International Journal of Engineering Science 47 (2009) 1149–1162

Contents lists available at ScienceDirect

International Journal of Engineering Science

journal homepage: www.elsevier.com/locate/ijengsci

On some contact problems for inhomogeneous anisotropic elastic materials

David L. Clements a,*, W.T. Ang b
a
School of Mathematics, University of Adelaide, North Terrace, Adelaide, SA 5005, Australia
b
School of Mechanical and Aerospace Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore

a r t i c l e i n f o a b s t r a c t

Article history: Some generalised plane strain contact problems are considered for a class of inhomoge-
Available online 6 February 2009 neous anisotropic elastic materials for which the elastic moduli vary continuously with
the spatial coordinates. Strip loading of a half-space and a layer on a rigid foundation are
considered and integral expressions for the displacement and stress are obtained. Numer-
PACS: ical results are obtained for some particular transversely isotropic and isotropic materials.
62.20.Dc Ó 2009 Elsevier Ltd. All rights reserved.
62.20.Mk

Keywords:
Contact problems
Anisotropy
Integral equations

1. Introduction

Contact problems involving the indentation of an elastic half-space by applied loads on its plane surface are the subject of
an extensive literature. The greater part of this literature is concerned with either a homogeneous isotropic half-space or a
half-space consisting of contiguous layers of homogeneous isotropic materials. Corresponding problems for anisotropic
homogeneous materials have also attracted the attention of a number of authors and a considerable literature exists on this
class of problems.
The solution of contact problems for an inhomogeneous half-space in which the elastic moduli vary continuously with the
spatial coordinates generally presents considerable difficulties compared with the corresponding problems for homogeneous
materials. Nevertheless a number of authors have succeeded in obtaining analytical solutions to contact problems involving
this class of materials. Such analytical solutions as do exist are restricted to particular types of inhomogeneous materials.
Thus, for example, in some of the early works in this area Gibson [1] and Gibson et al. [2] considered contact problems
for an incompressible half-space in which the elastic moduli varied linearly with the perpendicular distance from the plane
boundary of the half-space while Mossakovskii [3] considered problems where the elastic moduli varied exponentially with
the perpendicular distance form the plane boundary. Other more recent examples of solutions to problems for half-spaces
with continuously varying elastic moduli include the work of Clements and Ang [4], Azis and Clements [5] and Selvadurai
[6,7]. The latter two papers contain a number of references to papers which have addressed problems in this area in the latter
half of the 20th century.
The present work is concerned with the solution of some generalised plane contact problems for an anisotropic half-space
in which the elastic parameters are a quadratic function of the spatial variables. The problems considered involve a specified
displacement and a specified force on the boundary of the half-space. Solutions to these problems are obtained either in

* Corresponding author. Tel.: +61 8 83035948.

E-mail addresses: david.clements@adelaide.edu.au (D.L. Clements), mwtang@ntu.edu.sg (W.T. Ang).

0020-7225/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijengsci.2008.12.010
1150 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162

closed form or alternatively in terms of integrals which readily yield some analytical information regarding the solution and
also numerical values for the displacement and stress. The analysis is for general anisotropy and through a limiting
procedure also yields numerical results for the relevant class of isotropic materials. In the case of isotropic materials the class
of materials to which the analysis applies has a Poisson’s ratio of 1/4. The results obtained exhibit some similar character-
istics to the results obtained by Gibson [1] and Gibson et al. [2] for inhomogeneous isotropic materials with a Poisson’s ratio
1/2.

2. Statement of the problem

Referred to a Cartesian frame Ox1 x2 x3 consider an anisotropic elastic body with a geometry that does not vary in the Ox3
direction. Let the body occupy the region X which consists either of the half-space x2 > 0 or the slab lying in the region
0 < x2 < h, where h is a constant. On the plane boundary x2 ¼ 0 either the displacement or stress is specified and the slab
adheres to a rigid foundation so that for the slab the displacement is zero on x2 ¼ h. The problem is to determine the stress
and displacement throughout the elastic material.

3. Basic equations

The equilibrium equations governing small generalised plane deformations of an inhomogeneous anisotropic elastic
material may be written in the form
 
@ @uk ðxÞ
cijkl ðxÞ ¼ 0; ð1Þ
@xj @xl
where i; j; k; l ¼ 1; 2; 3; x ¼ ðx1 ; x2 Þ; uk denotes the displacement, cijkl ðxÞ the elastic moduli and the repeated summation con-
vention (summing from 1 to 3) is used for repeated Latin suffices. The stress displacement relations are given by
@uk
rij ðxÞ ¼ cijkl ð2Þ
@xl
and the stress vector Pi on a boundary with outward pointing normal n ¼ ðn1 ; n2 Þ is defined as
@uk
Pi ðxÞ ¼ rij nj ¼ cijkl nj : ð3Þ
@xl
For all points in X the coefficients cijkl ðxÞ are required to satisfy the usual symmetry condition
cijkl ¼ cijlk ¼ cjikl ¼ cklij ð4Þ
and also sufficient conditions for the strain energy density to be positive. This requirement ensures that the system of partial
differential equations is elliptic throughout X.
The coefficients in (1) are required to take the form
ð0Þ
cijkl ðxÞ ¼ cijkl gðxÞ; ð5Þ
ð0Þ
where the cijkl
are constants and gðx1 ; x2 Þ is a twice differentiable function of the variables x1 and x2 . Also in addition to the
ð0Þ
symmetry condition (4) the cijkl are required to satisfy the additional condition
ð0Þ ð0Þ
cijkl ¼ cilkj : ð6Þ

Eq. (17) may now be written in the form

 
ð0Þ @ @uk
cijkl g ¼ 0: ð7Þ
@xj @xl

Following Azis and Clements [5] consider a transformation of the dependent variables in the form

uk ¼ g 1=2 wk : ð8Þ
Use of (8) in (7) provides the equation
" #
ð0Þ @ 2 wk @g 1=2 @wk @g 1=2 @wk @ 2 g 1=2
cijkl g 1=2 þ   wk ¼ 0; ð9Þ
@xj @xl @xj @xl @xl @xj @xj @xl

where by virtue of (6) this equation reduces to

ð0Þ @ 2 wk 2 1=2
ð0Þ @ g
g 1=2 cijkl  wk cijkl ¼ 0: ð10Þ
@xj @xl @xj @xl
 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162 1151

Thus if

ð0Þ @ 2 wk
cijkl ¼0 ð11Þ
@xj @xl
and

ð0Þ @ 2 g 1=2
cijkl ¼ 0; ð12Þ
@xj @xl
then (10) will be satisfied. Thus when g satisfies the system (12) the transformation given by (8) transforms the linear system
with variable coefficients (7) to the linear system with constant coefficients (11).
As a result of the symmetry property cijkl ¼ cklij , Eq. (12) consists of a system of six constant coefficients partial differential
equations in the one dependent variable g 1=2 . In general this system will be satisfied by a linear function of the two indepen-
dent variables x1 ; x2 . Thus gðxÞ may be taken in the form

gðxÞ ¼ ðax1 þ bx2 þ cÞ2 ; ð13Þ

ð0Þ
where a; b and c are constants which may be used to fit the elastic moduli cijkl ðxÞ ¼ cijkl gðxÞ to given numerical data.
Now substitution of (5) and (8) into (3) yields
½g ½w
Pi ¼ Pik wk þ P i g 1=2 ; ð14Þ
where

½g @g 1=2ð0Þ
Pik ðxÞ ¼cijkl nj ; ð15Þ
@xl
½w ð0Þ @w
Pi ðxÞ ¼cijkl k nj : ð16Þ
@xl
Eq. (11) has the general solution (see Eshelby et al. [8] and Clements [9])
" #
X3
wi ¼ 2R Aia fa ðza Þ ; ð17Þ
a¼1

where R denotes the real part of a complex number, fa ðza Þ; a ¼ 1; 2; 3 are arbitrary analytic functions of the complex vari-
ables za ¼ x1 þ sa x2 , a ¼ 1; 2; 3, where sa are the three roots with positive imaginary part of the sextic in s
 
 ð0Þ ð0Þ ð0Þ ð0Þ 
ci1k1 þ ci2k1 s þ ci1k2 s þ ci2k2 s2  ¼ 0: ð18Þ

The Aia occurring in (17) are the solutions of the system

 
ð0Þ ð0Þ ð0Þ ð0Þ
ci1k1 þ ci2k1 sa þ ci1k2 sa þ ci2k2 s2a Aka ¼ 0: ð19Þ

Use of (17) in (14) to (16) provides a representation for P i in terms of the arbitrary functions fa ðza Þ in the form
3 h
X i
½g
Pi ¼ 2R Pik Aka fa ðza Þ þ g 1=2 Lija fa0 ðza Þnj ; ð20Þ
a¼1

where prime denotes differentiation with respect to the argument in question and
 
ð0Þ ð0Þ
Lija ¼ cijk1 þ sa cijk2 Aka ; ð21Þ

From (19) and (21) it follows that

Li1a ¼ sa Li2a : ð22Þ
Hence, in general it will only be necessary to consider the 3  3 matrix ½Li2a . The matrix ½Li1a  may then be determined
through (22).
From (8) and (17) the displacement in terms of the functions fa ðza Þ may be written in the form
" #
X3
uk ¼ 2g 1=2 R Aka fa ðza Þ : ð23Þ
a¼1

It is useful to have to have some alternative forms for Eqs. (20) and (23). Let
X
3
Aia fa ðzÞ ¼ hi ðzÞ; ð24Þ
a¼1
1152 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162

where the hi ðzÞ; k ¼ 1; 2; 3 are analytic functions of the complex variable z. The matrix ½Aia  is non-singular (see Stroh [10] and
Clements [9]) and hence from (24)
fa ðzÞ ¼ N ai hi ðzÞ; ð25Þ
where
X
3
dik ¼ Aia Naj ; ð26Þ
a¼1

where dij is the Kronecker delta. Substitution of (25) into (23) and (20) yields
" #
X3
1=2
uk ¼ 2g R Aka Naj hj ðza Þ ; ð27Þ
a¼1
3 h
X i
½g
Pi ¼ 2R Pik Aka Naj hj ðza Þ þ g 1=2 Lija Naj h0j ðza Þnj : ð28Þ
a¼1

In particular, on x2 ¼ 0 (27) and (28) become

uk ¼ g 1=2 ½hk ðx1 Þ þ hk ðx1 Þ; ð29Þ

h i
½g
Pi ¼ Pik ½hk ðx1 Þ þ hk ðx1 Þ  g 1=2 C ik h0k ðx1 Þ þ C ik h0k ðx1 Þ ; ð30Þ

where the bar denotes the complex conjugate and

X
3
C ik ¼ Li2a Nak : ð31Þ
a¼1

An alternative representation may be obtained by putting

X
3
Li2a fa ðzÞ ¼ vi ðzÞ; ð32Þ
a¼1

where the vk ðzÞ; k ¼ 1; 2; 3 are analytic functions of the complex variable z. The matrix ½Li2a  is non-singular (see Stroh [10]
and Clements [9]) and hence from (32)
fa ðzÞ ¼ M ia vi ðzÞ; ð33Þ
where
X
3
dik ¼ Li2a M ak : ð34Þ
a¼1

Substitution of (33) into (23) and (20) yields

" #
X3
1=2
uk ¼ 2g R Aka M aj vj ðza Þ ; ð35Þ
a¼1
3 h
X i
½g
Pi ¼ 2R Pik Aka Mar vr ðza Þ þ g 1=2 Lija M ar v0r ðza Þnj : ð36Þ
a¼1

In particular, on x2 ¼ 0 (35) and (36) become

h i
uk ¼ g 1=2 Bkj vj ðx1 Þ þ Bkj v
 j ðx1 Þ ; ð37Þ
½g

Pi ¼ Pik Bkr vr ðx1 Þ þ Bkr v  r ðx1 Þ  g 1=2 v0i ðx1 Þ þ v

 0i ðx1 Þ ; ð38Þ

where
X
3
Bkr ¼ Aka M ar : ð39Þ
a¼1

4. A half-space with specified boundary displacement

Consider an inhomogeneous elastic half-space x2 > 0 with the displacement uk prescribed on the boundary x2 ¼ 0. The
displacement and stress fields are required throughout the half-space. The boundary conditions on x2 ¼ 0 are
 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162 1153

uk ðx1 ; 0Þ ¼ U k ðx1 Þ; ð40Þ

where the U k ðx1 Þ; k ¼ 1; 2; 3 are given functions of x1 . For this problem the representation (27)–(30) is useful with hj ðzÞ given
by
Z 1
1
hj ðzÞ ¼ Gj ðpÞ expðipzÞdp; ð41Þ
2p 0

where the Gj ðpÞ; j ¼ 1; 2; 3 are functions of p which will be determined by the boundary conditions. From (29), (40) and (41)
it follows that
 Z 1 
1
g 1=2 ðx1 ; 0ÞR Gk ðpÞ expðipx1 Þdp ¼ U k ðx1 Þ: ð42Þ
p 0

Use of the inversion theorem for Fourier transforms provides

Z 1
Gk ðpÞ ¼ g 1=2 ðn1 ; 0ÞU k ðnÞ expðipnÞdn: ð43Þ
1

Substitution of (43) into (41) and changing the order of integration yields
Z 1 Z 1 Z 1
1 1 g 1=2 ðn; 0ÞU k ðnÞdn
hj ðzÞ ¼ g 1=2 ðn; 0ÞU k ðnÞdn expðipðz  nÞÞdp ¼ : ð44Þ
2p 1 0 2p i 1 nz
As a particular example consider the case when the surface displacement is given by
(
U k ða2  x21 Þ1=2
ð0Þ
for jx1 j < a;
U k ðx1 Þ ¼ ð45Þ
0 for jx1 j > a;
ð0Þ
where the U k for k ¼ 1; 2; 3 and a are constants. With gðxÞ given by (13), Eq. (44) yields
Z
U 0k a
ðan þ cÞða2  n2 Þ1=2 dn U 0k h i
hk ðzÞ ¼ ¼ ðaz þ cÞðz2  a2 Þ1=2  aðz2  a2 Þ  cz : ð46Þ
2pi a nz 2i

Use of (46) in (27) and (28) provides expressions for the displacement vector uk and the stress vector P i throughout the
the half-space when the surface displacement is given by (45).
Attention is now restricted to the case when U 1 ¼ U 3 ¼ 0 and a ¼ 0 so that Eq. (46) provides h1 ¼ h3 ¼ 0 and

U 02 c h 2 i
h2 ðzÞ ¼ ðz  a2 Þ1=2  z ; ð47Þ
2i " #
0 U 02 c z
h2 ðzÞ ¼ 1 : ð48Þ
2i ðz2  a2 Þ1=2

Commonly problems of this type are applicable for half-spaces in which the planes xi ¼ 0; i ¼ 1; 2; 3 are planes of elastic
ð0Þ
symmetry. If the anisotropy is restricted to materials which exhibit this symmetry the cijkl with an odd number
h iof ones, twos
½g
and threes in the suffices are zero. Thus from (15) and (13) (with a ¼ 0) on the boundary x2 ¼ 0 the matrix Pik has the form
2 ð0Þ
3
h i bc1212 0 0
6 7
Pik ¼ 6 7:
½g ð0Þ
4 0 bc2222 0 5 ð49Þ
ð0Þ
0 0 bc3232
Also in view of the symmetry it may be verified from the analysis of Section 3 that the matrix ½C kj  adopts the form (see Cle-
ments [9])
2 3
ic11 c12 0
6 7
½C kj  ¼ 4 c21 ic22 0 5; ð50Þ
0 0 ic33

where the cij ; i; j ¼ 1; 2, and c33 are real.

Use of (47) and (48) in (29) and (30) provides
(
ð0Þ
U 2 c12 c2 ½x1 ða2  x21 Þ1=2  for jx1 j < a;
P1 ðx1 Þ ¼ ð51Þ
0 for jx1 j > a;
(
U 2 bcc2222 ða2  x21 Þ1=2 þ U 2 c2 c22
ð0Þ ð0Þ ð0Þ
for jx1 j < a;
P 2 ðx1 Þ ¼ ð0Þ
ð52Þ
U 2 c22 2 ½jx1 jða2 c  x21 Þ1=2  1 for jx1 j > a:
1154 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162

Let b ¼ m=c where m P 0 is a constant. In terms of m and c, Eq. (5) together with (13) (with a ¼ 0) yields the elastic param-
eters in the form
ð0Þ
cijkl ¼cijkl ½c þ bx2  ð53Þ
 
ð0Þ m
¼cijkl c þ x2 ð54Þ
c
and the normal surface force in the form
(
U 2 mc2222 ða2  x21 Þ1=2 þ U 2 c2 c22
ð0Þ ð0Þ ð0Þ
for jx1 j < a;
P2 ðx1 Þ ¼ ð0Þ
ð55Þ
U 2 c22 2 ½jx1 jða2
c  x21 Þ1=2  1 for jx1 j > a:
As b ! 0 for a fixed and finite c > 0 it is apparent from (52) that, with the specified boundary displacement, the normal
force over the contact region tends to a constant value and when b ¼ 0 (so that the material is homogeneous) the normal
ð0Þ
force over the contact region assumes the constant value 2U 2 c2 c22 .
If c is sufficiently small for terms of order c2 to be ignored then (52) yields
(
U 2 bcc2222 ða2  x21 Þ1=2
ð0Þ ð0Þ
for jx1 j < a;
P2 ðx1 Þ ¼ ð56Þ
0 for jx1 j > a:

Thus for a fixed finite b with c sufficiently small for terms of order c2 to be ignored the surface force is concentrated over
the contact region. Over that region the surface force is a multiple of the specified surface displacement. Also the surface
force is order c and hence tends to zero as c ! 0.
For a fixed finite m as c ! 0 it is apparent from (55) that
(
U 2 mc2222 ða2  x21 Þ1=2
ð0Þ ð0Þ
for jx1 j < a;
P2 ðx1 Þ ! ð57Þ
0 for jx1 j > a:

Thus if the elastic parameters have the form given by (54) then as c ! 0 the boundary force outside the contact region tends
to zero and the normal force over the contact region is a constant finite multiple of the specified surface displacement.

5. A half-space with specified boundary force

Consider an inhomogeneous elastic half-space x2 > 0 with the stress vector P i prescribed on the boundary x2 ¼ 0. The dis-
placement and stress fields are required throughout the half-space. The boundary conditions on x2 ¼ 0 are
Pi ðx1 ; 0Þ ¼ Pi ðx1 Þ for i ¼ 1; 2; 3; ð58Þ
where the P i ðx1 Þ; i ¼ 1; 2; 3 are given functions of x1 . For this problem the representation (35)–(39) is useful with vj ðzÞ given
by
Z 1
1
vj ðzÞ ¼ Gj ðpÞ expðipzÞdp for j ¼ 1; 2; 3; ð59Þ
2p 0

where the Gj ðpÞ; j ¼ 1; 2; 3 are functions of p which will be determined by the boundary conditions. From (38), (59) and (58) it
follows that
 Z 1 h i 
1 ½g 1=2
R Pik Bkr  ig pdir Gr ðpÞ expðipx1 Þdp ¼ Pi ðx1 Þ: ð60Þ
p 0

If gðxÞ ¼ ðbx2 þ cÞ2 where b P 0 and c > 0 are constants then g and Pik are constant on x2 ¼ 0 and thus use of the inversion
½g

formula for Fourier transforms provides

Z 1
Dir Gr ðpÞ ¼ Pi ðnÞ expðipnÞdn; ð61Þ
1

where from (60) and (15)

½g 1=2 ð0Þ
Dir ¼ Pik Bkr  ig ð0Þpdir ¼ bci2k2 Bkr  icpdir : ð62Þ

Hence
Z 1
Gr ðpÞ ¼ Eri ðpÞ Pi ðnÞ expðipnÞdn; ð63Þ
1

where
dij ¼ Dir Erj : ð64Þ
 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162 1155

Use of (63) in (59) gives

Z 1 Z 1
1
vr ðzÞ ¼ Eri ðpÞ expðipzÞdp Pi ðnÞ expðipnÞdn: ð65Þ
2p 0 1

Substitution of (65) into (35) gives an expression for the displacement throughout the half-space in the form
X
3 Z 1 Z 1
1
uk ¼ R A ka M a r E rj ðpÞ expðipza Þdp Pi ðnÞ expðipnÞdn: ð66Þ
pg 1=2 a¼1 0 1

In particular the displacement on the surface x2 ¼ 0 is given by

 Z 1 Z 1 
1
uk ðx1 ; 0Þ ¼ R Bkr Erj ðpÞ expðipx1 Þdp Pi ðnÞ expðipnÞdn : ð67Þ
pc 0 1
ð0Þ ½g
Once the material constants cijkl ; b and c are known the constants Pik ; Aka , Lija ; M ar and Bkr may be calculated from the analysis
in Section 3 and then the Eij may be obtained through Eqs. (62) and (64). Then when P i ðx1 Þ is known (66) and (67) provide
equations which are suitable for the calculation of the displacement throughout the half-space. Also substitution of (65) into
(36) provides an expression from which the stress vector may be calculated.
Of particular interest is the case when the loading Pi ðx1 Þ on the surface of the half-space consists of a constant pressure
ð0Þ
P i over a strip of finite width so that
(
ð0Þ
Pi for jx1 j < a;
Pi ðx1 Þ ¼ ð68Þ
0 for jx1 j > a:

In this case it is convenient at this point to require b > 0 and then (66) and (67) yield convergent integral representations for
the displacement in the form
ð0Þ
" Z #
2Pj X3
sin pa 1
uk ¼ 1=2 R Aka M ar Erj ðpÞ expðipza Þdp ; ð69Þ
pg a¼1 0 p
ð0Þ  Z 1 
2Pj sin pa
uk ðx1 ; 0Þ ¼ R Bkr Erj ðpÞ expðipx1 Þdp : ð70Þ
pc 0 p
Commonly problems of this type are applicable for half-spaces in which the planes xi ¼ 0; i ¼ 1; 2; 3 are planes of elastic sym-
ð0Þ
metry. If the anisotropy is restricted to materials which exhibit this symmetry the cijkl with an odd numberh of ones,
i twos and
½g
threes in the suffices are zero. Thus from (15) and (13) (with a ¼ 0) on the boundary x2 ¼ 0 the matrix Pik has the form
given by Eq. (49). Also in view of the symmetry it may be verified from the analysis of Section 3 that the matrix ½Bkj  adopts
the form (see Clements [9])
2 3
ib11 b12 0
6 7
½Bkj  ¼ 4 b21 ib22 0 5; ð71Þ
0 0 ib33
where the bij ; i; j ¼ 1; 2, and b33 are real. Therefore from (62) and (64) the matrices ½Dir  and ½Erj  take the forms
2  ð0Þ

ð0Þ
3
i bb11 c1212  cp bb12 c1212 0
6   7
6 7
½Dik  ¼ 6 bb 21 c
ð0Þ
i bb22 c
ð0Þ
 cp 0 7; ð72Þ
4 2222 2222 5
ð0Þ
0 0 ibb33 c3232  icp
2 iðbb22 c0 cpÞ ðbb12 c01212 Þ 3
2222
D D
0
6 ðbb21 c02222 Þ iðbb11 c01212 cpÞ
7
½Eik  ¼ 6
4 0 7;
5 ð73Þ
D D
1
0 0 ðibb33 c03232 icpÞ

where
     
ð0Þ ð0Þ ð0Þ ð0Þ
D ¼  bb11 c1212  cp bb22 c2222  cp  bb21 c2222 bb12 c1212 : ð74Þ

Referred to the non-dimensional variables

ð0Þ 0
x0 ¼ x1 =a; p0 ¼ pa; c0ijkl ¼ cijkl =C; bij ¼ bij C;
ð0Þ
ð75Þ
b0 ¼ ba; P 0i ¼ Pi =C; B0kr ¼ Bkr C; E0rj ðp0 Þ ¼ Erj a:

Eq. (70) may be written in the form

1156 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162

0.2

Beta’=10000,
Gamma=0.0001
0.15

Surface Displacement
0.1 Beta’=2,
Gamma=0.5

0.05
Beta’=1,
Gamma=1

−0.05
0 5 10 15 20 25 30
10x’

Fig. 1. Antiplane surface displacement u3 ðx0 ; 0Þ=aP03 .

 Z 
2P0j 1
sin p0 0 0 0
uk ðx0 ; 0Þ=a ¼ R B0kr E0rj ðp0 Þ expðip x 1 Þdp ; ð76Þ
pc 0 p0
where C is a reference pressure and c is a non-dimensional constant.
With the assumed elastic symmetry the plane and antiplane problems separate out and, from Eq. (76), the antiplane dis-
placement on the surface x2 ¼ 0 is given by
" Z #
1
2 0 1 sin p0 0
u3 ðx 0
; 0Þ=aP03 ¼ b33 0
0 0
cosðp x Þdp : ð77Þ
pc 0 b0 b33 c03232  cp0 p0

In the case of a normal load with P01 ¼ 0 and P03 ¼ 0 the normal displacement is given by
Z 1
2 sin p0 0
u2 ðx0 ; 0Þ=aP02 ¼ Fðp0 Þ cosðp0 x0 Þdp ; ð78Þ
pc 0 p0
where
0  0  0  0 

 b21 b0 b12 c01212 þ b22 b0 b11 c02222  cp0

Fðp0 Þ ¼  0   0  0 0 : ð79Þ
b0 b11 c01212  cp0 b0 b22 c02222  cp0 þ b0 b21 c02222 b0 b12 c01212
Let
m0
b0 ¼ ; ð80Þ
c
where m0 is a constant. In terms of m0 and c, Eq. (5) together with (13) (with a ¼ 0) yields the elastic parameters in the form
 
cijkl m0 x2
¼ c0ijkl c þ ð81Þ
C c a
and the antiplane surface displacement in the form
" Z #
1
2 0 1 sin p0 0
u3 ðx 0
; 0Þ=aP03 ¼ b33 0
0 0
cosðp x Þdp : ð82Þ
p 0 m0 b33 c03232  c2 p0 p0

For a fixed m0 as c ! 0
Z ( 1
2 1
1 sin p0 0 m0 c03232
for jx0 j < 1;
u3 ðx 0
; 0Þ=aP03 ! cosðp0 x0 Þdp ¼ ð83Þ
p 0
0 0
m c3232 p0 0 for jx0 j > 1:
For the purpose of obtaining some numerical values for the antiplane surface displacement the relevant constants c0ijkl for a
particular transversely isotropic material with the x1 axis normal to the transverse planes are chosen as c01111 ¼ 18:1,
c02222 ¼ 16:2, c01212 ¼ c01122 ¼ c01133 ¼ c01313 ¼ 6:9, c03232 ¼ 5:4. With these values of the c0ijkl the analysis of Section 3 may be used
0 0 0 0 0 0
to determine the values of bij as b11 ¼ 0:1227, b12 ¼ 0:0416; b21 ¼ 0:01426, b22 ¼ 0:1297 and b33 ¼ 0:1638.
 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162 1157

For the material with these constants the the antiplane surface displacement u3 ðx0 ; 0Þ=aP03 given by (82) is plotted in Fig. 1
for various values of b0 and c with b0 c ¼ m0 ¼ 1. The plotted values indicate the movement as c ! 0 of the surface displace-
ment towards the values given by (83) in which in this particular case 1=ðm0 c03232 Þ ¼ 1=5:4 ¼ 0:1852.
Similarly in terms of m0 and c, Eq. (78) yields the normal surface displacement in the form
Z 1
2 sin p0 0
u2 ðx0 ; 0Þ=aP 02 ¼ F 0 ðp0 Þ cosðp0 x0 Þdp ; ð84Þ
p 0 p0
where
0  0  0  0 

0 0  b21 m0 b12 c01212 þ b22 m0 b11 c01212  c2 p0

F ðp Þ ¼  0 0 0  0  0 0 : ð85Þ
m b11 c1212  c2 p0 m0 b22 c02222  c2 p0 þ m02 b21 c02222 b12 c01212
For a fixed m0 as c ! 0
Z ( 1
2 1
sin p0 0 m0 c02222
for jx0 j < 1;
u2 ðx0 ; 0Þ=aP02 ! cosðp0 x0 Þdp ¼ ð86Þ
m0 c02222 p 0 p0 0 for jx0 j > 1:
For the purpose of obtaining some numerical values for the normal surface displacement the constants c0ijkl for a particular
transversely isotropic material are chosen as c01111 ¼ 5; c02222 ¼ 16:2, c01212 ¼ c01122 ¼ c01133 ¼ c01313 ¼ 2, c03232 ¼ 5:4. With these
0 0 0 0 0
values of the c0ijkl the values of bij are b11 ¼ 0:3787; b12 ¼ 0:0909, b21 ¼ 0:0909, b22 ¼ 0:2104 and b33 ¼ 0:3043.
0
For the material with these constants the surface displacement u2 ðx ; 0Þ=aP2 is plotted in Fig. 2 for various values of b0 and
0

c with b0 c ¼ m0 ¼ 1. The plotted values indicate the movement as c ! 0 of the surface displacement towards the values given
by (86) in which in this particular case 1=ðm0 c02222 Þ ¼ 1=16:2 ¼ 0:0617.
With the loading given by (68) the representations (69) and (70) are only valid in the inhomogeneous case when b > 0. In
the homogeneous case when b ¼ 0 it is apparent from (62) and (64) that the Dir and Erj assume the forms

Dir ¼ icpdir ; Erj ¼ ic1 p1 drj ð87Þ

and since the matrix with elements consisting of the imaginary part of Bij is non-singular (see Clements [9]) at least some of
the Bij will have a non-zero imaginary part and hence it is apparent that when b ¼ 0, Eq. (70) will contain divergent integrals.
This feature of the integral representations for the the displacement of an inhomogeneous half-space by a load profile
given by (68) is also present in the solution obtained by Gibson [1] to the corresponding problem for an isotropic incom-
pressible material with the elastic parameters varying linearly with depth.
In order to obtain information on the surface displacement for the homogeneous case the procedure followed by Gibson
[1] is employed. Specifically an expression for the shape of the deformed surface is obtained by determining the difference in
displacement Duk ðx1 ; x2 Þ ¼ uk ðx1 ; 0Þ  uk ð0; 0Þ. From (69) it follows that
ð0Þ
" Z #
2Pj X
3 1
Duk ðx1 ; x2 Þ ¼ uk ðx1 ; x2 Þ  uk ð0; 0Þ ¼ R Ak a M a r Erj ½expðipza Þ  1dp : ð88Þ
pc a¼1 0

In the homogeneous limit when b ¼ 0; Erj is given by (87) so that Eq. (88) yields

0.07

0.06 Beta’=10000,
Gamma=0.0001

0.05
Surface Displacement

0.04

0.03 Beta’=1,
Gamma=1

Beta’=2,
0.02 Gamma=0.5

0.01

−0.01
0 5 10 15 20 25 30
10x’

Fig. 2. Normal surface displacement u2 ðx0 ; 0Þ=aP02 .

1158 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162

ð0Þ
" Z #
2P j X3
sin pa 1
Duk ðx1 ; x2 Þ ¼ uk ðx1 ; x2 Þ  uk ð0; 0Þ ¼ R iAka M aj ½expðipza Þ  1dp ; ð89Þ
pc2 a¼1 0 p2
ð0Þ  Z 1 
2Pj sin pa
Duk ðx1 ; 0Þ ¼ uk ðx1 ; 0Þ  uk ð0; 0Þ ¼ R iB kj ½expðipx 1 Þ  1dp : ð90Þ
pc2 0 p2
Referred to the non-dimensional coordinates (75), Eq. (90) becomes
 Z 
2P 0j 0
1
sin p0 0 0 0
Duk ðx1 ; 0Þ=a ¼ ½uk ðx1 ; 0Þ  uk ð0; 0Þ=a ¼ R iB kj ½expðip x 1 Þ  1dp : ð91Þ
pc 2
0 p02
In the case considered previously when the planes xi ¼ 0; i ¼ 1; 2; 3 are planes of elastic symmetry the matrix ½Bij  is given by
(71) and the antiplane difference in surface displacement is given by
0 Z 1 
2b33 sin p0 0
Du3 ðx1 ; 0Þ=aP03 ¼ ½u3 ðx1 ; 0Þ  u3 ð0; 0Þ=aP 03 ¼  ½cosðp0 x01 Þ  1dp
pc 2
0 p02
0
b

¼ 332 ð1 þ x01 Þ log j1 þ x01 j þ ð1  x01 Þ log j1  x01 j : ð92Þ

pc
Similarly with this elastic symmetry and P 01 ¼ 0 and P 03 ¼ 0 the difference in normal surface displacement is given by
0 Z 1 
2b22 sin p0 0
Du2 ðx1 ; 0Þ=aP02 ¼ ½u2 ðx1 ; 0Þ  u2 ð0; 0Þ=aP 02 ¼  ½cosðp0 x01 Þ  1dp
pc2 0 p02
0
b22

¼ ð1 þ x01 Þ log j1 þ x01 j þ ð1  x01 Þ log j1  x01 j : ð93Þ

pc 2

The displacements in (92) and (93) are consistent with known results for homogeneous anisotropic materials (see, for exam-
ple, Clements [9]).

6. An elastic layer on a rigid foundation

Consider an inhomogeneous elastic layer occupying the region 0 < x2 < h with the stress vector P i prescribed on the
boundary x2 ¼ 0. The displacement and stress fields are required throughout the layer. The layer adheres to a rigid founda-
tion on x2 ¼ h. The boundary conditions are
Pi ðx1 ; 0Þ ¼ Pi ðx1 Þ; ð94Þ
ui ðx1 ; hÞ ¼ 0; ð95Þ
where the P i ðx1 Þ; i ¼ 1; 2; 3 are given functions of x1 . For this problem the representation (35)–(39) is useful with vj ðzÞ given
by
Z 1
1
vj ðzÞ ¼ ½Gj ðpÞ expðipzÞ þ Hj ðpÞ expðipzÞdp for j ¼ 1; 2; 3; ð96Þ
2p 0

where the Gj ðpÞ and Hj ðpÞ; j ¼ 1; 2; 3 are functions of p which will be determined by the boundary conditions. From (38), (94)
and (96) it follows that
Z 1 h    i
1 ½g 1=2 ½g 1=2
R Pik Bkr þ ig pdir Gr ðpÞ þ Pik Bkr þ ig pdir Hr ðpÞ expðipx1 Þdp ¼ Pi ðx1 Þ: ð97Þ
p 0

From (37), (95) and (96) it follows that

X3 Z 1h i
1
R ðAka M aj ÞGj ðpÞ expðisa phÞ  ðAka Maj ÞHj ðpÞ expðis
a phÞ expðipx1 Þdp ¼ 0: ð98Þ
pg 1=2
a¼1 0

Eq. (98) may be written in the form

Z 1
1

R Rkj ðpÞGj ðpÞ  Skj ðpÞHj ðpÞ expðipx1 Þdp ¼ 0; ð99Þ

pg 1=2 0

where
X
3
Rkj ðpÞ ¼ Aka Maj expðisa phÞ; ð100Þ
a¼1
X
3
Skj ðpÞ ¼ Aka Maj expðis
a phÞ: ð101Þ
a¼1
 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162 1159

Eq. (99) will be satisfied if

Hj ðpÞ ¼ T jn ðpÞRnl ðpÞGl ðpÞ; ð102Þ

where
Skj ðpÞT jn ðpÞ ¼ dkn : ð103Þ
Substitution of (102) in (97) provides
Z 1
1
R ½F ir ðpÞGr ðpÞ expðipx1 Þdp ¼ P i ðx1 Þ; ð104Þ
p 0

where
   
½g 1=2 ½g 1=2
F ir ðpÞ ¼  Pik Bkr þ ig pdir  Pik Bkm þ ig pdim T mk ðpÞRkr ðpÞ: ð105Þ

Use of the inversion formula for Fourier transforms now provides

Z 1
F ir ðpÞGr ðpÞ ¼ Pi ðnÞ expðipnÞdn: ð106Þ
1

If Pi ðx1 Þ satisfies (68) then

Z a
ð0Þ ð0Þ sin pa
F ir ðpÞGr ðpÞ ¼ Pi expðipnÞdn ¼ 2P i : ð107Þ
a p
Hence

ð0Þ sin pa
Gn ðpÞ ¼ 2U ni ðpÞP i ; ð108Þ
p
where
U ni ðpÞF ir ðpÞ ¼ dnr : ð109Þ
Hence

ð0Þ sin pa
Hj ðpÞ ¼ 2T jk ðpÞRkl ðpÞU li ðpÞPi : ð110Þ
p
The surface displacement is given by
Z 1
1

uk ðx1 ; 0Þ ¼ R Bkr Gr ðpÞ  Bkr Hr ðpÞ expðipx1 Þdp

pc 0
ð0Þ Z 1
2Pi
sin pa
¼R Bkn  Bkr T rl ðpÞRln ðpÞ U ni ðpÞ expðipx1 Þdp: ð111Þ
pc 0 p
With the same elastic symmetry as in the previous section the plane and antiplane parts of the problem separate with the
antiplane surface displacement given by
Z 1
2 ð0Þ sin pa
u3 ðx1 ; 0Þ ¼ R ½B33  B33 T 33 R33 U 33 P3 expðipx1 Þdp; ð112Þ
pc 0 p
where

B33 ¼ ib33 ; ð113Þ

R33 ¼ ib33 expðis1 phÞ; ð114Þ
S33 ¼ ib33 expðis
1 phÞ; ð115Þ
T 33 ¼ 1=ðS33 Þ ¼ iðb33 Þ1 expðis
1 phÞ; ð116Þ
   
ð0Þ ð0Þ
F 33 ¼ ibc3232 b33  icp þ ibc3232 b33  icp expðiðs1  s
1 ÞphÞ; ð117Þ
U 33 ¼ 1=F 33 : ð118Þ

Use of (113)–(118) in (112) provides

Z 1 ð0Þ
2b33 ½1  expðiðs1  s 1 ÞphÞ sinðpaÞ cosðpx1 ÞP3
u3 ðx1 ; 0Þ ¼   dp: ð119Þ
p 0
ð0Þ
bc3232 b33  cp ½1 þ expðiðs1  s 1 ÞphÞp
0
Referred to the non-dimensional coordinates of the previous section and with h ¼ ah . Eq. (119) becomes
1160 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162

0 Z
u3 ðx0 ; 0Þ 2b33 1
½1  expðiðs1  s 1 Þp0 h0 Þ sinðp0 Þ cosðp0 x0 Þ 0
¼ dp : ð120Þ
aP03 pc 0
0 0
ðb b33 c3232  cp Þ½1 þ expðiðs1  s
0 0 1 Þp0 h0 Þp0

Setting b0 ¼ m0 =c, where m0 is a non-dimensional positive constant, Eq. (120) yields

0 Z 1 0
u3 ðx0 ; 0Þ 2b33 ½1  expðiðs1  s 1 Þp0 h Þ sinðp0 Þ cosðp0 x0 Þ 0
¼ dp : ð121Þ
aP03 p 0
0 0
ðm b33 c3232  c p Þ½1 þ expðiðs1  s
0 2 0 1 Þp0 h0 Þp0

Since the imaginary part of s1 is chosen to be positive it follows from Eq. (119) that as c ! 0 and h ! 1:
Z ( 1
u3 ðx1 ; 0Þ 2 1
sinðp0 Þ cosðp0 x0 Þ 0 m0 c03232
for jx0 j < 1;
! dp ¼ ð122Þ
aP03 p 0 m0 c03232 p0 0 for jx0 j > 1:

7. An isotropic half-space and layer on a rigid foundation

For isotropic materials the non-zero elastic coefficients cijkl may be expressed in terms of the two Lamé coefficients k and
l, where
k ¼ kð0Þ gðxÞ; l ¼ lð0Þ gðxÞ; ð123Þ
ð0Þ
where k and l ð0Þ
are constants. The relevant coefficients cijkl are related to k and l by the equations
c1111 ¼ c2222 ¼ c3333 ¼ k þ 2l; ð124Þ
c1122 ¼ c1133 ¼ c2233 ¼ k; ð125Þ
c1212 ¼ c1313 ¼ c2323 ¼ l: ð126Þ
ð0Þ
The coefficients cijkl may be expressed in terms of the two constants kð0Þ and lð0Þ by the equations
ð0Þ ð0Þ ð0Þ
c1111 ¼ c2222 ¼ c3333 ¼ kð0Þ þ 2lð0Þ ; ð127Þ
ð0Þ ð0Þ ð0Þ
c1122 ¼ c1133 ¼ c2233 ¼ kð0Þ ; ð128Þ
ð0Þ ð0Þ ð0Þ
c1212 ¼ c1313 ¼ c2323 ¼ lð0Þ : ð129Þ

The coefficients cijkl must satisfy the condition (6) which requires that cijkl ¼ cilkj so that from 125, 126, 128 and 129 it follows
that

k ¼ l; kð0Þ ¼ lð0Þ : ð130Þ

Young’s modulus E and Poisson’s ratio m are related to the Lamé coefficients by the equations
lð3k þ 2lÞ k
E¼ ; m¼ : ð131Þ
kþl 2ðk þ lÞ
Hence, by virtue of Eq. (130) it follows that
5k 1
E¼ ; m¼ : ð132Þ
2 4
Thus the analysis of the previous two sections may be applied to an inhomogeneous isotropic material for which the Pois-
son’s ratio m is 1/4 (a frequently occurring value in rock materials – see Turcotte [11]) and the inhomogeneity is specified by
either Young’s modulus or one of the Lamé constants in the forms

E ¼ Eð0Þ ðc þ bx2 Þ; k ¼ kð0Þ ðc þ bx2 Þ; l ¼ lð0Þ ðc þ bx2 Þ; ð133Þ

where these alternative forms are related by
E ¼ 5k=2; k ¼ l; ð134Þ
so that

Eð0Þ ¼ 5kð0Þ =2; kð0Þ ¼ lð0Þ : ð135Þ

For the isotropic case the sextic (18) has equal roots. Also the plane and antiplane parts of the problem uncouple to form two
separate problems. In the plane case the two rows of the matrices ½Aka  and ½Li2a  associated with the plane problem are lin-
early dependent and hence their inverses do not exist. Thus for the plane problems the analysis of the previous three sections
is no longer valid. In order to employ the analysis of the previous sections to obtain numerical results for the isotropic case
the numerical values of the isotropic elastic moduli are perturbed slightly from their exact isotropic values in order to obtain
distinct roots for the sextic (18). Specifically the relevant elastic parameters cijkl 0 are chosen in the form
 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162 1161

c01111 ¼ c03333 ¼ 3k0 ; ð136Þ

c02222 ¼ 3k0 þ ; ð137Þ
c01122 ¼ c01133 ¼ c02233 ¼ k0 ; ð138Þ
c01212 ¼ c01313 ¼ c02323 ¼ k0 ; ð139Þ
0 ð0Þ
where k ¼ k =C and  is a small non-dimensional constant. As  ! 0 numerical values obtained from the relevant equations
for the displacement and stress vector in the previous three sections tend to the values for an isotropic material with Pois-
son’s ratio 1/4.
For illustrative purposes the values of the constants in (136)–(139) are chosen to be k ¼ 5:4 and  ¼ 0:001. With this
choice the analysis of Section 3 yields the numerical values
s1 ¼ i; b11 ¼ 0:13888; b12 ¼ 0:04629; b21 ¼ 0:04629; b22 ¼ 0:138888; b33 ¼ 0:18518:
Numerical values of the surface normal displacement for an inhomogeneous isotropic half-space with the above elastic mod-
uli were calculated using Eq. (84). The results are shown in Fig. 3. They may be compared with the corresponding results for
an anisotropic half-space displayed in Fig. 2. The results in Fig. 2 are for an anisotropic material whose constants c0ijkl are all
less than or equal to the corresponding isotropic constants as given by Eqs. (136)–(139) with k ¼ 5:4. For the anisotropic
material c02222 ¼ 16:2 while from (137) c02222 ¼ 16:201 in the isotropic case. Thus from Eq. (86) both the anisotropic and iso-
tropic surface displacement could be expected to be virtually identical as c ! 0. Figs. 2 and 3 verify that this is the case. For
larger values of c the lower elastic moduli in the anisotropic case give rise to a larger surface displacement than for the cor-
responding isotropic half-space.
Numerical values of the surface antiplane displacement for an isotropic layer on a rigid base with the above elastic moduli
were calculated using Eq. (121) for various values of the layer thickness and with m0 ¼ bc ¼ 1 and c ¼ 0:0001. The results are
shown in Fig. 4. The results indicate that for sufficiently small layer thickness the antiplane surface displacement is positive
over the contact region and increases as jx0 j increases from zero to one. Outside the contact region the surface displacement
becomes negative and tends towards zero as jx0 j ! 0. For this isotropic material the soft upper layers can easily move later-
ally and under the specified boundary conditions on x2 ¼ 0 and x2 ¼ h this capability facilitates a sharp change in the anti-
plane displacement from a small positive value to a smaller negative value in the vicinity of the edge of the loaded region.
This feature of the displacement profiles in Fig. 4 is similar to profiles obtained by Gibson et al. [2] for a related problem
involving an incompressible isotropic half-space with a linearly varying elastic modulus.

8. Final remarks

Some contact problems have been considered for an anisotropic half-space and layer in which the elastic moduli vary
quadratically with the spatial variables. For the class of problems considered formulas for the displacement are given in
either closed form or in integral formulations which readily yield numerical values for particular problems and provide
closed form formulas in limiting cases. The analysis can be used to consider the corresponding contact problems for isotropic
materials as a particular case of the general anisotropic analysis. The problems considered exhibit a number of characteris-
tics which are similar to the features observed by Gibson [1] and Gibson et al. [2] for the corresponding problems for an inho-
mogeneous incompressible isotropic half-space and layer in which the elastic moduli vary linearly with a spatial variable.

0.07

0.06 Beta’=10000,
Gamma=0.0001

0.05
Surface Displacement

0.04

0.03 Beta’=1,
Gamma=1

Beta’=2,
0.02 Gamma=0.5

0.01

−0.01
0 5 10 15 20 25 30
10x’

Fig. 3. Normal surface displacement u2 ðx0 ; 0Þ=aP 02 for an isotropic material.

1162 D.L. Clements, W.T. Ang / International Journal of Engineering Science 47 (2009) 1149–1162

0.25

Beta’=10000,
0.2 Gamma=0.0001,
h’=10

Surface Displacement
0.15

Beta’=10000,
0.1
Gamma=0.0001,
h’=2

0.05
Beta’=10000,
Gamma=0.0001,
0 h’=0.5

−0.05

−0.1
0 5 10 15 20 25 30
10x’

Fig. 4. Antiplane surface displacement u3 ðx0 ; 0Þ=aP 03 for an isotropic material.

Finally, the contact problems examined in this paper have involved problems in which either the stress or, alternatively,
the displacement is prescribed on the whole of the boundary x2 ¼ 0 of the half-space x2 > 0 and the layer h > x2 > 0. Prob-
lems involving mixed boundary conditions of the type which occur in crack and a number of contact problems have not been
considered. The analysis employed in the paper by Clements et al. [12] to solve some antiplane crack problems for an inho-
mogeneous isotropic material suggests that it may be possible to employ similar techniques to those contained in Clements
et al. [12] to extend the work in the current paper to solve generalised plane mixed boundary value problems for the half-
space and the layer either in terms of Fredholm integral equations or by employing iterative techniques.

References

[1] R.E. Gibson, Some results concerning displacements and stresses in a non-homogeneous elastic half-space, Geotechnique 17 (1967) 58–67.
[2] R.E. Gibson, P.T. Brown, K.R.F. Andrews, Some results concerning displacements in a non-homogeneous elastic layer, Z. Angew. Math. Phys. 22 (1971)
855–864.
[3] V.I. Mossakovskii, Pressure of a circular punch on an elastic half-space whose modulus of elasticity is an exponential function, J. Appl. Math. Mech. 22
(1958) 168–171.
[4] D.L. Clements, W.T. Ang, On the indentation of an inhomogeneous anisotropic elastic material by multiple straight rigid punches, Eng. Anal. Bound.
Elem. 30 (2006) 284–291.
[5] M.I. Azis, D.L. Clements, A boundary element method for anisotropic inhomogeneous elasticity, Int. J. Solids Struct. 38 (2001) 5747–5764.
[6] A.P.S. Selvadurai, The settlement of a rigid circular foundation resting on a half-space exhibiting a near surface elastic non-homogeneity, Int. J. Numer.
Meth. Geomech. 20 (1996) 351–364.
[7] A.P.S. Selvadurai, The analytical method in geomechanics, Appl. Mech. Rev. 60 (2007) 87–106.
[8] J.D. Eshelby, W.T. Read, W. Shockley, Anisotropic elasticity with applications to dislocation theory, Acta Metall. 1 (1953) 251–259.
[9] D.L. Clements, Boundary Value Problems Governed by Second Order Elliptic Systems, Pitman, Bath, 1981.
[10] A.N. Stroh, Dislocations and cracks in anisotropic elasticity, Philos. Mag. 3 (1958) 625–646.
[11] D.L. Turcotte, P. Schubert, Geodynamic Applications of Continuum Physics to Geological Problems, Wiley, New York, 1982.
[12] D.L. Clements, C. Atkinson, C. Rogers, Antiplane crack problems for an inhomogeneous elastic material, Acta Mech. 29 (1978) 199–211.