8

CHAPTER 2

LITERATURE REVIEW

2.1 INTRODUCTION

In the early years of the 20th century, the newly established Iowa
Engineering Experiment Station, directed by Anson Marston, began a
theoretical and experimental research on structural failure of drainage pipes.
Their aim was to establish a method for estimating the load and its
distribution on the buried pipe, and a method for determining the supporting
strength of the pipe. The research led to the well known Marston-Spangler
theory, which is still the current design method for buried pipes in different
parts of the world since the method is simple and easy.

Other researchers such as Burns and Richards (1964) and Hoeg

(1968) worked on more rigorous theoretical solutions for buried conduits.
They extended the original work of Mindlin on elastic circular pipes placed
in a deep infinite elastic media and subjected to vertical soil pressures
(Masada 1996).

The development of computers in the last few decades increased

the evolution of sophisticated numerical methods. As a result, analytical tools
such as finite element codes are now available for engineers and researchers
to calculate stresses and deformations of buried conduits.

Continued research and development in the pipe materials indicated

that the mechanical behaviour of buried pipe is essentially controlled by the
 9

soil stiffness, pipe stiffness, and depth of burial, unit weight of backfill and
method of installation. Some of the approaches and contributions that aided
understanding the behaviour of buried pipes are reviewed in this chapter
including classes of underground conduits in the following order.

(i) Classes of underground conduits.

(ii) Design methods for buried pipes.
(iii) Experimental studies on buried pipes.
(iv) Numerical studies

2.2 CLASSES OF UNDERGROUND CONDUITS

Underground conduits are divided into three major classes for the
purpose of load computation. These classes are ditch, projecting and
imperfect ditch conduits and are based on the construction methods that
influence the distribution of the load on the pipe. The projecting conduit class
is subdivided into positive projecting and negative projecting conduits.
Figure 2.1 illustrates the three main types of installation of conduits.

A ditch conduit (Figure 2.1a) is the one which is installed in a

relatively narrow trench in undisturbed soil, and then backfilled. The
projecting conduit is defined as a pipe installed above the natural ground
surface or in a relatively narrow and shallow trench, and then covered with an
embankment. There are two types of projecting conduits that are positive
(Figure 2.1b) and negative (Figure 2.1c) projecting conduits.

The conduit that is installed in shallow bedding with its crown

projecting above the natural ground surface is known as positive projecting
conduit and the conduit that is installed in relatively narrow and shallow
trench with its crown remaining below the natural ground surface is known as
negative projecting conduit.
 10

Figure 2.1 a) Trench b) Positive projecting c) Negative projecting

d) Imperfect ditch (Spangler and Handy 1982)

Imperfect ditch conduit (Figure 2.1d) is a special case of the

negative projecting conduit. In fact, it is an artificially made negative
projecting conduit. First, a certain height of well-compacted embankment
covers the pipe. Second, a narrow ditch is excavated over the pipe and
backfilled with loose soil and the embankment is completed to its final height.

Load on a buried pipe is not exactly equal to the weight of the soil
over the pipe, which is called the soil prism. The weight of the soil prism is
found to be the pipe outside diameter times the height of earth above the pipe
times the unit weight of the earth. The load acting on a pipe depends on the
movement of the soi1 prism relative to the soil on the sides. In the case of
ditch conduit, the backfill material and the pipe have a tendency to settle
downward relative to the sides of the trench. This relative movement
 11

mobilizes shearing forces, which act upward, along the sides of the trench.
These shearing forces, associated with horizontal forces from the sides of the
trench, reduce the vertical load on the pipe. This phenomenon is called the
arching effect and it is as permanent as any other form of shear resistance
(Petroff 1990).

2.3 DESIGN METHODS FOR BURIED PIPES

Literature review of the research works related to design of buried

pipes reveals that these can be grouped into two broad categories namely
classical and rigorous analytical methods. Typical design procedure adopted
is provided in Appendix 1. The main objective of the design methods was to
establish a procedure for estimating the load and its distribution on the buried
pipe. Some of these methods have provided the procedure to predict
deformation of pipe and their influence on load distribution on the pipe.

2.3.1 Classical Methods

Most of the methods referred here predict load on buried pipe using
theory of arching and limit equilibrium method.

2.3.1.1 Load on a Ditch Conduit

Marston (1930) applied the arching analysis by considering the

width of the ditch, in order to determine the vertical load on a pipe buried in a
narrow trench. Considering the value of cohesion as zero, the formula for
vertical load on top of the conduit was developed. The vertical load on any
horizontal plane at depth h of the backfill of buried pipe below ground
surface is
 12

V = Bd2 ((1- e -2K'(h/Bd)) /(2K') (2.1)

where, - unit weight of the backfill material ( kN/m3)

Bd - horizontal width of the ditch at the crown (m)
h - distance from the ground surface to any horizontal plane in
the backfill (m)
K - ratio of active lateral pressure to vertical pressure (Rankine's
ratio)
- friction angle of the fill material ()
' - friction angle between the fill material and the sides of the
ditch ()
- coefficient of interna1 fiction of the fill material (tan )
' - coefficient of fiction between the fil1 material and the sides
of the ditch (tan')

Christensen (1967), who undertook a further study on the value of

earth pressure ratio ( K) proposed the following formula:

K = (1+2tan2 )-1 = 1-sin 2 /1+sin2 (2.2)

The portion of the total load that is carried by the pipe depends on
the rigidity of the pipe. In the case of a rigid pipe, the side fills may move
downward relative to the soil prism causing the pipe to sustain the entire load
V. In the case of a relatively flexible pipe, the soil prism may move
downward relative to the side fills, because of the deflection of the pipe,
causing the pipe to sustain a reduced load.
 13

As a result, Marston determined the load of a rigid ditch conduit

(Wd) by the following formula,

Wd = Cd Bd2 (2.3)

where Cd = 1-(e-2K'(H/Bd) / 2k')

The formula for the relatively flexible pipe is the following:

Wc = CdBdBc (2.4)

BC = outside diameter of the conduit (m)

Equations 2.3 and 2.4 give the maximum load on a particular pipe
in service. Experiments and field observations have demonstrated that the
pipe may not reach the maximum load for a long period of time. In fact, the
load keeps building up for an extended period of time after the completion of
the backfill (Spangler and Handy 1982).

The width of the ditch (Bd) in the previous formulae is the width of
a normal trench with vertical sides. In the case of trenches with sloping side,
the parameter Bd must be the width of the trench at the crown of the pipe
(Schlick 1932).

2.3.1.2 Positive projecting conduit

As mentioned before, the crown of a positive projecting conduit

projects over the natural ground surface. The distance from the pipe's crown
to the ground surface is expressed by the term pBc, where p is the projection
ratio. The planes where the shear forces act in a positive projecting conduit
are assumed to be the vertical planes extending upward from the sides of the
conduit.
 14

The rigidity of the pipe has a large effect on the vertical load
applied on the pipe. The magnitude and direction of the movement of the soi1
prism relative to the side fills, are influenced by the settlements of the natural
ground, the settlement of the side fills and the deformation of the pipe.

In the early years of 20th century Marston derived expression for

the vertical load per unit length on positive projecting conduit. For the
complete ditch or projection condition the equation is

Wc = Cc Bc2 (2.5)

Cc = e+2K'(H/Bc)-1 /+2k (2.6)

where Cc - load coefficient for positive projecting conduit

The plus sign is used for the complete projection condition and the
negative sign is used for the complete ditch condition. Figure 2.2 presents the
variation of Cc for various values of (H/Bc) and four different types of
underground conduits.

Figure 2.2 Curves for determination of Cc for positive projecting

conduit (Moser 1990)
 15

2.3.1.3 Imperfect Ditch Conduit

In the imperfect ditch conduit, the objective is that the soil prism
above the pipe will settle more than the side fills (Figure 2.3). The imperfect
ditch system includes a second projection ratio, which is designated by p'. The
projection ratio p' is defined as the depth of the ditch divided by its width. In
addition, the critical plane is now the horizontal plane in the trench backfill
material at the level of the compacted backfill surface. The settlement ratio is
given by the following formula:

rsd = (Sg-(Sd+Sf+ dc)) /Sd (2.7)

where rsd - settlement ratio

Sg - settlement of the side fill of height pBc
Sf - settlement of the conduit into its foundation
d c - pipes deflection
Sd - settlement of the fill in ditch within height p'Bc

Spangler and Handy (1982) stated that it is tentatively

recommended that the settlement ratio be assumed to lie between -0.3 and
- 0.5 for the purpose of estimating loads.

Figure 2.3 Imperfect ditch conduit

 16

The formula for the calculation of the vertical Load on imperfect ditch
conduit is

Wc = CnB2c (2.8)

where Cn - load coefficient for imperfect ditch conduit (Figure 2.4), which is
function of (H/Bc), p' and rsd.

Figure 2.4 Curves for the determination of Cn (Moser 1990)

2.3.1.4 Negative projecting conduit

The load on negative projecting conduit is determined using

the same procedure as for the imperfect ditch conduit. Spangler (1960)
presented the complete theory on loads on negative projecting conduits
(Figure 2.5).
 17

Figure 2.5 Negative projecting conduit (Moser 1990)

2.3.1.5 Conduits in Wide Trenches

Research on the trench theory undertook by Schlick (1932) led to

the definition of a transition width. The transition width is defined as the
width at which the load calculated from the projecting theory is equal to the
load calculated from the trench theory. The transition width is a function of
the depth of cover H, and becomes smaller as the depth of cover increases. In
the application of transitional width concept, it is suggested to use the product
of the settlement ratio and the projection ratio.

2.3.1.6 Surface Loads

Wheel loads from trucks, airplanes or trains cause concentrated

loads on buried pipes. The AASHTO Bridge Specification has a simplified
procedure for determining the distribution of load in the ground resulting
from concentrated loads on the ground surface. In this procedure, the load is
assumed to attenuate with increasing depth at an angle of 41 with the vertical
in each direction. The American Concrete Pipe Association (ACPA) design
 18

practice distributes the earth pressure at the crown on a larger area in the
longitudinal direction of the pipe because of the beam strength of the pipe in
this specific direction. It is also possible to use Boussinesq formula to
determine the distribution of the load in the ground. The American Water
Works Association has its design approach for flexible pipe (AWWAC950).
The distributed surface surcharge loads may be calculated as an equivalent
additional layer of soil.

Figure 2.6 Distributed load over the pipe

In Figure 2.6, the point pressure is found by dividing the

rectangular surcharge area (ABCD) into four sub-area rectangles (a, b, c and
d), which have a common corner, E, in the surcharge area, and over the pipe.
The surcharge load is the sum of the four sub-area loads at the subsurface
point. Each sub-area load is calculated by multiplying the surcharge pressure
by an influence coefficient, Ic, as given in Table 2.1 for various M/H and N/H
values.
 19

Table 2.1 Influence coefficient, Ic for Distributed loads over pipe

2.3.2 Rigorous Analytical Methods

The analytical methods available to predict the theoretical

deformations of pipes are presented in this section. Some of the existing
theories which include the modified Iowa formula of Watkins, the modified
Iowa formula of Greenwood and Lang (1990) and the elastic solutions by
Hoeg (1968), used to predict the deformations are detailed in this section. The
prism load is used as the vertical applied load in the three methods.

Spangler (1941) first established relationships to define the

capability of a flexible pipe to resist ring deflection when not buried in the
soil.

The following are the relations:

Y = ( 0.149Fr3)/EI (2.9)

X = (0.136Fr3)/EI (2.10)

X = 0. 913 Y (2.11)
 20

where F - applied load

EI - stiffness factor
r - mid-wall radius
Y - measured change of the vertical inside diameter
X - horizontal deflection

Spangler incorporated the effects of the surrounding soi1 on the

pipe's deflection. In order to do that, Spangler used the Marston (1930) theory
for the determination of the load applied on the pipe. It is assumed that the
vertical load and reaction load are uniformly distributed. Further assumed that
the horizontal pressure is distributed parabolically over the middle 100, and
that the maximum unit pressure on the side fills is equal to the modulus of
passive resistance of the backfill material multiplied by one-half of the
horizontal deflection of the pipe. Based on the forces shown in Figure 2.7,
Spangler recommended following equation to determine horizontal deflection
of a buried pipe.

X = DL (KWcr3) /(EI+0.061er4) (2.12)

Where X - horizontal deflection

DL - deflection lag factor
K - bedding constant
Wc - Marston's load per unit length of pipe
r - mean radius of the pipe
E - modulus of elasticity of the pipe material
I - moment of inertia of the pipe wall per unit Length of pipe
e - empirical Modulus of Passive resistance of the side fill
further modified by Watkins

The original Iowa formula includes three empirical constants: K,

DL and e. The bedding constant ( K) obtained from Table 2.2 is related to the
 21

angle of the vertical support (bedding angle) of the pipe () or the uniform
soi1 reaction on the pipe (P) due to the overburden pressure (W). The
deflection lag factor, DL, considers the consolidation of the side fills with
time. The magnitude of the deflection lag factor for a conservative design
practice should be 1.5, as recommended by Spangler (1941). The experience
of Spangler has shown that deflections could increase over 40 Years (Uni-
Bell 1993).

Figure 2.7 Basis of Spanglers derivation of the Iowa formula (Uni-Bell

1993)

Table 2.2 Value of bedding constant K (Uni-Bell 1993)

Bedding angle () K
0 0.110
30 0.108
45 0.105
60 0.102
90 0.096
120 0.090
180 0.083
 22

In 1955, Watkins investigated the value of the modulus of passive

resistance and found that it could not be the property of the soil. As a result,
Watkins defined a new parameter, the modulus of soil reaction, E' = er. From
the work of Watkins, the modified Iowa formula for horizontal deflection is
as follows:

X = DL (KWcr3) / (EI+0.061E'r3) (2.13)

Iowa formula does not take into account the non-elliptical

deformation and the initial deformation due to construction stages. The
vertical elongation of a flexible pipe is significant and the earth pressure at the
completion of the backfill may not cause the pipe to extend horizontally. The
non-elliptical deformation is a result of the non-uniform earth pressure around
the pipe.

Howard (1977) reported typical values of modulus of soil reaction,

which varied with soil type and composition as well as the degree of soil
compaction. The moduli were deduced from laboratory tests by the US
Bureau of reclamation on range of pipe diameters and materials, and were
complemented by field test data. For clean sands, (less than 12% fines), the
reported E' values ranged between 6.9 and 20.7 Mpa with level of compaction
varying from Slight to high. The accuracy of the theoretical deflection
using values of Howard is + 2%.

Valsangkar and Britto (1978) tested the applicability of ring

compression theory for flexible pipes buried in trenches, largely through
centrifuge tests. If ring theory is applicable then membrane compression
stresses should dominate and flexural stresses should be insignificant. The
study concluded that for pipes in narrow trenches, where the side cover is less
than or equal to one diameter, the use of simple ring theory could not be
 23

justified. Therefore the Iowa equation should not be applied for ratios of
trench width to pipe diameter (B/D) of 2 or less.

Molin (1981) found that the vertical soil pressure, W, above a pipe
in an infinitely wide trench (e.g. under embankment fill) increased with the
stiffness of the pipe and so proposed that the average pressure at crown level
could be expressed by:

W = Cqo (2.14)

where q o = pressure at crown level without a pipe

C = load factor (minimum value of 1) and is given by ,
C = 36Sr (20Sr + 1)/(12Sr + 1)(36Sr +1) (2.15)

Sr = stiffness ratio = 8S/E'

S = stiffness of pipe = EI/D3, and
E' = horizontal modulus of soil reaction as defined by the Iowa
equation (Eq. 2.13)

The equation for C is a design approximation of the theoretical

cases of full slip between the pipe and the soil and no slip. Full slip gives rise
to maximum C values or the greatest pressures above pipes (Crabb and Carter
1985).

The German pipe design method allows calculation of pipe loads

for all types of pipe installations and incorporates the effects to pipe stiffness
and the variation of soil moduli in the vicinity of the pipe. The method is
semi-empirical although the basis of the method is similar to Marston theory.

Jeyapalan and Hamida (1988) provided an overview of the German

approach and showed that the Marston loads are always greater. Assuming
that the German approach leads to the correct loads, Jeypalan and Hamida
 24

concluded that even for relatively stiff, vitrified clay pipes Marston theory is
particularly conservative for small pipes backfilled with well compacted
granular material. The general expression for the load on a pipe is

WGDM = Cd LBD (2.16)

where WGDM = load at the pipe crown calculated by the German design
method.
L = load redistribution coefficient.
Cd = Marston load coefficient

Coefficient L, depends on soil moduli in the vicinity of the pipe, the

ratio of stiffness of the pipe to the side fill and the geometry of the buried pipe
installation.

Greenwood and Lang (1990) found out that the non-uniformity of

earth pressure is a function of soil type, degree of compaction, depth of
embedment and pipe stiffness. Greenwood and Lang also presented a
modified formula, which is more complete than the original formula of
Watkins. The modified formula includes the work of Leonhardt (1972-79)
who developed a factor to consider the soil resistance of the native soil.

The factor, which is applied to the soil resistance parameter in the

modified Iowa formula of Watkins, is a function of the trench width to pipe
diameter ratio and the embedment (backfill soil around the pipe) modulus to
native soi1 modulus. A pipe-soil interaction coefficient, which is an empiral
factor, is added to the soil resistance term to reflect the behaviour of flexible
pipes in the field. The modified formula of Greenwood and Lang (1990) is as
follows.

X = [(KH))/ (EI/r3 + 0.061CIE')] - vo (2.17)

 25

X - horizontal deformation
K - bedding factor (From Table 2.3)
- unit weight of the backfill
H - height of the backfill above the pipe
E - modulus of elasticity of pipe material
I - moment of inertia of the pipe wall per unit length of pipe
r - mean radius of the pipe
E' - Watkins modulus of soi1 reaction
vo - elongation due to compaction of the side fills
D - Pipe diameter
CI - Pipe-soi1 interaction coefficient defined by Greenwood and
Lang (1990) = a (EI/1250*D3)b
a, b - parameters provided in Table 2.4
- Leonhardt relationship

= (1.662 + (0.639( B/D -1))/(B/D-1)+[1.662-0.361(B/D)-1]E2/E3

E2 - Soil modulus of the embedment material

E3 - Soi1 modulus of the native soi1
B - Trench width

Table 2.3 Bedding factor values (Greenwood and Lang 1990)

Range of Backfill standard proctor density (%)

Soil group
fines % >95 85-95 70-84 <70
Clean gravel <5 0.083 0.083 0.083 0.083
5-12 0.096 0.096 0.083 0.083
Dirty gravel 12-50 0.103 0.103 0.096 0.083
Clean sand <5 0.103 0.103 0.096 0.083
5-12 0.083
Dirty sand 12-50 0.103 0.103 0.096 0.083
Inorganic clay >50 0.103 0.103 0.096 0.083
and silt
 26

Table 2.4 Values of parameters a and b for the pipe-soil interaction

coefficient (Greenwood and Lang 1990)

Backfill standard
a b
proctor density (%)
>95 1.24 0.180
85-95 0.938 0.245
70-84 0.643 0.353
<70 0.456 0.436

Watkins theory starts from the Iowa formula written in

dimensionless ratios:

Y/D = (PRs) /(EsARs+B) (2.18)

Y - measured change in vertical diameter

D - outside diameter
P - vertical pressure at the top of the pipe
Rs - stifness ratio : (12(EsD3))/(Et3)
Es - slope of stress-strain curve for soil in one-dimensional
consolidation test
- vertical soil strain
A,B - empirical constants including DIand K in the Iowa formula
t - thickness of the pipe

The equation can be rewritten as

Y/D = Rs /(ARs+B) (2.19)

The empirical constant B also includes the value of Es in the above

equation. However as discussed it is not necessary to calculate the values of A
and B.
 27

The ring deflection factor((Y / (D)) is a function of stiffness ratio.

Watkins observed that usually the deflection factor approaches the unity
because the stiffness ratio is usually greater than 300. As a result, the ring
deflection becomes as much as the side fills settlement. It is then possible to
evaluate the pipe's deflection from the vertical soil strain in the fill.

Burns and Richards (1964) published a theoretical solution for an

elastic pipe placed in an infinite elastic medium and subjected to both vertical
and horizontal loads. Figure 2.8 illustrates the elastic pipe-soil problem
considered by Burns and Richards. The solution was based on the condition
of full-bond or free-slip at the soil/pipe interface. In both the conditions, it
was assumed that no gap takes place at the pipe-soil interface.

Figure 2.8 Elastic pipe/soil problem considered by Burns and Richards

(1964)

Hoeg (1968) proposed an elastic solution to analyse the magnitude

and distribution of static stresses around horizontal cylinders. The study was
limited to plane strain condition. The soil was assumed to behave like a
linearly elastic, isotropic and homogenous material. The pipe is also assumed
 28

to be elastic material. In addition, his solution considers two extreme interface

conditions namely full bonding and free slip conditions. The mathematical
solution is expressed in terms of two ratios. The compressibility ratio (C),
which is the compressibility of the structural cylinder relative to
compressibility of the solid soil cylinder, and the flexibility ratio (F), which
relates the flexibility of the structural cylinder to the compressibility of the
solid soil cylinder. The equations for compressibility ratio and flexibility ratio
are given as

C = 0.5 (1/(1-))(M/(Ec/1- c2)) (D/t) (2.20)

F = 0.5 ((1-2 ) /(1-))(M/(Ec/1- c2)) (D/t) (2.21)

where D = diameter of the cylinder

t = cylinder thickness
= Poissons ratio for the medium
c = Poissons ratio for the material in cylinder wall.

Chua and Lytton (1989) proposed a visco- elastic solution including

the time dependence of the deflections, stresses and strains in the pipe-soil
system. Their study is based on the linear elastic solution of Hoeg (1968) and
modified by Shmulevish and Galili (1985) to include the effect of bedding.

In the solution of Chua and Lytton (1989), it is assumed that there

is a full adhesion of the soil to the outer wall of the pipe. The solution
indicates that the additional deflection caused by the soil sliding around the
pipe is minimal.

Masada (2000) revised the classical work of Spangler (1941) to

derive a modified formula for estimating the vertical deflection of flexible
pipe. During the derivation process the approach suggested by Spangler was
faithfully followed. No new assumptions or empirical elements were
 29

introduced beyond those originated by Spangler. The formula for vertical

deflection was established and its reliability was verified against field
deflection measurement data of Spangler. A few parametric studies were
conducted using the pipe-soil interaction problem as revealed by the work of
Spangler. It was concluded by pointing a few key issues that need to be
remembered when applying the classical approach pioneered by Spangler.

Carrier (2005) studied the influence of deflection and buckling on

flexible pipe design and concluded that the design of buried flexible pipe is
controlled by deflection rather than buckling. However there are two cases in
which buckling may control the design (1) shallow cover with an internal
vacuum pressure and (2) shallow cover submerged in deep water with
atmospheric internal pressure. In these cases the standard short term
deflection criterion of 3% should be reduced sufficiently in order to ensure
that the deflection controls the design.

2.3 SUMMARY

The methods available for design have sprung from elastic thin ring
theory. It is also seen that the Iowa equation is inadequate for calculation of
pipe deflections. The work of Greenwood and Lang (1990) has presented
more rational approaches. The approaches have tackled industry concerns for
flexible pipe over observed non elliptical deformations and the threat of pipe
buckling.

In general the design of pipes subjected to external loading suffers

from a number of uncertainties as follows:

1. The amount of load that reaches the pipe due to trafficking of a

backfilled surface, paved or un- paved.
 30

2. The variations of density that can occur in the various zones

due to pipe installation.

3. The contribution of the natural soil in a trench installation to

the stiffness of the surround soil.

4. Deformations of the pipe to loading due to the backfilling

process.

5. The applicability of pipe stiffness to pipe installations.

6. The bond that exists between the pipe and the surrounding soil.

7. Acceptable deflection criteria for pipes.

2.4 EXPERIMENTAL STUDIES ON BURIED PIPES

Literature reviewed in the earlier section focused on theoretical

analyses and classical theories developed for the design of buried flexible
pipes. Such analyses need experimental validation to know the accuracy and
the validity of the methods adopted. Hence some researchers have worked in
this direction and pointed out the limitations of certain analytical methods.
Some of the experimental studies on buried pipelines reported in the literature
are reviewed in this section.

Howard (1972) investigated the soil structure interaction of flexible

pipes (made of steel, FRP, RPM, PE and PVC) of diameters 450mm, 600mm
and 750mm buried in a large container using clay as material for bedding and
backfill and subjecting them to surcharge loads. Tests were conducted in
backfills compacted at 90% and 100% Proctor densities. Measurements of
soil pressures on the pipe and on the container walls changing pipe
dimensions, soil movement around the pipe and the strain on the inner surface
of the pipe were made during the one day loading sequence and reported. It
was found that the elliptical pipe deformation created high compressive
 31

strains on the inner surface of the pipe at the horizontal diameter (90 and
270) of the pipe. Rectangular pipe deformations created high compressive
strains at four locations generally at 45, 135, 225, and 315. These points of
high strains became the sites for the formation of plastic hinges. Strain gauge
readings on the inner circumference surface of the pipe provided an early
indication of how the pipe would deform elliptically and rectangularly. The
pipe in the 90% backfill failed at vertical deflections between 16 and 22% of
the surface pressure. The pressures at the top and bottom decreased after the
pipe had reached 6-7% of the surface pressure. Since the soil apparently
arched over the flattening pipe and the soil was no longer in firm contact with
the pressure cell. These tests were useful in evaluating the limiting factors of
pipe flexibility and soil densities for pipe design.

Abel et al (1973) used the experimental and numerical techniques

to analyse soil structure interaction of two configurations of elliptical buried
pipes: (i) a shallow pipe where the soil surface is subjected to a concentrated
load simulating early stage of construction and (ii) a deep pipe with a
distributed loading representing the completed installation under a dead
weight. The results form the experimental and numerical analyses were
reduced to field plots of isochromatics (equal values of 1- 2), isopachis
(equal values of 1+ 2), principal stress isobars and isostatic (principal stress
trajectories) for the soil. In addition the normal, shearing and hoop stresses at
the soil-pipe interface were plotted for compressible bedding, thrust beam and
bonded pipe conditions. Some of the significant factors that have not been
accounted for in this analysis are the incremental construction procedure, the
inelastic nature of the soil, the inability of the soils to sustain appreciable
tension, non-linearity associated with possible large deflections and the
potential buckling of the conduit wall. Nevertheless the following
conclusions were drawn.
 32

1. The development of soil arching plays a significant role in an

efficient elliptical soil-structure system as it does for a flexible
conduit system.

2. Slippage between the pipe and soil is also quite beneficial to

the arching action wherein the soil remains the basic load
carrying structure.

3. Compressible under bedding has neither favorable nor

deleterious effects on the elastic stress field.

Howard and Selander (1974) conducted special laboratory tests on

buried sections of steel, reinforced plastic mortar (RPM), fibre reinforced
plastic (FRP), polyethylene (PE) and PVC pipes to understand the
performance of these pipes under external loads. The test pipes were buried in
a large steel container and surcharge loads were applied by a universal testing
machine. Measurements of the changing dimensions of the pipe, soil pressure
on the soil container walls, the soil movement around the pipes and the strain
on the inner circumference of the pipe were made. Pipe deflections were
measured on one end of the pipe with inside micrometers and on the other end
with a revolving dial gauge. A circumferential ring of SR-4 type strain gauge
was placed at one third point of the pipe and measured the inner
circumferential strains. The pipes were tested in lean clay compacted to 90%
and 100% Proctor density. Few tests were conducted in sand backfill with
70% relative density. From the results of testing it was concluded that
reinforce plastic mortar (RPM) pipe and fibre reinforced plastic pipe (FRP)
deflect two to three times more than the steel pipe of the same stiffness and
special care has to be taken in evaluating the stiffness of the polyethylene
(PE) pipe.
 33

Kataria and Kameswara Rao (1982) conducted experimental studies

on shallow buried flexible pipes subjected to surface pressures applied over
finite areas. Aluminium and PVC circular pipes were buried in loose and
dense conditions of sand bed and their response was studied for soil cover to
diameter of pipe ratios (D/d) of 1, 1.5 and 2. The selection of relatively
important dimensionless parameters was done by dimensional analysis. The
twelve relevant parameters for the study were identified as : 1, the hoop strain
developed in buried pipe when subjected to uniform surface pressure p ; the
unit weight of soil ; D, the soil cover on the pipe ; ID, the relative density and
the angle of internal friction of soil; p and s the possions ratios of pipe
and soil respectively ; Ep and Es the elastic moduli of pipe and soil
respectively ; d, the diameter of pipe ; t , the thickness of pipe ; and c , the
crown deflection of pipe. The variation of (1 ) with respect to (p/d) for
different values of (D/d) has been presented in Figure 2.9 (a).

Figure 2.9(a) The variation of (1) with respect to (p/d) for 90 mm PVC
pipe in Sand (after Kataria and Kameswara Rao 1982)
 34

The variation of 1 with (p/d) was found to decrease as D/d value

increases from 1 to 2. The variation of (c/d) with (p/d) was studied for
different D/d ratios and the elastic moduli Ep/Es. A typical graph has been
shown in Figure 2.9(b).

Figure 2.9(b) Variation of (c/d) with (p/d) for 90 mm PVC pipe in

Sand (after Kataria and Kameswara Rao 1982)

The variation of (c/d) with (p/d) was found to be nonlinear. Based

on the experimental studies the following conclusions were drawn.

1. The hoop strain decreases as D/d ratio increases.

2. The strain values were more for loose sand bed than for dense
sand bed.

3. The crown deflection was directly proportional to the hoop

strain in the pipe for both (loose and dense) materials. The
effect of pipe diameter and the depth of burial on surface
deflection are negligible for both loose and dense sand beds.

4. The surface deflection was found to be more for loose sand bed
than that for dense sand bed.
 35

Jeyapalan and Boldon (1986) presented a detailed account of the

existing design procedures for rigid and flexible pipelines and standards. In
addition the warranties provided by various manufacturers were examined
with the backfill requirements. The following clear guidelines for proper pipe
selection were presented for the benefit of city and municipal engineers and
other design engineers.

1. Low stiffness pipes are more susceptible to failures through

excessive pipe deflection, pipe elongation, pipe buckling,
squaring, and excessive wall strains than high stiffness pipes.
Hence low stiffness pipes should be examined with great care
prior to use, to ensure suitability.

2. Pipe manufacturers who do not provide clear specifications for

their products and cannot guarantee that the product is suitable
for the intended use should be avoided.

3. Rigid pipe require high quality embedment material in

comparison to flexible pipes. The cost difference due to this
must be offset by pipe price, lower installation costs or project
design life differences to make the product cost competitive.

Jeyapalan et al (1987) made simple modifications based on the

results from pipe-soil interaction finite element analyses of very flexible pipes
used to predict the vertical deflection and pipe wall strain. The finite element
results were also compared with the field data and the following equation is
proposed for predicting pipe deflection (y).

y = Dc . ys (2.22)

in which ys is obtained using Spanglers formula and the deflection

correction factor, Dc, is given by
 36

Dc = ((800+120H)) 220 (EI/R3))/ E' (2.23)

where H is depth of embedment ( ft), E is modulus of the pipe (psi), I moment

of inertia (in 4/in), R is radius of the pipe(in), and E' is modulus of soil reaction
(psi). The correlation coefficient for the data fit with this equation was given
as 0.93.

The equation for predicting strain is as follows:

= Sc. w (2.24)

in the above equation, the strain w, is calculated using Watkins strain formula
and the correction factor, Sc, is given by

Sc = (6 - (2EI/R3)) (2.25)

The correlation coefficient for the data fit with this equation was
given as 0.87.

From the research, the following conclusions were made.

1. Field performance of very flexible plastic pipes indicated that

the strains and deflections are higher than those computed by
manufacturers and designers using Watkins and Spanglers
equations.

2. The use of Spanglers equation will yield unrealistic

predictions of deflections for very flexible pipes embedded in
soft backfills.

3. The new equations proposed involved prediction of deflection

and strain correction factors. These factors are easy to compute
knowing the properties of the backfill and the pipe and
 37

recommended to estimate the field performance of pipes more

reliably.

Moore (1989) reviewed various theoretical models for predicting

the buckling strength of buried flexible pipes and examined the experimental
data to identify more suitable method for design of buried flexible pipes.
Among the methods reviewed the linear multiwave buckling solution based
on elastic continuum representation of the ground considered to be a
promising design method. An examination of both theory and experiment
confirms that maximum rather than average hoop thrust should be used to
characterize the loads in the buried structures. Various scalar correction
factors are recommended for use in calibrating the buckling theory so that it
can be used in the design of buried flexible tubes.

Neelam and Vipulanandan (1998) investigated the behavior of

flexible PVC pipes with sand and controlled low strength materials (CLSM)
as backfill materials in the soil box. A series of tests were conducted with
pipes under standard dimension ratio (SDR) of 55 to determine the load-
deflection relationship under incremental backfill loads. A special deflection
measuring sensor and pressure sensor were fabricated and used in the soil box
tests. Pipe deflections were compared against the modified Iowa formula and
appropriate modification was recommended for CLSM. It was found that the
horizontal deflection and vertical deflection were less compared to the
modified Iowa formula. A factor was introduced in the modified Iowa
formula to better represent the test data. The values are 0.1 and 0.5 for vertical
and horizontal deflections respectively.

Alferink and Wavin (1999) focused on the performance of gravity

pipes. Extensive field trials have been performed in a project to obtain well
documented information about the behaviour of buried thermoplastic pipes
and to obtain a workable design method. Tests were carried out using
 38

different pipe materials (PVC, PE, STEEL) having different pipe stiffness
buried at depths varying between 1.15 and 3m using clay as well as sand as
side fill and by installing them in a compacted as well as in a dumped fill.
Pipe deflections and strains were measured when loading the pipes with soil,
traffic and internal hydrostatic pressure. European design experts were put to
work with the soil and pipe parameters and asked to carry out the design using
their own or their national design method. The results showed that
consistency between installation and design practice is one of the major
factors to be considered. Based on the results as found in the field studies a
design approach which safely predicts the behaviour of buried pipes was
recommended.

Reddy and Ataoglu (2000) conducted tests on HDPE pipes to

establish their long term behaviour. Tests were conducted in soil chambers for
three levels of service loading and failure mechanism for the prediction of
service life. The long term behaviour was accelerated with super ambient
temperatures of 40 and 50c to provide the data for design life prediction and
the ambient temperature of 20c based on 7.5% vertical deflection as the
failure criterion. Strains and deflections of pipes were monitored for 10,000
hours or upto failure in each test. From the data stresses and bending moments
were estimated. The maximum effective stress was 2.614 Mpa, which is less
than yield stress value of 20.68 Mpa, hence change in diameter is the
governing factor for the design of HDPE pipes. A 7.5% vertical change in
diameter is the failure criterion observed at approximately 3200 hours for the
specimens tested at 50c and subjected to maximum service loading.

Sargand and Masada (2000) presented the field test results of a

1.07m diameter honey comb design, HDPE pipe, buried under 15.85 m high
fill at a highway construction site in Ohio. The pipe was instrumented with six
biaxial strain gages to monitor strains during initial backfilling and earth
 39

pressure cells for measuring load for about one year. A portable linear
variable displacement transducer was used to detect changes in vertical and
horizontal diameters at mid length section. The pipe performance was
monitored for 386 days. The vertical and horizontal deflection of the test pipe
stabilized at -10% and +3% respectively. The pipe exhibited localized short
wave deformation and inner wall tearing at springline due to combined action
of bending and ring compression. Elastic solutions of Burns and Richards
(1964) and a finite element computer code , CANDE 89 were applied with
long term moduli specified for the pipe material to evaluate their analytical
results in relation to the measured field pipe performance.

Neelam and Vipulanandan (2003) conducted the study using soil

boxes of size to investigate the parameters affecting the deflection and wall
loading of buried plastic pipes and to evaluate pipe behaviour under various
backfills. The plastic pipes of diameter 203 mm and length 609 mm with SDR
of 35 and 55 were selected for the study. Eighteen strain gages were glued to
each pipe. Eight on the inside mid section, eight on the outside mid section
along the circumferential direction. Two strain gages were placed along the
longitudinal direction. External loading was applied by tightening a set of four
rods on the side of the chamber. Two strain gages were mounted on each rod
to estimate the load magnitude. Deflections were monitored within the mid
section by a special device fabricated for this purpose. Pipe wall strains
recorded indicated that placing and excavation of the backfill have noticeable
effects on the deformations of the pipe as compared to the deflection during
sustained crown loading. Further from the studies the following conclusions
were made.

1. The horizontal deflection and the vertical deflection are less

compared to the modified Iowa formula.

2. A factor was introduced in the modified Iowa formula to

better represent the test data, for E' = 500 psi, vertical deflection
 40

= 0.1 and Horizontal deflection = 0.05 corresponded to the

measured experimental results.

Cho and Vipulanandan (2004) conducted experiments to

understand the interaction between soil backfill and flexible plastic pipes in a
soil box which represented the trench condition. Earth pressure transducers
strain guages and deflectometer were used to determine the stress distribution,
hoop strain distribution and deflection of flexible pipe in the soil box. The
sand was used as the backfill material and had a relative density of 65%.
Surface pressure was applied and the measurements were taken during
backfilling and when surface pressure was applied. Earth pressure, hoop strain
and side wall friction were analysed and discussed. It was found that the
vertical stress distribution obtained from the experimental results agreed with
the finite element analysis program PLAXIS. It was inferred that the pipe
properties, soil box dimension, side wall and soil interface angle played an
important role in the development of stresses in the flexible pipe.

Galli and Prisco (2006) studied the mechanical interaction between

a buried pipe and the surrounding soil. Focus of this study was on evaluating
the stresses along the pipes induced by slope instabilities, fault displacement
and settlements due to liquefaction of sand strata. The interaction was usually
modeled by means of uncoupled elastoplastic rings. In this study several
inclined loading directions were taken into account and a remarkable coupling
effect was observed among the different load components both when
experimental and numerical analyses were performed. In particular an
experimental campaign on a small scale model in plane strain condition was
presented. In order to describe the influence of the main geometrical and
geotechnical parameters, two different densities were tested as well as several
depths to diameter ratios. To simulate numerically this coupling effect, 2D
FEM numerical analyses were performed and the coupling between the
 41

vertical and horizontal directions was reproduced. As far as the longitudinal

direction is concerned a distinct element model was employed. From the
analyses it was shown that in axial direction no remarkable coupling is
evident among the load components since the system was mainly governed by
the soil pipe interface behaviour. Finally few cyclic numerical tests were
performed to capture the mechanical response of the system to seismic
actions. In this case the interpretation of the mechanical system was more
complex because different and alternate failure mechanisms may take place.

Chapman et al (2007) conducted a series of laboratory tests on thin

walled uPVC ( i.e very flexible ) pipes buried in sand. The tests were
conducted in a glass fronted test tank having internal dimensions of 1.5 m
wide, 1.0 m long and 1.5 m deep. uPVC is having external diameters of 250
mm and 168 mm and corresponding standard dimension ratios (SDR) of 83
and 48 respectively were tested, the pipe being positioned up against the glass
with its longitudinal axis perpendicular to the glass face. This allowed direct
observation of the sand pipe interactions. Photographs were taken through the
glass allowing discrete measurement of pipe and soil displacements during
pipe installation and subsequent surface loading. The surface loading was
simulated using a water bag arrangement positioned on the carefully leveled
surface of the sand. The pipe response to the installation methods, cover depth
and pipe stiffness for increasing static surface stress was studied and the
following conclusions were made.

1. The load transferred to the pipe was consistently lesser for the
well placed dense (WPD) sand than the well placed loose
(WPS) sand surrounds due to more effective (positive) arching
induced by stiffer side fill.
 42

2. The results of cover /diameter ratios of 1and 2 were almost

identical indicating that the arching mechanism developed in the
immediate vicinity of the pipe crown and shoulders.

3. The effect of increasing cover depth confirmed the previous

research findings regarding the influence of the ground surface
on the pipe performance. The results clearly demonstrated the
valuable insight by direct observation of the pipe soil interaction
during installation and the subsequent loading of flexible pipes.

Warnakulasuriya (1999) studied the pipe deformations, pipe strains,

normalized and shear stresses at the soil pipe interface of a 2 mm thick, 150
mm diameter glass reinforced plastic pipe in a 2 m long, 1.5 m high soil box
with variable width. The movable wall facility of the soil box permitted tests
in trench width of 350, 450, 600, 750 and 800 mm. The backfill used was
river sand. The instrumentation was provided for direct measurement of the
horizontal and vertical soil stresses at the boundaries of the trench. The soil
box test results were complemented with the numerical test results using a
three dimensional finite difference programme (FLAC 3D) and the following
conclusions were drawn.

1. When the first layer of fill above the cover to diameter ratio of
0.649 is placed on a very flexible pipe, a distinctive soil/pipe
system is formed with the development of a soil arch, for any
trench width.

2. Trench width parameter is shown to have a significant effect on

the behaviour of buried flexible pipes. The magnitude of such
effects on soil load and radial deflection has increasing effect
depending on the trench width. The rate of increase of a soil
load and strain at the crown, with increasing fill height is seen to
 43

decrease with increasing trench width parameter (W/D). Wide

trench fills can minimize the influence of boundary soil on the
pipe behaviour.

3. The trench width controls the degree of soil arching which plays
a major role in the behaviour of buried flexible pipes. As the
trench width decreases for example in the narrowest study
trench (W/D = 350/150 = 2.33), the proportionate sharing of soil
load between the pipe and the side fill is increased due to the
weakening of soil arch resulting from the limited soil fill at the
springline of pipe. Thus a very flexible pipe buried with similar
boundary conditions will be subjected to even higher soil loads
and consequently higher displacement than those buried in a
wider trench. In the case of a narrow trench this zone is in close
proximity to the trench boundary act as a separate pillar for the
soil arch. The effect of soil arching is reduced with increasing
stiffer pipes.

Gallage et al (2008) examined the behaviour of a 2 m long

polyethylene pipe buried and tested in a box filled with reactive clay. The
pipe had inner and outer diameters of 85 mm and 110 mm respectively. The
box measured 1015 mm in length, 720 mm in width and 1900mm in length
and the pipe was accessible from both ends. The clay had an initial water of
13% by using wet condition method and achieved a dry density of 1.15 g/cm 3.
The wetting was continued for a period of 136 days during which moisture
content, suction and pipe and soil displacement were monitored. The
deflection was measured along the pipe inserting a specially designed device
into the pipe and using a surveying level. A simplified numerical analysis was
then conducted on the pipe buried in the soil box using PROKON (structural
analysis software) which led to the comparison of computed results with
measured values leading to the following conclusions.
 44

1. Water content and suction changed simultaneously as wetting

proceed, but the swelling movement is dependant on the
overburden pressure, confinement and also displayed a certain
time lag.

2. Both free swelling displacement and swelling increase with the

increase in soil density.

3. The numerical study has provided information on the pipe

bending mechanisms which proved that the pipe movement was
directly related to the properties of pipe material and the soil.

4. The soil box experiment provided a useful simulation of a

buried pipe bending due to change of soil moisture content in a
laboratory situation while the simplified theoretical analysis was
found to predict reasonably the measured values and may be
applied approximately to simulate field conditions.

From the discussion it is clear that the experimental studies on the

performance of buried pipes have not received wider attention as any other
geotechnical problem from the researches. The studies are limited to static
loading. The deflections and the strains occurring in pipes when subjected to
surface pressures in different types of backfill and at different densities have
been monitored using suitable instrumentation and the results are validated
using a suitable finite element code. Comparison has been made on the
behaviour of pipes with different materials and stiffness and the compatibility
with the existing theories has also been reported. However the method of
reducing the deformations and stresses on the pipes buried at shallow depths
with suitable soil reinforcement techniques and thereby minimizing the depth
of burial in pipelines has not been given consideration. Therefore it is
proposed to study the performance of the buried flexible plastic pipe through
 45

1g model tests in the laboratory and parametric analyses using PLAXIS FE

code.

2.5 NUMERICAL STUDIES ON BURIED PIPES

The finite element analyses have proved to be very useful in the

analyses of buried structures. Many finite element programs are available in
the market and each of them has their own advantage depending on the
problem. Some of these programs are PLAXIS, which is used in this thesis,
PIPE 5 which is a version of SAP (Wilson 1971), modified by Utah state
university researchers for the analysis of the flexible pipe, CANDE (Culvert
Analysis and Design) (Allgood 1976 and Katona 1980), which is mostly used
in the U.S., SPIDA (Heger et al 1985) which is used for the analysis of rigid
concrete pipes and some others such as ABAQUS (1998), ADINA and
SIGMA/W. Some of the numerical studies carried out by the researchers are
reviewed in this section.

Doderer (1970) presented both finite element and photoelastic

results of circular buried pipes, which include stresses and displacements,
pipe-soil medium interface stresses, pipe stresses and displacements as
functions of load type and position, depth and material properties of both pipe
and soil medium. The results were presented graphically in both dimension
and dimensionless forms and have afford quantitative and qualitative
information on soil- pipe medium interaction. This and similar finite element
studies provided examples of the only strategy for elastic analysis of shallow
pipe.

Burns and Richards (1964) developed a flexibility solution based

on deformation compatibility at the interface to find the stress and
displacement fields for a deeply buried thin walled elastic pipe. Their Fourier
series expressions for pipe displacement and stress resultants in terms of pipe
 46

medium interface stresses constitute an advanced closed form solution for this
case. The effect of concentrated surface loads on shallow buried pipe was
approximated by transforming the pipe into an equivalent ring of soil and the
results were compared with finite element solutions. The results showed that
the transformation solution gives a valid picture of the interaction in a general
sense upto D/R =2 and more particularly for stiffer pipes (T/R = 0.2, Ep/Es
>100) which attract a sizeable portion of the load.

Crofts et al (1977) proposed a method for estimating the horizontal

movement of a long shallow buried pipeline due to nearby excavation and
backfilling of a long deep trench parallel to the pipeline. An elastic model of
the problem consisting of a beam embedded in an elastic foundation which is
locally displaced laterally was analysed and an approach was suggested for
estimating the risk of pipe fracture.

Provakar (1983) analysed the effects of excavation on adjacent

buried pipes using a three dimensional finite element model having elastic
material properties. The bending moments and displacements induced only in
pipes running parallel to open, unsupported rectangular trenches were
considered. The results showed that small diameter pipes such as those having
diameters and longitudinal bending stiffness smaller than that of a 6 inch
diameter cast iron pipe, closely follow the movements of the ground and for
practical purposes the moments induced in such pipes may be assumed to be
directly proportional to their flexural rigidities and inversely proportional to
the elastic modulus of the soil. These proportionalities break down for large
diameter pipes which are less flexible and restrain the ground from moving
freely. The ground movements and pipe strains predicted by the analysis have
been found to be of the same order as those measured in several field tests.

Selvadurai and Shinde (1993) examined the problem of flexural

interaction between a long distance buried pipelines embedded in soil medium
 47

that experiences differential frost heave. The modeling takes into

consideration of the interaction at a transition zone between a frozen region
and a frost susceptible region that experiences a time dependent growth of a
frost bulb around the buried pipeline. The heave that accompanies the
development of a frost bulb induces the soil-pipeline interaction process. The
analysis focuses on the development of a computational scheme that
addresses the three dimensional nature of the soil-pipeline interaction
problem. The numerical results presented illustrated the influence of the
heave process and the creep behaviour of the frozen soil on the displacement
and stresses in the buried pipeline.

Zhan and Rajani (1997) conducted finite element analyses to assess

the effects of different trench backfill materials, pipe burial depths and pipe
materials on the amount of traffic load transferred to buried pipe. The main
objectives were i) to compare the measured field responses of vehicular
loading with the results of finite element analyses and ii) to determine the
sensitivity of load transfer to buried pipes with different trench backfill
material, pipe burial depths and pipe materials using finite element modeling.
The PVC pipes were instrumented with electrical resistance strain gauges.
Strain gauges were oriented in both circumferential and longitudinal
directions to measure hoop and axial strains. The trench were backfilled with
native clay, clean sand and unshrinkable fill (or CLSM). Earth pressure cells
were placed in the backfills to monitor the earth pressures during the cold
season. The surface load was represented by a uniformly distributed load of
100 kPa over the width of trench rather tire load of 140 kN with a dynamic
factor of 1.5. The magnitude of the uniform surface load of 100 kPa was
established by consideration of equivalent stress induced by either loads at
depth of 2 m which is typical of water main burial depth in Canada. The
analyses showed that the use of trench backfills such as CLSM instead of
traditional materials such as sand and clay resulted in significantly reduced
 48

stresses in PVC pipe under traffic loading. This finding was in agreement
with respect to truck load tests carried out in the city of Edmonton where
strains were monitored on buried PVC water mains. A pipe buried under
CLSM was expected to have higher uniform hoop stress and less bending
stress than in a similar pipe buried under sand and clay backfills and this was
confirmed through the finite element results. The protection of buried pipe
under or in CLSM backfill from traffic loading became more significant with
the decrease in pipe burial depth and stiffness. It was observed that high
elastic modulus and strength of CLSM as well as uniform load transfer along
the longitudinal axis of the pipe were the reasons behind the difference in load
transfer between the traditional backfills and CLSM.

Suleiman et al (2002) modeled the soil pipe system using ANSYS,

a general finite element program. Small and large deflection theories of
ANSYS were used in the analyses of several case studies and the results were
compared to those of CANDE, one of the most commonly used programs for
buried pipe analysis. Also a code was written to run within ANSYS to include
the following soil parameters: the hyperbolic tangent modulus, the power bulk
modulus, and the hyperbolic bulk modulus. Results obtained using ANSYS
with the modified soil models were in good agreement, less than 10% except
in one case with CANDE results for 6.1 m of soil cover above the springline
for the case of 610 mm pipe diameter with SM and ML soils. Use of large
deflection theory resulted in an insignificant effect, less than 5% when
compared with ANSYS small deflection theory results for soil heights upto
6.1 m above the spring line which proves that small deflection theory is
adequate for these cases.

Comparing CANDE and ANSYS for 1200 mm diameter

polyethylene pipes with experimental results showed that the small deflection
theories in ANSYS more accurately described the PE pipe behaviour for cases
 49

of 9 m of soil cover or more and the large deflection theories for both PE and
PVC pipes. The difference between the small and large deflection theories for
both pipe materials become significant, say more than 10% at a vertical
deflection percentage of 4%.

Sivakumar et al (2003) presented a critical appraisal of the

mechanical behaviour of buried flexible pipes and proposed a design
methodology for prediction of performance of buried flexible pipes using
finite element analysis program NISA. A design chart for the analysis of a
pipe section was presented for simple loading and boundary conditions. It was
concluded that the behaviour of flexible steel pipes was affected by the
modulus of the backfill. The study also indicated that the assumptions
involved in Spanglers formula are not realistic and deflections predicted
using this formula need to be treated with caution. This observation was
found to be in agreement with the opinion of previous investigators such as
Gumbel (1983) and Davis and Bardet (2000).

Cameron (2005) studied the performance of flexible pipes buried

beneath dry sand backfill in shallow trench conditions and subjected to traffic
loading. To understand the complex soil-structure interaction between pipe
and soil a comprehensive study was undertaken through twenty two buried
pipe tests on spirally wound uPVC pipes of nominal diameters ranging
between 300 and 450 mm. The majority of the tests were conducted in a
laboratory soil box facility with poorly graded dry sand used as the fill
material. Half the buried pipe installations were instrumented with soil
pressure cells, which provided useful information in the analyses of the
performance of the flexible plastic pipes. Loads were applied directly onto the
backfill surface through a rectangular steel plate representing the area of the
backfill surface stressed by traffic loading.
 50

In the two dimensional analysis carried out using FEM code

ADINA the pipe was represented by beam column elements with properties
developed from manufacturers specifications. With the availability of the
three dimensional version the buried pipes were reanalysed to validate the
results obtained from the two dimensional analysis. The 3D FEA was limited
to some extent by choice of elements and processing time. A 3D interface
element was not available so analyses were undertaken with completely rough
side wall conditions and the following conclusions were made.

1. Sufficient cover height is essential in protecting the pipe from

trafficking of the backfill surface.

2. The rigidity of the sidewall boundary for the laboratory soil box
test was called into question after comparing the relative
performance of the pipes tested in the field.

3. Two dimensional FEA could not adequately model the observed

behaviour of the soil pipe systems.

4. Although the 3D analysis improved the predictions of behaviour

when compared with the 2D analysis the assumption of rigid
side wall with rough wall condition could not adequately
simulate the observed load-deflection behaviour of the pipes
tested.

5. The predicted distribution of vertical soil stress by FEA was non

uniform with the bulk of the stress concentrated closer to the
pipe crown.

6. Hoegs theory produced more conservative design estimates

which however extended the useful range of application and
good estimates of the lateral pipe response.
 51

Guo (2005) studied the pipe soil interaction for pipelines subjected
to combined horizontal and vertical (upward) movements in the oblique
direction. An associative hardening elasto- plasticity model was proposed in
the horizontal load space for pipes buried in clay. In particular the model
includes a failure surface, a family of yield or loading loci, and a plastic flow
rule in which the plastic potential is assumed to be similar to the loading
surface. The model was verified using the results from continuum finite
element analysis within the framework of continuum elasto-plasticity. Good
agreement was obtained between the model predictions and the results of
finite element analysis.

Guo and Stolle (2005) investigated the pipe-soil interaction for

pipes buried in cohesionless soil when subjected to lateral ground movement.
A review and reappraisal of the literature was conducted to identify some
inconsistencies and shortcomings of previous studies. The effects of model
scale, stress level, burial depth ratio, and soil properties are systematically
addressed through finite element analyses. The study concludes that the
effects of pipe size and burial depth must be taken into account to properly
estimate the maximum pipe-soil interaction forces induced by lateral ground
movement. A unique equation considering scale and effect of burial depth is
proposed (Eq. 2.26) to determine the maximum pipe-soil interaction forces at
various burial depth ratios. The equation is given as

Nh = Nhof(D/Do) (2.26)

where Nh, Nho = the maximum dimensional force, or horizontal bearing

capacity factor and D, Do = Pipe diameter. It was concluded that the proposed
equation which matches the experimental data in the literature could be
incorporated into the analytical models recommended in current pipeline
design guidelines to take into account the scale and burial depth effects.
 52

Sargand et al (2005) presented the field performance of 1524 mm

diameter, corrugated high density polyethylene (HDPE) pipes, installed at the
Ohio- University ORITE deep burial site. The pipes were subjected to 6.1 m
to 12.2 m high embankment fill for about 2 years. Structural performance of
each pipe was monitored with various sensors during initial backfilling,
during embankment construction and after embankment construction. The
data obtained in the field indicated that these pipes were performing
satisfactorily showing little signs of structural distress. Examination of the
field data revealed some valuable insights into the large diameter flexible pipe
performance under deep soil cover. In addition selected analytical models
were evaluated using the field data. An improved approach based on the
modified Iowa formula was generally effective in estimating the horizontal
deflection. Elastic solutions showed a limited ability in describing the field
performance of large diameter HDPE pipes. The nature of the finite element
computer program CANDE 89 was recognized but not fully demonstrated in
numerical solutions of the field installation conditions because of a few short
comings in CANDE 89 computer analysis.

Vorster et al (2005) presented a method of estimating the maximum

bending moment for continuous (i.e. rigidly joined) pipelines affected by
tunnel induced ground movements. The estimation was made based on the
knowledge of tunnel and pipeline geometries, the stiffness of the soil and
pipelines, tunnel induced ground deformation at the pipe line level. The
method takes account of the soil nonlinearity by an equivalent linear approach
in which the stiffness of the soil is evaluated based on an average deviatoric
strain developed along the pipeline. The approach is conservative and
promises that the bending moment is not underestimated. The validity of the
method as an upper bound approximation was evaluated against centrifuge
test results.
 53

Kawabata et al (2006) presented the behaviour of an instrumented

flexible pipe buried under a 47.1 m deep fill. For filling above 20 m the
vertical stress above the pipe exhibited a concave distribution corresponding
to 90% and 110% of the average vertical pressure at the mid section and
edges respectively. The results suggested that a triangular lateral pressure
distribution can lead to conservative and uneconomical results for high fills
while Spanglers analysis is unconservative. Based on the measured results, a
revised vertical and lateral earth pressure diagram as shown in Figure 2.10
was proposed for the design of flexible pipe under high fills (> 20 m). The
study proposed a closed form method for estimating the moments and
displacements of the pipe subjected to high fill earth pressures which agreed
well with the measured pipe displacements and strains.

Figure 2.10 Earth pressure diagram: a) Spanglers analysis, b) Simplified

analysis and c) proposed analysis (after Kawabata et al
2006)

Sivakumar et al (2006) reviewed the behaviour of buried flexible

pipes considerably influenced by uncertainties in external load; pipe and soil
properties, soil stiffness and load were considered as random variables.
Reliability assessment of deflection of a typical pipe section with typical soil
properties was conducted using the point estimate method with respect to two
reliability measures, factor of safety and safety margin. Within the framework
 54

of analysis, results showed that for flexible pipes, the conventional tolerable
deflection limit of 5% of the diameter of the pipe with a factor of safety of 4
is conservative.

Kang et al (2007) identified the variables that significantly affect

earth loads, as well as the effects of bedding and side fill treatments. An
optimum geometry for imperfect trench installations with regard to a soft
material zone is presented to maximize the earth load reduction effects. The
optimization process was based on parametric studies of the geometry and
location of the soft zone bedding and side fill treatments to reduce earth
pressures. Equations of earth load reduction rates were formulated
incorporating the optimum geometry for the soft zone bedding and side fill
requirement and the following conclusions were derived from the study.

1. The maximum earth pressure on a pipe installed with the

AASHTO types 1, 2 or 3 standard embankment installations
occurs at about 25 from the invert and emphasizes the
importance of the material and the construction quality of the
haunch area in the installation of concrete pipes.

2. Vertical arching factors for buried concrete pipes are affected by

backfill height as well as installation practices for bedding and
side fills.

3. The modulus of elasticity of the light weight materials in the

soft zone is the single most important factor affecting the earth
load reduction rate in the imperfect trench installation.

Trickey and Moore (2007) investigated the three dimensional

response of buried pipes under circular surface loading using ANSYS FE
code. Previous work by Poulos and Davies in 1974 has been reexamined
considering the longitudinal behaviour under surface loading. Analyses were
 55

performed for pipes of varying stiffness and embedment depths and the
following conclusions were presented.

1. The burial depth located close to the ground surface has little
impact on the peak deflection. However flexible pipe
deflections decrease significantly as the embedment depth
increases.

2. Peak moment increases with pipe stiffness and decreases as the

pipes become more remote from the ground surface.

3. The comparison of results with those of Poulos and Davies

(1974) indicates that Mindlin solution calculation are some what
conservative relative to the finite element solutions for deeply
buried pipes, but unconservative at shallow burial.

2.6 SUMMARY

The Marstons load theory is still used in the current practice. The
theory accounts for the shearing forces between the native soil and the
backfill in a trench installation and between the soil directly above the pipe
and the side fill in an embankment installation. It is assumed that the shear
planes are vertical and that the ratio of lateral pressure to vertical pressure is
expressed by the Rankines ratio, Ka. These assumptions are subjected to
discussion. For instance, there is no proof that the horizontal stresses acting
on the vertical planes are the active Rankine pressures, and that the shear
planes are always vertical.

Spangler derived a formula which is known as Iowa formula in

order to predict the deflection of flexible pipes. He assumed that a flexible
pipe deflects elliptically and that the Marstons load theory applies. Therefore,
he made additional assumptions. If the ratio of soil stiffness to pipe stiffness is
 56

high, then the deformation of the pipe may be rectangular instead of elliptical.
Moreover, Watkins determined that the constant of proportionality of
Spangler could not be a true property of the soil. As a result, Watkins defined
the modulus of soil reaction, E'. The modulus of soil is not a constant and
varies with the depth of soil over the pipe, the size of the pipe, the stiffness of
the pipe relative to the soil, the trench width, the soil type, the compaction
density, etc.

Greenwood and Lang (1990) presented a modified Iowa formula

which was more complete than the original formula. A pipe soil interaction
coefficient, which is an empirical factor, is added to the soil resistance term to
reflect the behaviour of flexible pipes in the field.

Hoegs (1968) method assumes that the soil behaves like an elastic
material, which is isotropic and homogenous. This is an unrealistic
assumption because the soil behaviour is always nonlinear. In addition, the
Hoegs method does not take into account the installation practice.

Limited experimental studies have been carried out in the area of

buried pipelines considering the parameters such as crown deflection of pipe,
surface pressure, modulus of the pipe and soil, depth of embedment, Poissons
ratio of the pipe and the relative density of the soil. However the use of
geogrid reinforcement to reduce the deformations and stresses on buried pipes
has not received the attention of the research workers.

The finite element seems to be a realistic method in predicting the

deformations of buried pipes because it avoids several assumptions that have
been made in the development of analytical methods. It also allows the
consideration of construction stages in complex trench configurations and to
simulate complex loading patterns. However, the accuracy of the results
depends on the soil models available.