131

NOTE / NOTE

Tunneling effects on jointed pipelines

Assaf Klar, Alec M. Marshall, Kenichi Soga, and Robert J. Mair

Abstract: The problem of tunneling effects on existing jointed pipelines is studied using the boundary integral method.
Normalized solutions to evaluate the maximum bending moments and rotations are presented. They depend on tunnel-
induced ground settlements at pipeline level, relative soil–pipe stiffness, relative pipe-joint stiffness, and the location
of the joints in relation to the tunnel centerline. A jointed pipeline generally experiences smaller bending moments
than a continuous one owing to the joint rotation. The solutions indicate that there are certain cases where hinged
systems result in greater bending moments than continuous ones. However, these cases rarely occur in reality.
Key words: pipelines, tunneling, soil–structure interaction, continuum solute.
Résumé : Au moyen de la méthode intégrale des frontières, on traite du problème des effets du creusage de tunnels sur
des pipelines avec joints. On présente les solutions normalizées pour évaluer les moments fléchissants et les rotations
maximum. Ils dépendent des tassements du sol induits par le tunnel au niveau du pipeline, de la rigidité relative sol –
tuyau, de la rigidité tuyau-joint, et de la localization des joints par rapport à l’axe central du tunnel. Un pipeline avec
joints va en général subir de plus petits moments fléchissants qu’un pipeline continu à cause de la rotation dans les joints.
Les solutions indiquent qu’il y a certains cas où des systèmes avec charnières donnent des moments fléchissants plus
grands que des systèmes continus. Cependant, ces cas se produisent rarement en réalité.
Mots-clés : pipelines, creusage de tunnel, interaction sol – structure, solution continue.
[Traduit par la Rédaction]

Introduction tion based on a pipeline within an elastic soil continuum,

Construction of a tunnel may cause damage to buried
structures, such as pipelines. As a tunnel is excavated, it in-
duces deformation in the soil above. These displacements
will cause additional loading on a pipeline positioned above
the tunnel, as illustrated in Fig. 1. The behavior of the pipe-
line depends on the stiffness of the pipeline sections, the po-
sition and behavior of the pipe joints, and the nature of the
ground deformations. Solutions to this problem are pre-
sented in this note so that potential risk of pipeline damage
can be evaluated.
The general problem of tunneling effects on pipelines was
covered extensively by the excellent monograph of Attewell
et al. (1986). They provided solutions for continuous and
jointed pipelines using a Winkler ground model and ground
movements derived using a Gaussian curve. Klar et al. (2005)
obtained an exact closed form solution for the Attewell et al.
(1986) Winkler problem and provided a more rigorous solu-
Received 4 September 2006. Accepted 23 July 2007. Published
on the NRC Research Press Web site at cgj.nrc.ca on
21 February 2008.
A. Klar.1 Faculty of Civil Engineering, Israel Institute of
Technology, Technion City, Haifa 32000, Israel.
A.M. Marshall, K. Soga, and R.J. Mair. University of
Cambridge, Department of Engineering, Trumpington Street,
Cambridge CB2 1PZ, UK.
1Corresponding author (e-mail: klar@technion.ac.il).
again using a Gaussian settlement curve. They compared
these two solutions and found that the continuum one resulted
in higher bending moments, hence, the continuum solution
may be considered more conservative. Vorster et al. (2005)
extended the continuum solution to allow more freedom in de-
fining the shape of the settlement trough by using a modified
Gaussian curve. The solutions presented by Vorster et al.
(2005) were used to develop a conservative design procedure
based on the linear equivalent method for soil stiffness. Their
solutions were verified with centrifuge model tests.
The continuum solutions presented by Klar et al. (2005)
and Vorster et al. (2005) dealt solely with continuous pipe-
lines and did not address the problem of jointed pipelines. In
reality, pipelines are composed of sections that are con-
nected to each other by joints. If the joints are relatively stiff
compared to the flexural rigidity of the pipe, the solutions of
continuous pipes, mentioned above, may be considered ap-
propriate. However, more often than not, joints are not rigid
or lose their rigidity as they deform. Attewell et al. (1986),
using the Winkler model, presented normalized solutions for
the problem of hinged pipelines (i.e., zero joint stiffness). In
addition, they suggested a complicated iterative procedure to
deal with cases where the joints possess stiffness.
This technical note extends the continuum solution to the
case of jointed pipelines and provides normalized solutions
that allow for simple evaluation of joint rotation and bend-
ing moments of a jointed pipeline that is affected by tunnel
construction underneath.

Can. Geotech. J. 45: 131–139 (2008) doi:10.1139/T07-068 # 2008 NRC Canada

132 Can. Geotech. J. Vol. 45, 2008

Fig. 1. Illustration of the problem.

Fig. 2. Idealized joint behavior. Table 1. Values of the parameters for bilinear joint model (modi-
fied from Attewell et al. 1986).

Pipe diameter Small rotation

(cm) angle,
s (rad) 1 (kNm/rad) 2 (kNm/rad)
Lead-yarn hardened jute packing
10 0.00875 1100 375
15 0.00875 2500 900
20 0.00875 2500 900
Lead-yarn with soft jute packing
7.5 0.00875 100 50
10 0.00875 160 80
15 0.0175 330 165
20 0.035 635 320
25 0.035 970 480
30 0.035 1320 660
Engineering considerations for jointed Rubber gasket push-in type
pipelines 10 0.005 1.5 0.3
15 0.005 3.5 0.5
Although the focus in design of pipelines is concentrated
20 0.0037 12 1.75
on the cross-sectional behavior, attention should also be
25 0.006 14 2.6
given to the problem of settlement-induced loading, such as
by tunneling. Pipelines are always built using discrete sec-
tions, but they may be categorized as continuous or jointed,
depending on their joint behavior. For example, high-pressure
gas-transmission pipelines may be considered continuous as integrity is allowable tensile stress, one may use only the
they are built of steel sections rigidly connected by either bending moment solution to obtain a conservative evaluation
welded or flanged joints. Design criteria for these joints in- of tensile stresses for joints located in the sagging area
clude limits on allowable stresses across the joints (BS EN (above the tunnel centerline) where compressive axial forces
14161. 2003). Current wastewater pipelines are commonly exist. For joints located in the hogging area (offset from the
built of concrete sections joined by flexible rubber push- tunnel centerline) where tensile axial forces exist, the bend-
on joints. Design criteria for these joints include maximum ing moment solution alone is not sufficient to make a con-
allowable rotations (BS EN 545. 2002). Historically, cast iron servative evaluation of tensile stresses, and therefore one
pipelines with semi-rigid push-on joints sealed with lead- also needs to consider an axial-force solution (such as that
yarn were commonly used in the UK during the Victorian given by Attewell et al. 1986). Nonetheless, as shown by
time and these pipelines are still extensively used. The ro- Vorster (2005), in many cases the value of tensile stress cal-
tational stiffness of these joints is nonlinear and can be ap- culated from the maximum sagging bending moment alone
proximated by a bilinear relationship presented in Fig. 2. is higher than that calculated using the combination of both
Table 1 summarizes the parameters published by Attewell maximum hogging bending moment and tensile forces. In
et al. (1986) for the bilinear model. which case, the evaluation of tensile stresses based on max-
When a jointed pipeline is subjected to tunneling-induced imum sagging bending moments is conservative for continu-
settlement, both stresses and joint rotation need to be esti- ous pipelines (i.e., rigid joints). For flexible joints, tensile
mated to ensure pipeline serviceability. Generally, these stresses may not necessarily have to be considered because
entail the evaluation of both axial forces and bending mo- these joints generally allow for longitudinal movement and
ments. If the joint is considered rigid (i.e., the pipeline is release the axial forces. Moreover, the design criteria for
continuous) and the design criterion to maintain its structural these joints focus on joint rotation.
# 2008 NRC Canada
Klar et al. 133

Fig. 3. Representation of a pipeline as beam elements (right bottom) and joint elements (left bottom).

Fig. 4. Odd and even joint configurations. ½1
½½S
þ ½K 
þ ½K 
½ s
½S

fug ¼ ½K 
fuCAT g

½K 
i;j ¼ 1=Gi;j for i ¼ j; 0 for i 6¼ j

½ s
i;j ¼ Gi;j for i 6¼ j; 0 for i ¼ j
where [S] is the stiffness matrix of the pipeline, {u} is the
displacement of the pipe, Gi,j is the Green’s function that de-
fines the elastic continuum displacement at point i due to a
unit loading at point j, and {uCAT} is the additional displace-
ment at the pipeline level caused by the tunneling. By
employing assumption (1), {uCAT} becomes the green field
displacement (i.e., the soil displacement at the level of the
pipe if the pipe did not exist).
In the previous works (Klar et al. 2005; Vorster et al.
2005), the stiffness matrix of the pipeline was composed of
standard beam elements. In the current work, it is extended
to include joint elements, as described in Fig. 3.
For the bending behavior solution, the global pipeline
stiffness matrix, [S], is composed of the beam element ma-
trix [KB] and the joint element matrix [KJ], defined as fol-
lows (Weaver and Gere 1990):
2 3
12EI 6EI 12EI 6EI
6 L3 
The following sections provide normalized solutions for the 6 e L2e L3e L2e 7
7
bending moments in the pipeline and rotations at the joints. 6 7
6 6EI 4EI 6EI 2EI 7
6  7
6 L2e Le L2e Le 7
6 7
Formulation ½KB
¼ 6 12EI 6EI 12EI 6EI 7
6  2  2 7
The formulation essentially follows that presented by Klar 6 7
6 L3e Le L3e Le 7
et al. (2005) with modifications to include the effects of ½2
6 7
6 6EI 2EI 6EI 4EI 7
joints. The formulation is based on the following assump- 4  2 5
tions: (1) the tunnel is not affected by the presence of the L2e Le Le Le
2 3
pipeline; (2) the pipeline is buried in an elastic homogene- kjS 0 kjS 0
ous soil; (3) the soil response to loading, at the level of the 6 0 kjM 0 kjM 7
pipe, is unaware of the tunnel (this relaxing assumption al- ½KJ
¼ 6
4 kjS
7
0 kjS 0 5
lows the use of Mindlin’s (1936) Green’s function for load- 0 kjM 0 kjM
ing in a semi-infinite half space to construct barrel loads that
are used in the analysis); and (4) the pipe remains in contact where EI is the bending stiffness of the pipeline sections, Le
with the soil. is the length of the beam elements that constitute the pipe-
Using these assumptions, Klar et al. (2005) derived the line sections (each pipe section is composed of several beam
following equation, which can be solved to obtain the pipe elements), and kjM and kjS are the joint stiffness for rotation
behavior: and shear, respectively. Typical kjM values for different
# 2008 NRC Canada
134 Can. Geotech. J. Vol. 45, 2008

Fig. 5. Distribution along the pipeline for R = 10 and Lj/i = 1.

Fig. 6. Distribution along the pipeline for R = 10 and Lj/i = 2.

# 2008 NRC Canada

Klar et al.
Fig. 7. Distribution along the pipeline for R = 10 and Lj/i = 4.
joints are presented in Table 1 as 1 and 2. It should be
noted that the above matrices refer to the behavior of the
pipe in flexure, ignoring the axial response. If necessary,
the axial response can be solved either by extending the
stiffness matrices and adding the longitudinal degree of
freedom or by solving independently the axial response. For
elastic joints considered in this study, these two will result
in the same behavior if the vertical soil response to loading
is decoupled from the horizontal response (a commonly
used assumption in elastic soil-structure interaction).
The following section presents normalized solutions for
elastic joints, for which the stiffness is constant and equal
for all joints, and is independent of the axial behavior.
Normalized solution for jointed pipelines
affected by tunnel excavation
In the present work, we consider the case of a pipeline
transverse to the tunnel line, which is believed to be the
worst case for bending moments and rotations. For this
case, the error function (Peck 1969) may be considered a
good estimate of the vertical green field displacements, Sv:


1 x2
½3
Sv ðxÞ ¼ Smax exp  2
2i
where Smax is the maximum settlement at the pipeline level, x
is the transverse distance from the tunnel centerline, and i is
the distance to the inflection point. Several parameters are
used in the normalization of the results: (1) the relative
pipe–soil rigidity factor R = EI/Esi3r0 (where EI is the long-
135
itudinal bending stiffness of the pipe sections, Es is the
Young’s modulus of the soil, and r0 is the radius of the
pipe), (2) the joint stiffness ratio T = kjM/(EI/i), and (3) the
joint spacing ratio Lj/i (where Lj is the spacing between the
joints). By using the ratio R for normalization, the solution
becomes practically independent of the ratio i/r0. It should
be noted that all normalized solutions given in this paper cor-
respond to a Poisson’s ratio of 0.25 and an embedment depth
ratio Z/r0 = 7 (Z is the depth to the pipe axis from the ground
surface). Nevertheless, the solutions are not significantly sen-
sitive to these parameters, as shown by Klar et al. (2005).
In the current analysis, kjS is set to a very high value, prac-
tically forcing an infinite shear stiffness of the joint, leading
to zero relative vertical displacements across the joints.
Two possible symmetric cases are considered: (1) where
the tunnel centerline is located beneath a joint (referred to
as ‘‘odd configuration’’) and (2) where the tunnel centerline
is located in between the joints (referred to as ‘‘even config-
uration’’). The terms ‘‘odd’’ and ‘‘even’’ are derived from
the fact that a symmetric pipeline system with respect to
the tunnel has either an odd or an even number of joints, re-
spectively, as illustrated in Fig. 4.
Different trends of the pipeline behavior may be seen in
Figs. 5–7, which show the normalized bending moment
(Mi2/EISmax) and the normalized settlements (Sp/Smax) of the
pipeline as a function of the joint rigidity for the cases of
R = 10 and Lj/i = 1.0, 2.0, and 4.0.
As joint stiffness decreases (T decreases), the solutions
shift from that of a continuous pipeline to that of a hinged
one. This shift results in either an increase or a decrease in
# 2008 NRC Canada
136
Fig. 8. Normalized solution for Lj/i = 1.
bending moments, depending on the joint normalized spacing
and configuration. This can be seen by comparing the even
configurations in Figs. 5–7. For Lj/i = 1 and 2, the shift from
continuous to hinged behavior is associated with a reduction
in the maximum bending moments, whereas for Lj/i = 4 an
increase in the maximum bending moment is observed.
The importance of the joint configuration is a function of
the normalized spacing: for small normalized spacing (i.e.,
Lj/i = 1), the overall decrease in bending moment is similar
for the even and odd configurations; whereas for the larger
Lj/i of 2 and 4, there is a significant difference in bending
moment for the even and odd configurations. The pipeline
displacement follows more closely to the green field settle-
ment trough while the normalized joint spacing decreases,
approaching more and more a piecewise linear fit.
The design criteria for jointed pipelines that are influ-
enced by nearby tunnel excavation may be related to both
rotations at the joints and bending moments along the pipe.
A jointed pipeline behaves between the two limit cases of a
continuous and a hinged pipeline. For a given jointed pipe-
line with either odd or even configuration, the maximum
bending moment and the maximum joint rotation are a func-
tion of the relative rigidities R and T. Figures 8–10 show
normalized solutions for the maximum bending moments
and the joint rotations along the pipeline as a function of
both R and T.
For a given T, the solution is closer to the hinged behav-
ior at small R, but it becomes close to the continuous pipe-
Can. Geotech. J. Vol. 45, 2008
line behavior at large R. For very small R values in even
configuration, the joints are too far away from the tunnel
centerline to influence the value of the maximum bending
moment (located above the tunnel centerline), and the pipe-
line bends according to the green field displacements.
As noted earlier and can also be seen in Fig. 10, a hinged
pipeline may result in higher bending moments than a con-
tinuous one. This happens in the case of the even configura-
tion when the two innermost joints are located close to the
area where a continuous pipeline would experience hogging
moments. The critical condition for which the hinged solu-
tion may result in higher bending moments than the continu-
ous one is given in Fig. 11. Figure 11a shows the ratio of
maximum bending moment in a hinged system to that in a
continuous one as a function of the normalized joint spacing
Lj/i for varying R values. Figure 11b recognizes the points
on Fig. 11a where the hinged system results in a higher
bending moment and creates a boundary line relating to
joint spacing and soil–pipe relative rigidity; an example is
given for the point of R = 3.16. Although a point (Lj/i versus
R) in Fig. 11b may be located to the right of the critical con-
dition line, it does not mean that the hinged solution will re-
sult in higher bending moments than the continuous one, as
the innermost joints may be farther away from the hogging
position.
It is worth noting that Lj/i = 4 is an extreme case because
precast sections of pipes are rarely long enough to result in
this ratio unless the pipeline is very close to the tunnel in
# 2008 NRC Canada
Klar et al.
Fig. 9. Normalized solution Lj/i = 2.
which case the value of i is relatively small. In such a case,
there may be significant tunnel–pipeline interactions that do
not follow the assumptions used to derive these solutions,
and it may be necessary to perform a more rigorous interac-
tion analysis using a finite element method, for example.
Nonetheless, this extreme case is included here to demon-
strate the full range of potential behavior of pipelines.
Use of normalized graphs (Figs. 8–10)
Although the normalized graphs (Figs. 8–10) allow for a
wide range of input parameters, they do not cover all possi-
ble cases. That is, the tunnel centerline does not necessarily
have to lie symmetrically between joints (i.e., the even and
odd configuration). In addition, if the joints exceed the small
rotation region (see Fig. 2), each one will have a different
stiffness, but the given normalized solutions assume equal
stiffness values for all joints. However, since the exact loca-
tions of the joints are rarely known, the worst case scenarios
need to be considered and these are captured by the solu-
tions of the odd and even geometrical configurations pro-
vided within the normalized graphs above. Additionally,
engineers should be cautious when applying the presented
solutions to problems giving normalized parameters outside
the ranges, as shown here.
To obtain an estimation of the maximum bending moment
and rotation values for a general jointed pipeline system, one
may use the normalized solution in the following manner.
137
For a given Lj/i and R, if a hinged (zero stiffness joint)
system results in smaller bending moments than a continu-
ous one (left side of line in Fig. 11b), the small rotation
stiffness may be used to evaluate the maximum bending
moment. To evaluate the maximum rotation value, one may
use an iterative procedure, where a linear equivalent rota-
tional stiffness is applied to all joints based on the maxi-
mum value of rotation in the previous iteration. If the
joints do not exceed the small rotation region, no iteration
is required and the solution of the bending moment and ro-
tation may be considered correct. If the joint rotation ex-
ceeds the small rotation region, this iterative procedure is
considered to provide a conservative estimation of the cor-
rect values.
In the rare geometrical configuration where a hinged sys-
tem may result in greater bending moments than a continu-
ous system (right of the line in Fig. 11b), bending moments
shall also be evaluated based on the hinged system solu-
tions, although the difference between the hinged and con-
tinuous solutions will be marginal.
Summary and conclusions
This study addresses the problem of soil–pipe interaction
due to tunneling by extending the previous published contin-
uum solution to include joints within the pipeline. The re-
sults are given in normalized graphs, as a function of the
# 2008 NRC Canada
138
Fig. 10. Normalized solution Lj/i = 4.
Fig. 11. Boundary line between critical hinged–continuous system.
Can. Geotech. J. Vol. 45, 2008
Gaussian settlement trough parameters Smax and i, the joint
spacing Lj/i, the relative pipe–soil stiffness R = EI/Esr0i3,
and the relative joint-pipe stiffness T = kjM/(EI/i).
When a tunnel excavation induces ground settlements to
pipelines that lie above, a jointed pipeline experiences a
smaller maximum bending moment than a continuous one.
However, there are certain cases where hinged systems re-
sult in a greater bending moment than continuous ones and
these cases, even though rare, are classified within this note.
The given solutions assume linear elasticity of the soil. In
reality, soils rarely behave this way, and therefore the solu-
tions must be considered with judgement. The value of soil
stiffness should be selected with care considering the stiff-
ness degradation due to tunnel-induced ground settlement.
Moreover, the solutions are based on additional input param-
eters involved in the description of the soil settlement trough
in a green field (Smax, i). The effect of these parameters on
the bending moment and joint rotation is even greater than
that of the soil stiffness. It is therefore recommended that a
parametric study, covering the possible range of input pa-
rameters, be undertaken in any engineering use of the solu-
tions presented in this paper.
Acknowledgments
The authors wish to express their gratitude to the Natural
Sciences and Engineering Research Council of Canada
(NSERC) and the Cambridge-MIT Institute (CMI) for their
financial support.
# 2008 NRC Canada
Klar et al. 139

References List of symbols

Attewell, P.B., Yeates, J., and Selby, A.R. 1986. Soil movements EI bending stiffness of the pipe
induced by tunnelling and their effects on pipelines and struc- Es Young’s modulus of the soil
tures. Blackie & Son Ltd., London, UK. F nodal force
BS EN 545. 2002. Ductile iron pipes, fittings, accessories and their Gi,j Green’s function (flexibility coefficient of point i to j)
joints for water pipelines – requirements and test methods. Brit- i distance to the inflection point
ish Standard Institute. [KB] beam element matrix
[KJ] joint element matrix
BS EN 14161. 2003. Petroleum and natural gas industries – pipe-
kjM rotational stiffness of the joint
line transportation systems. British Standard Institute. kjS shear stiffness of the joint
Klar, A., Vorster, T.E.B., Soga, K., and Mair, R.J. 2005. Soil–pipe Lj spacing between joints
interaction due to tunnelling: comparison between Winkler and Le length of the beam elements
elastic continuum solutions. Géotechnique, 55: 461–466. M bending moment
Mindlin, R.D. 1936. Forces at a point in the interior of a semi- Mmax maximum bending moment along the pipeline
infinite soild. Physics, 7: 195–202. doi:10.1063/1.1745385. R relative pipe–soil rigidity factor
Peck, R.B. 1969. Deep excavation and tunneling in soft ground. In r0 radius of the pipe
Proceedings of the 7th International Conference on Soil Me- [S] global stiffness matrix of the pipeline
chanics and Foundation Engineering, Mexico City, Mexico. Smax maximum green field settlement at the pipeline level
pp. 266–290. Sp pipeline vertical displacement
Vorster, T.E.B. 2005. The effect of tunnelling on buried pipes. Sv Green field vertical settlement at the pipeline depth
Ph.D. thesis, Engineering Department, University of Cambridge, T relative joint-pipe stiffness
Cambridge, UK. {u} pipeline displacement
Vorster, T.E.B., Klar, A., Soga, K., and Mair, R.J. 2005. Estimating {uCAT} additional continuum displacement due to tunnel
the effects of tunneling on existing pipelines. Journal of Geo- excavation
x transverse distance from the tunnel centerline
technical and Geoenvironmental Engineering, 131: 1399–1410.
Z depth to pipe centerline
doi:10.1061/(ASCE)1090-0241(2005)131:11(1399).
1, 2 small and large rotational stiffness of the joint
Weaver, W., and Gere, J.M. 1990 Matrix analysis of framed struc-
joint rotation (rad)
tures. Van Nostrand Reinhold, New York.
max maximum joint rotation along the pipeline

s limit of small rotation stiffness

# 2008 NRC Canada