Вы находитесь на странице: 1из 14

997

Uncoupled axial, flexural, and circumferential


pipe–soil interaction analyses of partially
supported jointed water mains
Balvant Rajani and Solomon Tesfamariam

Abstract: Pipelines used in the distribution of potable water are a vital part of everyday life. The pipelines buried in
soil–backfill are exposed to different deleterious reactions; as a result, the design factor of safety may be significantly
degraded and, consequently, pipelines may fail prematurely. Proactive pipeline management, which entails optimal
maintenance, repair, or replacement strategies, helps increase the longevity of pipelines. The effect of different deterio-
ration mechanisms and operating conditions needs to be understood to develop good proactive management practices.
In this paper, a Winkler-type analytical model is developed to quantify the contributions of different stress drivers, e.g.,
pipe material type and size, bedding conditions, and temperature. Sensitivity analyses indicate that the extent of the un-
supported length developed as a result of scour has a significant influence on the flexural pipe–soil response. As well,
plastic pipes tolerate less loss of support than metallic pipes.
Key words: jointed water mains, Winkler model, pipe–soil interaction, elastoplastic soil.

Résumé : Les conduites utilisées pour la distribution de l’eau potable constituent une partie vitale de la vie de tous les
jours. Les conduites enfouies dans un remblai de sol sont exposées à différentes réactions nuisibles, et il en résulte une
dégradation du coefficient de sécurité utilisé pour le calcul et en conséquence les conduites se brisent prématurément.
La gestion proactive des conduites qui comporte des stratégies optimales de maintenance, de réparation, ou de rempla-
cement aide à accroître la longévité des conduites. L’effet de différents mécanismes de détérioration et conditions
d’opération doit être bien compris de façon à développer de bonnes pratiques de gestion proactive. Dans cet article, on
a développé un modèle analytique de type Winkler pour quantifier les contributions de différentes sources de contrain-
tes, e.g., type de matériau et dimension de la conduite, conditions du coussin, et température. Des analyses de sensibi-
lité indiquent que l’importance de la longueur non supportée qui se développe à la suite de l’érosion a une influence
significative sur la réaction conduite–sol en flexion. Également, les conduites en plastique tolèrent moins de perte
d’appui que les conduites métalliques.
Mots clés : conduites maîtresses d’eau articulées, modèle Winkler, interaction sol–conduite, sol élasto-plastique.

[Traduit par la Rédaction] Rajani and Tesfamariam 1010

Introduction period right after installation, in which breaks occur mainly


as a result of faulty installation or faulty pipes. These breaks
Understanding failure mechanisms of deteriorating infra- emerge gradually and are fixed in a declining frequency.
structure is paramount for the proactive management of in- Once the system is purged of these “early” problems, the
frastructure assets. Exposure of water mains to aggressive pipeline enters phase two, in which the pipe operates rela-
environmental conditions and deleterious reactions can lead tively trouble free, with a low failure frequency resulting
to significant deterioration so as to undermine their ability to from random phenomena such as unusual heavy loads and
reliably deliver safe drinking water. The life cycle of a typi- third-party interference. The third phase, also called the
cal buried pipe is described by the so-called “bathtub” curve
“wear-out phase,” depicts a period of increasing failure fre-
as shown in Fig. 1. It describes the instantaneous failure
quency due to pipe deterioration and ageing. Not every pipe
probability (hazard function), and the bathtub curve often
experiences every phase, and the length of the phases may
distinguishes between three phases in the life of a pipe. The
vary dramatically for various pipes and under various condi-
first phase, also known as the “burn-in” phase, describes the
tions. Alternatively, the various phases in the deterioration
of structural reliability (expressed here as the factor of
safety) that ultimately lead to the failure of the water main
Received 26 August 2003. Accepted 13 April 2004. Published
on the NRC Research Press Web site at http://cgj.nrc.ca on are shown in Fig. 2. Older water mains are usually made of
29 October 2004. pit- or spun-cast iron (CI), and the newer mains are largely
made of ductile iron (DI) or polyvinyl chloride (PVC). In an
B. Rajani1 and S. Tesfamariam. Institute for Research in aggressive environment, corrosion in CI takes the form of
Construction, National Research Council Canada, Ottawa, ON graphitization (Makar and Rajani 2000), and pitting takes
K1A 0R6, Canada.
place in DI pipes. PVC water mains have not been used long
1
Corresponding author (e-mail: Balvant.Rajani@nrc-cnrc.gc.ca). enough to establish a definite deterioration mechanism.

Can. Geotech. J. 41: 997–1010 (2004) doi: 10.1139/T04-048 © 2004 NRC Canada
998 Can. Geotech. J. Vol. 41, 2004

Fig. 1. The “bathtub” curve of the life cycle of a buried pipe. stalled years ago. Rajani et al. (1996), in analyzing pipe
breakage, did not include an assessment of the influence of
unsupported length, which could result as a consequence of
water leaks and possible shrinkage due to moisture changes
in bedding materials. In most cases, shrinkage-susceptible
native material (predominantly clayey and silty soils in most
urban areas in Canada) was used as backfill and as bedding
material for CI pipe installations.
Water mains used in water distribution systems have bell
and spigot connections and are referred to here as jointed
pipes; typical pipe lengths are 6 m (20 ft). Elastomer gaskets
at bell and spigot connections (joints) prevent leaks while
permitting axial movement and slight rotation (3–4°) to
accommodate limited movement of soil bedding. In the
analysis described here, boundary conditions dictated by a
bell–spigot joint are assumed to be ideal, i.e., free to move
longitudinally and permit rotation. In practice, these move-
ments are likely to be restrained as a consequence of ageing
Fig. 2. Decrease in factor of safety with time. of both pipe joint and gasket materials. Longitudinal and ro-
tational movements are probably restrained to a greater ex-
tent in the case of cast iron pipes than for modern pipe
materials. Structural design of water mains usually provides
sufficient (with an accepted margin of safety) resistance
against circumferential (in-plane) stresses imposed by soil
overburden loads, live traffic loads, and internal pressure.
Loads imposed on pipes as a result of temperature changes,
soil pressures induced by frost heave, and loss of support
from bedding are largely unaccounted for in the axial, flex-
ural (longitudinal bending), or circumferential response
analyses. The practice to exclude some of these circum-
stances from routine analyses and design is acceptable as
long as the margin of safety is adequate and pipes do not de-
teriorate with time, leak, and develop locations with loss of
bedding support or differential settlement. Ageing water dis-
tribution systems, however, do indicate that pipes deteriorate
Nonetheless, whatever the pipe materials and the associated
with time and that there is a marked increase in the number
deterioration mechanisms, it is assumed that structural
of water main breaks with dramatic temperature differentials
strength will decrease, say, as a consequence of corrosion
between the water in the pipe (1–2 °C) and the adjacent soil
pitting, graphitization, fracture, creep, or material softening.
(10–12 °C).
It is further assumed that a leak will develop as soon as the
pipe wall is breached. The leak will in turn scour the sur- The intent of this paper is to quantify the impact of the
rounding bedding and undermine the soil support to the unsupported length and soil elastoplasticity on the axial and
pipe. The size of the corrosion pit and the extent to which flexural responses considering pipe–soil interaction using
the bedding support is lost (“unsupported” length) will lower the Winkler model. Further, the axial, flexural, and cir-
the design factor of safety. An arch is likely to form in cumferential stress components are consolidated from the
sandy soils (backfills); if the leak occurs at the pipe invert or analyses described here, and previously reported (Rajani et
crown, analysis is not required because the pipe is not al. 1996; Ugural and Fenster 1987), to provide an overall
loaded. A leak at the crown in clayey soils is likely to un- picture of the response of buried water mains under the in-
load the pipe initially, but as the soil becomes saturated the fluence of earth and live loads, water pressure, temperature
clay is likely to cave in and impose earth loads on the pipe. differential, and pipe–soil interaction. Though the extent of
On the other hand, the proposed model allows for any load loss of support cannot be determined in the field today, sev-
(q) on the unsupported pipe length, i.e., if arching merits eral attempts have been made by Makar (1999) for sewers
consideration, then the load can be reduced. Consequently, using nondestructive techniques.
the model considers what can be referred to as a “worst- The axial and flexural responses of buried jointed pipe is
case” scenario. considered to be uncoupled, and it is assumed that the pipe
Three ingredients required to develop frost heave and deformations are small and always within the elastic range.
hence frost load are availability of water, thermal gradient, For simplicity, it is also assumed that the leak produces
and soil with appropriate particle size and hence pore size scour (undermined bedding) at the centre of jointed pipe of
(Rajani and Zhan 1996). The presence of a water main leak length 2L, probably representing a worst-case scenario. The
contributes to the first of the three ingredients. All three lead soil or bedding in the pipe–soil interaction analysis is con-
to increased stresses on the pipe which were not considered sidered as an elastoplastic Winkler model. Generally, the
in the original structural design when the pipes were in- Winkler model has several shortcomings, e.g., it assumes no
© 2004 NRC Canada
Rajani and Tesfamariam 999

interaction through the soil from location to location and Fig. 3. Typical axial soil resistance versus axial displacement.
no interaction through shear nor volumetric effects, and the
model relies on a definition of soil pressure in terms of ab-
solute displacement of the pipe, not the displacement of the
pipe relative to the soil. Nevertheless, given all uncertainties
in modelling pipe–soil interaction, it is an acceptably simple
model to permit the consideration of axial effects, longitudi-
nal bending and radial effects associated with overburden
pressures, internal pressure, frost loads, and thermal effects.
It is important to note that the circumferential response cor-
responds to that of a rigid pipe but could easily be extended
to that of a flexible pipe. This consideration is appropriate
for the pipe materials (CI and DI) and pipe sizes of interest
here. In the analysis that follows it is assumed that thrust is
positive when it results in tensile stresses in the pipe wall
and negative when it results in compressive stresses. Simi-
larly, a moment (longitudinal bending) is positive when tional angle between the pipe material and the surrounding
there are tensile stresses on the pipe invert and negative backfill.
when there are compressive stresses on the pipe invert. Ten- If the soil is represented as an elastoplastic material
sile stresses in the circumferential direction are treated as (Fig. 3) and the soil deformation exceeds the limiting axial
positive. displacement, u = ux , then the governing equilibrium equa-
Typical pipe–soil systems are considered in the sensitivity tion is
analyses to illustrate the applicability of the proposed mod-
els. Pipe and soil properties and the operating conditions se- ∂σax Fx
[3] − =0 u ≥ ux
lected in these analyses are typical of those likely to be ∂x t
encountered in practice.
where Fx is the ultimate axial resistance and is given by
Axial pipe–soil interaction [4] Fx = ksa ux
Rajani et al. (1996) described the Winkler model for axial Equations [1] and [3] are applicable to a pipe that is fully
pipe–soil interaction of a jointed buried pipe in an elastic supported longitudinally and radially (normal to the cir-
soil (medium). The equilibrium equation is governed by cumferential direction). As discussed earlier, the bedding of
a leaky pipe is likely to be scoured, causing the pipe to lose
∂σax ksa support. This situation necessitates that the problem be
[1] − u=0 u < ux
∂x t solved piece-wise in three regions, i.e., unsupported pipe
length (b), pipe embedded in an elastic soil (d ), and remain-
where σax is the axial stress, u is the axial displacement, ksa is ing pipe length in plastic soil ( f ) (Fig. 4). The soil is repre-
the axial pipe–soil reaction modulus, t is the pipe wall thick- sented as a Winkler elastoplastic material.
ness, and ux is the displacement required to develop ultimate The axial response (solution to eq. [1]) of the supported
axial resistance (Committee on Gas and Liquid Fuel Life- pipe length d is described by
lines 1984). The axial pipe–soil reaction modulus of the soil
can be estimated using the elastic properties as suggested by [5] ud = C1a exp(−γx d ) + C 2a exp(+γx d )
Scott (1981) or empirical relationships for sand and clay as
where ud is the axial displacement of the pipe embedded in
suggested by the Committee on Gas and Liquid Fuel Life-
the elastic soil; C1a and C 2a are axial pipe–soil interaction
lines (1984). These relationships are as follows:
constants; and γ is the reciprocal of the axial characteristic
Gs length, given by
[2a] ksa = elastic soil
4(1 − υs) D / 2 ksa
[6] γ2 =
α su Ep t {[1 + υp Es/ 2β1(1 +
2
υs) Ep ](1 + D / t)}
[2b] ksa = ux = 2.5 − 10.0 mm for clay
ux where υp is Poisson’s ratio of the pipe, β1 is a constant as de-
0.5 (γ s H)(1 + Ko) tan δ fined in Rajani et al. (1996), Es is the elastic modulus of the
ksa = soil, and Ep is the elastic modulus of the pipe.
[2c] ux The axial response of the portion of the pipe supported by
ux = 2.5 − 5.0 mm for sand elastoplastic soil is given by
Fx x 2f
where D is the external diameter of the pipe, Gs is the soil [7] uf = + F1a x f + α p∆T x f + F2a
shear modulus, υs is the soil Poisson’s ratio, α is the adhe- 2tEp
sion coefficient, su is the undrained shear strength of clay, γ s
is the submerged unit weight, H is the burial depth of water where u f is the axial displacement of the pipe embedded in
mains from the surface to the centreline of the pipe, Ko is the elastoplastic soil, Ep is the pipe elastic modulus, and F1a
the coefficient of active resistance at rest, and δ is the fric- and F2a are constants. Temperature change in the soil is cap-
© 2004 NRC Canada
1000 Can. Geotech. J. Vol. 41, 2004

B1a ⎛ b ⎞ ⎛ ⎞
⎜ + 1 ⎟ − Pi ⎜ χ 2 ⎟
Fig. 4. Schematic model for axial and radial pipe–soil interaction.
[11] C 2a =
Ep ⎝ 2 2γχ1 ⎠ Ep ⎝ 2γχ1 ⎟⎠
⎜ ⎟ ⎜

⎛b χ ⎞
+ α ∆T ⎜⎜ + 3 ⎟⎟
⎝ 2 2γχ1 ⎠
−Fz f
[12] F1a =
t
[13] F2a = C1a exp(−γd ) + C 2a exp(+γd )
where χ1, χ 2, and χ 3 are as defined by Rajani et al. (1996) to
account for axial pipe movement, internal water pressure,
and temperature change, respectively; Pi is the internal water
pressure; α p is the linear thermal expansion coefficient of
the pipe material; and Fz is the maximum lateral soil resis-
tance per unit length. The constants χ 4 , χ 5, and χ 6 are as
follows:
[14a] χ 4 = (1 + b γχ1) exp(γd ) + (1 − b γχ1) exp(−γd )
[14b] χ 5 = −2 + exp(γd ) + exp(−γd )
[14c] χ 6 = −2χ 3 + (b γχ1 + χ 3) exp(γd )
− (b γχ1 − χ 3) exp(−γd )
tured by the term α p∆T x f , where α p is the linear thermal ex-
pansion coefficient of the pipe material, ∆T is the maximum The axial stress responses in the three different regions of
temperature difference between the water and the surround- soil behaviour can be expressed in terms of “stress drivers,”
ing soil, and x f is the longitudinal coordinate system for soil i.e., axial pipe movement, internal water pressure, and tem-
in plastic region (Fig. 4). perature change:
Similarly, the unsupported pipe length b is an axially Fx
loaded prismatic element, and its response is described by [15a] σax = (x f − f ) for (b + d ) ≤ x < (b + d + f )
t

ub = B1a x + α p∆T x b + Ba2 ∂ud


[8] [15b] σax = χ1 Ep + χ 2 Pi − χ 3 Epα p∆T
∂x
where ub is the axial displacement of the unsupported pipe, for b ≤ x < (b + d )
B1a and Ba2 are constants, and x is the distance along the pipe
−2 Fx d χ 5χ 2 χ
from the centreline. Symmetry considerations require that [15c] σax = + Pi − 6 Epα p∆T
ub(x = 0) = 0, and consequently Ba2 = 0 because unsupported t χ4 χ4 χ4
length is assumed to form at the centre of the pipe segment. for 0≤x <b
The other boundary conditions are essentially compatibility
requirements for deformations at transition points from por- where σax is the axial stress. If the soil yields beyond the
tions of pipe within different regions of soil behaviour, i.e., plastic limit, i.e., ud > ux, the temperature change will not
ub (x b = b) = ud (x d = 0), ub′ (x b = b) = u d′ (x d = 0), ud (x d = d) have any influence on the axial stress as shown by eq. [15a].
= u f (x f = 0), ud′ (x d = d) = u f′ (x f = 0), and σxa (x f = f ) = 0, This solution is similar to that developed by Rajani et al.
where σxa is the axial stress. The unknown constants B1a , C1a , (1996) except that the soil supporting medium is considered
C 2a , F1a , and F2a can be obtained by applying these boundary as elastoplastic and there is an added consideration for the
conditions: loss of bedding support as a consequence of scour developed
from a leaky pipe. The solution reverts to that previously ob-
−2 Fx d χ 5 χ 2 χ tained by Rajani et al. (1996) when the unsupported length,
[9] B1a = + Pi − 6 Epα p∆T b, is zero and the soil is elastic.
t χ4 χ4 χ4
Sensitivity analyses: axial stress
A sensitivity analysis was carried out to investigate the
Ba ⎛ b 1 ⎞⎟ Pi ⎛⎜ χ 2 ⎞⎟
[10] C1a = 1 ⎜⎜ − + axial response of buried pipe to different pipe materials, pipe
Ep ⎝ 2 2γχ1 ⎟⎠ Ep ⎜⎝ 2γχ1 ⎟⎠ diameters, soil types, and soil temperatures close to the pipe.
A 150 mm (6 in.) CI pipe buried in medium sand with ksa =
⎛b χ ⎞ 125 MPa/m was selected as a reference case (Table 1) to
+ α ∆T ⎜⎜ − 3 ⎟⎟ gauge the sensitivity of the axial pipe–soil system to varia-
⎝ 2 2γχ1 ⎠ tions of different parameters.

© 2004 NRC Canada


Rajani and Tesfamariam 1001

Table 1. Reference data for typical cast iron (CI), ductile iron (DI), and polyvinyl chloride
(PVC) water mains, soil properties, and operating conditions.
(A) Reference data for pipe material
CI DI PVC
Pipe diameter (mm) 150 (6 in.) 100 (4 in.) 200 (8 in.)
External diameter, D (mm) 175.26 121.92 229.87
Wall thickness, t (mm) 10.92 10.16 10.16
Pipe length, 2L (m) 6.0 6.0 6.0
Elastic modulus, Ep (MPa) 206 000 165 000 2250
Ultimate tensile strength (MPa) 207 290 48
Poisson’s ratio, υp 0.26 0.28 0.42
Thermal coefficient, α p (×10–6/°C) 10.5 11.0 79.0
(B) Soil properties
Medium sand Very soft clay
Elastic modulus, Es (MPa) 100 1.5
Poisson’s ratio, υs 0.3 0.3
Soil unit density, ρ (kg/m3) 2344 1988
Lateral foundation modulus, ks′ (MPa) 50
Axial foundation modulus, ksa (MPa/m) 125
(C) Operating conditions
Water pressure, Pi (kPa) 345 (50 psi)
Installation temperature (°C) 24
Temperature amplitude, AT (°C) 15
Annual mean temperature (°C) 15
Overburden load, q (N/mm) 27.13
Frost load factor, ffrost 0.5

The axial stress profile for a 150 mm diameter CI main ation of elastoplasticity increases the axial stresses induced
shown in Fig. 5 indicates that the axial stress decreased as in the pipe compared with elastic analysis only. Elastoplastic
the unsupported length increased, since the pipe length that response of the soil was artificially induced in this example
is constrained from movement decreases. Similar responses by increasing the temperature difference (–56 °C) by four
are observed as unsupported pipe lengths increase and pipe times the value (–14 °C) used in the rest of analyses to illus-
diameters decrease (Fig. 6) or if the soil temperatures close trate the fact that a large temperature differential would have
to the pipe decrease (Fig. 7). Axial stresses induced at the to exist to induce significant stress.
pipe–soil interface will be lower in large-diameter pipes be-
cause of a proportional increase in contact surface area. This
is expected because the remaining pipe length that is sub- Flexural pipe–soil interaction
jected to axial restraint decreases as the unsupported pipe A pipe buried at constant depth in soil backfill or bedding
length increases. with uniform geotechnical properties should not be normally
Figure 8 shows the maximum axial stress (expressed as a subjected to flexural or bending deformations or stresses in
percentage of the ultimate tensile strength) induced in differ- the longitudinal direction. Loss of bedding support near the
ent pipe materials (CI, DI, and PVC) as a consequence of pipeline under circumstances described earlier, however,
temperature differential (between soil temperature and tem- subjects the pipe to flexural stresses. The total load, q, im-
perature of water in the pipe). Although there are no appre- posed on the unsupported length of the pipe is the earth load
ciable differences between the axial responses of CI and DI together with the traffic and frost loads. The soil near the un-
pipes, the PVC pipe shares a significant proportion (7% of supported pipe region may exceed the elastic displacement
the ultimate stress at ∆T of –14 °C) of the stress because of limit (vu) and develop plastic deformations if the overburden
its higher thermal expansion coefficient and lower elastic loads or unsupported length are high enough. Prior to the de-
modulus. velopment of solutions for a jointed pipe, simple solutions
Axial stresses in the pipe increase (Fig. 9) as the soil be- considering the supported portion of the pipe as an infinite
comes stiffer, i.e., higher axial foundation moduli, for any beam (when λ[L – (b + c)] > π, where λ is the reciprocal of
specific unsupported pipe length. As explained earlier, this the flexural characteristic length as defined in eq. [18]) on
increase is accentuated at shorter unsupported lengths be- an elastoplastic foundation (Fig. 11) were obtained to deter-
cause more of the pipe surface in contact with the soil is mine if there was merit to incorporating soil elastoplasticity.
restrained. The role of elastoplastic behaviour of the soil These simple solutions indicated that soil elastoplasticity is
shown in Fig. 10 corresponds to a pipe with an unsupported significant for typical pipe–soil characteristics encountered
length of b = 500 mm and clearly illustrates that consider- in the water distribution systems when the unsupported

© 2004 NRC Canada


1002 Can. Geotech. J. Vol. 41, 2004

Fig. 5. The effect of unsupported length, b, on axial stress. Fig. 7. The effect of seasonal pipe temperatures on axial stress.

Fig. 6. The effect of pipe size and unsupported length on axial


stress.
Fig. 8. The effect of pipe material on axial stress as a result of
temperature difference.

length is greater than 2 m and rigid body movement is per-


mitted. Thus, two distinct regions of soil in contact with the
where I zz is the second moment of inertia of the pipe around
pipe can be characterized as plastic and elastic, where xc and
the z axis, v e is the pipe vertical displacement in the elastic
xe represent the coordinate axes of regions c and e, respec-
soil region, and ks′ is the lateral elastic foundation modulus
tively, in Fig. 11. The pipe in these circumstances can be
of the soil. The general solution for v e is
modelled as a beam on an elastoplastic foundation (bepf)
with partial support.
[17] v e = exp(λx e )( E1 cos λx e + E2 sin λx e )
Frost load (Rajani and Zhan 1996) can be accounted for in
a simple form as a multiple of the frost load factor ( ffrost) of + exp(−λx e )( E3 cos λx e + E4 sin λx e )
the soil load. The frost load multiple, ffrost, ranges from 1,
where there is no frost load, to 2, the maximum expected where E1, E2, E3, and E4 are constants; and the reciprocal of
frost load. λ is the reciprocal of the flexural characteristic length:
The equilibrium equation for the portion of the pipe sup- ks′
ported by elastic bedding is represented by a Winkler beam [18] λ=4
on an elastic foundation (bef) (Hetényi 1946): 4 Ep I zz

The lateral elastic foundation modulus of the soil, ks′, in


d4v e
[16] Ep I zz + ks′ v e = 0 terms of elastic soil properties can be estimated as suggested
dx 4 by Vesic (1961):

© 2004 NRC Canada


Rajani and Tesfamariam 1003

Fig. 9. The effect of soil stiffness, ksa , and unsupported length on Fig. 11. Schematic model for partially supported pipe on
axial stress. elastoplastic foundation.

Fig. 10. The effect of elastoplastic analysis on axial stress.

Fig. 12. Typical lateral soil resistance versus displacement.

where Fz is the maximum lateral soil resistance per unit


length.
The maximum lateral soil resistance per unit length, cor-
0.65 Es 12 EsD4 responding to the undrained state, is computed as proposed
[19] ks′ =
1 − υ2s EpI zz by Trautmann et al. (1985) and Poulos and Davis (1980) for
sand and clay as follows:

Differentiation of vertical displacement v with respect to x [21] Fz = γ sDHN z for sand


gives the slope, moment, and shear, respectively, as v′ = tan θ,
−Ep I zz v′′ = M x , and −Ep I zz v′′′ = Q, where θ is the slope, M x [22] Fz = N c Dsu for clay
is the bending moment, and Q is the shear force. where N z is a dimensionless resistance factor for sand, and
As stated earlier, loss of support is likely to induce soil N c is the bearing capacity type factor.
deformations that are beyond the elastic limit, vu, and hence The factor N c depends on the ratio of burial depth (H) to
will induce plastic deformations in the soil immediately ad- pipe diameter (D). For typical pipe sizes and burial depths,
jacent to the point of lost support. The limiting elastic defor- the ratio H/D can vary between 3 and 20. Rowe and Davis
mation, vu, when the plastic deformations of the soil initiate (1982) have shown that the value of N c (= 11.42) is essen-
(Fig. 12) is given by tially constant for H/D > 3. Using Vesic’s equation, Das
(1995) showed that Nz is approximately 18.4 when the angle
[20] vu = Fz / ks′ of internal friction φ is 30°.

© 2004 NRC Canada


1004 Can. Geotech. J. Vol. 41, 2004

The equilibrium equation for the portion of the pipe em- Fig. 13. The effect of pipe size and unsupported length on flex-
bedded in soil with plastic deformations (region c in Fig. 11) ural stress.
is given by
d4v c
[23] Ep I zz + Fz = 0
dx 4
The displacement response, v c , for the eq. [23] is
−Fz x c4 C x3 C x2
[24] vc = + 1 c + 2 c + C 3 xc + C 4
24 Ep I zz 6 2

where C1, C2, C3, and C4 are constants.


The vertical deflection (vb) along the unsupported length
of the pipe subjected to a uniform load (soil and live), q, can
be described by the fourth-order polynomial
q x b4 B x3 B x2
[25] vb = + 1 b + 2 b + B3 x b + B4
24 Ep I zz 6 2

where B1, B2, B3, and B4 are constants.


It is assumed that there is no vertical movement at the
bell–spigot joint relative to the adjacent jointed pipe and, as
indicated earlier, bell–spigot joints permit rotation and hence Fig. 14. The effect of pipe size and unsupported length on the
no moment will develop. Therefore, the corresponding maximum flexural stress.
boundary conditions at the bell–spigot joint are v e (x e = L) =
0 and v′′e (x e = L) = 0. Symmetry considerations at the centre-
line dictate the boundary conditions, zero rotation, v b′ (x b = 0) =
0, and zero shear, v′′′ b (x b = 0) = 0. Consequently, B1 and B3
are equal to zero. At xb = b and xc = 0, to satisfy continuity
and compatibility, the boundary conditions are vb(x = b) =
vc(xc = 0), v b′(xb = b) = v c′(xc = 0), v′′b(xb = b) = v′′c (xc = 0),
and v′′′ b (xb = b) = v′′′
c (xc = 0). Similarly, at xc = c and xe = 0, to
satisfy continuity and compatibility, the boundary conditions
are vc(xc = c) = ve(xe = 0), v c′ (xc = c) = v e′ (xe = 0), v′′c (xc = c) =
v′′e (xe = 0), and v′′′c (xc = c) = v′′′e (xe = 0). The unknowns in
eqs. [17], [24], and [25] can be determined by applying
these boundary conditions. Detailed derivations and descrip-
tions of the unknown terms are given in Appendix A. The
unknown terms of eq. [17], B2, and B4 are given in eqs. [26]
and [27]:
Fzc / 2 Ep I zzλ3 + qb / 2 Ep I zzλ3 + θ3 + θ2 + α 6 − α 4
[26] B2 =
− θ4 − θ1 + α 7 − α 5
q b4 B b2
[27] B4 = C 4 − − 2 at the centreline and in the soil at the location where the
24 Ep I zz 2
pipe support ends. As expected, small-diameter mains expe-
where α 4 – α 7 and θ1– θ4 are constants. rience higher stresses (Figs. 13, 14) than large-diameter
mains, and the stress differences increase as the unsupported
length, b, increases (Fig. 14). The 100 mm (4 in.), 150 mm
Sensitivity analyses: flexural stress (6 in.), and 200 mm (8 in.) diameter CI pipes approach ulti-
A cursory look indicates that a number of variables influ- mate strength at b = 1150, 1760, and 2180 mm, respectively.
ence the flexural response of a partially unsupported pipe on The flexural stresses in CI and DI pipes are comparable
an elastoplastic bedding. Sensitivity analyses are carried out (Fig. 15), as both pipes have higher moduli of elasticity than
to identify the role of the principal variables such as soil PVC. The 150 mm PVC, CI, and DI pipes approach their re-
type, pipe material, and pipe diameter. The same 150 mm spective ultimate strengths (Fig. 16) at unsupported lengths
(6 in.) CI pipe used for the sensitivity study of axial pipe– of b = 1000, 1760, and 1880 mm. This highlights the fact
soil interaction is used to study the flexural pipe–soil inter- that CI and DI pipes can tolerate much higher unsupported
action except that the lateral foundation modulus for lengths than PVC pipes.
medium sand is selected as ks′ = 50 MPa. The lateral elastic foundation modulus of the soil, ks′ , is
For a given unsupported length, two locations (Fig. 13) influenced by the pipe rigidity, EpIzz, and elastic modulus of
along the length of the pipe are susceptible to high stresses: the soil. The foundation moduli, ks′ , can range between 24
© 2004 NRC Canada
Rajani and Tesfamariam 1005

Fig. 15. The effect of pipe material and unsupported length on Fig. 17. The effect of soil type on flexural stress.
flexural stress.

Fig. 16. The effect of pipe material and unsupported length on Fig. 18. The effect of soil stiffness on flexural stress.
the maximum flexural stress.

can be significant to conduct elastoplastic pipe–soil


and 52 MPa for medium sand and between 0.08 and interaction analysis for flexural behaviour, depending on the
0.55 MPa for very soft clay. The flexural responses (Fig. 17) specific design variables.
of the same pipe but buried in different soil types can be
dramatically different. This difference in response is further Circumferential (or hoop) pipe–soil
demonstrated in Fig. 18, where the flexural stresses decrease interaction
with an increase in ks′ . The pipe effectively behaves like a
simply supported beam when ks′ is very small. The flexural It was stated earlier that it is appropriate to consider most
stress response does not change (Fig. 18) significantly, how- common pipe materials as rigid, for pipe sizes of interest in
ever, when the lateral foundation modulus exceeds 50 MPa. the water industry. The overburden (earth) load together with
The role of elastoplastic behaviour of the soil on flexural traffic and frost loads induce thrust and moment in the pipe
bending stress is shown in Fig. 19 for a pipe with an unsup- in the circumferential (or hoop) direction. The rigid-pipe
ported length of b = 1000 mm. It can be seen that slightly assumption considers there is no interaction support from
higher bending stresses are obtained when the bedding is the soil, and there is zero lateral earth pressure (Ko = 0).
modelled as an elastoplastic medium. The soil near unsup- This assumption will assure conservative results for the CI
ported pipe develops plastic deformations as the unsupported pipes and very conservative results for the PVC pipes. The
length or overburden load increases. Though the sensitivity hoop stress, σw θ , for the overburden loads, q (combination of
analysis does not show a dramatic difference in response, it earth, traffic, and frost load), is
© 2004 NRC Canada
1006 Can. Geotech. J. Vol. 41, 2004

Fig. 19. The effect of soil elastoplasticity on flexural stress. Fig. 20. Variation of hoop stress, σPθi , with D/t and Ep/Es for cast
iron pipes.

⎛ 3D ⎞
θ = q⎜
σw ⎟
[28] Fig. 21. Variation of hoop stress with change in soil temperature
⎜ πt 2 ⎟
⎝ ⎠ close to the pipe.

(Watkins and Anderson 1999).


The hoop stress (σPθi ) in a thin pipe as a result of the inter-
nal pressure is given by
⎛D −t⎞
[29] σPθi = (Pi − Pe ) ⎜⎜ ⎟

⎝ 2t ⎠
where Pe is the external radial pressure. The radial displace-
ment is constrained by the radial stiffness of the surrounding
soil, and the force–displacement relation for a pipe in an in-
finite medium is given by
Es
[30] Pe = krθu r = ur
(D / 2) (1 + υs)
where krθ is the radial soil stiffness, and ur is the radial dis-
placement. There is no radial restraint in the region where
the pipe is unsupported, and hence Pe = 0. It is assumed that
for the typical radial displacement, the soil remains in the
elastic range. where ri is the distance from the centre of the pipe to the in-
As a consequence of the Poisson’s effect on longitudinal ner wall, ro is the distance from the centre of the pipe to the
stress, the bending hoop stress σfθ is outer wall, and r is the distance from the centre of the pipe
⎛M D⎞ to any point (Ugural and Fenster 1987).
[31] σfθ = −υpσxf = −υp ⎜ x ⎟ The total hoop stress because of external loads, internal
⎜ 2 I zz ⎟
⎝ ⎠ pressure, temperature differential, and longitudinal bending is
where σfx is the stress in the longitudinal direction due to [33] σTotal
θ = σw
θ + σθ + σθ + σθ = σθ + σθ + σθ − v p σx
Pi T f w Pi T f

flexural action. The temperature differential (∆T ) between


the inside of the pipe and the surrounding soil induces a Sensitivity analyses: hoop stress
thermal hoop stress σTθ :
The nondimensional form of hoop stress derived using
α p Ep∆T eq. [29] for CI is shown in Fig. 20 (similar plots for PVC
[32] σTθ =
2 (1 − υp) ln(ro/ ri ) and DI were provided by Rajani et al. 1996). Unlike for
PVC and DI pipes, the analyses show that CI pipes exhibit

r r2 ⎛ r 2 ⎞ ⎛ r ⎞⎤ little sensitivity to the surrounding soil stiffness, which is al-
× ⎢1 − ln o − 2 i 2 ⎜⎜1 + o2 ⎟⎟ ln ⎜ o ⎟⎥
⎢ r (ro − ri ) ⎝ r ⎠ ⎜⎝ ri ⎟⎠⎥ together not surprising because CI pipes are considered as
⎣ ⎦ rigid.
© 2004 NRC Canada
Rajani and Tesfamariam 1007

Table 2. Longitudinal stress components from axial and flexural responses.


Internal Unsupported External Pipe–soil
Domain pressure Temperature pipe length load interaction
(b + d) ≤ x < (b + d + f ) — — Mx D ⎛ 3D ⎞ Fx
(x f − f )
2 I zz υp ⎜⎜ 2 ⎟⎟ q t
⎝ πt ⎠
b < x < (b + d) χ 2 Pi χ 3E p α p ∆T Mx D ⎛ 3D ⎞ ∂ud
υp ⎜⎜ 2 ⎟⎟ q χ1 E p
2 I zz ∂x
⎝ πt ⎠
x<b ⎛χ ⎞ χ6 Mx D — −2 Fx d
χ 2 Pi ⎜⎜ 5 ⎟⎟ E p α p ∆T
χ4 2 I zz t χ4
⎝ χ4 ⎠

Table 3. Hoop stress components from axial, flexural, and circumferential responses.
Domain x>b x≤b
Unsupported pipe length M D M D
υp x υp x
2 I zz 2 I zz
External load ⎛ 3D ⎞ —
⎜ ⎟
⎜ πt2 ⎟ q
⎝ ⎠
Internal pressure ⎛ D − t⎞ ⎧⎪ β exp( −γx ′′) + exp( γx ′′) ⎫⎪ ⎛ D − t⎞
Pi ⎜ ⎟
⎜ ⎟ Pi ⎨1 − − υp E s χ 2
2
⎬ ⎝ 2t ⎠
⎝ 2t ⎠ ⎪⎩ β1 β1 χ1 E p (1 + υs )[exp( −γf ) + exp( γf )] ⎪⎭

⎛ D⎞ ⎧ exp( −γx ′′) + exp( γx ′′) ⎫ α p E p ∆T


⎜ ⎟ α p ∆T⎨−E p η + υp E s χ 3
Temperature
⎜ 2t⎟ ⎬ 2(1 − υp ) ln (ro /ri )
⎝ ⎠ ⎩ β1 β1 χ1 (1 + υs )[exp( −γf ) + exp( γf )] ⎭
⎡ ⎛ 2⎞ ⎤
α p E p ∆T ⎡ r r2 ⎛ r 2 ⎞ ⎛r ⎞⎤ r r2
× ⎢1 − ln o − 2 i 2 ⎜1 + ro ⎟ ln ⎛⎜ ro ⎞⎟ ⎥
+ ⎢1 − ln o − 2 i 2 ⎜1 + o ⎟ ln ⎜ o ⎟ ⎥
2(1 − υp ) ln (ro /ri ) ⎢⎣ ri ro − ri ⎜ r 2 ⎟⎠ ⎜⎝ ri ⎟⎠ ⎥⎦ ⎢ r ro − ri ⎜ r 2 ⎟ ⎜⎝ ri ⎟⎠ ⎥
⎝ ⎣ ⎝ ⎠ ⎦

The effect of temperature on hoop stress is a combination cult to ascertain the precise cause because all the operational
of the radial restraint and the thermal difference between the data are not always known, e.g., surge pressure, pipe condi-
inside and outside surfaces of the pipe (eq. [32]). As de- tion, unsupported length, pit geometry. There is a high
picted in Fig. 21, the radial restraint is almost zero, whereas degree of uncertainty associated with all the factors contrib-
the thermal stress calculated using eq. [32] is significant uting to pipe failure because of the great spatial variability
(and in compression). The reason for this is that when the (even in a moderate size network), especially with corrosion
inside temperature is greater than the outside pipe tempera- rates. The analytical procedures developed here for jointed
ture, the hoop stress on the outer surface is negative (com- water mains and in combination with those developed earlier
pressive). provide a means to identify the impact of different interven-
ing variables on axial, flexural, and circumferential stress re-
sponses. The contributions towards these stress responses
Conclusions from internal water pressure, temperature differential, un-
Current structural design of new water mains is based pri- supported pipe length, external loads, and pipe–soil interac-
marily on circumferential (hoop) stresses imposed by inter- tion are summarized in Tables 2 and 3.
nal pressure and external loads. This design process is valid Sensitivity analyses indicate that the extent of the unsup-
as long as the pipe is uniformly supported along its length. ported length developed as a result of scour has a significant
The deterioration (corrosion) of water mains with time, how- influence on the flexural pipe–soil response, but the axial
ever, dictates that axial and flexural (longitudinal bending) pipe–soil response is not negatively affected in terms of
responses should be considered together with the circumfer- higher stress. In general, plastic pipes tolerate less loss of
ential response. Furthermore, Rajani et al. (1996) showed support than metallic pipes. In most practical situations, it is
the importance of considering temperature differential on the appropriate to ignore soil elastoplasticity, since its influence
axial response to explain the increased number of water on pipe response is minor.
main breaks during periods of late spring – early winter and The physical deterministic model as described in this
late winter – early spring, i.e., when the temperature differ- paper can be used to conduct postfailure analysis. As previ-
ence between the water and the soil–backfill close to the ously mentioned, there is a high degree of uncertainty asso-
pipe is likely to be the highest. ciated with all the factors contributing to pipe failure
In most cases a combination of circumstances leads to the because of the great spatial variability. The physical deter-
failure of any one particular water main, and it is very diffi- ministic model for pipe–soil interaction proposed here
© 2004 NRC Canada
1008 Can. Geotech. J. Vol. 41, 2004

provides point estimates (or fixed values) to determine the Poulos, H.G., and Davis, E.H. 1980. Pile foundation analysis and
factor of safety, which is generally not sufficient. Therefore, design. John Wiley and Sons, Toronto, Ont.
the model requires further development to include uncertain- Rajani, B., and Zhan, C. 1996. On the estimation of frost loads.
ties so that the probability of pipe failures at a given time Canadian Geotechnical Journal, 33: 629–641.
can be quantified to plan maintenance and repair strategies. Rajani, B., Zhan, C., and Kuraoka, S. 1996. Pipe–soil interaction
Possible approaches to do this are Monte Carlo simulations analysis of jointed water mains. Canadian Geotechnical Journal,
(Sadiq et al. 2004) and fuzzy-based methods (Guynnet et al. 33: 393–404.
2000) to evaluate the time-dependent reliability, i.e., hazard Rowe, R.K., and Davis, E.H. 1982. The behaviour of anchor plates
in clay. Géotechnique, 32: 9–23.
function of time to failure of buried pipes, and to identify
Sadiq, R., Rajani, B., and Kleiner, Y. 2004. Probabilistic risk anal-
key modelling and input parameters that contribute to the re-
ysis of corrosion associated failures in cast iron water mains.
duction in factor of safety. Reliability Engineering & System Safety, 86(1): 1–10.
Scott, R.F. 1981. Foundation analysis. Prentice Hall Inc., Engle-
References wood Cliffs, N.J.
Trautmann, C.H., O’Rourke, T.D., and Kulhawy, F. 1985. Uplift
Committee on Gas and Liquid Fuel Lifelines. 1984. Guidelines for
force–displacement response of a buried pipe. Journal of Geo-
the seismic design of oil and gas pipeline systems. America So-
technical Engineering, ASCE, 111(9): 1061–1076.
ciety of Civil Engineering, New York.
Ugural, A.C., and Fenster, S.K. 1987. Advanced strength and ap-
Das, B.M. 1995. Principles of foundation engineering. PWS Pub-
plied elasticity. Elsevier, New York.
lishing Company, Toronto, Ont.
Vesic, A.S. 1961. Bending of beams resting on isotropic elastic
Guynnet, D.G., Come, B., Perrochet, P., and Parriaux, A. 2000.
solid. Proceedings of the American Society of Civil Engineers,
Comparing two methods for uncertainty in risk assessments.
87(EM2): 35–51.
Journal of Environmental Engineering, ASCE, 125(7): 660–666.
Watkins, R.K., and Anderson, L.R. 1999. Structural mechanics of
Hetényi, M. 1946. Beams on elastic foundations. University of
buried pipes. CRC Press, New York.
Michigan Press, Ann Arbor, Mich.
Makar, J. 1999. Diagnostic techniques for sewer systems. Journal
of Infrastructure Systems, 5(2): 69–78. Appendix A
Makar, J.M., and Rajani, B.B. 2000. Gray cast-iron water pipe
metallurgy. Journal of Materials in Civil Engineering, ASCE,
For a medium-length beam, the governing equations given
12(3): 245–253. in eq. [8] and the corresponding derivatives are as follows:

v e = exp(λx e )( E1 cos λx e + E2 sin λx e ) + exp(−λx e )( E3 cos λx e + E4 sin λx e )


v e′ = λ{exp(λx e )[ E1(cos λx e − sin λx e ) + E2(cos λx e − sin λx e )]
− exp(−λx e )[ E3(cos λx e + sin λx e ) − E4(cos λx e − sin λx e )]}
v′′e = 2λ2[− exp(λx e )( E1 sin λx e − E2 cos λx e ) + exp(−λx e )( E3 sin λx e − E4 cos λx e )]
v′′′
e = 2λ {− exp(λx e )[ E1(cos λx e + sin λx e ) − E2(cos λx e − sin λx e )]
3

+ exp(−λx e )[ E3(cos λx e − sin λx e ) + E4(cos λx e + sin λx e )]}


It is assumed that because of the bell and spigot connection the vertical displacement and the moment are zero at the finite
length of the pipe.
Similarly, at the unsupported beam and the beam supported by a plastic soil connection, to satisfy the continuity, the fol-
lowing boundary conditions are solved at x = b and x′ = 0:

Beam resting on
Condition plastic foundation Unsupported beam
vc = vb C1 q b4 B b2
+ 2 + B4
24E p I zz 2
vc′ = vb′ C3 q b3
+ B2 b
6 E p I zz
vc′′ = vb′′ C2 q b2
+ B2
2 E p I zz
vc′′′ = vb′′′ C1 qb
2 E p I zz

© 2004 NRC Canada


Rajani and Tesfamariam 1009

At the intersection of the beam supported by plastic soil and elastic soil, to satisfy the continuity, the following boundary
conditions are solved at xc = c and xe = 0:

Condition Beam on elastic foundation Unsupported beam


ve = vc E1 + E3 −Fz c 4 C c3 C c2
+ 1 + 2 + C3c + C4
24E p I zz 6 2
ve′ = vc′ λ(E1 + E2 + E3 + E4) −Fz c 3 C c2
+ 1 + C2c + C3
6 E p I zz 2
ve′′ = vc′′ –2λ2(E2 – E4) −Fz c 2
+ C1 c + C 2
2 E p I zz
ve′′′ = vc′′′ 2λ3(–E1 + E2 + E3 + E4) −Fz c
+ C1
E p I zz

Using the aforementioned governing equations and bound- and θ1, θ2, θ3, and θ4 are
ary conditions, the solution can be shown to be
α 5 exp(2λe) + α 7(1 − 2 cos 2 λe)
Fzc / 2 Ep I zzλ + qb / 2 Ep I zzλ + θ3 + θ2 + α 6 − α 4
3 3 θ1 =
B2 = 2 cos λe sin λe
−θ4 − θ1 + α 7 − α 5
α 4 exp(2λe) + α 6(1 − 2 cos 2 λe)
4 2 θ2 =
B4 = C 4 −
qb B b
− 2 2 cos λe sin λe
24 Ep I zz 2
θ3 = θ2 exp(−2λe) − α 4 tan λe − α 6 exp(−2λe) tan λe
E1 = θ3 + θ4B 2
θ4 = θ1 exp(−2λe) − α 5 tan λe − α 7 exp(−2λe) tan λe
E2 = α 4 + α 5 B 2
E3 = −θ2 − θ1B 2 List of symbols
E4 = α 6 + α 7B 2 AT temperature amplitude
b unsupported pipe length
qb
C1 = Bia , C ia , Dia , Fia axial pipe–soil interaction constants (i = 1, 2)
2 Ep I zz c extent of soil in elastoplastic state (lateral)
d extent of soil in elastic state (axial)
q b2
C2 = + B2 D outside pipe diameter
2 Ep I zz e extent of soil in elastic state
Ei flexural pipe–soil interaction constants (i = 1–4)
q b2 Ep elastic modulus of pipe
C3 = + B 2b
2 Ep I zz Es elastic modulus of soil
f extent of soil in elastoplastic state (axial)
Fz c4 C c3 C c2 ffrost frost load factor
C 4 = ( E1 + E3) + − 1 − 2 − C3 c Fx ultimate axial soil resistance
24 Ep I zz 6 2
Fz maximum lateral soil resistance per unit length
where α4, α5, α6, and α7 are Gs soil shear modulus
H burial depth of water mains
Fz c qb Izz pipe second moment of inertia around the z
α4 = (−c2λ2 − 3 cλ − 3) + axis
24 Ep I zz λ3
24 Ep I zz λ3
k θr radial soil stiffness
× (3 + 3c2λ2 + 6 cλ + 3bc λ2 + 3bλ + b2λ2) ksa axial pipe–soil reaction modulus
ks′ lateral soil elastic foundation modulus
c b 1
α5 = + + 2 Ko coefficient of active resistance at rest
4λ 4λ 4λ L half jointed pipe length
Fz c qb Mx bending moment
α6 = (−c2λ2 + 3 cλ − 3) + Nz dimensionless resistance factor for sand
24 Ep I zz λ3 24 Ep I zz λ3 Nc bearing capacity type factor for clay
× (3 + 3c2λ2 − 6 cλ + 3bc λ2 − 3bλ + b2λ2) Pe external radial pressure
Pi pipe internal pressure
c b 1
α7 = + − q overburden load
4λ 4λ 4λ2 Q shear force

© 2004 NRC Canada


1010 Can. Geotech. J. Vol. 41, 2004

r distance from the centre of the pipe to any α i flexural pipe–soil interaction constants (i = 4–7)
point α p linear thermal expansion coefficient of pipe
ri distance from the centre of the pipe to the in- material
ner wall β1 axial pipe–soil interaction constants
ro distance from the centre of the pipe to the δ frictional angle between pipe material and
outer wall surrounding backfill
su undrained shear strength of clay ∆T maximum temperature difference between wa-
t pipe wall thickness ter and surrounding soil
u axial displacement φ soil internal friction angle
ub , ub′ pipe axial displacement and strain in the un- γ reciprocal of the axial characteristic length
supported region γs soil submerged unit weight
ud , ud′ pipe axial displacement and strain in the elas- η ratio of elastic pipe and soil properties as de-
tic region fined by Rajani et al. (1996)
uf , uf′ pipe axial displacement and strain in the λ reciprocal of the flexural characteristic length
elastoplastic region υs Poisson’s ratio of soil
ur pipe radial displacement υp Poisson’s ratio of pipe
ux soil displacement at ultimate axial resistance θ slope
v vertical displacement θi flexural pipe–soil interaction constants (i = 1–4)
vb , vb′ , vb′′, vb′′′ pipe vertical displacement, slope, curvature ρ unit density soil
and change in curvature in the unsupported re- σax axial stress
gion σfx stress in longitudinal direction due to flexural
vc , vc′ , vc′′, v c′′′ pipe vertical displacement, slope, curvature action
and change in curvature in the elastoplastic re- σPθi hoop stress due to internal pressure
gion σfθ hoop stress due to the Poisson’s effect of lon-
ve , ve′ , ve′′, ve′′′ pipe vertical displacement, slope, curvature gitudinal bending
and change in curvature in the elastic region σTθ hoop stress due to temperature differential
vu soil displacement at the ultimate lateral resis- σTotal
θ total hoop stress
tance σwθ hoop stress due to overburden loads
x distance along pipe from the centreline χ i axial pipe–soil interaction constants (i = 1–6)
xb, xc, xd, xe, xf longitudinal coordinate systems for axial
(Fig. 4) and lateral (Fig. 11) pipe-soil interac-
tion
α adhesion coefficient

© 2004 NRC Canada

Вам также может понравиться