See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236143558

Surface settlements due to deformation of a tunnel in an elastic half plane

Article  in  Géotechnique · January 1996

DOI: 10.1680/geot.1998.48.5.709

CITATIONS READS

200 1,830

2 authors, including:

Arnold Verruijt
Delft University of Technology
108 PUBLICATIONS   2,526 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

SoilMechanics View project

All content following this page was uploaded by Arnold Verruijt on 30 May 2014.

The user has requested enhancement of the downloaded file.

Verruijt, A. & Booker, J. R. (1996). GeÂotechnique 46, No. 4, 753±756

TECHNICAL NOTE

Surface settlements due to deformation of a tunnel in an elastic half

plane

A . V E R RU I J T  a n d J. R . B O O K E R {

KEYWORDS: deformation; elasticity; stress analysis;

tunnels.

INTRODUCTION
Deformations of a tunnel may result in settlements
of the ground surface. In engineering practice these
surface settlements are often described by empiri-
cal formulae, based upon ®eld observations, for
instance a normal (Gaussian) distribution curve
(Peck, 1969; Attewell & Woodman, 1982). In this
note an approximate analytical solution is presented,
considering the soil as a linear elastic material.
Although it is realized that this is a very poor
schematization of the real behaviour of soils, an
elastic solution may well serve to investigate some
of the characteristics of the resulting ®elds of stress
and strain, and it may also serve as a reference for
more re®ned (numerical) computations.
Two basic deformation mechanisms of the
tunnel are considered: a uniform radial displace-
ment (representing, in ®rst approximation, the
ground loss that may occur during construction
of the tunnel), and an ovalization of the tunnel,
see Fig. 1. Even though in practice the tunnelling
process may be executed very carefully and
appropriate engineering techniques may be applied
to minimize the deformations, for instance the
injection of grout into the soil surrounding the
tunnel, it remains of interest to study the defor-
mations and stresses caused by both ground loss
and ovalization.
The method used is an extension of a method
suggested by Sagaseta (1987) for the case of
ground loss in an incompressible soil. The starting
point is the solution for a singularity at a point of
an elastic half plane (at the axis of the tunnel). By
adding the image solution for a singularity at a
Manuscript received 4 July 1995; revised manuscript
accepted 30 October 1995.
Discussion on this technical note closes 3 March 1997;
for further details see p. ii.
 Delft University of Technology
{ University of Sydney
Fig. 1. Ground loss and ovalization of a tunnel
point located symmetrically above the soil surface,
the shear stresses at the surface are made equal to
zero. In order to balance the normal stresses at the
surface, a third solution is added, by solving a
Boussinesq-type problem. The approximation in
establishing the solution is that, in balancing the
normal and tangential stresses on the surface, the
presence of the tunnel, with its prescribed dis-
placements, is disregarded. Although an exact
solution might well be obtainable by the methods
outlined by Mindlin (1939), using bipolar coordi-
nates, such a solution is very complicated. The
present approximate solution will be seen to be
very simple and easy to apply.
The solution given in this technical note is a
generalization of Sagaseta's solution in that it gives
the solution for the case of ground loss not only
for the incompressible case (with Poisson's ratio
equal to 0´5), but for arbitrary values of Poisson's
ratio, and that it includes the effect of ovalization.
The process of ovalization has been studied before
for a tunnel in an in®nite elastic medium (Muir
Wood, 1975). Here the surface displacements for a
tunnel in a semi-in®nite medium are considered, as
well as the displacements and stresses throughout
the half space.
THE SOLUTION
The (approximate) solution of the problem is
supposed to consist of three parts. The ®rst two

753
754 VERRUIJT AND BOOKER

parts are singular solutions from the theory of stresses ózz induced by the two singular solutions
elasticity in the points x ˆ 0, z ˆ h and x ˆ 0, are equal, and add up to
z ˆ ÿh, (Fig. 2). In these solutions the displace-
ments are of order O(1/r), so that they correspond x2 ÿ h2
z ˆ 0 : ó zz ˆ q(x) ˆ ÿ4ìåR2
to the removal of a ®nite volume, and the stresses (x2 ‡ h2 )2
are of order O(1/r2 ). They are well-known basic
8ìmäR2 h2 (3x2 ÿ h2 )
solutions from the theory of elasticity (Timoshenko ÿ (3)
& Goodier, 1951). All other possible singular m ‡ 1 (x2 ‡ h2 )3
solutions are of larger negative orders of r, and
where ì is the shear modulus of the elastic
thus will be small compared to the present one if
material, and m is an auxiliary elastic constant,
the radius R of the tunnel is small enough.
related to Poisson's ratio by
Taken together, the expressions for the displace-
ment components ux and uz for the singular parts 1
of the solution and their images are mˆ (4)
1 ÿ 2í
 
x x In order to satisfy the boundary condition that the
ux ˆ ÿåR2 2 ‡ 2
r r normal stress ózz ˆ 0 at the surface z ˆ 0, a third
1 2 2 2  solution must be added, which must balance the
x(x ÿ kz1 ) x(x2 ÿ kz22 )
‡ äR2 ‡ (1) stress distribution q(x). This third part of the
r41 r42 solution is the solution of a Boussinesq problem,
 
z1 z2 for the half plane z . 0, with the boundary
uz ˆ ÿåR2 2 ‡ 2 conditions
r1 r2
 2  z ˆ 0 : ózz ˆ ÿq(x), ózx ˆ 0: (5)
2
2 z1 (kx ÿ z1 ) z2 (kx2 ÿ z22 )
‡ äR ‡ (2)
r41 r42 Problems of this type can be solved most
conveniently by using Fourier transforms (Sneddon,
where å and ä are parameters indicating the relative 1951). It can be shown that, if the load function is
displacement of the tunnel surface, for the uniform even, q(ÿx) ˆ q(x), the solution of the problem
radial displacement case (å) and the ovalization with the boundary conditions (5) can be written as
case (ä) respectively. In these solutions z1 ˆ z ÿ h, …1
z2 ˆ z + h, and r1 and r2 are the distances from the ux ˆ C(1 ÿ máz) exp (ÿáz) sin (áx) dá (6)
singular point and its image (Fig. 2). It can easily 0
be veri®ed that these solutions satisfy the basic …1
differential equations from the theory of plane uz ˆ ÿ C(1 ‡ m ‡ máz) exp (ÿáz) cos (áx) dá
strain elasticity. 0

Because of the symmetry of these two solutions, (7)

the shear stress ózx and the vertical displacement
where C is de®ned by the integral
uz will be zero at the surface z ˆ 0. The normal
…
1 1 1
Cˆÿ q(x) cos (áx) dx (8)
mðì á 0

−h In this case, with the function q(x) de®ned by (3)

the function C is found to be

2åR2 2äR2
Cˆÿ exp (ÿáh) ÿ áh exp (ÿáh)
m m‡1
x (9)
r2
Substitution of this result into equations (6) and (7)
r1
gives, after evaluation of the Fourier integrals, using
known integrals from Bateman (1954) and some
h results that can easily be derived from them,
 
2åR2 x 1 2mzz2
ux ˆ ÿ ÿ
m r22 r42
 
z 4äR2 xh z2 mz(x2 ÿ 3z22 )
ÿ ‡ (10)
Fig. 2. A singularity and its image m ‡ 1 r42 r62
 SURFACE SETTLEMENTS DUE TO TUNNEL 755
 
2åR2 (m ‡ 1)z2 mz(x2 ÿ z22 ) interesting to note that for other values of í the
uz ˆ ÿ total area below the settlement curve is larger than
m r22 r42
 2  the ground loss. This initially somewhat surprising
x ÿ z22 m 2zz2 (3x2 ÿ z22 ) result must be attributed to the deformations at
ÿ 2äR2 h ‡
r42 m‡1 r62 in®nity. For í ˆ 0 the total area is double that for
(11) an incompressible material, and thus Sagaseta's
solution will in general lead to an underestimation
The complete solution of the problem is the sum of of the settlement under drained conditions. It may
this solution and the expressions (1) and (2). also be concluded that it is of great practical
All stresses and strains can be directly derived importance to keep the ground loss as small as
from the solution given above. The vertical possible, in order to prevent surface settlements, a
displacement of the surface z ˆ 0, denoted as u0 , fact long recognized in tunnel construction.
is of particular interest. Taking z ˆ 0 in equation The shape of the settlement curve is indepen-
(11) gives dent of í, but its values increase if í decreases
2 2
from 0´5 to 0. In a consolidating, porous, elastic
m‡1 h 2 h(x ÿ h ) medium this would mean that the immediate
u0 ˆ 2åR2 2
ÿ 2äR
m x2 ‡ h (x2 ‡ h )2
2
settlement is equal to the ground loss, with the
(12) consolidation settlement leading to a gradual but
approximately conformal increase of the settle-
This is the total settlement of the surface, because ments. This will be the subject of a subsequent
for z ˆ 0 the two singular solutions in equation (2) publication.
cancel. It may also be interesting to note that the
The shape of the two settlement functions is average value of the ovalization term in the
shown in Fig. 3. The solid line is the settlement solution is zero. In the zone x , h there is a
curve for the ®rst term, due to the ground loss. settlement, but in the zone x . h there is some
The dashed line is the settlement curve for the heave, and the two of these appear to balance on
second term, due to the ovalization of the tunnel. average.
It appears that the width of the settlement trough In order to assess the accuracy of the approxi-
is considerably smaller for the ovalization case mation, the total displacements at the tunnel's
than for the ground loss case. This may be radial circumference may be determined. The ®rst
one explanation for the rather narrow settlement term of the solution represents the constant radial
troughs usually observed in tunnelling practice. displacement due to the singularity in an in®nite
Another explanation may of course be that the soil medium. The other two solutions disturb this
behaviour is non-linear, exhibiting plastic effects. displacement, and the relative magnitude of this
The factor (m + 1)/m in equation (12) can also disturbance is a measure of the accuracy of the
be written as 2(1 ÿ í). For í ˆ 0´5 this factor is 1. solution. This can be evaluated by comparing the
For that value of í the ®rst term of the solution radial displacements at the tunnel face due to the
has been presented earlier by Sagaseta (1987). second and the third parts of the solution with
The total area (A) of the settlement trough is those due to the ®rst part. It appears that the error
found by integrating equation (12) from ÿ1 to is less than 10%, provided that the radius of the
+1. The result is tunnel is less than one half of its depth.
m‡1
Aˆ 2åðR2 ˆ 4(1 ÿ í)åðR2 (13)
m STRESSES
For í ˆ 0´5 this is precisely the ground loss, 2åðR2 , The strain components can easily be derived by
a result also mentioned by Sagaseta (1987). It is differentiation from the expressions for the dis-
placements, and then the stress components can be
obtained with the aid of Hooke's law. The isotropic
−5 0 5 stress ó0 ˆ 12(óxx + ózz ) is found to be, for instance,
x/h

z22 ÿ x2
ó0 ˆ ÿ4ìåR2
r42
2  4
äR z1 ÿ x4 z42 ÿ x4
‡ ‡
1ÿí r61 r62

4hz2 (z22 ÿ 3x2 )
ÿ (14)
u0
r62

Fig. 3. Surface settlements A graphical representation of the isotropic stress ó0

756 VERRUIJT AND BOOKER
is shown in Fig. 4. The left half of the ®gure shows
contours of ó0 for the ground loss problem, and the
right half shows contours of ó0 for the ovalization
problem. The thick lines indicate the contours for
ó0 ˆ 0, and the characters + and ÿ indicate zones
of tension and compression respectively. The con-
tour intervals are 0´2ìå(R/h)2 and 0´2ìä(R/h)2 /
(1 ÿ í). Although in Fig. 4 it has been assumed
that R/h ˆ 0´5, the contours are valid for any value
of R/h. It is interesting to note that in the case of
ground loss the stresses around the tunnel are all
compressive, whereas in the case of ovalization
there are zones of tension above and below the
tunnel. The construction of contours of other stress
Fig. 4. Contours of isotropic stress
components is a relatively simple and straight-
forward exercise.
CONCLUSIONS
It appears to be possible to derive approximate
analytic solutions for two important cases of tunnel
deformation in an elastic soil: a uniform radial
displacement due to ground loss, and an ovaliza-
tion of the tunnel. Simple analytical expressions
can be given for the displacements at arbitrary
points of the half plane, and for the stress com-
ponents.
REFERENCES
Attewell, P. B. & Woodman, J. P. (1982). Predicting the
dynamics of ground settlement and its derivatives by
tunnelling in soil. Ground Engineering 15, 13±22.
Bateman, H. (1954). Tables of Integral Transforms. New
York: McGraw-Hill.
Mindlin, R. D. (1939). Stress distribution around a
tunnel. Proc. ASCE 65, 619±642.
Muir Wood, A. M. (1975). The circular tunnel in elastic
ground. GeÂotechnique 25, 115±127.
Peck, R. B. (1969). Deep excavations and tunneling in
soft ground, state of the art report. 7th Int. Conf. on
Soil Mech. and Fdn Eng., Mexico City, 225±290.
Sagaseta, C. (1987). Analysis of undrained soil deforma-
tion due to ground loss. GeÂotechnique 37, 301±320.
Sneddon, I. N. (1951). Fourier Transforms. New York:
McGraw-Hill.
Timoshenko, S. P. & Goodier, J. N. (1951). Theory of
Elasticity, 2nd edn. New York: McGraw-Hill.
Surface settlements due to deformation of a tunnel in an elastic half
plane
A. Verruijt and J.R. Booker

Géotechnique 46 (1996) 753-756

Erratum

In equations (1) and (2) : k = ν/(1 − ν).

 Surface settlements due to ground loss
and ovalization of a tunnel
A. Verruijt and J.R. Booker
Géotechnique, 46, 753-756, 1996.
Details of the analysis, and an alternative

The problem
The paper gives an approximate analytic solution for a tunnel in a homogeneous elastic half
space, caused by ground loss and ovalization. The two basic processes are illustrated in
figure 1. This note presents some additional material and derivations.
.......................................................................................................................................................................................................................................................................
.... ..... ...... .... ..... ...... .... ..... ...... .... ..... ...... .... ..... .... ...... ..... .... ...... ..... .... ...... ..... .... ...... ..... ...
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
... ... .... ... ... .................. ...... ... . ... .... ... ... ... .... ... ...... .......................... ... ... .... ... ..
... ... .... ... ... ..................... .................. .... .... ... ... ... .... ...................... ... .. .................... . . . . . . . .
... ... .... ... ............ .... . . . . . . . . ........ ......... ... .... ... ..
............ ... ... ... . ... ..... . . . . . . .
... ... .... ... ........ ... . . . . . . . ..... . .... . . . . . .
... ... .... .......... ... ...... ... ... ...... . .. ....... ...... ..... ...
... ... .... ... ... .. .. .. ... ... ... . .
. . . . .... .... ... ..
... ... .... ......... . ..... ... ... .... ...
.
. ... ..... .... ... ..
... ... .... ........ .. ... . . . . . .. . . .. . .
... ... .... ........... ............ .... ..... .......... ... ...... .... ... ..
. .
... ... .... ... ... ..... .
... .. . . . . . . ...... . .. ... .... ... ..
. .
... ... .... ... ...............
... ............. ...... .... ..... .... ...............
. . .. ............. .... ...... ..... ...
... ... .... ... ... ........ ............................ ........ .... ... ... ... .... ..................................................... ... ... .... ... ..
... ... .... ... ... ............... ............... ... .... ... ... ... .... ... ............ ...... ...... .... ... ... .... ... ..
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .

Figure 1: Ground loss and Ovalization of a tunnel.

Basic solution for ground loss

The first solution, describing the ground loss, is a purely radial displacement, decreasing
inversely proportional to the distance r from the point x = 0, y = 0.
εR2
ur = − , (1)
r
where ur is the displacement in radial direction, and ε is a parameter indicating the relative
displacement at the circumference of a tunnel of radius R. The total ground loss (per unit
length) is 2επR2 , and the relative ground loss, per unit volume of the tunnel, is 2ε. The
horizontal and vertical components of the displacement are
εR2 x
ux = − , (2)
r2
εR2 y
uy = − , (3)
r2
It can easily be verified that these expressions satisfy the basic differential equations for a
homogeneous isotropic linear elastic material, deforming under plane strain conditions,
∂e ∂ 2 ux ∂ 2 ux
m + + = 0, (4)
∂x ∂x2 ∂y2

1
 ∂e ∂ 2 uy ∂ 2 uy
m + 2
+ = 0, (5)
∂y ∂x ∂y2
where e is the volume strain,
∂ux ∂uy
e= + , (6)
∂x ∂y
and the elastic parameter m is defined by
1
m= , (7)
1 − 2ν
where ν is Poisson’s ratio.
In this case the volume strain appears to be zero,
e = 0. (8)
The stresses can be determined using Hooke’s law for the case of plane strain,
σxx ∂ux ν ∂ux
= 12 (m − 1) e + = e+ , (9)
2µ ∂x 1 − 2ν ∂x
σyy ∂uy ν ∂uy
= 12 (m − 1) e + = e+ , (10)
2µ ∂y 1 − 2ν ∂y
σxy 1 ∂ux ∂uy

= 2
+ . (11)
2µ ∂y ∂x
This gives
σxx εR2
= 4 (x2 − y2 ), (12)
2µ r

σyy εR2
= 4 (y2 − x2), (13)
2µ r

σxy εR2
= 4 2xy. (14)
2µ r

Basic solution for ovalization

The basic form of the second type of singularity, describing the ovalization, is
δR2x(x2 − ky2 )
ux = , (15)
r4
δR2y(kx2 − y2 )
uy = , (16)
r4
where δ is a measure of the degree of ovalization (the relative increase and decrease of the
radius R in the two coordinate directions), and where k is an elastic parameter, defined as
ν λ
k= = . (17)
1−ν λ + 2µ
Again it can be verified that this solution satisfies the basic differential equations (4) and (5).

2
In this case the volume strain is
∂ux ∂uy δR2 δR2 1 − 2ν 4
e= + = 6 (k − 1)(x4 − y4 ) = − 6 (x − y4 ). (18)
∂x ∂y r r 1−ν
The stresses are, with equations (9), (10) and (11),
σxx δR2
=− x2 (x2 − 3y2 ), (19)
2µ (1 − ν)r6

σyy δR2
=− y2 (3x2 − y2 ), (20)
2µ (1 − ν)r6

σxy δR2
=− 2xy(x2 − y2 ). (21)
2µ (1 − ν)r6

Solution for two singularities

...
...
...
...
...
.
.....
−h .............
......
... ...
... ...
... ...
.... ...
... ...
... ...
... ...
..
... ..
................................................................................................................................................................................................................................ x
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ...... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ........ .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ....... ... ... .... ............. .2 r . .. . .. .
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ............ . ...... .... ..... .... ...
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ............ .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... ... ............ ...1.... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... .......... ... ... .... ... ... ... ..
r
. ... ... ... .... ... ... .... ... ... ....... ............ .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... . ............. ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ................. ... .... ... ... .... ... ... ... ..
h
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ..... ... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ....... ... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ... ... .... ... ... ... ..
...
...
..
.......
.....
y .

Figure 2: A singularity and its image.

Superposition of two singularities, with ground loss and ovalization, at y = h and at y = −h,
see figure 2, leads to the displacement field
x x  2 2
2 x(x − ky1 ) x(x2 − ky22 )
ux = −εR2 + + δR + , (22)
r12 r22 r14 r24

 y1 y2  y1(kx2 − y12 ) y2 (kx2 − y22 )

uy = −εR2 2
+ 2 + δR2 + , (23)
r1 r2 r14 r24

where y1 = y − h and y2 = y + h, and r1 and r2 are the distances from the singular point
and its image, see figure 2, so that
r12 = x2 + (y − h)2, r22 = x2 + (y + h)2 . (24)
The stresses due to the two singularities are
σxx  x2 − y12 x2 − y22 δR2  x2 (x2 − 3y12 ) x2(x2 − 3y22 )
= εR2 4 + 4 − + , (25)
2µ r1 r2 1−ν r16 r26

σyy  x2 − y12 x2 − y22 δR2  y12 (3x2 − y12 ) y22 (3x2 − y22 )
= −εR2 4 + 4 − + , (26)
2µ r1 r2 1−ν r16 r26

3
 σxy  2xy1 2xy2 δR2  2xy1 (x2 − y12 ) 2xy2(x2 − y22 )
= εR2 4
+ 4 − + . (27)
2µ r1 r2 1−ν r16 r26
At the surface y = 0: y1 = −h and y2 = h, so that it follows from equations (23) and (27)
that on that surface uy = 0 and σxy = 0. This is a consequence, of course, of the symmetry
of the solution. The normal stresses σyy induced by the two parts of the solution are equal,
and add up to
x2 − h2 4µδR2 h2(3x2 − h2 )
y = 0 : σyy = q(x) = −4µεR2 − , (28)
(x2 + h2 )2 1 − ν (x2 + h2 )3
In order to satisfy the boundary condition that the normal stress σyy = 0 at the surface
y = 0, a third solution must be added, which should balance the stress distribution q(x).

The third part of the solution

The third part of the solution is a problem for an elastic half plane with zero shear stresses
at the surface y = 0, and normal stresses
x2 − h2 4µδR2 h2(3x2 − h2 )
y = 0 : σyy = −q(x) = 4µεR2 + , (29)
(x2 + h2 )2 1 − ν (x2 + h2 )3
This is a Boussinesq problem, which can be solved using the Fourier transform method.

The Fourier transform method

The Fourier transform method for solving problems of elasticity for a half plane (Sneddon,
1951) leads to the general type of solution
Z ∞
ux = C(1 − mαy) exp(−αy) sin(αx) dα, (30)
0
Z ∞
uy = − C(1 + m + mαy) exp(−αy) cos(αx) dα, (31)
0
where, as before, m is defined by
1
m= . (32)
1 − 2ν
It can easily be verified that these equations satisfy the basic differential equations (4) and
(5).
It follows from equations (30) and (31) that
Z ∞
∂ux ∂uy
e= + =2 Cα exp(−αy) cos(αx) dα, (33)
∂x ∂y 0
Z ∞
∂ux ∂uy
f= − = −2m Cα2y exp(−αy) cos(αx) dα, (34)
∂x ∂y 0
Z ∞
∂ux ∂uy
g= + = 2m Cα2y exp(−αy) cos(αx) dα. (35)
∂y ∂x 0
Using these equations the stresses can be expressed as
Z ∞
σxx 1 1
= 2 me + 2 f = m Cα(1 − αy) exp(−αy) cos(αx) dα, (36)
2µ 0
Z ∞
σyy
= 12 me − 12 f = m Cα(1 + αy) exp(−αy) cos(αx) dα, (37)
2µ 0
Z ∞
σxy
= 12 g = m Cα2y exp(−αy) cos(αx) dα. (38)
2µ 0

4
Ground loss
In the present problem the boundary condition for the stress σyy is, with equation (29), and
considering the ground loss case only,
σyy x2 − h2
y=0 : = 2εR2 2 . (39)
2µ (x + h2)2
With equation (37) this leads to the condition
Z ∞
2εR2 x2 − h2
Cα exp(−αy) cos(αx) dα = . (40)
0 m (x2 + h2)2
Using the inversion theorem for the Fourier cosine integral (Sneddon, 1951) it follows that
Z
4εR2 ∞ x2 − h2
Cα = cos(αx) dx. (41)
πm 0 (x2 + h2)2
Using the Fourier transform (114) from the Appendix gives
2εR2
C=− exp(−αh). (42)
m
Substitution into the expressions (30) and (31) for the displacements now gives
Z
2εR2 ∞
ux = − (1 − mαy) exp[−α(y + h)] sin(αx) dα, (43)
m 0
Z
2εR2 ∞
uy = (1 + m + mαy) exp[−α(y + h)] cos(αx) dα. (44)
m 0
These integrals are all given in the Appendix. The final results are, using the definition of
m, see equation (32),
n (1 − 2ν)x 2xyy o
2
ux = −2εR2 − , (45)
r22 r24
n 2(1 − ν)y y(x2 − y22 ) o
2
uy = 2εR2 − , (46)
r22 r24
where y2 = y + h and r22 = x2 + (y + h)2 .
With equation (42) the expressions for the stresses become
Z ∞
σxx
= −2εR2 α(1 − αy) exp(−αy2 ) cos(αx) dα, (47)
2µ 0
Z ∞
σyy 2
= −2εR α(1 + αy) exp(−αy2 ) cos(αx) dα, (48)
2µ 0
Z ∞
σxy 2
= −2εR α2y exp(−αy2 ) cos(αx) dα, (49)
2µ 0

where y2 = y + h.
All the integrals in these expressions are also given in the Appendix. This gives
σxx n x2 − y 2 2yy2 (3x2 − y22 ) o
2
= 2εR2 − , (50)
2µ r24 r26
σyy n x2 − y 2 2yy2 (3x2 − y22 ) o
2
= 2εR2 + , (51)
2µ r24 r26
σxy 2xy(x2 − 3y22 )
= 2εR2 . (52)
2µ r26

5
Ovalization
For the ovalization case the boundary condition for the stress σyy is, with equation (29) and
ε = 0,
σyy 2δR2 h2 (3x2 − h2 )
y=0 : = , (53)
2µ 1 − ν (x2 + h2 )3
With equation (37) this leads to the condition
Z ∞
1 − 2ν h2(3x2 − h2 )
Cα exp(−αy) cos(αx) dα = 2δR2 . (54)
0 1 − ν (x2 + h2 )3
Using the inversion theorem for the Fourier cosine integral (Sneddon, 1951) it follows that
Z
4δR2 1 − 2ν ∞ h2(3x2 − h2 )
Cα = cos(αx) dx. (55)
π 1−ν 0 (x2 + h2 )3
This integral is determined in the Appendix, see equation (116). In this case the result is
2δR2
C=− αh exp(−αh), (56)
m+1
in which the factor 2/(m + 1) can also be written as (1 − 2ν)/(1 − ν).
Substitution of (56) into the expressions (30) and (31) for the displacements now gives
Z ∞
2δR2
ux = − αh(1 − mαy) exp[−α(y + h)] sin(αx) dα, (57)
m+1 0
Z ∞
2δR2
uy = αh(1 + m + mαy) exp[−α(y + h)] cos(αx) dα. (58)
m+1 0
The Fourier integrals in these expressions are given in the Appendix. This gives
4δR2xh n y2 my(x2 − 3y2 )2 o
ux = − + , (59)
m + 1 r24 r26
n x2 − y 2 2m yy2 (3x2 − y22 ) o
2
uy = −2δR2h + . (60)
r24 m+1 r26
With equation (42) the expressions for the stresses become
Z
σxx 2δR2m ∞ 2
=− α h(1 − αy) exp(−αy2 ) cos(αx) dα, (61)
2µ m+1 0
Z
σyy 2δR2m ∞ 2
=− α h(1 + αy) exp(−αy2 ) cos(αx) dα, (62)
2µ m+1 0
Z
σxy 2δR2m ∞ 3
=− α hy exp(−αy2 ) cos(αx) dα, (63)
2µ m+1 0
where y2 = y + h.
All the integrals in these expressions are also given in the Appendix. This gives
σxx 2δR2h n y2 (3x2 − y22 ) 3y(x4 − 6x2 y22 + y24 ) o
= + , (64)
2µ 1−ν r26 r28
σyy 2δR2h n y2(3x2 − y22 ) 3y(x4 − 6x2y22 + y24 ) o
= − , (65)
2µ 1−ν r26 r28
σxy 24δR2h xyy2 (x2 − y22 )
= . (66)
2µ 1−ν r28

6
An alternative solution for ovalization
It has been observed by Strack (2002) that in the solution for the ovalization case presented in
the paper considered above, the tangential displacement ut does not vanish at the boundary
of the cavity, in contrast with his complex variable solution. For the purpose of comparison
an alternative may be used, in which ut = 0 along the cavity boundary. This alternative is
..
.
... ... .... ... ... .... ... ... .... ...... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ...... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ..... ... .... ... ... ... .... ... ... ..
. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . .
.................................................................................................................................................................................................................................................. x
... ... .... ... ... .... ... ... .... ......... ... .... ... ... ... .... ... ... ..
. . . . . . . . . . . . . . . .. .. . . . . . . . . . . . . . .
.... ..... ...... .... ..... ...... .... ..... ...... ................ ...... .... ..... .... ...... ..... .... ...
... ... .... ... ... .... ... ... .... ..... .......... ... ... ... .... ... ... ..
. . . . . . . . . . . . . . . .. . ....... . . . . . . . . . . . .
.... ..... ...... .... ..... ...... .... ..... ...... .................................. .... ..... .... ...... ..... .... ...
θ
. . . . . . . . . . . . . . . .. . . . .. . . . . . . . . . . .
.... ..... ...... .... ..... ...... .... ..... ...... ....... ..... .............. ..... .... ...... ..... .... ...
... ... .... ... ... .... ... ... .... ...... ... .... ...... ... ... .... ... ... ..
. . . . . . . . . . . . . . . .. . . . . . .. . . . . . . . . .
.... ..... ...... .... ..... ...... .... ..... ...... ....... ..... ...... ............. .... ...... ..... .... ...
... ... .... ... ... .... ... ... .... ..... ... .... ............. ... .... ... ... ..
... ... .... ... ... .... ... ... .... .... ... .... ... .... ... .... ... ... ..
... ... .... ... ... .... ... ... .... ...... ... .... ... ... ... .... ... ... ..
r
. . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . .
.
........
...
y
Figure 3: Cartesian and polar coordinates

presented here.
The general solution for ovalization of a circular cavity in an infinite elastic plane is, in
polar coordinates, see figure 3,
BR2 (m + 1) DR4
ur = { + } cos 2θ, (67)
mr 3r3
BR2 DR4
ut = {− + } sin 2θ, (68)
mr 3r3
where, as before, m = 1/(1 − 2ν).
The condition that ut = 0 for r = R gives
3B
D= . (69)
m
The solution now reduces to
BR2 R2
ur = {(m + 1) + 2 } cos 2θ, (70)
mr r
BR2 R2
ut = {−1 + 2 } sin 2θ. (71)
mr r
The relation between polar coordinates and cartesian coordinates is
ux = ur sin θ + ut cos θ, (72)

uy = ur cos θ − ut sin θ. (73)

7
With
BR δR
=− , (74)
m m+2
one obtains
δR2 x n y2 − x2 R2 (3y2 − x2) o
ux = 1 − m − , (75)
(m + 2)r2 r2 r4
δR2 y n y2 − x2 R2 o
uy = − 1 + m + 2 . (76)
(m + 2)r2 r2 r
It can easily be seen that
x = R, y = 0 : ux = δR, uy = 0, x = −R, y = 0 : ux = −δR, uy = 0, (77)

x = 0, y = R : ux = 0, uy = −δR, x = 0, y = −R : ux = 0, uy = δR, (78)

which is in agreement with the assumed ovalization of the circle of radius R.
It follows from equations (75) and (76) that
∂ux δR2 n y2 − x2 x4 − 6x2y2 + y4 3R2(x4 − 6x2y2 + y4 ) o
= − m − , (79)
∂x (m + 2)r2 r2 r4 r6
∂ux 2δR2xy n x2 − 3y2 6R2(x2 − y2 ] o
= 4
−1 + m 2
− , (80)
∂y (m + 2)r r r4
∂uy 2δR2xy n x2 − 3y2 2R2 o
= 1 − m + . (81)
∂x (m + 2)r4 r2 r2
∂uy δR2 n y2 − x2 x4 − 6x2y2 + y4 R2(x4 − 2x2y2 − 3y4 ) o
= + m − , (82)
∂y (m + 2)r2 r2 r4 r6
It now follows that
∂ux ∂uy 2δR2 n y4 − x4 2R2x2 (x2 − 5y2 ) o
e= + = − (83)
∂x ∂y (m + 2)r2 r4 r6
∂ux ∂uy 2δR2 n x4 − 6x2y2 + y4 R2(x4 − 8x2 y2 + 3y4 ) o
f= − =− m + (84)
∂x ∂y (m + 2)r2 r4 r6
∂ux ∂uy 8δR4 xy(2y2 − x2)
g= + = . (85)
∂y ∂x (m + 2)r4 r4
Using these relations the stresses are found to be
σxx δR2 n 2x2(x2 − 3y2 )
= 12 (me + f) = − m
2µ (m + 2)r2 r4
2R x (x − 5y2 ) R2(x4 − 8x2y2 + 3y4 o
2 2 2
+m + , (86)
r6 r6

σyy δR2 n 2y2 (y2 − 3x2)

= 12 (me + f) = m
2µ (m + 2)r2 r4
2R x (x − 5y2 ) R2(x4 − 8x2y2 + 3y4 o
2 2 2
−m + , (87)
r6 r6
σxy 4δR4 xy(2y2 − x2)
= 12 g = . (88)
2µ (m + 2)r4 r4

8
 ...
...
...
...
..
...
............
−h .......
.... ....
... ...
... ...
... ...
.... ...
.. ...
... ...
..
.... ..
.
................................................................................................................................................................................................................................
.
x
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ..... ... .... ... ... ... .. .
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ......... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ..... ... ... .... ............ .2 . r . .. . .. .
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ............ . ...... .... ..... .... ...
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ............ .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ....... ... ... . ........... ...1.... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... .......... ... ... .... ... ... ... .. .
r
.
. ... ... ... .... ... ... .... ... ... ...... .......... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... . ............... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ................. ... .... ... ... .... ... ... ... ..
h
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ....... ... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ... ... .... ... ... ... ..
. ... ... ... .... ... ... .... ... ... ...... ... ... .... ... ... .... ... ... ... ..
...
...
..
.......
....
y

Figure 4: A singularity and its image.

Solution for two singularities

Superposition of two of these ovalization singularities, at y = h and at y = −h, see figure 4,
leads to the displacement field
δR2x n 1 y12 − x2 R2(3y12 − x2) 1 y2 − x2 R2(3y22 − x2 ) o
ux = 2 −m 4 − 6 + 2 −m 2 4 − ,(89)
m + 2 r1 r1 r1 r2 r2 r26

δR2 n y1 y1 (y12 − x2 ) R2 y1 y2 y2 (y22 − x2) R2 y2 o

uy = − 2 +m 4 + 4 + 2 +m + 4 , (90)
m + 2 r1 r1 r1 r2 r24 r2

where y1 = y − h, y2 = y + h, r12 = x2 + y12 and r22 = x2 + y22 .

The stress field for these two singularities is
σxx δR2 n 2x2(x2 − 3y12 ) 2R2 x2(x2 − 5y12 ) R2(x4 − 8x2y12 + 3y14 )
=− m + m +
2µ m+2 r16 r18 r18
2 2
2x (x − 3y2 ) 2
2R x (x − 5y2 ) R (x − 8x y2 + 3y24 ) o
2 2 2 2 2 4 2 2
+m + m + , (91)
r26 r28 r28

σyy δR2 n 2y12 (y12 − 3x2) 2R2 x2(x2 − 5y12 ) R2(x4 − 8x2y12 + 3y14 )
= m − m +
2µ m+2 r16 r18 r18
2y2 (y2 − 3x2) 2R2 x2(x2 − 5y22 ) R2 (x4 − 8x2y22 + 3y24 ) o
+m 2 2 6 −m + , (92)
r2 r28 r28
σxy 4δR4x n y1 (2y12 − x2 ) y2 (2y22 − x2) o
= + . (93)
2µ m+2 r18 r28
At the free surface y = 0, so that y1 = −h, y2 = h and r12 = r22 = x2 + h2 . It then follows
that The vertical displacement uy and the shear stress σxy on that surface vanish,
y = 0 : uy = 0, (94)

y = 0 : σxy = 0. (95)

This is a consequence of the symmetry of the problem, of course.

The vertical normal stress on the surface y = 0 is found to be
y = 0 : σyy = 2µf(x), (96)

9
where now
2δR2 n 2h2(h2 − 3x2) 2R2x2 (x2 − 5h2) R2 (x4 − 8x2h2 + 3h4) o
f(x) = m − m + .(97)
m+2 (h2 + x2 )3 (h2 + x2 )4 (h2 + x2)4
This can also be written as
2δR2 n h2 h4
f(x) = −6m 2 2 2
+ 8m 2
m+2 (h + x ) (h + x2)3
R2 R2 h2 R2h4
−2m 2 2 2
+ 14m 2 2 3
− 12m 2
(h + x ) (h + x ) (h + x2 )4
R2 R2h2 R2h4 o
+ 2 2 2
− 10 2 2 3
+ 12 2 . (98)
(h + x ) (h + x ) (h + x2 )4
Written in this form the numerators no longer contain terms dependent upon x. This makes
the Fourier integrals somewhat easier to determine.

The third part of the solution

The third part of the solution is a problem for an elastic half plane with zero shear stresses
at the surface y = 0, and normal stresses
y = 0 : σyy = −2µf(x). (99)
This is a Boussinesq problem, which can be solved using the Fourier transform method.
With equation (37) this leads to the condition
Z ∞
m Cα cos(αx) dα = −f(x). (100)
0

Application of the inversion theorem for the Fourier cosine integral (Sneddon, 1951) now
gives
Z ∞
2
Cα = − f(x) cos(αx) dα. (101)
mπ 0
Using some of the Fourier integrals derived in the Appendix the value of C is found to be
4δR2 n 1 2 R2
C=− (αh) + [2 + 2(αh) + (αh)2 − (αh)3 ]
(m + 2)αh 2 8h2
R2 o
2 3
+ [2 + 2(αh) + (αh) + (αh) ] exp(−αh). (102)
8mh2
Although this form may be used to derive the displacements and stresses for the third part
of the solution, it is assumed here that the radius of the tunnel is small, R/h  1. In that
case equation (102) can be approximated by
2δR2
C=− αh exp(−αh). (103)
m+2
This is the same solution as obtained in the original problem, see equation (56), except that
the factor m + 1 in the denominator must be replaced by m + 2.
It follows that the additional expressions for the displacements in this case are
4δR2xh n y2 my(x2 − 3y2 )2 o
ux = − + , (104)
m + 2 r24 r26

10
 2δR2h n x2 − y22 yy2 (3x2 − y22 ) o
uy = − (m + 1) 4
+ 2m . (105)
m+2 r2 r26

And the expressions for the stresses are

σxx 4mδR2h n y2 (3x2 − y22 ) 3y(x4 − 6x2y22 + y24 ) o
= + , (106)
2µ m+2 r26 r28

σyy 4mδR2h n y2 (3x2 − y22 ) 3y(x4 − 6x2y22 + y24 ) o

= − , (107)
2µ m+2 r26 r28

σxy 48mδR2h xyy2 (x2 − y22 )

= . (108)
2µ m+2 r28

Reference
O.E. Strack, Analytic Solutions of Elastic Tunneling Problems, Ph.D. Thesis, Delft, 2002.

11
APPENDIX : SOME INTEGRAL TRANSFORMS
In this appendix a number of integral transforms are assembled, mainly from the literature.
Many of these integrals are used in this paper.

A well known Laplace transform (Churchill, 1972), which can easily be derived by using
partial differentiation, is
Z ∞
s
cos(at) exp(−st) dt = 2 . (109)
0 s + a2

By a simple change of notation this integral can also be written as a Fourier cosine transform,
Z ∞
h
exp(−αh) cos(αx) dα = 2 . (110)
0 h + x2

Using the Fourier inversion theorem (Sneddon, 1951), it now follows that
Z
2 ∞ h
cos(αx) dx = exp(−αh), (111)
π 0 h2 + x2
or
Z ∞
1 π
cos(αx) dx = exp(−αh). (112)
0 h2 + x2 2h

Differentiation of (112) with respect to h gives, after division by −2h,

Z ∞
1 π
2 + x2 ) 2
cos(αx) dx = 3 (1 + αh) exp(−αh). (113)
0 (h 4h

Because h2 − x2 = 2h2 − (h2 + x2 ) it follows from (112) and (113) that

Z ∞
h2 − x2 πα
2 + x2 ) 2
cos(αx) dx = exp(−αh). (114)
0 (h 2

Differentiation of (113) with respect to h gives, after division by −4h,

Z ∞
1 π
2 + x2 ) 3
cos(αx) dx = 5
[3 + 3αh + (αh)2] exp(−αh). (115)
0 (h 16h

Because h2 − 3x2 = 4h2 − 3(h2 + x2) it follows from (113) and (115) that
Z ∞ 2
h − 3x2 πα2
cos(αx) dx = exp(−αh). (116)
0 (h2 + x2)2 4h

Differentiation of (115) with respect to h gives, after division by −6h,

Z ∞
1 π
2 + x2 ) 4
cos(αx) dx = 7
[15 + 15αh + 6(αh)2 + (αh)3 ] exp(−αh). (117)
0 (h 96h

Now returning to the second integral, eq. (110), it follows, by differentiation with respect to
h, that
Z ∞
h2 − x2
α cos(αx) exp(−αh) dα = 2 . (118)
0 (h + x2)2

12
 Differentiating this equation once more with respect to h gives
Z ∞
2h(h2 − 3x2)
α2 cos(αx) exp(−αh) dα = . (119)
0 (h2 + x2)3
It may be noted that (118) and (119) are the inverse Fourier transforms of (114) and (116).
Differentiating equation (119 with respect to h gives
Z ∞
6(h4 − 6h2 x2 + x4)
α3 cos(αx) exp(−αh) dα = . (120)
0 (h2 + x2)4

Another well known Laplace transform is

Z ∞
a
sin(at) exp(−st) dt = 2 . (121)
0 s + a2

Written in the notation of a Fourier sine transform this integral is

Z ∞
x
sin(αx) exp(−αh) dα = 2 . (122)
0 h + x2

Differentiation with respect to h gives

Z ∞
2hx
α sin(αx) exp(−αh) dα = 2 . (123)
0 (h + x2 ) 2

Repeating the differentiation with respect to h gives

Z ∞
2x(3h2 − x2 )
α2 sin(αx) exp(−αh) dα = . (124)
0 (h2 + x2 )3
Another differentiation with respect to h gives
Z ∞
24xh(h2 − x2)
α3 sin(αx) exp(−αh) dα = . (125)
0 (h2 + x2 )4

13

View publication stats