depth of fill for an ML90 backfill increases by 1.7 m (5.5 ft) in
uniform soil and 0.8 m (2.5 ft) in non-uniform soil vs. those of Table 1. Overall, other effects, such as strains from
the construction process, counter-act such beneficial effects as decreases in the peak thrust strain due to
circumferential slippage at the pipe-soil interface.
Test Methods and the Capacity Curve
Alternative to the detailed finite element analyses, test methods could be employed for determining the capacity
curve. The parallel plate test predicts the maximum bending strain that the pipe profile will maintain when supported
against a rigid foundation (the steel platens) this provides an upperbound approximation of the positive bending
strain capacity. The stub compression test (3), or soil cell hoop tests, can provide the capacity in pure thrust. These
two values along with reasonable strain limits could be used to form a basic capacity curve.
Factor of Safety
The analysis for factor of safety indicates that, given installation in accordance with the design conditions, the
ultimate depth of fill exceeds the allowable depth of fill by more than the factor of safety of about 2 targeted by
AASHTO, and traditionally applied to all flexible pipe. From an alternate point of view, looking at backfill or
compaction not meeting specified levels, a pipe designed for SW95 backfill with a design depth of fill 6.4 m (21.1
ft) would not reach an ultimate condition as long as the backfill properties exceeded those of ML90. This is a
substantial allowance for non-conforming installation. Given the substantial uncertainties in the soil and installation,
and to a lesser extent in the pipe performance, systematic methodologies for determining the reliability (and
resulting factor of safety) for the pipe-soil system are still needed.
CONCLUSIONS
A computational method was presented for determining the allowable depth of fill of 1500 mm (60 in.) diameter
profile wall (corrugated) HDPE pipe and compared with recently adopted AASHTO design methods for the same
pipe. The method relies on strain demands from a two-dimensional plane strain model of the pipe in an embankment
condition under increasing fill depths and determination of the ultimate strain capacity from a three-dimensional
model of a pipe and soil segment under expected end actions: thrust and/or bending. The developed soil stresses,
direct pipe strain demands, and change in pipe diameter as depth of fill is increased all indicate the prominent role of
thrust strains (as opposed to bending strains) in the behavior of these pipe in the embankment condition. Analysis of

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

the pipe-soil segment under increasing thrusts and/or bending indicate a bounding capacity curve that exists due to
local buckling of the pipe profile. For the examined pipe this capacity curve is in generally good agreement with the
strain limits employed in the newly adopted AASHTO design method. Further, comparison of the depth of fill
predictions between the new AASHTO design method and the simulations indicates that agreement is remarkably
good: depth of fill predictions are within 10% of one another when uniform soil distribution is considered, and
within 20% of one another when a soft haunch and other soft soils are considered in the embankment. The
simulation model presented provides a general means for assessing capacity of buried pipe, and serves to provide
additional validity to the recently adopted AASHTO specification methods.
ACKNOWLEDGEMENTS
The authors would like to gratefully acknowledge the financial support of Hancor, Inc. in developing the work
presented herein. In addition, Dan Valentine, formerly of Simpson Gumpertz & Heger, Inc. aided in the calculations
for the design depth of fill by the recently adopted AASHTO method.
REFERENCES
1.

Heger, F.J., (1988), "New Installation Designs for Buried Concrete Pipe," Pipeline Infrastructure
Proceedings of the Conference, American Society of Civil Engineers, New York, NY, pp. 117-135.

2.

McGrath, T.J., Moore, I.D., Selig, E.T., Webb, M.C. and Taleb, B., (2002) "Recommended Specifications for
Large Span Culverts," Tansportation Research Board, National Cooperative Highway Research Program
(NCHRP). Report (number will be available by September).

3.

McGrath, T.J., Sagan, V.E. (2000). LRFD Specifications for Plastic Pipe and Culverts. Tansportation
Research Board, National Cooperative Highway Research Program (NCHRP) Report 438.

4.

AASHTO (2001). AASHTO-LRFD Bridge Design Specification, 1998 with 2nd Edition with 2001 Interim
Specifications

5.

CANDE-89 Culvert Analysis and Design computer program User Manual. Publication No. FHWA-RD-89-169.
Research, Development, and Technology, Turner-Fairbank Highway Research Center, Federal Highway
Administration, U.S. Department of Transportation

6.

Duncan, J.M. and Chang, C-Y. (1970) Nonlinear Analysis of Stress and Strain in Soil, Journal of the Soil
Mechanics and Foundations Division, ASCE, Vol. 96, No. SM5: 1629-1653.

7.

Heger, F.J., Liepins, A.A., and Selig E.T., (1985), "SPIDA: An Analysis and Design System for Buried
Concrete Pipe," Advances in Underground Pipeline Engineering Proceedings of the International
Conference, American Society of Civil Engineers, pp. 143-154.

8.

McGrath, T.J., and Schafer, B.W. (2003) Parallel Plate Testing and Simulation of Corrugated Thermoplastic
Pipe Submitted for presentation at the 2003 Transportation Research Board Annual meeting, Washington D.C.

9.

ABAQUS Version 6.1. (1998) Hibbitt, Karlsson & Sorensen, Inc. Pawtucket, RI.

10. AWWA (1996). AWWA M45 Manual for Fiberglass Pipe Design.

TRB 2003 Annual Meeting Initial Submittal for Review

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

Schafer and McGrath

10

12-20 surcharge load

425.

11

400.
375.

10

350.

ML90

325.
9

300.
275.

250.
225.
200.
3, pipe

ML90
7
CL85

175.
2

CL50

150.

125.
ML90
100.

75.
50.
25.

in situ Medium Stiffness

1

0.

INC. DEPTH(FT)
1 -5.2
2 -5.2
3 -5.2
4 -3.6
5 -2.6
6 -0.6
7 1.6
8 7.1
9 12.4
10 18.7
11 20.8
12 26.0
13 31.2
14 36.4
15 41.6
16 46.8
17 52.0
18 57.2
19 62.4
20 67.6

0. 25. 50. 75. 100.125.150.175.200.225.250.275.

Figure 1 CANDE model for strain demand calculation showing construction increments and soil zones

Figure 2 ABAQUS model for strain capacity calculation showing pipe and local ring of soil

TRB 2003 Annual Meeting Initial Submittal for Review

Schafer and McGrath

11

DEP TH = 41.6 133 FT

12
Uniform ML90, E pipe =E initial
Uniform ML90, E pipe =E 50

10
axial strain (%)

NON-Uniform ML90, E
8

NON-Uniform ML90, E

pipe
pipe

=E
=E

initial
50

6
4
2
0
0

20

40

60

20

40

60

80
100
theta (deg. 0=crown)

120

140

160

180

120

140

160

180

outs ide fiber bending stra in (% )

4
3
2
1
0
-1
-2
-3
-4
80

100

theta (deg.) 0 o =crown

Figure 3 CANDE predictions for applied axial and bending strains around the circumference of the pipe
(note, axial strain refers to circumferential hoop strain, not longitudinal effects)
VERTICAL DIAMETER CHANGE

HORIZONTAL DIAMETER CHANGE

6

ve rtic al dia m ete r cha nge (% ID)

Uniform ML9 0, E pipe =E initial

Uniform ML9 0, E pipe =E 50

-2

NON-Uniform ML9 0, E pipe =E initial

NON-Uniform ML9 0, E pipe =E 50

-4

-6

-8

horiz ontal dia m ete r cha nge (% ID)

-10

5
4
3
2
1
0
-1
-2

10

20

30
de pth (ft)

40

50

60

VERT. DIAM CHANGE DUE TO CIRCUMFERENTIAL S HO RTENING

0

-2

-4

-6

-8

-10

10

20

30
depth (ft)

40

50

60

VERT. DIAM CHANGE DUE TO BE NDING

0
vertica l dia m ete r c ha nge (%ID)

vertica l dia m ete r c ha nge (%ID)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

-2

-4

-6

-8

-10
0

10

20

30
de pth (ft)

40

50

60

10

20

30
depth (ft)

40

50

60

Figure 4 Vertical and horizontal diameter change versus depth of fill based on CANDE demands

TRB 2003 Annual Meeting Initial Submittal for Review

Schafer and McGrath

12

8
6

(a)

outside bending strain (%)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

(b)

(c)
(d)

0
-2

(e)

-4
-6

(f)
-8
0

0.5

1.5

2.5

3.5

4.5

thrust strain (%)

Figure 5 Result of pipe-soil segment analyses under varying amounts of applied bending and thrust, envelope
of the results is the buckling capacity curve for the studied pipe diameter (Figure 6 provides deformations at
peak capacity for points (a) (f) identified in the figure)

(a) peak positive bending capacity

(b) peak positive bending capacity under moderate thrust

(c) peak positive bending capacity under large thrust

(d) peak thrust capacity

(e) peak negative bending capacity under moderate

thrust

(f) peak negative bending capacity

Figure 6 Cut-away view of pipe-soil deformation at peak capacity (identified in Figure 5)

TRB 2003 Annual Meeting Initial Submittal for Review

Schafer and McGrath

13

OUTSIDE STRAIN AT DEPTH OF FILL = 6.4235 FT.

OUTSIDE STRAIN AT DEPTH OF FILL = 12.484 FT.

6
Buckling
Yielding

Buckling
Yielding

)
%
(

outsidenai
rt
s
bending
g
strainndi
(%) neb
e
di
st
u
o

)
%
(

outsidenai
r
bendinggst
strain ndi
(%) neb

0
180 30
150

120
60 90

e
di
st
u
o

-2

-4

-6

0
180

30
120 90
60

150
-2

-4

0.5

1.5

(a)

2
2.5
3
thrust strain (%)

3.5

4.5

-6

0.5

1.5

2
2.5
3
thrust strain (%)

(b)

OUTSIDE STRAIN AT DEPTH OF FILL = 15.605 FT.

3.5

Buckling
Yielding

)
%
(

e
di
st
u
o

)
%
(

outsidenari
t
bendinggs
strain ndi
n
(%) eb

0
180

30
120
150

60 90

e
di
st
u
o

-2

-4

-6

6
Buckling
Yielding

outsidenari
t
bendinggs
strain ndi
n
(%) eb

4.5

OUTSIDE STRAIN AT DEPTH OF FILL = 18.726 FT.

0
180

30
120
60 90

150
-2

-4

0.5

1.5

(c)

2
2.5
3
thrust strain (%)

3.5

4.5

-6

0.5

1.5

(d)

2
2.5
3
thrust strain (%)

3.5

4.5

Figure 7 Strain demands vs. capacity as depth of fill Increases (Applied strain demands shown here are for
the CANDE model with non-uniform ML90 backfill, Figure 1, and the long-term pipe modulus, E50)
Uniform ML90, E pipe =E 50

2.5

2.5
outs ide fiber s train (%)

outs ide fiber s train (%)

Uniform ML90, E pipe =E initial

2
1.5
1
Uniform ML90, E pipe =E initial
S hape Fa ctor Df=6 (AWWA)

0.5

2
1.5
1
Uniform ML90, E pipe =E 50
S hape Fac tor Df =6 (AWWA)

0.5

S hape Fa ctor Df=5 (AWWA)

S hape Fac tor Df =5 (AWWA)

0
0

10

20

30
depth (ft)

40

50

60

10

NON-Uniform ML90, E pipe =E initial

20

30
de pth (ft)

40

50

60

50

60

NON-Uniform ML90, E pipe =E 50

3
NON-Uniform ML90, E pipe =E initial
S hape Fa ctor Df=6 (AWWA)

NON-Uniform ML90, E pipe =E 50

S hape Fac tor Df=6 (AWWA)

2.5
outs ide fibe r s tra in (%)

2.5
outs ide fibe r s tra in (%)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

S hape Fa ctor Df=5 (AWWA)

2
1.5
1
0.5

S hape Fac tor Df=5 (AWWA)

2
1.5
1
0.5

0
0

10

20

30
depth (ft)

40

50

60

10

20

30
de pth (ft)

40

Figure 8 Comparison of shape factor approach for applied strain demands with CANDE models

TRB 2003 Annual Meeting Initial Submittal for Review