Journal of Thermal Stresses, 28: 595–648, 2005

Copyright # Taylor & Francis Inc.

ISSN: 0149-5739 print/1521-074X online
DOI: 10.1080/01495730590932706

Sixth International Congress on Thermal Stresses,

TS 2005, Vienna, May 26–29, 2005 Invited Lecture
COMPLEX VARIABLE METHOD FOR THERMAL
STRESS PROBLEM

Norio Hasebe and Xianfeng Wang

Downloaded by [University of California Davis] at 10:13 11 April 2014

Department of Civil Engineering, Nagoya Institute of Technology, Gokiso-cho,

Showa-ku, Nagoya, Japan

The two-dimensional thermal stress problem is one of the important problems in the field of
solid mechanics. In this paper, the complex variable method for solving this kind of problem is
stated. The basic equations of complex variable functions in conjunction with a rational map-
ping function are considered. The formulation of the rational mapping function and its accu-
racy are indicated. The general solutions for the external force, displacement, and mixed
boundary value problems under uniform heat flux and under a point heat source are stated.
The Green’s functions for a point heat source and sink are suggested for the three kinds of
boundary value problems. The equation to obtain the stress intensity factor for the crack pro-
blems and the stress intensity of debonding for a partially debonded rigid inclusion are shown.
The interaction problem between a hole and a line crack under uniform heat flux is analyzed.

Keywords: Complex variable method; Thermal elasticity; Green’s function; Heat flux; Point heat source;
Heat source couple; Mixed boundary value problem; Stress intensity factor

The theory of thermoelasticity, as a classical problem in the field of solid mechanics, has
long been developed, since the 19 century. The thermal stress problem still continues to
attract research interest due to some new engineering applications such as in the very
large scale integration (VLSI) systems, new materials, nuclear engineering, and the field
of aeronautics in recent years. Heating or nonuniform temperature fields make failures
or reduce the service life of materials and structures. The investigations of the related
problems have resulted in a considerable number of research works, both theoretical
and experimental, treating various aspects of thermal stresses encountered in practice.
Nowacki [1], Muskhelishvili [2], Parkus [3], and Timoshenko and Goodier et al.
[4] developed the basic theory for treating this kind of problem, in which the two-
dimensional (20) thermal stress problem is a typical branch. There are usually two
kinds of methods in solving the thermal stress problems [3]; one is the real variable
method, the other is the complex variable one [2,3]. Both of them need to introduce a
thermoelastic potential. In this paper, a series of results that the authors solved for
two-dimensional, static, and thermoelastic problems are reviewed.

Received 22 January 2005; accepted 9 February 2005.

Address correspondence to Norio Hasebe, Department of Civil Engineering, Nagoya Institute of
Technology, Gokiso-cho, Showa-ku, Nagoya 466-8555, Japan. E-mail: [email protected]

595
 596 N. HASEBE AND X. WANG

The basic equations of the complex variable method in conjunction with a

rational mapping function are considered. The formulation of the rational mapping
function and its accuracy are indicated. The temperature and the thermoelastic fields
are analyzed. From the point of view of boundary condition, the external force, the
displacement and the mixed boundary value problems are systematically stated.
With regard to the thermal condition, the uniform heat flux and the point heat
source (Green’s function) are considered. The configuration we focused on is an infi-
nite plane with a hole of arbitrary shape, of which the elliptic hole is a special case.
As one of the applications, the interaction problem between a hole and a line crack is
analyzed. The equations used to obtain stress intensities of debonding and the stress
intensity factors are also shown.
Some assumptions and the features of the analysis used in this paper are as
Downloaded by [University of California Davis] at 10:13 11 April 2014

follows:

1. The treated target is the elastic body which is homogeneous, isotropic, satisfying
the generalized Hooke’s law and the small deformation assumption, which is so-
called linear elastic.
2. The problems considered are static.
3. The method is the complex variable method. One of the biggest advantages of
using the complex variable method is that a mapping function can be introduced
for the analysis. Moreover, its formulation is not complicated compared to the
real variable method.
4. The introduction of the rational mapping function enables us to obtain the ana-
lytical solutions for not only the particular shapes but also a comparatively arbi-
trary shape of boundary.
5. It is generally not easy to solve the Fredholm integral equation that is necessary
in the analysis of the 2D elasticity problem. However, when the mapping function
is a rational one, the solution can be obtained in the closed form.
6. The exact solution is obtained for the shape that is expressed by the rational map-
ping function. Especially for the problems in which the accuracy is not guaran-
teed in the numerical computation: the stress distribution with a significant
variation such as the stress concentration for a notch, the stress intensity factors
for a crack or a bimaterial interface can be obtained with good accuracy.
7. The rational mapping function is formed by the sum of fractional expressions.
The implementation needs some mathematical procedures.
8. Since the solution is obtained in the general form, what should be done becomes
the fabrication of the rational mapping function of the configuration, which will
be analyzed and the determination of the corresponding boundary condition.

BASIC EQUATIONS IN THE PHYSICAL PLANE

Temperature Field
The temperature function Tðx; yÞ for an elastic body should be a harmonic
function satisfying the following heat conduction equation
@2T @2T
þ 2 ¼0 ð1Þ
@x2 @y
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 597

When complex variables z ¼ x þ iy;
z ¼ x
iy are used, the temperature function

Tðx; yÞ can be presented as a real part of a complex function X0 ðzÞ as
1h i
Tðx; yÞ ¼ X0 ðzÞ þ X0 ðzÞ ð2Þ
2
The components of heat flux qx ; qy in the x and y directions can be expressed as

qx
iqy ¼
kX00 ðzÞ ð3Þ

where k denotes the heat conductivity of the material. The boundary condition for
the given temperature can be obtained from Eq. (2) as

X0 ðzÞ þ X0 ðzÞ ¼ 2Tðx; yÞ ð4Þ

Downloaded by [University of California Davis] at 10:13 11 April 2014

The heat flux boundary condition is

Z
2i
X0 ðzÞ
X0 ðzÞ ¼
qn ðsÞds þ const: ð5Þ
k
where qn ðsÞ represents the normal heat flux for the point s along the boundary [15].
Eqs. (4) and (5) can be expressed in a unified form from the mathematical point
of view (see Eq. (22)).

Thermal Stress Field

The basic equations in complex variable forms for the thermoelastic problem
are stated. First, the stress components can be expressed in terms of Airy stress func-
tion F ðx; yÞ as

@2F @2F @2F

rx ¼ ry ¼ sxy ¼
ð6Þ
@y2 @x2 @x@y

By using Eq. (6), the compatibility equation becomes biharmonic, as follows:

@4F @4F @4F

þ 2 þ ¼0 ð7Þ
@x4 @x2 @y2 @y4

of which the solution is given by

1h i
F ðx; yÞ ¼ Re½
zuðzÞ þ vðzÞ ¼
zuðzÞ þ zuðzÞ þ vðzÞ þ vðzÞ ð8Þ
2
where uðzÞ; vðzÞ are analytic functions in the considered region [2]. The stress
components are then expressed using v0 ðzÞ  wðzÞ as
h i
rx þ ry ¼ 2 u0 ðzÞ þ u0 ðzÞ ¼ 4Re½u0 ðzÞ
ð9Þ
ry
rx þ 2isxy ¼ 2½
zu00 ðzÞ þ wðzÞ

For the thermoelastic problem, the temperature in Eq. (2) may generate a
thermal displacement. The entire displacement including the thermal effect can be
 598 N. HASEBE AND X. WANG

obtained as
Z
1 h i
u þ iv ¼ juðzÞ
zu0 ðzÞ
wðzÞ þ a0 X0 ðzÞdz ð10Þ
2G
where G represents the shear modulus of the material. j and a0 are the coefficients
consisting of the Poisson’s ratio n and the linear thermal expansion a0 as

plane strain state j ¼ 3
4n a0 ¼ ð1 þ n Þa0 ð11Þ

3
n
plane stress state j¼ a0 ¼ a0 ð12Þ
1þn
The resultant force between two points in an elastic body can be obtained by the fol-
Downloaded by [University of California Davis] at 10:13 11 April 2014

lowing equation
Z
uðzÞ þ zu0 ðzÞ þ wðzÞ ¼ i ðpx þ ipy Þds þ const: ð13Þ

When px and py are applied to the boundary, this equation becomes the external
force boundary condition and the const. term can be taken as zero. The displacement
boundary condition is obtained from Eq. (10) as
Z
0 0
juðzÞ
zu ðzÞ
wðzÞ ¼ 2Gðu þ ivÞ
2Ga X0 ðzÞdz ð14Þ

BASIC EQUATIONS IN THE MAPPED PLANE

The basic equations for the temperature and thermoelastic field analysis are
given here when a mapping function z ¼ xðfÞ is introduced.

Temperature Field
The complex temperature function becomes

X0 ðzÞ ¼ X0 ½xðfÞ  XðfÞ ð15Þ

The temperature Tðx; yÞ and the heat flux components qx ; qy are then
1h i
Tðx; yÞ ¼ XðfÞ þ XðfÞ ð16Þ
2

qx
iqy ¼
kX0 ðfÞ=x0 ðfÞ ð17Þ

where k signifies the heat conductivity. The heat flux components in the curvilinear
coordinates generated by the mapping function are

fx0 ðfÞ
qr þ iqh ¼ ðqx þ iqy Þ ð18Þ
jfx0 ðfÞj
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 599

The net heat flux along a line between two points can be given by [15]
Z
2i
XðfÞ
XðfÞ ¼
qn ðsÞds þ const: ð19Þ
k
When considering the boundary point f ¼ r, the boundary conditions for tempera-
ture and heat flux are as follows:
For temperature boundary condition,

XðrÞ þ XðrÞ ¼ 2Tðx; yÞ  2Tðr; r

Þ ð20Þ

For heat flux boundary condition,

Z
2i
Downloaded by [University of California Davis] at 10:13 11 April 2014

XðrÞ
XðrÞ ¼
qn ðsÞds þ const: ð21Þ

k

Eqs. (20) and (21) are mathematically identical when they are used for determination
of function XðfÞ. When the right-hand sides in Eqs. (20) and (21) are zero, they can
be combined into a single equation as
XðrÞ
CXðrÞ ¼ const: ð22Þ
where C ¼ 1 for the adiabatic condition, and C ¼
1 for the isothermal condition.

Thermal Stress Field

Introducing the mapping function, the complex stress functions become
uðzÞ ¼ u½xðfÞ  / ðfÞ
ð23Þ
v0 ðzÞ ¼ v0 ½xðfÞ  wðfÞ
The stress components in Eq. (9) and the displacements in Eq. (10) can be expressed as

0 
/ ðfÞ
rx þ ry ¼ 4Re 0
x ðfÞ
" #
 0
xðfÞ /0 ðfÞ w0 ðfÞ
ry
rx þ 2isxy ¼ 2 0 þ 0 ð24Þ
x ðfÞ x0 ðfÞ x ðfÞ
" # Z
1 xðfÞ 0
u þ iv ¼ j/ðfÞ
/ ðfÞ
wðfÞ þ a0 XðfÞx0 ðfÞdf
2G x0 ðfÞ

And the components of stress and displacements in the curvilinear coordinates ðr; hÞ
expressed by the mapping function are [2]

rr þ rh ¼ rx þ ry ð25Þ
f2 x0 ðfÞ
rh
rr þ 2isrh ¼ ðry
rx þ 2isxy Þ
jfj2 x0 ðfÞ
fx0 ðfÞ
ur þ ivh ¼ ðu þ ivÞ
jfx0 ðfÞj
 600 N. HASEBE AND X. WANG

Since the principle of superposition is applicable, without loss of generality, the

external forces or displacements on the boundary may be let zero for the thermo-
elastic problem. The stress and displacement boundary conditions are, respectively,

xðrÞ
/ðrÞ þ /0 ðrÞ þ wðrÞ ¼ 0
x0 ðrÞ
Z ð26Þ
xðrÞ 0 0 0
j/ðrÞ
/ ðrÞ
wðrÞ ¼
2Ga XðrÞx ðrÞdr
x0 ðrÞ

where r represents f on the boundary.

Downloaded by [University of California Davis] at 10:13 11 April 2014

RATIONAL MAPPING FUNCTION

Although the thermoelastic problem of an infinite plane containing a hole
is considered in this paper, the rational mapping function for an arbitrary shape is
stated here. The rational mapping function is expressed by the sum of fractional
expressions.
The shapes of two-dimensional elastic bodies can be classified into three cases:

1. A finite region mapped into a unit circle

XN
Ek
z ¼ xðfÞ ¼ þ Ec ðf  1Þ ð27Þ
f
f
k¼1 k

or
XN
Ek
z ¼ xðfÞ ¼ E0 f þ þEc ðf  1Þ ð28Þ
f
k¼1 k

f

In the case of mapping a finite region, it is convenient to map the region into a
unit circle to understand the corresponding points on the physical and mapping
regions. This mapping function was used for analyzing a strip with cracks [5–7].
A circular region is mapped into a unit circle when using E0 6¼ 0; Ek ¼ 0
ðk ¼ 1; 2; 3; . . . ; nÞ in Eq. (28).
2. A semi-infinite region mapped into a unit circle

E0 X
N
Ek
z ¼ xðfÞ ¼ þ þ Ec ðf  1Þ ð29Þ
1
f k¼1 fk
f

This mapping function is used for a semi-infinite region with a notch or a mound.
The infinite point on the physical plane corresponds to f ¼ 1 on the unit circle. A
flat half plane is mapped into a unit circle when the coefficients,
E0 6¼ 0; Ek ¼ 0 ðk ¼ 1; 2; 3; . . . ; nÞ are used in Eq. (29). This mapping function
was used in the analysis of a semi-infinite plane with a notch or a crack [8–10].
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 601

3. An infinite region with a hole mapped onto the exterior of a unit circle
XN
Ek
z ¼ xðfÞ ¼ E0 f þ þ Ec ðf  1Þ ð30Þ
f
f
k¼1 k

XN
Ek ENþ1
z ¼ xðfÞ ¼ E0 f þ þ þ Ec ðf  1Þ ð31Þ
f
k¼1 k

f f

The infinite point in the physical plane corresponds to the infinite point in the
mapped plane. It is convenient to determine the corresponding points by
mapping an infinite region with a hole onto the exterior of a unit circle. The map-
ping functions of the coefficients, E0 6¼ 0; Ek ¼ 0 ðk ¼ 1; 2; 3; . . . ; nÞ and
Downloaded by [University of California Davis] at 10:13 11 April 2014

E0 6¼ 0; E1 6¼ 0; Ek ¼ 0 ðk ¼ 2; 3; . . . ; nÞ are ones of the infinite region outside a

circular hole and an elliptical hole, respectively, onto the exterior of a unit circle.
Eq. (31) is a special case of Eq. (30). This mapping function was used in the
analysis of an infinite plane containing cracks or holes [11–13].

Since problems of an infinite region with a hole are stated in this paper, the fabri-
cation of this kind of rational mapping function is shown by simple examples.

Example 1 [13]
The mapping of the exterior of the cracked elliptical hole in the z-plane onto
the exterior of a unit circle in the f-plane is considered as shown in Figure 1, in which
parameters e, a and b denote the length of edge crack, the half-axes of the elliptical
hole in the x and y directions, respectively. The shape parameters e1 ¼ e=a and
k ¼ b=a are used in the analysis in which k ¼ 1 represents a circular hole, while
k 6¼ 1 represents an elliptical hole.
From Figure 1(a, b), we have the mapping
qffiffiffiffiffiffiffiffiffiffiffiffiffi
z ¼ f1 þ k f21
1 ð32Þ

Through coordinate translation and scaling, we obtain

f1 ¼ ð1 þ h1 Þf2 þ h1 ð33Þ
where

e21
h1 ¼ ðk ¼ 1Þ
4ð1 þ e1 Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ðe1 þ k2 Þ
2k k2 þ 2e1 þ e21
h1 ¼ ðk 6¼ 1Þ ð34Þ
4ð1
k2 Þ
Mapping the slit onto the unit circle, we have
 
1 1
f2 ¼ fþ ð35Þ
2 f
 602 N. HASEBE AND X. WANG
Downloaded by [University of California Davis] at 10:13 11 April 2014

Figure 1 Mapping function for an elliptical hole with an edge crack.

The mapping function is obtained by combination of the procedure (32)–(35) as

     
1 þ h1 1 1 þ h1 c1 0:5 c2 0:5
z ¼ xðfÞ ¼ fþ þ h1 þ k ðf þ 1Þ 1
1
ð36Þ
2 f 2 f f

where
 pffiffiffiffiffi
c1 1
h1 2 h1
¼ i ð37Þ
c2 1 þ h1 1 þ h1
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 603

The irrational terms in (36) are replaced by the rational ones expressed by the sum of
fractional expressions as
  X12  
c1 0:5
Aj
1
¼1þ þ Aj ð38Þ
f j¼1
1
c1 aj =f

The expressions for c2 can be obtained by merely changing c1 with c2 in the above
equation. The determination of coefficients Aj ; aj can be referred to Appendix 1 or
Refs. [5,9,10]. Substituting Eq. (38) and the expression for c2 into Eq. (36), and
making some arrangement, the general form of Eq. (31) is obtained. In this case,
we chose N ¼ 24.
Downloaded by [University of California Davis] at 10:13 11 April 2014

Example 2 [17]
As shown in Figure 2, the mapping function that maps the exterior of a square
hole onto the infinite region (S+) outside a unit circle (in the f-plane) can be obtained
through the Schwarz–Christoffel transformation as follows:
Z  
1 0:5
z¼K 1
4 df ð39Þ
f

where K is a constant related to the scale of the original configuration. The mapping
function is conformal at all exterior points except at the corners on the boundary.
Such nonconformity may cause the transformation not to be reversible and single-
valued, which gives rise to difficulties in the analysis. To deal with this problem,
the mapping function needs to be rationalized. The above expression can be
expanded to a series as:
 
a1 a2 a3
z ¼ xðfÞ ¼ K f þ 3 þ 7 þ 11 þ    ð40Þ
f f f

Figure 2 Rectangular hole and a unit circle.

 604 N. HASEBE AND X. WANG

Now, a fractional expression with n terms is considered:

( )
X n
Bj bj =f3 Xn
Bj bj Bj b2j Bj b3j
4
¼ þ 7 þ 11 þ    ð41Þ
j¼1 1
bj =f j¼1 f3 f f

 
Here, we have bj  < 1 because Eq. (41) can be expanded to a series. Coefficients Bj
and bj can be determined by equalizing the terms except the first one in Eq. (40) with
the fractional terms in Eq. (41). A comparison of coefficients leads us to establish the
following equations:

X
n
Bj bkj ¼ ak ðk ¼ 1; 2; 3; . . . ; 2nÞ ð42Þ
Downloaded by [University of California Davis] at 10:13 11 April 2014

j¼1

The following simultaneous linear equations with ak being the coefficients in n

unknowns are considered

a1 x1 þ a2 x2 þ    þ an xn þ anþ1 ¼ 0
a2 x1 þ a3 x2 þ    þ anþ1 xn þ anþ2 ¼ 0
.. .. ð43Þ
. .
an x1 þ anþ1 x2 þ    þ a2n
1 xn þ a2n ¼ 0

bj can be obtained from the following equation of power n, of which the coefficients
x1 ; x2 ; . . . ; xn are chosen as the solution of Eq. (43) as

bn þ xn bn
1 þ    þ x2 b þ x1 ¼ 0 ð44Þ

Substituting the n solutions bj obtained from Eq. (44) into Eq. (42), Bj can be
obtained from a system of n linear equations. Finally, the mapping function can
be expressed in terms of Bj ; bj and rearranged in the general form as

z ¼ xðfÞ
( )
X
n
Bj bj =f3
¼K fþ þ const
1
bj =f4
j¼1
2 8 pffiffiffi pffiffiffi pffiffiffi pffiffiffi 9 3
Xn < B j bj Bj b j Bj b j Bj b j =



¼ K 4f þ 4
þ 4
þ 1=44 þ 4
þ const5
j¼1
: b
1=4

f
b
1=4

f ib
f
ib
1=4

f ;
j j j j

X
N
Ek
 E0 f þ þ E
1 ðN  4nÞ ð45Þ
f
f
k¼1 k

In [17], n ¼ 12 was used.

For a more complicated configuration, the mapping function is obtainable
using the methods shown in examples 1 and 2 [5,7].
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 605

ANALYSIS OF TEMPERATURE FIELD

Problems for Uniform Heat Flux
The problems of an infinite plane with a hole under uniform heat flux, as
shown in Figure 3, are considered [15, 16]. The mapping functions (30) and (31)
for an infinite region containing a hole are used.

Adiabatic boundary condition The condition that a heat flow does not pass
through a hole boundary, i.e., the adiabatic condition, is considered in this section.
Since we have known that the uniform heat flux is applied at infinity, the
complex temperature function to be determined can be expressed in two parts as:

XðfÞ ¼ X1 ðfÞ þ X2 ðfÞ ð46Þ

Downloaded by [University of California Davis] at 10:13 11 April 2014

where X1 ðfÞ is obtained from Eq. (17) by using condition of uniform heat flux as
q
X1 ðfÞ ¼
e
id xðfÞ ð47Þ
k

Here q represents the heat flux per unit area, d denotes the flux direction. Substitut-
ing Eqs. (46) and (47) into the boundary condition (21) with considering the
condition qn ¼ 0, multiplying dr=½2piðr
fÞ on both sides and carrying out
the Cauchy integration, X2 ðfÞ is obtainable. Finally, Eq. (46) is obtained by some
arrangement as follows:
 
q
id E 0 id
XðfÞ ¼
e E0 f þ e þ const: ð48Þ
k f

The constant term can be determined from the temperature at a standard point.

Isothermal boundary condition Next, we consider the condition that the

temperature is known on the boundary. Here, typically, the isothermal condition
(constant temperature along the boundary) is considered. Substituting Eqs. (46)
and (47) into the boundary condition Eq. (20) with considering the condition
Tðr; r
Þ ¼ 0, multiplying by dr=½2piðr
fÞ and carrying out the Cauchy integration,

Figure 3 Kinked crack under uniform heat flux.

 606 N. HASEBE AND X. WANG

X2 ðfÞ is obtainable. Finally, Eq. (46) can be expressed as follows:

 
q
id E 0 id
XðfÞ ¼
e E0 f
e þ const: ð49Þ
k f

The constant term can be determined from the temperature at a standard point.
Eqs. (48) and (49) can be combined into a single equation by using boundary
condition (22) as
 
q
id E 0 id
XðfÞ ¼
e E0 f þ C e þ const: ð50Þ
k f
Downloaded by [University of California Davis] at 10:13 11 April 2014

where C ¼ 1 for the adiabatic condition, while C ¼
1 for the isothermal condition.

Point Heat Source Problems [14,17]

The problem of a heat source located at an arbitrary point in an infinite plane
with a hole is considered as shown in Figure 4.
The temperature function for a point heat source at fa in the mapped plane is

M
X1 ðfÞ ¼
logðf
fa Þ ð51Þ
2pk

where M is the intensity of the heat source (see Appendix 2). þM denotes the gen-
erating heat source and
M represents the heat sink source. The intensity of M in the
above equation can be determined from the condition that the net heat flux around
the heat source is M in Eq. (5). In the follows, the isothermal and the adiabatic
boundary conditions are considered.

Heat source at an arbitrary position and heat sink at infinity The tem-
perature function to be determined is expressed by Eq. (46), in which X1 ðfÞ is given
by Eq. (51). Substituting Eq. (46) into Eq. (21), multiplying by dr=½2piðr
fÞ and

Figure 4 Heat and sink sources in an infinite plane with a hole.

 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 607

carrying out the Cauchy integration, X2 ðfÞ is obtainable. Finally, Eq. (46) can be
expressed as follows:
  
M f
f0a
XðfÞ ¼
logðf
fa Þ þ C log þ const: ð52Þ
2pk f

where f0a  1=
fa .

Heat source and heat sink at respective arbitrary positions Consider a

point heat source and a point heat sink located at fa ; fb , respectively (Figure 4).
X1 ðfÞ can be obtained by the superposition of the cases of þM and
M in Eq.
(51) as
Downloaded by [University of California Davis] at 10:13 11 April 2014

 
M f
fa
X1 ðfÞ ¼
log ð53Þ
2pk f
fb

The temperature function can be obtained with the same procedure as in (a). It is
also obtainable from the superposition of Eq. (52) and the equation for a point heat
sink.
    
M f
fa f
f0a
XðfÞ ¼
log þ C log þ const: ð54Þ
2pk f
fb f
f0b

where f0a  1=
fa ; f0b  1=
fb .

Mixed Boundary Value Problem

As shown in Figure 5, the analysis of the mixed boundary value problem is sta-
ted, in which the heat flux is given on one part of the boundary and the temperature
is given on the remaining part of the boundary.

Figure 5 Mixed heat conduction boundary.

 608 N. HASEBE AND X. WANG

Uniform heat flux [18] The mixed heat boundary value problem under uni-
form heat flux is considered for that one part of the boundary, denoted by L, is
assumed adiabatic, while the remaining part of the boundary, denoted by M, is
assumed isothermal. The boundary condition can be expressed as follows:

XðrÞf1
2DðrÞg
XðrÞ ¼ const


0 on L ð55Þ
DðrÞ ¼
1 on M

where boundaries L and M correspond to Eqs. (21) and (20), respectively. The func-
tion to be determined can be expressed in the form of Eq. (46). Substituting Eq. (46)
Downloaded by [University of California Davis] at 10:13 11 April 2014

with (47) into Eq. (55) and making some arrangement, we have

X2 ðrÞf1
2DðrÞg
X2 ðrÞ ¼
X1 ðrÞf1
2DðrÞg þ X1 ðrÞ þ const: ð56Þ

Consider the Plemelj function

vðfÞ ¼ ðf
aÞ0:5 ðf
bÞ0:5 ð57Þ

which has the following relationship on the boundary:

v
ðrÞ ¼ vþ ðrÞf1
2DðrÞg ð58Þ

where a; b denote the coordinates of joint points for boundaries L and M on the unit
circle. Multiplying Eq. (56) by dr=½2piðr
fÞv
ðrÞ and carrying out the Cauchy in-
tegration, the temperature function is obtained as

0
q
id eid E
XðfÞ ¼
vðfÞ e E0 þ þ const: ð59Þ
k vð0Þf

Point heat source and point heat sink for elliptical hole [19] When a
point heat source is applied along with a point heat sink, the temperature function
for the mixed boundary value problem with an adiabatic and an isothermal bound-
aries and the temperature is zero at infinity, is given as follows:
8 0qffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffi1 0qffiffiffiffiffiffiffiffi ffi1
qffiffiffiffiffiffiffi 9
> fa
a fa
b f0a
a 0
fa
b  0 >
M< B f
a þ f
b C B f
a þ 1
f
b C ðf
fa Þðf
fa Þ =
XðfÞ ¼ log@qffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiA þ log@qffiffiffiffiffiffiffi
ffi q ffiffiffiffiffiffiffi
ffi A
log
pk >
: fb
a
þ fb
b f0b
a
þ b
f0
b 2 ðf
fb Þðf
f0b Þ >
;
f
a f
b f
a f
b
8 0qffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffi19
> pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffi 0 0 >
M< fa
a þ fa
b B fa
a þ fa
bC
=

log pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffi
log@qffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiA ð60Þ
pk >
: fb
a þ fb
b f0b
a þ f0b
b ;
>
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 609

EXTERNAL FORCE BOUNDARY VALUE PROBLEM

An infinite plane containing a hole under uniform heat flux or point heat
sources is analyzed in this chapter. The mapping function (30) is used. Without loss
of generality, the hole boundary can be treated as traction free.

Uniform Heat Flux Problem [15,16,36]

The temperature function for the adiabatic boundary condition is given by Eq.
(48) (Figure 3). The external force boundary condition is expressed by the first equa-
tion in Eq. (26).
Substituting Eq. (48) into the 3rd equation in Eq. (24) and carrying out the
Downloaded by [University of California Davis] at 10:13 11 April 2014

integration, the logarithmic function appears. This term is of a multiple value that
does not return to the starting value when moving round the hole, i.e., it causes
the dislocation in the displacement. Since a displacement must be single valued, a
stress function that may cancel this dislocation is considered as follows [20,21]:

/1 ðfÞ ¼ A log f
ð61Þ
w1 ðfÞ ¼ B logf

Substituting the above equation into Eq. (26), the resultant force must return
to the starting value when moving round the hole. Furthermore, substituting these
equations into the 3rd equation in Eq. (24), the displacement must return to the
starting value when moving round the hole. From these single-valued conditions,
the following equation are obtained:

B¼A
!
a0 qGR XN
A¼ E0 e
id
0
Ek þ eid E
2k k¼1 ð62Þ
1þn
R¼ Plane strain state
1
n
R¼1þn Plane stress state

Regarding Eq. (61), the stress functions to be determined can be expressed as

follows:

/ðfÞ ¼ /1 ðfÞ þ /2 ðfÞ

ð63Þ
wðfÞ ¼ w1 ðfÞ þ w2 ðfÞ

Substituting the above equation into the first equation in Eq. (26), multiplying by
dr=½2piðr
fÞ and carrying out the Cauchy integration, we have

X
N
Ak Bk XN
fk Bk
/2 ðfÞ þ ¼
A ð64Þ
f
k¼1 k

f f
k¼1 k

f
 610 N. HASEBE AND X. WANG

where Bk  Ek =x0 ðf0k Þ, Ak  /02 ðf0k Þ; f0k  1=
fk . Assuming f ¼ f0j ðj ¼ 1; 2; 3; . . . ; NÞ in

the derivative of Eq. (64), a system of algebraic equations Aj ¼ /02 ðf0j Þ regarding the
real and imaginary parts of the unknowns Ak can be obtained and then Ak are
decided. Function wðfÞ can be obtained from the following procedures: taking the
conjugate form of the external force boundary condition, 1st equation in Eq. (26),
multiplying by dr=½2piðr
fÞ and carrying out the Cauchy integration, the 2nd
equation in Eq. (63) is finally obtained as

XN
0 XN
2

ð1=fÞ 1
x B k fk
ð1=fÞ 0
x Bk Ak f0k
wðfÞ ¼ A log f
A
A 0
/2 ðfÞ
0 þ const:
x0 ðfÞ f f
f
k¼1 k
x0 ðfÞ k¼1
fk
f
ð65Þ
Downloaded by [University of California Davis] at 10:13 11 April 2014

However, it can be directly obtained from the analytical continuation on the traction
free boundary on which no external force is applied [2].

x 1=
f 0
wðfÞ ¼
/ 1=
f
0 / ðfÞ ð66Þ
x ðfÞ

Uniform Heat Flux with Mixed Heat Boundary Condition [18]

The stress function for the mixed heat boundary condition under uniform heat
flux can be derived by the same way in the above section (see Figure 5). However the
coefficient A in (62) is given by the following equation:
!
a0 qGRE0 X
N
eid E 0 ða þ bÞ 1
id
A¼ e
id
Ek

e E0 ða
bÞ2 ð67Þ
2k k¼1
2vð0Þ 8

where a; b; vð0Þ are given in Eq. (57).

Heat Source and Heat Sink at Arbitrary Positions [14,17]

Consider a heat source and a heat sink at fa ; fb in the mapped plane, respect-
ively (see Figure 4). The temperature function is given by Eq. (54). The stress func-
tions to be determined are expressed by Eq. (63).
The terms for a heat source in an infinite plane and the terms to remove the
thermal dislocation in the displacement are considered as follows (see Appendix 2):

a0 MGR
/1 ðfÞ¼ ½fxðfÞ
xðfa Þglogðf
fa Þ
fxðfÞ
xðfb Þglogðf
fb ÞþAlogf
4pk ð68Þ
a0 MGR h i
w1 ðfÞ¼
xðfa Þlogðf
fa Þ
xðfb Þlogðf
fb Þ þBlogf
4pk

The coefficients A and B of the function log are determined from the condition that
the stresses in the external force boundary condition should be single-valued when
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 611

moving round the hole,

B¼A ð69Þ

Further, substituting Eq. (68) into Eq. (24), A is obtained from the condition that the
displacement is single-valued when moving round the hole,
( )
a0 MGR XN
Ek X
N
Ek
0 0
A¼
CE0 ðfa
fb Þþ
ð70Þ
4pk f
fa k¼1 fk
fb
k¼1 k

Substituting Eq. (63) into the external force boundary condition, multiplying by
dr=½2piðr
fÞ and carrying out the Cauchy integration, we have
Downloaded by [University of California Davis] at 10:13 11 April 2014

   
a0 MGR 1 1
/2 ðfÞ ¼
fxðfÞ
xðfa Þg log
fa
fxðfÞ
xðfb Þg log
fb
4pk f f

 X N  
f Ek f
fa

E0 f log
a þ log k
fb f
k¼1 k

f fk
fb
( )#
XN
Bj X N
E
k E
k
þ

j¼1 j
f
f k¼1 ð
fk

f0j Þð
fk

fa Þ ð
fk

f0j Þð
fk

fb Þ
X
N
k þ A
ðA
fk ÞBk

þ const: ð71Þ
k¼1
fk
f

where Ak  /02 ðf0k Þ; f0k  1=
fk . The real and the imaginary parts of the unknown con-
stants Ak are determined from a system of algebraic linear equations obtained by as-
suming f ¼ f0j ð j ¼ 1; 2; 3; . . . ; NÞ in the derivative Aj ¼ /02 ðf0j Þ. The final stress
function /ðfÞ is expressed by the 1st equation in Eq. (63), while function wðfÞ can
be directly obtained from analytical continuation by Eq. (66).
From the stress function /ðfÞ, we can see that the point of the heat source will
not be a singular point when the heat source is located on the boundary, i.e., the
stresses on the traction free boundary will not become infinite due to no displace-
ment restraint [14,17].

Heat Source at an Arbitrary Position and Heat Sink at Infinity [14,17]

Consider a heat source located at fa in the mapped plane. The temperature
function is given by Eq. (52). The stress functions to be determined are expressed
by Eq. (63), in which /1 ðfÞ; w1 ðfÞ are expressed as follows (see Appendix 2):

a0 MGR
/1 ðfÞ ¼ fxðfÞ
xðfa Þgflogðf
fa Þ
1g þ A log f
4pk ð72Þ
a0 MGR
w1 ðfÞ ¼
xðfa Þ logðf
fa Þ þ B log f
4pk
 612 N. HASEBE AND X. WANG

The coefficients A and B are

( )
a0 MGR XN
Ek
A¼
CE0 f0a þ
4pk f
fa
k¼1 k ð73Þ
B¼A

Using the same procedure as before, we obtained

"

a0 MGR 1    X N
Ek
/2 ðfÞ ¼
fxðfÞ
xðfa Þglog
fa
E0 flog
fa
2
4pk f f
k¼1 k

f
( )#
XN XN X N k
Downloaded by [University of California Davis] at 10:13 11 April 2014

Ek Bj 0 þ E
þ logðfk
fa Þ þ E
f
f f
f  0  
k¼1 k j¼1 j k¼1 ðfk
fj Þðfk
fa Þ
XN k þA
ðA  fk ÞBk

þ const: ð74Þ
k¼1
fk
f

where the real and imaginary parts of constants Ak can be determined from the
relations Ak  /02 ðf0k Þ; f0k  1=fk . Finally, the stress function /ðfÞ is expressed by the
1st equation in Eq. (63). Function wðfÞ can be obtained from analytical continuation
by Eq. (66).
The stress functions (68) and (71) for a point heat source and sink at fa ; fb ,
respectively, are also obtainable by the superposition of the function with an inten-
sity of heat M and sink (
M) at fa and fb , respectively, using Eqs. (72) and (74).
However, inversely, the results for a sink source at infinity cannot be obtained from
the stress functions (68) and (71).

Heat Source at Arbitrary Position and Heat Sink at Infinity for

Elliptical Hole [14]
The mapping function that maps the exterior of an elliptical hole onto the
exterior of a unit circle can be expressed as a special case of Eq. (30) as

E1 aþb a
b
z ¼ xðfÞ ¼ E0 f þ ; E0 ¼ ; E1 ¼ ð75Þ
f 2 2

a; b are the half axes of the ellipse, respectively, at the x and y directions. Using this
mapping function and the method previously stated, the stress function /ðfÞ can be
obtained as

a0 MGR
/ðfÞ ¼ A log f þ fxðfÞ
xðfa Þgflogðf
fa Þ
1g
4pk

 
1   E1 E1

fxðfÞ
xðfa Þg log
fa þ E0 f log
fa
logð
fa Þþ ð76Þ
f f f
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 613

where
 
a0 MGR 0 E1
A¼
CE0 fa þ ð77Þ
4pk fa

The solutions for a point heat source and heat sink at respective arbitrary positions
can be obtained by the superposition stated in the previous section Eqs. (68)–(71).
Function wðfÞ can be obtained from analytical continuation by Eq. (66).

DISPLACEMENT BOUNDARY VALUE PROBLEM

Consider an infinite plane with the displacement restraint boundary under uni-
Downloaded by [University of California Davis] at 10:13 11 April 2014

form heat flux or point heat sources. For simplicity, the zero displacement boundary
condition is used since the principle of superposition is applicable. Such a boundary
condition corresponds to the models of a rigid reinforced ring around a hole or a
rigid inclusion in the elastic plate. It is assumed that the heat flux does not flow in
or out through the boundary in either of the cases. The rigid inclusion is assumed
fixed in the plane without rigid body rotation. The relative displacement between
the inclusion and the point heat source is not restrained. The solutions for the
inclusion with a rigid body rotation and with zero relative displacement between
the inclusion and the heat source have been considered in [22].

Uniform Heat Flux Problem [23]

The mapping function is expressed by the general formula Eq. (30). The
temperature function for the adiabatic boundary condition is given by Eq. (48).
The displacement boundary condition is expressed by the 2nd equation in Eq. (26).
The stress functions to be determined are expressed by Eq. (63), in which
/1 ðfÞ; w1 ðfÞ can be obtained from Eqs. (61) and (62). Substituting Eq. (63) into
the 2nd equation of the displacement boundary condition (26), multiplying it by
dr=½2piðr
fÞ and carrying out the Cauchy integration, we obtain

XN
fk B k X
N
Ak Bk 2Ga0 q
j/2 ðfÞ ¼ A
þ F ðfÞ
f
f k¼1 fk
f
k¼1 k
k
!
X
N
f X
N
E k fk

id
F ðfÞ ¼ E0 e
Ek log þ ð78Þ
k¼1
fk
f k¼1 fk
f
!
X
N
Ek f XN
Ek 1

0e
þE id
log þ
2 f
f f f
f
k¼1 fk k k¼1 k k

where Bk  Ek =x0 ðf0k Þ, Ak  /02 ðf0k Þ; f0k  1=
fk . The unknown constants Ak are
determined as follows: Using the Aj ¼ /02 ðf0j Þ, ðj ¼ 1; 2; 3; . . . ; NÞ, the real and the
imaginary parts of Aj ; Aj are determined by solving the resulting 2N linear algebraic
equations. Function wðfÞ can be obtained from the following procedures: taking the
conjugate form of the displacement boundary condition Eq. (26), multiplying it by
dr=½2piðr
fÞ and carrying out the Cauchy integration, the final function wðfÞ is
 614 N. HASEBE AND X. WANG

obtained as

XN
k f0 2 XN
f0k B

k a0 GqE
2 eid 1

log f
x 1=f /0 ðfÞ

wðfÞ ¼ A
Ak B k

A
0
ð79Þ
0 0
x0 ðfÞ k¼1
f k
f k¼1
f k
f k f2

However, it can be directly obtained from the analytical continuation on the zero
displacement boundary condition as
 
Z 
1 x 1=
f 0 0 0

wðfÞ ¼ j/

0 / ðfÞ þ 2Ga XðfÞx ðfÞd f ð80Þ
f x ðfÞ f
¼1=f
Downloaded by [University of California Davis] at 10:13 11 April 2014

Heat Source at Arbitrary Position and Heat Sink at Infinity [22]

Consider a heat source located at fa in the mapped plane. The temperature
function is given by Eq. (52). The stress functions to be determined are expressed
by Eq. (63), in which /1 ðfÞ; w1 ðfÞ have the same forms as those in Eqs. (72) and
(73). Substituting the stress functions Eq. (63) into the displacement boundary con-
dition, multiplying it by dr=½2piðr
fÞ and carrying out the Cauchy integration
along the unit circle, we have

1X N
A
k Bk a0 MGR X

fk þ A N
Ek 1 a0 MGR
/2 ðfÞ ¼ þ þ
j k¼1 fk
f 4pk k¼1 fk
f j 4pk
"  
XN
Ek 1
 logðfk
fa Þ þ fxðfÞ
xðfa Þg log
f
a
E0 f log
f
a
k¼1 k
f
f f
8 93
Z
XN
Ek X N
Bj <
XN
k
E =
5 1 1 F ðrÞ

þ E0 þ
dr
k¼1 k
f
f j¼1 fj
f :

k¼1 fk
fj

f
k
f
a
0 ; j 2pi r
f
Z "     
1 F ðrÞ a/ MGR f 0 f
dr ¼ ð1 þ jÞ CE0 f log
fa log
2pi r
f 4pk f
f0a f
f0a
X N   XN  
Ek f Ek f
þC log
C log
f
f
k¼1 k
f
f0a k¼1 k
f f
fk
XN   X N  #
Ek f Ek f
f0a
þ log þC log
k¼1 k
f
fa f
fk k¼1 k
f
f0a f
fk
a/ MGR XN
Ek

C ð1 þ jÞ ð81Þ
4pk f
k¼1 k

f

where Bk  Ek =x0 ðf0k Þ, Ak  /02 ðf0k Þ; f0k  1=f
k and C is a constant term in Eq. (52).
Solving a system of 2N algebraic equations Aj ¼ /02 ðf0j Þ ð j ¼ 1; 2; 3; . . . ; NÞ, the real
and the imaginary parts of the unknown constants Aj ; A
j can be determined. Finally,
function /ðfÞ can be obtained from Eq. (63). Similar to the preceding section, func-
tion wðfÞ can be obtained from the analytical continuation over the zero displace-
ment boundary by Eq. (80).
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 615

Heat Source and Heat Sink at Respective Arbitrary Positions

The solution can be obtained by the superposition of the stress functions in the
preceding section Eq. (81) with the original forms and the stress functions for a heat
source of intensity (
M) locating at fb . Also similar to the preceding section, func-
tion wðfÞ can be given by Eq. (80).

MIXED BOUNDARY VALUE PROBLEM UNDER UNIFORM HEAT FLUX

The problems under the mixed boundary condition are considered, in which
the external force is applied to one part of the boundary and the remaining part
of the boundary is given by displacement. An infinite plane with a hole subjected
to uniform heat flux or point heat sources is analyzed. The external force and the
Downloaded by [University of California Davis] at 10:13 11 April 2014

displacement on the boundary can be treated as zero without loss of generality. Zero
displacement can be seen as a model of a partially rigid reinforcement or a partially
debonded rigid inclusion. The heat flux is assumed not to pass through the rigid in-
clusion. The rigid body rotation of the inclusion is zero here. In case that it is taken
into consideration, the solution is obtained in such a way that the sum of the result-
ant moment generated on the inclusion boundary must be zero. From this, the
rotation angle of the inclusion is also determined. The relative displacement between
the point heat source and the inclusion is assumed not restrained.

An Infinite Plane With an Arbitrarily Shaped Hole Under Uniform

Heat Flux [24,25]
The problem of an infinite plane with a hole subjected to uniform heat flux is ana-
lyzed. The number of the segment of the external force and the displacement are
given, respectively, on the boundary is one. The mapping function of Eq. (31) is
used. The solution for the mapping function (30) can be obtained by making
ENþ1 ¼ 0 in the stress function obtained. The temperature function is given by
Eq. (48). The stress functions to be determined are
/ðfÞ ¼ A log f þ /2 ðfÞ
wðfÞ ¼ B log f þ w2 ðfÞ
!
a0 qGR X N
ð82Þ
A¼ E0 Ek e
id þ E 0 eid
ENþ1 e
id
2k k¼1
B¼A

Here the coefficients A and B, as expressed in Eq. (61), are determined using the
conditions that the stress and the displacement components around the hole are
single-valued.
Substituting the function wðfÞ given by Eq. (66) into the external force bound-
ary condition Eq. (26) with the replacement f ¼ r, and r
¼ 1=r, we have
/þ

2 ðrÞ
/2 ðrÞ ¼ 0 ð83Þ

where þ and
indicate the limit values of the function /2 ðfÞ on the unit circle as
its variable approaches the boundary from the physical region (outside the unit circle
 616 N. HASEBE AND X. WANG

in this case) and the nonphysical region (inside the unit circle), respectively. The
displacement boundary condition (2nd equation in Eq. (26)) becomes
Z
þ

j/2 ðrÞþ/2 ðrÞ¼
ð1þjÞAlogr
2Ga XðrÞx0 ðrÞdr
0

"
XN 
0  Ek
a0 qGR E2 E
¼ ð1þjÞ 0 e
id r2 þ E0 e
id fk þ eid
2k 2 k¼1
fk fk
r
! #
XN
0
E

E0 e
id
2 eid Ek flogr
logðfk
rÞg
k¼1 fk
þconst:
Downloaded by [University of California Davis] at 10:13 11 April 2014

ð84Þ

Eqs. (83) and (84) form the Riemann–Hilbert problem of which the solution is
expressed by
Z
vðfÞ HðrÞ
/2 ðfÞ¼ drþQðfÞvðfÞ ð85Þ
2pi MþL vðrÞðr
fÞ

where M and L denote the displacement and the external force boundaries, respect-
ively. Function HðrÞ is taken as the right-hand side of Eqs. (83) and (84) on L and
M, respectively. The Plemelj function vðfÞ is

vðfÞ¼ðf
aÞm ðf
bÞ1
m ð86Þ

m¼0:5
iðlogjÞ=ð2pÞ ð87Þ
where a;b are the joint points of boundaries L and M. The unknown function QðfÞ is
determined from the regularity of function w2 ðfÞ. Finally, the stress function /ðfÞ in
Eq. (82) is expressed as

X
N
k þA
ðA
fk ÞBk a0 qGR
/ðfÞ ¼ A log f
vðfÞ þ
k¼1
vðfk Þðfk
fÞ 2k

XN 
0 
E02
id 2

id E
 e f
vðfÞðf þ ma þ ð1
mÞbÞ
E0 e fk þ eid
2 fk
  ( k¼1 )
0
Ek vðfÞ E0 ENþ1 id 1 v ð0Þf
vð0Þ

1 þ e 2 1 þ vðfÞ
fk
f vðfk Þ 2 f vð0Þ2
!  Z fk 
XN
E
0 dr

id id

E0 e
2 e Ek log f
logðf
fk ÞþvðfÞ ð88Þ
k¼1 fk 0 vðrÞðr
fÞ

where Bk  Ek =x0 ðf0k Þ;f0k  1=f
k . Evaluation of the derivative of the integral in the
above equation can be referred to Appendix 3. The unknown constants Ak are
determined from a system of 2N algebraic equations: Ak  /02 ðf0k Þ. Therefore, Ak
and the stress components can be obtained without numerical quadrature. However,
the determination of displacement needs numerical procedure [27].
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 617

Elliptical Hole Problem under Uniform Heat Flux [26]

The mapping function for an elliptical hole problem is given by Eq. (75).
The temperature function is expressed by Eq. (48). The problem having n boundary
segments of the external force and the displacement, respectively, is considered. The
external force boundaries are on Lj ,

xðrÞ
/ðrÞ þ /0 ðrÞ þ wðrÞ ¼ Cj ð j ¼ 1; 2; 3; . . . ; nÞ on Lj ð89Þ
x0 ðrÞ

and the displacement boundaries are on Mj

Z
Downloaded by [University of California Davis] at 10:13 11 April 2014

xðrÞ 0 0
j/ðrÞ
/ ðrÞ
wðrÞ ¼
2Ga XðrÞx0 ðrÞdr ð j ¼ 1; 2; 3; . . . ; nÞ on Mj ð90Þ
x0 ðrÞ

The constants Cj in Eq. (89) correspond to the resultant forces resulting on the
displacement boundary segments Mj . One of them can be treated as zero due to
the equilibrium condition.
The stress function to be determined can be expressed as

/ðfÞ ¼ /1 ðfÞ þ /2 ðfÞ /1 ðfÞ ¼ A log f

ð91Þ
wðfÞ ¼ w1 ðfÞ þ w2 ðfÞ w1 ðfÞ ¼ B log f

Since the mapping function of Eq. (75) is used, constants A and B can be obtained
from Eq. (62) as

a0 qGR
0 eid
E1 e
id

A¼ E0 E
2k ð92Þ


B¼A

Substituting function wðfÞ in Eq. (66) into the stress and the displacement boundary
condition (26) and using Eq. (91), the following Riemann-Hilbert problem can be
obtained,

/þ

2 ðrÞ
/2 ðrÞ ¼ Cj on Lj
 2
0 E1 
a0 qGR E0
id 2 E ð93Þ
j/þ
2 ðrÞ þ /

2 ðrÞ ¼ ð1 þ jÞ e r þ eid  HðrÞ on Mj
2k 2 2r2

where þ and
indicate the limit values of the function on the unit circle as its vari-
able approaches the boundary from the outside and inside of the unit circle, respect-
ively. Its solution is
Z Z
vðfÞ X
n
HðrÞ vðfÞ X
n
Cj
/2 ðfÞ ¼ dr þ dr ð94Þ
2pi j¼1 Mj vðrÞðr
fÞ 2pi j¼1 Lj vðrÞðr
fÞ
 618 N. HASEBE AND X. WANG

where the Plemelj function is

Y
n
vðfÞ ¼ ðf
aj Þm ðf
bj Þ1
m ð95Þ
j¼1

m ¼ 0:5
iðlog jÞ=ð2pÞ ð96Þ

Here, aj ; bj ð j ¼ 1; 2; 3; . . . ; nÞ are the joint points of boundary segments Lj and Mj .

Finally, /ðfÞ is obtained
" #
a0 qGR E02 e
id
0 E1 v0 ð0Þf
vð0Þ
E
/ðfÞ ¼ A log f þ vðfÞ Vþ eid
2k 2 2 vð0Þ2 f2
Downloaded by [University of California Davis] at 10:13 11 April 2014


0 E  vðfÞ X n Z
a0 qGR E02
id 2 E id Cj
þ e f þ 2 e þ dr ð97Þ
2k 2 2f 2pi j¼1 Lj
vðrÞðr
fÞ

where constant V is determined by the number of boundary segments n

n ¼ 1 V ¼
ff þ ma1 þ ð1
mÞb1 g

n ¼ 2 V ¼
1 ð98Þ
n3 V ¼0

The undetermined constants Cj are determined from the condition that the stresses at
infinity should be zero, i.e., /0 ð1Þ ¼ 0 (see Eqs. (136)–(138)).

MIXED BOUNDARY VALUE PROBLEM FOR AN INFINITE PLANE WITH

AN ELLIPTICAL HOLE SUBJECTED TO HEAT SOURCE
In this section, as shown in Figure 6, a heat source (Green’s function), replac-
ing the uniform heat flux in the previous chapter, applied to an infinite plane with an
elliptical hole is considered. The Green’s function of the external force boundary
value problem is used in the derivation of this mixed boundary value problem. It
can be seen as a model of a rigid inclusion embedded in a plane with debondings.

Figure 6 Elliptic rigid inclusion with debondings with a heat source and sink.
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 619

The rigid body rotation of the inclusion and the restraint of the relative displacement
between the point heat source and the inclusion are not considered.

Heat source and heat sink at respective arbitrary positions [27] Con-
sider a heat source and a heat sink at fa ; fb , respectively, in the mapped plane, in
which n segments of zero displacement and traction free, respectively, are given on
the boundary. The temperature function is given by Eq. (54). The stress functions
to be determined are

/ðfÞ ¼ /1 ðfÞ þ /2 ðfÞ

ð99Þ
wðfÞ ¼ w1 ðfÞ þ w2 ðfÞ
Downloaded by [University of California Davis] at 10:13 11 April 2014

where /1 ðfÞ is the Green’s function of the external force boundary value problem.
The solution for a heat source and a heat sink is obtained by the superposition of
the respective solutions using Eq. (76) as

  
a0 MGR f
/1 ðfÞ ¼ A logðfÞ þ fxðfÞ
xðfa Þg logðf
fa Þ þ log
4pk f
f0a
    0 
f E1 fa fb

fxðfÞ
xðfb Þg logðf
fb Þ þ log
log
f
f0b f f0a fb
ð100Þ

where
 
1 1
A ¼
CE0 ðf0a
f0b Þ þ E1
ð101Þ
fa fb
Substituting the function wðfÞ of Eq. (66) into the stress and the displacement
boundary condition (26) and using Eq. (99), the Riemann–Hilbert problem for deter-
mination of /2 ðfÞ is obtained,

/þ

2 ðrÞ
/2 ðrÞ ¼ Cj on Lj
þ
ð102Þ
j/2 ðrÞ þ /2 ðrÞ ¼ ð1 þ jÞHðrÞ on Mj
Finally, the solution is

 
a0 MGR  f 
/2 ðfÞ ¼ xðfa Þ þ Cxðf0a Þ log 0
xðfb Þ þ Cxðf0b Þ
4pk f
fa
     0
f f
f0a E1 ff
 log þ ð1 þ CÞxðfÞ log þ log a0 b
f
f0b f
f0b f fa fb
    0  Z f0
E1 vðfÞ f f a xðf Þ þ Cxðf0 Þ

log a þ C log a0 þ vðfÞ a a
dr
fvð0Þ fb fb 0 vðrÞðr
fÞ
Z f0b Z f0b 
xðfb Þ þ Cxðf0b Þ xðrÞ

dr þ ð1 þ CÞ dr
0 vðrÞðr
fÞ f0a vðrÞðr
fÞ
n Z
1 X Cj
þ dr ð103Þ
2pi j¼1 Lj vðrÞðr
fÞ
 620 N. HASEBE AND X. WANG

The derivative of the 1st, 2nd, and 4th integral terms in the above equation can be
obtained up to n ¼ 2 without need of integration (see Appendix 3). The numerical
quadrature is simpler in evaluating the integrals for more than n ¼ 3. Evaluation
of the 3rd integral term needs numerical procedure [27]. This term will be canceled
under the isothermal condition due to C ¼
1. The undetermined constants Cj are
determined from the condition that the stress function /2 ðfÞ should be regular at
infinity. Cj , physically stands for the resultant force on the jth displacement bound-
ary segment. One of them can be treated as zero due to the equilibrium condition
that the entire resultant force on the inclusion is zero. We take

Cn ¼ 0 ð104Þ
Since the stress function /2 ðfÞ is regular at infinity, the coefficients of the power
Downloaded by [University of California Davis] at 10:13 11 April 2014

terms of f in the expanded series must be zero. Consequently, (n
1) equations

can be obtained as (see Eqs. (136)–(138)).
Z f0a
 Z f0b 
xðfa Þ þ Cxðf0a Þ rk
1 xðfb Þ þ Cxðf0b Þ rk
1
dr
dr
0 vðrÞ 0 vðrÞ
Z f0b     0 
xðrÞrk
1 E1 fa f
þ ð1 þ CÞ dr þ dðkÞ log þ C log a0
0
fa vðrÞ vð0Þ fb fb
Xn Z k
1
2ki Cj r

dr ¼ 0 ðk ¼ 1; 2; 3; . . . ; n
1Þ ð105Þ
a0 MGR j¼1 Lj vðrÞ

where

1 k¼1
dðkÞ ¼ ð106Þ
0 k¼
6 0

Heat source at arbitrary position and heat sink at infinity [27] The stress
functions in this case are also expressed by Eq. (99), in which the Green’s function of
the external force boundary value problem expressed by Eq. (76) is adopted as the
function /1 ðfÞ.
The stress function /2 ðfÞ is expressed as

 
a0 MGR  f
/2 ðfÞ¼ xðfa ÞþCxðf0a Þ
ð1þCÞxðfÞ log
4pk f
f0a
  0  Z f0a
E1 fa xðfa ÞþCxðf0a Þ

Cþlog þvðfÞ dr
f fa 0 vðrÞðr
fÞ
Z f0a   0 
r E1 vðfÞ f

ð1þCÞE0 vðfÞ drþ Cþlog a
0 vðrÞðr
fÞ vð0Þf fa
I 0  Xn Z
ð1þCÞE1 logðrÞ
logðr
fa Þ vðfÞ Cj

vðfÞ dr þ dr
2pi R rvðrÞðr
fÞ 2pi j¼1 Lj
vðrÞðr
fÞ
ð107Þ
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 621

The integral of the 3rd term in the above equation can be replaced by an integral of
arbitrary path that includes the origin of the coordinates and f0a [27]. The stress func-
tion wðfÞ is given by Eq. (66).

MIXED BOUNDARY VALUE PROBLEM FOR AN INFINITE PLANE WITH

AN ARBITRARY HOLE SUBJECTED TO A POINT HEAT SOURCE:
ONE DISPLACEMENT RESTRAINT
The problem with respective one segment of displacement boundary and exter-
nal force boundary is discussed as special case of Figure 7. The mapping function is
given by Eq. (30). The temperature function for a heat source at an arbitrary position
Downloaded by [University of California Davis] at 10:13 11 April 2014

and a heat sink at infinity is given by Eq. (52).

A Heat Source at an Arbitrary Position and a Heat Sink

at Infinity [28]
The stress functions can be derived based on a different form of the boundary
condition from Eq. (102). The boundary condition Eq. (26) can be expressed in the
following form as

xðrÞ
/ðrÞf1
dðrÞð1 þ jÞg þ /0 ðrÞ þ wðrÞ ¼ f ðrÞ ð108Þ
x0 ðrÞ

where
(
0 on L
dðrÞ ¼ ð109Þ
1 on M

Figure 7 Rigid inclusion with debondings and a unit circle.

 622 N. HASEBE AND X. WANG

and
8
<0 on L
Z
f ðrÞ ¼ ð110Þ
: 2Ga0 XðrÞx0 ðrÞdr on M

The stress functions to be determined can be expressed as

/ðfÞ ¼ /1 ðfÞ þ /2 ðfÞ

ð111Þ
wðfÞ ¼ w1 ðfÞ þ w2 ðfÞ
Downloaded by [University of California Davis] at 10:13 11 April 2014

where

a0 MGR
/1 ðfÞ ¼ fxðfÞ
xðfa Þgflogðf
fa Þ
1g þ A logðfÞ
4pk
a0 MGR
w1 ðfÞ ¼
xðfa Þ logðf
fa Þ þ B logðfÞ ð112Þ
4pk !
a 0 MGR XN
E k
A¼B
¼
E0 Cf0a
4pk f
fa
k¼1 k

Here /ðfÞ and w1 ðfÞ are the functions for an infinite plane with a heat source at an
arbitrary position and a heat sink at infinity. The terms A log f, B log f are intro-
duced to cancel the dislocation generated in the displacement and the stresses
(see Eq. (61)). Substituting Eq. (111) into Eq. (108) and making some arrangement,
we have

xðrÞ
/2 ðrÞf1
dðrÞð1 þ jÞg þ /02 ðrÞ þ w2 ðrÞ ¼ f ðrÞ þ /1 ðrÞdðrÞð1 þ jÞ
x0 ðrÞ
xðrÞ

/1 ðrÞ
/01 ðrÞ
w1 ðrÞ ð113Þ
x0 ðrÞ

Here consider the Plemelj function

vðfÞ ¼ ðf
aÞ1
m ðf
bÞm

ð114Þ
m ¼ 0:5 þ ic; c  lnðjÞ=2p

that has the following relation on the boundary

v
ðrÞ ¼ vþ ðrÞf1
dðrÞð1 þ jÞg ð115Þ

where þ and
indicate the limit values of the function as z approaches the bound-
ary from the physical region (outside the unit circle in this case) and the nonphysical
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 623

region (inside the unit circle), respectively. Multiplying both sides of Eq. (113) by
dr=½2piðr
fÞv
ðrÞ and carrying out the Cauchy integration, we obtain

"   X  
 N
a0 MGR 0 f f
/2 ðfÞ ¼ C xðfa Þ
xðfÞ log þ D1k log
4pk f
f0a k¼1
f
fk
X Ek
N Z f0a
 1
þ þ xðfa Þ þ Cxðf0a Þ
ð1 þ CÞxðfÞ vðfÞ dr
k¼1
f k
f 0 vðrÞðr
fÞ
Z fk #
XN
1 XN
D2k

D0 vðfÞ þ D1k vðfÞ dr þ vðfÞ
k¼1 0 vðrÞðr
fÞ f
f
k¼1 k
X
Downloaded by [University of California Davis] at 10:13 11 April 2014

N
f k þ gk Þ
Ek ðA

vðfÞ ð116Þ
0 ðf0 Þðf
fÞ
k¼1 vðfk Þx k k

where

gk  /02 ðf0k Þ; f0k  1=fk ð117Þ

Z f0a
1 ie
cp
D0 ¼ ð1þCÞE0 dr ¼ ð1þCÞE0
0 vðrÞ m

 0 m      
a
fa a
f0a a m a
 2 F1 m; m; 1þm;
2 F1 m; m; 1þm; ð118Þ
a
b a
b a
b a
b

 
C 1 C
D1k ¼ Ek

ð119Þ
fk fk
fa fk
f0a

Z f0a
1
D2k ¼
ð1þCÞEk dr
" 0 vðrÞðr
fk Þ
Ek f f
f
þ ð1þCÞlog k 0
log a 0 k
vðfk Þ fk
fa fa
( )#
1 XN
Ek
þ1
E0 þ
x0 ðf0k Þ 0
j¼1 ðfj
fk Þðfj
fa Þ
"
Ek f f
f
¼
ð1þCÞEk hð0;f0a ;fk Þþ ð1þCÞlog k 0
log a 0 k
vðfk Þ fk
fa fa
( )#
1 XN
Ek
þ1
E0 þ ð120Þ
x0 ðf0k Þ 0
j¼1 ðfj
fk Þðfj
fa Þ
 624 N. HASEBE AND X. WANG

The integral terms in D0 ; D2k can be expressed in terms of the hypergeometric func-
tions as
"   X  
N
a0 MGR
0
 f f
/2 ðfÞ ¼ C xðfa Þ
xðfÞ log þ D1k log
4pk f
f0a k¼1
f
fk
XN
Ek

þ þ xðfa Þ þ Cxðf0a Þ
ð1 þ CÞxðfÞ vðfÞhð0;f0a ;fÞ
D0 vðfÞ
f
f
k¼1 k
#
XN XN
D2k XN
f k þ gk Þ
Ek ðA
þvðfÞ D1k hð0;fk ;fÞ þ vðfÞ
vðfÞ
k¼1
f
f
k¼1 k
0 0
k¼1 vðfk Þx ðf Þðfk
fÞ
k
ð121Þ
Downloaded by [University of California Davis] at 10:13 11 April 2014

The constants in Eq. (117) can be determined using the derivatives of Eqs. (117) and
(116) (or Eq. (121)) with respect to f. Thus the real and the imaginary parts of gk are
obtained from a system of algebraic equations. Finally, /ðfÞ and wðfÞ are given by
the 1st equation in Eq. (111) and Eq. (66), respectively. The derivatives of the inte-
gral terms with respect to f in Eq. (116) can be referred to Appendix 3. It also can be
expressed by the hypergeometric function as follows:
Z d  
1 ie
cp a
df
a m
dr ¼ hðc; d; fÞ ¼
c vðrÞðr
fÞ mða
fÞ a
b f
d
   
a
df
b a
cf
a m
 2 F1 m; m; 1 þ m;

a
bf
d a
bf
c
 
a
cf
b
2 F1 m; m; 1 þ m; ð122Þ
a
bf
c

Z    
d
1 ie
cp a
d m a
d
dr ¼ 2 F1 m; m; m þ 1;
c vðrÞ m a
b a
b
 m  
a
c a
c

2 F1 m; m; m þ 1; ð123Þ
a
b a
b

The first- and the second-order derivatives of the hypergeometric function are
expressed as
d C1 C2
f2 F1 ðC1 ; C2 ; C3 ; fÞg ¼ 2 F1 ðC1 þ 1; C2 þ 1; C3 þ 1; fÞ ð124Þ
df C3

d C1 C2 ðC1 þ 1ÞðC2 þ 1Þ
f F ðC1 ; C2 ; C3 ; fÞg ¼
2 2 1 2 F1 ðC1 þ 2; C2 þ 2; C3 þ 2; fÞ ð125Þ
df C3 ðC3 þ 1Þ

Heat Source and Heat Sink at Respective Arbitrary Positions [28]

The solution for the problem of a heat source þM at fa and a heat sink
M at
fb can be obtained by the superposition of the previous solution. The temperature
function is given by Eq. (54). /ðfÞ can be expressed by Eq. (111), in which /1 ðfÞ
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 625

is given by Eqs. (68) and (70), /2 ðfÞ is

"      
a0 MGR 0 f 0 f f
f0a
/2 ðfÞ ¼ Cxðfa Þ log
Cxðfb Þ log þ CxðfÞ log
4pk f
f0a f
f0b f
f0b
XN  
f 
þ D1k log þ xðfa Þ þ Cxðf0a Þ vðfÞhð0; f0a ; fÞ
k¼1
f
fk


xðfb Þ þ Cxðf0b Þ vðfÞhð0; f0b ; fÞ
D0 vðfÞ
ð1 þ CÞxðfÞvðfÞ
#
0 0
 X N XN
D2k
 hð0; fa ; fÞ
hð0; fb ; fÞ þ D1k hð0; fk ; fÞ þ
k¼1
f
f
k¼1 k
X
N
Ek ðAfk þ gk Þ
Downloaded by [University of California Davis] at 10:13 11 April 2014

vðfÞ ð126Þ
k¼1 vðfk Þx0 ðf0k Þðfk
fÞ

Z f0a
1 ie
cp
D0 ¼ ð1 þ CÞE0 dr ¼ ð1 þ CÞE0
f0b vðrÞ m

 0 m  
a
fa a
f0a
 2 F1 m; m; 1 þ m;
a
b a
b
 m  
a
f0b a
f0b

2 F1 m; m; 1 þ m; ð127Þ
a
b a
b

 
1 C 1 C
D1k ¼ Ek

0 þ þ ð128Þ
fk
fa fk
fa fk
fb fk
f0b

D2k ¼
ð1 þ CÞEk hðf0b ; f0a ; fÞ

(
Ek f
f0b f
fa
þ ð1 þ CÞ log k
log k
vðfk Þ fk
f0a fk
fb
!)
f0a 1 X N
Ek 1 1
þ log 0

ð129Þ
fb x0 ðf0k Þ j¼1 ðfj
f0k Þ ðfj
fa Þ ðfj
fb Þ

wðfÞ is given by Eq. (66).

MIXED BOUNDARY VALUE PROBLEM FOR AN INFINITE PLANE WITH

AN ARBITRARY HOLE SUBJECTED TO A POINT HEAT SOURCE:
n DISPLACEMENT RESTRAINTS [29]
This chapter, as a generalization of the previous chapter, considers n boundary
segments of the displacement and the external force, as shown in Figure 7. The map-
ping function is given by Eq. (30). The temperature function for the problem of a
heat source at an arbitrary position and a heat sink at infinity is given by Eq. (52).
 626 N. HASEBE AND X. WANG

A Heat Source at an Arbitrary Position and a Heat Sink at Infinity

Method 1 The boundary conditions are expressed by Eqs. (108) and (109).
Eq. (110) is given by


Cj R on Lj
f ðrÞ ¼ ð130Þ
2Ga0 XðrÞx0 ðrÞdr on Mj

The stress functions to be determined can be expressed by Eqs. (111) and (112). The
boundary condition is expressed by Eq. (113). Here, the Plemelj function is

Y
n
Downloaded by [University of California Davis] at 10:13 11 April 2014

vðfÞ ¼ ðf
aj Þm ðf
bj Þ1
m
j¼1 ð131Þ
m ¼ 1=2
i lnðjÞ=2p  1=2
ic

Multiplying both sides of Eq. (113) by dr=½2piðr
fÞv
ðrÞ with regard to Eq. (115)
and carrying out the Cauchy integration, we obtain

"   X  
 N
a0 MGR f f
/2 ðfÞ¼ C xðf0a Þ
xðfÞ log þ D 1k log
4pk f
f0a k¼1
f
fk

X N Z f0a XN Z fk
Ek xðfa ÞþCxðf0a Þ D1k
þ þvðfÞ drþvðfÞ dr
f
f
k¼1 k 0 vðrÞðr
fÞ k¼1 0
vðrÞðr
fÞ
Z f0a #
ð1þCÞxðrÞ XN
D2k XN
fk þgk Þ
Ek ðA

vðfÞ drþvðfÞ
vðfÞ
0 vðrÞðr
fÞ f
f
k¼1 k
0 0
k¼1 vðfk Þx ðf Þðfk
fÞ
k
n Z
X
vðfÞ Cj
þ dr ð132Þ
2pi j¼1 Lj vðrÞðr
fÞ

where

gk /02 ðf0k Þ; f0k 1=fk ð133Þ

 
C 1 C
D1k ¼Ek

ð134Þ
fk fk
fa fk
f0a


Ek f f
f
D2k ¼ ð1þCÞlog k 0
log a 0 k þ1
vðfk Þ fk
fa fa
( )#
1 XN
Ek

E0 þ ð135Þ
x0 ðf0k Þ 0
j¼1 ðfj
fk Þðfj
fa Þ
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 627

The constants Cj are determined from the condition that the stress function /2 ðfÞ is
regular at infinity and the equilibrium condition that the entire resultant force on the
inclusion is zero. Consider the terms which are not holomorphic at infinity (positive
power in the series) in /2 ðfÞ
Z b X 1 Z b j
1
1 r
dr¼
f
j dr ð136Þ
a vðrÞðr
fÞ j¼1 a vðrÞ

Z b X1 Z b
xðrÞ xðrÞrj
1
dr¼
f
j dr ð137Þ
a vðrÞðr
fÞ j¼1 a vðrÞ
Downloaded by [University of California Davis] at 10:13 11 April 2014

1 X1
¼
ðfk Þj
1 f
j ð138Þ
fk
f j¼1

Since the preceding equation and the Plemelj function have n-th power at infinity,
the coefficients of the terms with n
jðj ¼1;2;3;...;n
1Þ power in /2 ðfÞ at infinity
must be zero,
" Z
 f0a
a0 MGR rj
1
xðfa ÞþCxðf0a Þ
ð1þCÞxðrÞ dr
4pk 0 vðrÞ
#
XN Z fk X n Z
D1k rj
1 N
j
1 1 X Ck rj
1 dr
þ drþ D2k ðfk Þ þ
k¼1 0
vðrÞ k¼1
2pi k¼1 Lk vðrÞ
X
N
fk þgk Þ
Ek ðA

ðfk Þj
1 ¼0 ðj ¼1;2;...;n
1Þ ð139Þ
0 0
k¼1 vðfk Þx ðfk Þ

Cn ¼0 ð140Þ

where the constants gk satisfy Eq. (133). Accordingly, gk and Cj can be determined
from a system of equations.
The derivative of the 1st, 2nd, and 4th integral terms in Eq. (132) can
be obtained up to n ¼ 2 without need of integration (see Appendix 3). However,
the numerical quadrature is simpler in evaluating the integrals for more than
n ¼ 3. Evaluation of the 3rd integral term needs numerical procedure [27].

Method 2 The Riemann–Hilbert problem that is obtained from Eq. (66) of

analytical continuation that has been stated in the mixed boundary value problem
of an infinite plane with an elliptical hole under a heat source [27] is considered.
The stress functions to be determined are

/ðfÞ ¼ /1 ðfÞ þ /2 ðfÞ

ð141Þ
wðfÞ ¼ w1 ðfÞ þ w2 ðfÞ
 628 N. HASEBE AND X. WANG

where /1 ðfÞ is the stress function of the external force boundary value problem for a
point heat source [17]

/1 ðfÞ ¼ /11 ðfÞ þ /12 ðfÞ ð142Þ

xð1=f
Þ 0
w1 ðfÞ ¼
/1 ð1=f
Þ
0 / ðfÞ ð143Þ
x ðfÞ 1

a0 MGR
/11 ðfÞ ¼ ½fxðfÞ
xðfa Þgflogðf
fa Þ
1g þ A logðfÞ ð144Þ
4pk
Downloaded by [University of California Davis] at 10:13 11 April 2014

"   X #
N
a0 MGR f Ek Bk
/12 ðfÞ ¼ fxðfÞ
xðfa Þg log

4pk f
f0a f
f
k¼1 k
X
N
Ek g1k

ð145Þ
0 0 f
k¼1 x ðfk Þ k

f

X
N
Ek
A ¼
E0 Cf0a þ ; f0k  1=fk ð146Þ
f
fa
k¼1 k

( )
fa
fk 1 XN
E k
Bk ¼ log
2þ
fk þ E 0 þ
A ð147Þ
f0a x0 ðf0k Þ 0
j¼1 ðfj
fk Þðfj
fa Þ

g1k  /012 ðf0k Þ ð148Þ

/1 ðfÞ is also the Green’s function for the external force boundary value problem. The
constants g1k are determined from a system of linear equations using Eq. (148). Sub-
stituting the function wðfÞ which is given by Eq. (66) into the stress and the displace-
ment boundary condition (26) and using Eq. (141), the Riemann-Hilbert problem for
determination of /2 ðfÞ is obtained,

/þ

2 ðrÞ
/2 ðrÞ ¼ Cj on Lj
Z
ð149Þ
j/þ
2 ðrÞ þ /

2 ðrÞ ¼
2a 0
XðrÞx0 ðrÞdr
ð1 þ jÞ/1 ðrÞ  HðrÞ on Mj

Its solution is

Z Z
vðfÞ X
n
HðrÞ vðfÞ X
n
Cj
/2 ðfÞ ¼ dr þ dr þ QðfÞvðfÞ ð150Þ
2pi j¼1 Mj vðrÞðr
fÞ 2pi j¼1 Lj vðrÞðr
fÞ
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 629

The Plemelj function is given by Eq. (131). The unknown function QðfÞ is determ-
ined from the regularity of w2 ðfÞ at the exterior of the unit circle, i.e., there is no
singular term of heat source in w2 ðfÞ.
Finally, we have
"  
a0 MGR f
/2 ðfÞ ¼ xðfa Þ þ Cxðf0a Þ

ð1 þ CÞxðfÞ log
4pk f
f0a
XN   X N
f Ek ðBk þ 1Þ
þ C1k log þ
k¼1
f
f k k¼1
fk
f
Z f0a X N Z fk
xðfa Þ þ Cxðf0a Þ C1k
þ vðfÞ dr þ vðfÞ dr
Downloaded by [University of California Davis] at 10:13 11 April 2014

0 vðrÞðr
fÞ k¼1 0
vðrÞðr
fÞ
Z f0a #
ð1 þ CÞxðrÞ XN
C2k

vðfÞ dr þ vðfÞ
0 vðrÞðr
fÞ f
f
k¼1 k
X
N
Ek g1k X
N
Ek ðg1k þ g2k Þ
þ
vðfÞ
0 0 vðfk Þx0 ðf0k Þðfk
fÞ
k¼1 x ðfk Þðfk
fÞ k¼1
X n Z
vðfÞ Cj dr
þ ð151Þ
2pi j¼1 Lj vðrÞðr
fÞ

where

g2k  /02 ðf0k Þ ð152Þ

 
C 1 C
C1k ¼ Ek

ð153Þ
fk fk
fa fk
f0a

 
Ek fk
C2k ¼ ð1 þ CÞ log
Bk
1 ð154Þ
vðfk Þ fk
f0a

The advantage of this method is in determination of QðfÞ that the Cauchy integral is
not needed in the derivation. However, it is necessary to obtain the Green’s function
of the external force boundary value problem in advance. The constants
g2k ðk ¼ 1; 2; 3; . . . ; NÞ are determined from Eq. (152) and the regularity of function
/2 ðfÞ at infinity (see Eqs. (136)–(138)). Constants Ck ðk ¼ 1; 2; 3; . . . ; n
1Þ are de-
termined from the following equations:
"Z 
f0a
a0 MGR xðfa Þ þ Cxðf0a Þ
ð1 þ CÞxðrÞ rj
1
dr
4pk 0 vðrÞ
#
XN Z fk XN
C1k rj
1 j
1
þ dr þ C2k ðfk Þ
k¼1 0
vðrÞ k¼1
 630 N. HASEBE AND X. WANG

Z
1 X n
Ck rj
1 dr X N
Ek ðg1k þ g2k Þ
þ

2pi k¼1 vðrÞ 0 0
Lk k¼1 vðfk Þx ðf Þ k ð155Þ
j
1
 ðfk Þ ¼0 ðj ¼ 1; 2; . . . ; n
1Þ

Cn ¼ 0 ð156Þ

THE TEMPERATURE AND THE THERMOELASTIC MIXED BOUNDARY

VALUE PROBLEM FOR AN INFINITE PLANE WITH AN ELLIPTICAL
HOLE SUBJECTED TO A HEAT SOURCE [19]
Downloaded by [University of California Davis] at 10:13 11 April 2014

The temperature and the thermo elastic mixed boundary value problems for
a point heat source are considered here, in which the temperature field has been
analyzed previously in the Analysis of Temperature Field. The stress functions to
be determined are expressed by Eq. (99), in which /1 ðfÞ is given by Eq. (100).
The temperature function is expressed by Eq. (60). The constant A is


A ¼ vðfa Þ
vðfb Þ
E0 vðfa Þ þ vðf0a Þ
vðfb Þ
vðf0b Þ ð157Þ

where the Plemelj function vðfÞ is given by Eq. (57). Here, instead of the Riemann–
Hilbert boundary condition expressed in Eq. (102), we consider an alternative form
of the boundary condition (26) as follows:

xðrÞ
/ðrÞf1
dðrÞð1 þ jÞg þ /0 ðrÞ þ wðrÞ ¼ f ðrÞ ð158Þ
x0 ðrÞ

where

0 R on L
f ðrÞ ¼ ð159Þ
2Ga0 XðrÞx0 ðrÞdr on M

Substituting Eq. (99) into Eq. (109),

xðrÞ
/2 ðrÞf1
dðrÞð1 þ jÞg þ /02 ðrÞ þ w2 ðrÞ ¼ f ðrÞ þ /1 ðrÞdðrÞð1 þ jÞ ð160Þ
x0 ðrÞ

Here, introduce the Plemelj function

vs ðfÞ ¼ ðf
aÞm ðf
bÞ1
m ð161Þ

m ¼ 0:5
iðlog jÞ=ð2pÞ ð162Þ

 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 631

which satisfies the following relation on the boundary

v
þ
s ðrÞ ¼ vs ðrÞf1
dðrÞð1 þ jÞg ð163Þ

 
Multiplying both sides of Eq. (160) with dr= 2piðr
fÞv

s ðrÞ and carrying out the
Cauchy integration using the above relation, function /2 ðfÞ is obtained,

Z
1þj b hðrÞ
/2 ðfÞ ¼ vs ðfÞ dr
2pij a vþs ðrÞðr
fÞ

 
a0 MGR  r
Downloaded by [University of California Davis] at 10:13 11 April 2014

hðrÞ ¼ xðfa Þ þ xðf0a Þ log

4pk r
f0a
 
 r

xðfb Þ þ xðf0b Þ log
r
f0b
   0 
r
f0a E1 fa fb þ
þ2xðrÞ log þ log 0 þ J ðrÞ
r
f0b r fa fb
8 0qffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffi1 0qffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffi19
>
< fa
a
þ fa
b f0a
a f0a
b > =
B r
a r
b C B r
a þ r
b C
JðrÞ ¼
2xðrÞ log@qffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiA þ log@qffiffiffiffiffiffiffi ffi q ffiffiffiffiffiffiffi
ffi A
>
: fb
a fb
b f0b
a f0b
b > ;
r
a þ r
b r
a þ r
b
8 0qffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffi19
>
< pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffi 0 0 >
=
fa
a þ fa
b B f a
a þ f a
bC
þ 2xðrÞ log pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffi þ log@qffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiA
>
: fb
a þ fb
b >
f0b
a þ f0b
b ;
ð164Þ
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
r
a r
b
þ 2xðfa Þ log þ
rðfa
aÞ rðfa
bÞ
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
r
a r
b

2xðfb Þ log þ
rðfb
aÞ rðfb
bÞ
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi s ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi!
r
a r
b
þ 2xðf0a Þ log þ
rðf0a
aÞ rðf0a
bÞ
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!
r
a r
b

2xðf0b Þ log þ
rðf0b
aÞ rðf0b
bÞ
rffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffi!
r
a  r
b

2E0 vðfa Þ þ vðf0a Þ
vðfb Þ
vðf0b Þ
log þ
r r
  rffiffiffiffiffiffiffiffiffiffiffi s ffiffiffiffiffiffiffiffiffiffiffi!
2E1 vðfa Þ vðf0a Þ vðfb Þ vðf0b Þ r
a r
b

þ 0

0 log þ
xð0Þ fa fa fb fb
ra
rb
 632 N. HASEBE AND X. WANG

Evaluation of the integral in the above equation yields

 
a0 MGR 0
 f
/2 ðfÞ ¼ xðfa Þ þ xðfa Þ log
4pk f
f0a
 
0
 f

xðfb Þ þ xðfb Þ log
f
f0b
 0  0
f
fa E1 fa fb
þ 2xðfÞ log þ log
f
f0b f f0a fb
Z f0a
xðfa Þ þ xðf0a Þ
þ vs ðfÞ dr
0 vþ s ðrÞðr
fÞ
Z f0b
xðfb Þ þ xðf0b Þ E1 vs ðfÞ
Downloaded by [University of California Davis] at 10:13 11 April 2014

vs ðfÞ þ ðrÞðr
fÞ
dr

0 v s fvs ð0Þ
 0 Z f0b
f f xðrÞ
 log a0 b þ 2vs ðfÞ þ
dr
fa fb fa vs ðrÞðr
fÞ
0

Z b 
1þj J þ ðrÞ
þ v ðfÞ dr ð165Þ
2pij s þ
a vs ðrÞðr
fÞ

It is necessary to compute the derivative of /2 ðfÞ in evaluating the stresses. The deri-
vatives of the first two integral terms in the above equation can be evaluated without
involving into any integration. Whereas the last two integral terms are expressed in
the form of series
"  
0 a0 MGR xðfb Þ þ xðf0b Þ f0b xðfa Þ þ xðf0a Þ f0a
/2 ðfÞ ¼

4pk fðf
f0b Þ fðf
f0a Þ
   0
f
f0a 2xðfÞðf0a
f0b Þ E1 ff
0
þ 2x ðfÞ log þ
log a0 b
f
f0b ðf
f0a Þðf
f0b Þ f2 f fb
 a
 1 1
þ xðfa Þ þ xðf0a Þ vt ðfÞ þ 0

vt ðfa Þðf
f0a Þ fvt ð0Þ
 
0
 1 1

xðfb Þ þ xðfb Þ vt ðfÞ þ 0

vt ðfb Þðf
f0b Þ fvt ð0Þ
 0
 E1 ff
þ vs ðfÞ
fv0s ðfÞ 2 log a0 b
f vs ð0Þ fa fb
#
X1 

v0s ðfÞ
vs ðfÞj=f Ij f
j ð166Þ
j¼1

where

vt ðfÞ ¼ ðf
aÞm
1 ðf
bÞ
m ð167Þ

Z f0b Z b
xðrÞrj
1 1þj J þ ðrÞrj
1
Ij ¼ 2 dr þ dr ð168Þ
f0a vs ðrÞ 2pij a vþs ðrÞ
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 633

The above integral is evaluated by the numerical computation [27]. Function wðfÞ can
be obtained from analytical continuation on the traction-free boundary condition by
Eq. (66).

INTERACTION BETWEEN A HOLE AND A CRACK IN AN INFINITE PLANE

UNDER UNIFORM HEAT FLUX [30]
The problems for an infinite plane with a hole under uniform heat flux or a heat
source have been discussed so far. Using these solutions, the interaction between a
hole and a crack in an infinite plane under uniform heat flux is analyzed as shown
in Figure 8. It is not easy to obtain the analytical solution for the problem involved
in multiply connected domain. To overcome this difficulty, the problem is analyzed
Downloaded by [University of California Davis] at 10:13 11 April 2014

using the superposition of the solutions for a heat source and a dislocation in such a
way that the boundary condition on the crack surface must be satisfied. The original
problem is denoted by A, the three subsidiary problems B, C, and D are superposed
as shown in Figure 8. The mapping function is expressed by Eq. (31) with N ¼ 24.
An interaction problem between a rigid inclusion and a crack can be also solved by a
similar way, but it is no stated here.

Figure 8 Interaction between a hole and a crack under uniform heat flux.
 634 N. HASEBE AND X. WANG

Problem B
Problem B is specified in Figure 8, where the temperature function XB ðfÞ under
uniform heat flux is given by Eq. (48). The stress function /B ðfÞ is expressed by Eqs.
(61), (62), and (63). The constant A in Eq. (62) is given by
!
a0 qGR X
N¼24

id id
id
A¼ E0 Ek e þ E0 e
E25 e ð169Þ
2k k¼1

The stress function wB ðfÞ is expressed by Eq. (66).

Downloaded by [University of California Davis] at 10:13 11 April 2014

Problem C
In problem C, the temperature and the stress functions for a heat source couple
are needed. Making use of the formulation stated in Appendix 2 can derive the tempera-
ture and the stress functions. They, Tðf; fa Þ and /C ðfÞ, are for a heat source couple are
" 2
#
P eik e
ik f0a
T ðf; fa Þ ¼


2pk x0 ðfa Þðf
fa Þ x0 ðfa Þ f
f0a

 2
C 1
ffa  x
C  ðfa Þ
x f0a f0a
/C ðfÞ ¼ log þ
2p fðf
fa Þ 2p x 0 ð fa Þ f
f0a
!
X24 
þ k þ A
A  fk
C 1 Bk
þ A log f ð170Þ
2p 0
f0 f
fk
f
k¼1 k
( 2
)
1 X 24
Ek C E0 f0a C  E25 C
A¼ þ

2p k¼1 x0 ðfa Þðfk
fa Þ2 x0 ðfa Þ x0 ðfa Þf2a
a0 GRP ik
C¼ e
2k

where Bk  Ek =x0 ðf0k Þ; k denotes the direction angle of the heat source couple from the
positive x-axis (see Figure 9). P and fa represent its intensity and the position, respect-
ively. The constant A is determined from the single-valuedness condition of the displace-
ment when moving round the hole. The stress function wC ðfÞ is expressed by Eq. (66).

Problem D [13,34,35]
In problem D, a point dislocation exists in an infinite plane with a hole. The
stress function is expressed as follows:

 2
D 1
ff0  x
D  ðf0 Þ
x f00 f00
/D ðfÞ ¼ log þ
2p fðf
f0 Þ 2p x0 ðf0 Þ f
f00
!
X24 
þ k
D
A
1 Bk
ð171Þ
0
k¼1
2p fk
f0 f
fk
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 635
Downloaded by [University of California Davis] at 10:13 11 April 2014

Figure 9 Heat source couple.

where D ¼ D1 þ iD2 ; subscripts 1 and 2 represent the x and y directions, respectively,

of the dislocation. f0 signifies the dislocation point and f00  1=f
0 . The stress function
wD ðfÞ is expressed by Eq. (66).

Problem A
The solution of this problem is obtained by the superposition of that for pro-
blems B, C and D, in which the thermal adiabatic and the traction free conditions are
satisfied on the crack surface. Supposing the heat source couple located at t of which
coordinate is taken on the crack (see Figure 8), the distribution density cðtÞ can be
expressed by PðtÞ ¼ cðtÞdt in Eq. (170). The temperature function for problem C
can be obtained by the integration along the crack line AB using Tðf; fa Þ of
P ¼ 1 in Eq. (170) as
Z f
XC ðfÞ ¼ cðtÞTðf; tÞdt ð172Þ

f

where f denotes the crack length. Consequently, the temperature function for the
problem A is expressed
Z f
XA ðfÞ ¼ XB ðfÞ þ XC ðfÞ ¼ XB ðfÞ þ cðtÞTðf; tÞdt ð173Þ

f

where XB ðfÞ is expressed by Eq. (48). From the adiabatic condition (21), we have

Z f Z f 
Im XB ðsÞ þ cðtÞTðs; tÞdt
XB ðs0 Þ
cðtÞTðs0 ; tÞdt ¼ 0 ð174Þ

f
f

where s0 is an arbitrary point (standard point) on the crack face AB and s is the co-
ordinate on the face AB. Usually, the analytical solution of the unknown function
 636 N. HASEBE AND X. WANG

cðtÞ in the above equation cannot be obtained, and thus we treat it by the numerical
approach. Since this is a crack problem, the heat flux and the temperature should
have the singularity
0:5 and 0.5 power, respectively, at the crack tips. Function
cðtÞ can be rewritten in the following form for the convenience of the numerical
integration
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cðtÞ ¼ GðtÞ f 2
t2 ð175Þ

Taking the replacement t ¼
f cos h in Eq. (174), the equation becomes easier to
deal with
Z p
Downloaded by [University of California Davis] at 10:13 11 April 2014

GðtÞ½Tðs; tÞ
Tðs0 ; tÞ f 2
t2 dh ¼
½XB ðsÞ
XB ðs0 Þ ð176Þ

0

The distribution density GðtÞ of the heat source couple is obtained by the Gauss–
Chebyshev method numerically (see Appendix 4).
Next, consider the stress boundary condition on the crack surface. The edge
dislocation density is defined as

2G d n þ o
hj ðtÞ ¼ uj ðtÞ
u

j ðtÞ j ¼ n; s ð177Þ
1 þ j dt

where uþ

j ðtÞ
uj ðtÞ is the relative displacement at point t between the upper and the
lower surfaces of the crack in the j direction.
From the condition of the single-valuedness of the displacement around the
crack, it has

Z f Z f
hðtÞdt  fhn ðtÞ þ ihs ðtÞgdt ¼ 0 ð178Þ

f
f

From the traction-free condition on the line crack surfaces, the singular integral
equation for the determination of hj ðtÞ is obtained:
Z f
fhn ðtÞ½Nn ðs; tÞ þ iTn ðs; tÞ þ hs ðtÞ½Ns ðs; tÞ þ iTs ðs; tÞgdt

f ð179Þ
¼
fNB ðsÞ þ NC ðsÞ þ i½TB ðsÞ þ TC ðsÞg

where Nj ðs; tÞ and Tj ðs; tÞ ð j ¼ n; sÞ stand for the stress components ND ðsÞ and
TD ðsÞ, which are, respectively, normal and tangential traction components to the
crack line at point s caused by the unit point dislocation in the j-direction located
at point t. The tractions NB ðsÞ þ iTB ðsÞ and NC ðsÞ þ iTC ðsÞ represent the normal
and tangential traction components at point s on the crack obtained from problems
B and C.
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 637

Further, for a crack problem, the functions hn ðtÞ and hs ðtÞ are expressed as:

Hj ðtÞ
hj ðtÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi j ¼ n; s ð180Þ
f 2
t2

The integral equations (178) and (179) can be expressed as follows by the variable
replacement t ¼
f cos h
Z p
½Hn ðtÞ þ iHs ðtÞdh ¼ 0
0
Z p
fHn ðtÞ½Nn ðs; tÞ þ iTn ðs; tÞ þ Hs ðtÞ½Ns ðs; tÞ þ iTs ðs; tÞgdh ð181Þ
0
Downloaded by [University of California Davis] at 10:13 11 April 2014

¼
fNB ðsÞ þ NC ðsÞ þ i½TB ðsÞ þ TC ðsÞg

The density function Hn ðtÞ and Hs ðtÞ on the crack can be obtained by solving the
above equation. The Gauss–Chebyshev method is used in the numerical procedure.
The stress intensity factors at the crack tips are obtained using this density function
(see Appendix 4).

STRESS INTENSITY FACTOR

The initiation and extension of a crack by the thermal stress are of interest in
the engineering practice. To this end, it is necessary to obtain the stress intensity fac-
tor of the crack. The stress intensity factor can be readily obtained from the rational
mapping function and the stress function for a configuration with a crack [6,9,31].
The stress intensity factors of modes I and II, KI, KII, can be expressed by
pffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
KI
iKII ¼ 2 2p lim fxðfÞ
xðr1 Þge
ik /0 ðfÞ x0 ðfÞ
f!r1
ð182Þ
pffiffiffi
ik=2 0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ 2 pe / ðr1 Þ x00 ðr1 Þ ¼ 2 p/0 ðr1 Þ eik x00 ðr1 Þ

where k is the angle between the crack direction and the x-axis; r1 is the point on the
unit circle corresponding to the crack tip. To obtain the limiting value for the above
equation, the expanded series for the mapping function in the vicinity of r1 is used:
1 00
xðfÞ ¼ xðr1 Þ þ x0 ðr1 Þ þ x ðr1 Þ þ    ð183Þ
2
Regarding the fact x0 ðr1 Þ ¼ 0 at the crack tip, we have Eq. (182). However, rigor-
ously speaking, x0 ðr1 Þ 6¼ 0 because the mapping function is rational. This makes
the crack tip round at the end. The stress intensity factor can be obtained in a good
accuracy when the ratio q=a of the curvature radius q at the crack tip by the crack
length a is smaller than 10
6 10
5. Generally, there is no problem for the mapping
functions formed by the fractional expressions unless the crack is extremely short.
For example, the stress intensity factors can be obtained with accuracy for the case
of the cracked circular hole as shown in Figure 1 with the ratio being about
e=a ¼ 0:0001. Stress intensity factors under uniform heat flux and heat sources are
 638 N. HASEBE AND X. WANG

obtained in [15, 16, 18, 24, 30, 36–38] and [14, 19, 27], respectively. Stress intensity
factor at a rigid line inclusion is also shown in [23].

STRESS INTENSITY OF DEBONDING AND ENERGY RELEASE

RATE [26,32]
The model of the zero external force and zero displacement in the mixed
boundary value problem may correspond to a rigid inclusion with a debonding.
The stresses have the singularity of
0:5 þ i logðjÞ=ð2pÞ at the debonding tip be-
tween an elastic body and a rigid one (see Figure 10). Mathematically, it is induced
by the derivative of the Plemelj function. The stress components can be expressed in
Downloaded by [University of California Davis] at 10:13 11 April 2014

terms of the distance r from the debonding tip on the x-axis,

    
1 þ j j~ a0 j h0
ry ¼ 0:5 cos  c log r
r ~ j
jb 0 h00
    
3
j j~ a0 j h0
rx ¼ 0:5 ~ j cos h0  c log r ð184Þ
r jb 0 0
    
1 þ j j~ a0 j h0
sxy ¼
0:5 sin  c log r
r ~
jb0 j h00
where ~a0 ; b~ are the stress intensity factors of the interfacial crack. To distinguish from
0
that for the homogeneous material case, it is called stress intensity of debonding (SID)
here and c ¼ ðlog jÞ=ð2pÞ; h0 ¼ argð~ a0 Þ; h00 ¼ argðb ~ Þ. The oscillatory stresses at the
0
debonding tip come from the term log(r) in the trigonometric function in Eq. (184).
~ ~ can be computed using the stress function as follows:
a0 ; b 0

jx0 ðaÞða
bÞj1
m gðaÞ ið1
mÞðhb
ha Þ

a0 ¼
~ e
x0 ðaÞða
bÞ
ð185Þ
0 1
m
~ ¼ jx ðbÞðb
aÞj gðaÞ eimðhb
ha Þ
b 0
x0 ðbÞðb
aÞ

Figure 10 Debonding tip of a rigid and elastic bodies.

 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 639

where m ¼ 0:5
ic; gðaÞ; gðbÞ are that of gðfÞ ¼ ðf
aÞ1
m ðf
bÞm /0 ðfÞ with the
substitution of n by a; b, respectively. ha ; hb are the respective arguments of the
debonding tips a; b. Using the angle ffaob on the unit circle, we have relation
hb
ha ¼ p
ffaob=2, which is used in Eq. (185).
The energy release rates g at the debonding tip are expressed by

pjðj þ 1Þ
ga ¼ j~a0 j2
2G ð186Þ
pjðj þ 1Þ ~ 2
gb ¼ jb0 j
2G

It is seen that the stress intensity of debonding is proportional to the root of the
~ are the same as the energy
Downloaded by [University of California Davis] at 10:13 11 April 2014

energy release rate. From this, it can be said that ~a0 ; b 0

release rate when they are used to indicate the debonding extension.
Stress intensities of debonding under uniform heat flux are obtained in [25,26].

SUMMARY AND CONCLUDING REMARKS

1. The maximum advantage of the complex variable method is that the map func-
tion can be used. The infinite region with a hole, the semi-infinite and finite
regions with a notch or a mound can be analyzed using the rational mapping
function which is formed by the fractional expressions. Especially, the shapes
of strip or convex, which is usually difficult to deal with by other methods,
can be analyzed.
The polygonal shapes can be formed by Schwarz–Christoffel’s formulation
from which the rational mapping of fractional expressions can be formed sys-
tematically. The mapping function (30) can represent a circular or an elliptical
hole as the special case.
2. The solutions for the external force, displacement, and the mixed boundary
value problem were shown using the mapping function. An arbitrary hole can
be treated by merely changing the coefficients in the mapping function. The
main task is to form the rational function to be analyzed after the stress func-
tions have been derived. The mapping function for an elliptical hole can be
obtained by the operation of limit f1 ! 0 in Eq. (30). However, it seems to be
easier to induce the solution as an elliptical hole. The solution for a crack can
be obtained by treating it as a special case of an elliptical hole.
3. The temperature problems under the isothermal and the adiabatic boundary
condition are mathematically the same. Basically, the external force and the dis-
placement boundary value problem are also mathematically the same. However,
the displacement one should be a little more complicated because the displace-
ment term induced by the heat is contained in the boundary condition. The
stress function wðfÞ can be obtained by analytic continuation for the respective
problems.
4. The displacement and the mixed boundary value problem are also the model of
a rigid inclusion. The rigid body rotation of the inclusion is not considered here.
In case that it is taken into consideration, the solution is obtained in such a way
 640 N. HASEBE AND X. WANG

that the sum of the resultant moment generated on the inclusion boundary must
be zero [22]. Moreover, there is no restraint between the rigid inclusion and the
point heat source in the heat source problems. When taking into account the
relative restraint, the solution is obtained by the superposition of the resultant
force applied on the inclusion [22].
5. The analytic approaches for the various problems include the one depending
on the Cauchy integral, the induction of Riemann-Hilbert problem, and
the one using the boundary condition (108). The external force and the dis-
placement boundary condition are the special cases of (108). The mixed
boundary value problem with two or more displacement constraints is more
complicated because it is necessary to determine the constants in the bound-
ary condition.
Downloaded by [University of California Davis] at 10:13 11 April 2014

6. The solution for a point heat source is the Green’s function in the thermo elas-
ticity. The solution in the special case that the point heat source is on the bound-
ary is also obtainable. In this case, the heat source point is not a singular point
because there is no displacement constraint on the boundary.
7. The stress functions for the problem of a heat and sink sources at arbitrary posi-
tions can be obtained by the superposition of the stress functions of problem for
a heat source at an arbitrary position and a heat sink at infinity. However the
inverse case can not been available.
8. The interaction between the hole and a crack can be analyzed by using these
Green’s functions. In this case, the boundary condition of the hole is completely
satisfied. The analysis of the interaction problem under uniform heat flux needs
the solution for a uniform heat flux, the one for a heat source couple, and for a
point dislocation. A variety of thermo elastic problems can be analyzed by using
these solution. The interaction problem between a rigid inclusion and a crack
can be also solved in a similar way.
9. The stress intensity factor of the crack and the stress intensity of debonding are
obtained with good accuracy by the rational mapping function. When consider-
ing the fracture such as the extension of the crack and debonding, the tempera-
ture distribution of the configuration is needed because the fracture toughness
value depends on the temperature in general.
10. As an application, a problem of an arbitrary heat boundary condition may be
solved, since the stress functions of a point heat source on the boundary have
been obtained.

REFERENCES
1. W. Nowacki, Thermoelasticity, Pergamon Press, Oxford, UK, 1962.
2. N. I. Muskhelishvili, Some Basic Problems of Mathematical Theory of Elasticity, 4th
edition, Noordhoff, the Netherlands, 1963.
3. H. Parkus, Thermoelasticity, Blaisdell Publishing Co., Germany, 1968.
4. S. P. Timoshenko and J. N. Goodier, Theory of Elasticity, 3rd edition, McGraw-Hill Book
Co., New York, 1970.
5. N. Hasebe and Y. Horiuchi, Stress Analysis for a Strip with Semi-Elliptical Notches
of Crack on Both Sides by Means of Rational Mapping Function, Ingenieur Archiv,
vol. 47, pp. 169–179, 1978.
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 641

6. N. Hasebe and M. Miura, Cracks Initiating at the End of an Embedded Strip, Trans.
of the Japan Society of Mech. Engineers, vol. 47A, no. 423, pp.1129–1136, 1981, (in
Japanese).
7. N. Hasebe, S. Matsuura, and N. Kondo, Stress Analysis of a Strip with a Step and a
Crack, Eng. Fracture Mech., vol. 20, pp. 447–462, 1984.
8. N. Hasebe, Stress Analyses of a Semi-infinite Plate with a Triangular Notch or Mound,
Trans. of Japan Society of Civil Engineers, no.194, pp. 29–40, 1971–10, (in Japanese).
9. N. Hasebe and J. Iida, A Crack Originating from a Triangular Notch on a Rim of a Semi-
Infinite Plate, Eng. Fracture Mech., vol. 10, pp. 773–782, 1978.
10. N. Hasebe and S. Inohara, Stress Analysis of a Semi-Infinite Plate with an Oblique Edge
Crack, Ingenieur Archiv, vol. 49, pp. 51–62, 1980.
11. N. Hasebe and M. Ueda, Crack Originating from a Corner of a Square Hole, Eng.
Fracture Mech., vol. 13, pp. 913–928, 1980.
Downloaded by [University of California Davis] at 10:13 11 April 2014

12. N. Hasebe, L. M. Keer, and S. Nemat-Nasser, Stress Analysis of a Kinked Crack Initiat-
ing from a Rigid Line Inclusion I. Formulation, Mech. Materials, vol. 3, pp. 131–145,
1984.
13. N. Hasebe and Y. Z. Chen, Stress Intensity Solutions for the Interaction between a Hole
Edge Crack and a Line Crack, Int. J. Fracture, vol. 77, pp. 351–366, 1996.
14. K. Yoshikawa and N. Hasebe, Heat Source in Infinite Plane with Elliptic Rigid Inclusion
and Hole, ASCE J. Eng. Mech., vol. 125, pp. 684–691, 1999.
15. N. Hasebe, K. Tamai, and T. Nakamura, Analysis of a Kinked Crack Under Uniform
Heat Flow, ASCE, J. Eng. Mech., vol. 112, pp. 31–42, 1986.
16. N. Hasebe, A. Tomida, and T. Nakamura, Thermal Stresses of a Cracked Circular Hole
Due to Uniform Heat Flux, J. Thermal Stresses, vol. 11, pp. 381–391, 1988.
17. K. Yoshikawa and N. Hasebe, Green’s Function for a Heat Source in an Infinite Region
with an Arbitrary Shaped Hole, ASME J. Appl. Mech., vol. 66, pp. 204–210, 1999.
18. J. J. Han and N. Hasebe, Thermal Stress Problem for Mixed Heat Condition Boundary
Around an Arbitrarily Shaped Hole with Crack Under Uniform Heat Flux, J. Thermal
Stresses, vol. 24, pp. 725–735, 2001.
19. J. J. Han and N. Hasebe, Green’s Function for a Thermomechanical Mixed Boundary
Value Problem of an Infinite Plane with an Elliptic Hole, J. Thermal Stresses, vol. 24,
pp. 903–916, 2001.
20. A. L. Florence and J. N. Goodier, Thermal Stress at Spherical Cavities and Circular Holes
in Uniform Heat Flow, ASME, J. Appl. Mech., vol. 26, pp. 293–294, 1959.
21. A. L. Florence and J. N. Goodier, Thermal Stresses Due to Disturbance of Uniform Heat
Flow by an Insulated Ovaloid Hole, ASME, J. Appl. Mech., vol. 27, pp. 635–639, 1960.
22. K. Yoshikawa and N. Hasebe, Green’s Function of the Displacement Boundary Value
Problem for a Heat Source in an Infinite Plane with an Arbitrary Shaped Rigid Inclusion,
Arch. Appl. Mech., vol. 69, pp. 227–239, 1999.
23. N. Hasebe, A. Tomida, and T. Nakamura, Solution of Displacement Boundary Value
Problem Under Uniform Heat Flux, J. Thermal Stresses, vol. 12, pp. 71–81, 1989.
24. N. Hasebe, H. Irikura, and T. Nakamura, A Solution of the Mixed Boundary Value
Problem for an Infinite Plate with a Hole Under Uniform Heat Flux, ASME, J. Appl.
Mech., vol. 58, pp. 996–1000, 1991.
25. K. Yoshikawa and N. Hasebe, Analysis of the Debonding Propagation Initiated on the
Square Rigid Boundary Under Uniform Heat Flux, Trans. of the Japan Society of Mech.
Engineers, Series A, vol. 65, pp. 1600–1606, 1999 (In Japanese).
26. N. Hasebe and K. Yoshikawa, Solution of an Elliptical Rigid Inclusion with Debonding
in an Infinite Plane Under the Uniform Heat Flux, J. Thermal Stresses, vol. 22,
pp. 189–212, 1999.
 642 N. HASEBE AND X. WANG

27. N. Hasebe and J.-J. Han, Green’s Function of Thermoelastic Mixed Boundary Value
Problem for an Elliptic Hole, ASCE, J. Eng. Mech., vol.127, pp. 800–807, 2001.
28. J.-J. Han and N. Hasebe, Green’s Function for Thermal Stress Mixed Boundary
Value Problem of an Infinite Plane with an Arbitrary Hole Under a Point Heat Source,
J. Thermal Stresses, vol. 25, pp. 1147–1160, 2002.
29. J.-J. Han and N. Hasebe, Green’s Functions of Point Heat Source in Various Thermo-
elastic Boundary Value Problems, J. Thermal Stresses, vol. 25, pp. 153–167, 2002.
30. T. Saito, N. Hasebe, and X. F. Wang, Interaction Between A Cracked Elliptical Hole and
a Line Crack Under Uniform Heat Flux, Trans. of the Japan Society of Mech. Engineers,
Series A, vol. 70, pp. 1110–1115, 2004 (In Japanese).
31. N. Hasebe, Uniform Tension of a Semi-Infinite Plate with a Crack at an End of a
Stiffened Edge, Ingenieur Archiv, vol. 48, pp. 129–141, 1979.
32. N. Hasebe, S. Tsutsui, and T. Nakamura, Debondings at a Semielliptic Rigid Inclusion on
Downloaded by [University of California Davis] at 10:13 11 April 2014

the Rim of a Half Plan, ASME, J. Appl. Mech., vol. 55, pp. 574–579, 1988.
33. N. Hasebe, S. Tsutsui, and T. Nakamura, A Mixed Boundary Value Problem for Debond-
ing of a Semielliptic Rigid Inclusion on the Rim of a Half Plane, Arch. Appl. Mech.,
vol. 62, pp. 306–315, 1992.
34. N. Hasebe, X. F. Wang, and M. Kondo, Interaction Between Crack and Arbitrarily
Shaped Hole with Stress and Displacement Boundaries, Int. J. of Fracture, vol. 119,
pp. 83–102, 2003.
35. N. Hasebe, X. F. Wang, and M. Kondo, Green’s Functions for Plane Problem under
Various Boundary Conditions and Applications, Int. J. of Solids and Structures, vol. 40,
pp. 5037–5049, 2003.
36. M. Tsuji and N. Hasebe, Thermal Stresses of Crack Originating from a Corner of a
Rhombic Hole in a Plate Due to Uniform Heat Flux, Trans. of the Japan Society of Mech.
Engineers, Series A, vol. 57, pp. 1569–1574, 1991 (In Japanese).
37. N. Hasebe, H. Irikura, and T. Nakamura, Stress Intensity Factors of Cracks Initiating
from a Rhombic Hole Due to Uniform Heat Flux, Eng. Fracture Mech., vol. 42,
pp. 331–337, 1992.
38. N. Hasebe, K. Tamai, and T. Nakamura, Direction of Kinked Crack for Heat Flux,
ASCE, J. Eng. Mech., vol. 114, pp. 553–557, 1986.

APPENDIX 1: RATIONAL MAPPING FUNCTION IN EXAMPLE 1 [5,10]

Equation (38) can be expressed in the following form and further expanded to a
power series

X
1
aða
1Þ    ða
k þ 1Þ
ð1
fÞa ¼ ð
1Þk fk ¼ 1
a1 f
a2 f2
a3 f3 þ    ðA1Þ
k¼0
k!

This term represents the shape with a projection, where coefficient a ð0 < a < 1Þ is
determined by the angle of the projection.
Consider the following fractional expression and the expansion of the power
series corresponding to Eq. (A1):

Xn
 Xn

Aj
1þ þ Aj ¼ 1
ðAj aj f þ Aj a2j f2 þ Aj a3j f3 þ   Þ ðA2Þ
j¼1
1
a j f j¼1
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 643

where jaj j < 1; The signs in front of Aj is determined from the condition that
the signs of the coefficients in Eqs. (A1) and (A2) are identical. The 2n
unknowns Aj ; aj are determined by approximately equating the corresponding
coefficients of the terms with the same power in the series of Eqs. (A1) and (A2),
respectively,
X
n
Aj akj ¼ ak ðA3Þ
j¼1

The selection of 2n numbers of ak depends on the convergence of the expanded

power series, which is treated separately for the cases of rapidly and slowly
convergent terms. Here, considering the slow convergence of the series in Eq. (A1),
Downloaded by [University of California Davis] at 10:13 11 April 2014

the monotonous decrease of its coefficients, and the convenience of the numerical
calculation, ak are selected up to high order of power for f in the following way:

1st block a1 ; a2 ; a3 ; a4 ; a5 ; a6 ; a7 ; a8
2nd block a2M ; a3M ; a4M ; a5M
3rd block a2M 2 ; a3M 2 ; a4M 2 ; a5M 2
ðA4Þ
4th block a2M 3 ; a3M 3 ; a4M 3 ; a5M 3
5th block a2M 4 ; a3M 4 ; a4M 4 ; a5M 4
6th block a2M 5 ; a3M 5 ; a4M 5 ; a5M 5

28 of ak (in this case) was selected in six blocks of which the particular terms
are defined by M. Eq. (A3) is solved by the iteration. In this case, the number
of the fractional expressions is n ¼ 14. The number of blocks may be increased
or decreased due to the convergence of the series in Eq. (A1). The value of M is
related to the interval of ak . M ¼ 8 or its vicinity 7 or 9 is selected due to our
experience for the good agreement between (A1) and (A2) and the good conver-
gence of the numerical iteration. That more numbers of ak are adopted in the
1st block than others is to make the coefficients of low power agree well.
The nonlinear equations (A3) are solved by the numerical iteration in the fol-
lowing way: Chiefly, A1 ; a1 and A2 ; a2 of Aj ; aj are determined from the 6th block
and A3 ; a3 ; A4 ; a4 from the 5th block and so forth. The calculation is started from
the 6th block. All initial values of Aj ; aj except A1 ; a1 and A2 ; a2 which will be determ-
ined from the 6th block can be treated as zero in the first iterative calculation. By
5
subtracting the influence of A3 ; a3 ; A4 ; a4 , in the 5th block, i.e., A3 a2M 3 þ
5 5 5 5 5 5 5
A4 a2M
4 ; A3 a3M
3 þ A4 a3M
4 ; A3 a4M
3 þ A4 a4M
4 ; A3 a5M
3 þ A4 a5M
4 , from the 4 num-
bers of a2M 5 ; a3M 5 ; a4M 5 ; a5M 5 in the 6th block, A1 ; a1 and A2 ; a2 can be obtained
by solving the following equations expressed in terms of a02M 5 ; a03M 5 ; a04M 5 ; a05M 5 :

5 5 5 5
A1 a2M
1 þ A2 a2M
2 ¼ a2M 5
A3 a2M
3 þ A4 a2M
4  a02M 5
5 5 5 5
A1 a3M
1 þ A2 a3M
2 ¼ a3M 5
A3 a3M
3 þ A4 a3M
4  a03M 5
5 5 5 5
ðA5Þ
A1 a4M
1 þ A2 a4M
2 ¼ a4M 5
A3 a4M
3 þ A4 a4M
4  a04M 5
5 5 5 5
A1 a5M
1 þ A2 a5M
2 ¼ a5M 5
A3 a5M
3 þ A4 a5M
4  a05M 5
 644 N. HASEBE AND X. WANG

Eq. (A5), corresponding to the case of n ¼ 2 and k ¼ 4 in Eq. (42), can be solved in
5
M5
the same way as for Eq. (42). aM 1 ; a2 are obtained by successively solving the linear
equations in two variables and a square equation. After that, A1 ; A2 are obtained
from the 1st two equations in Eq. (A5). Next, consider the 5th block to determine
A3 ; a3 ; A4 ; a4 . By subtracting the effect of A5 ; a5 ; A6 ; a6 and A1 ; a1 ; A2 ; a2 in the 4th
and 6th block, respectively, which corresponds to a2M 4 et al., from the 4 numbers
4
M4
of a2M 4 et al., in the 5th block, A3 ; aM 3 ; A4 ; a4 can be obtained by solving four sim-
ultaneous equations similar equation (A5) to the 6th block. In case of treating the
4th block, the effect of the 3rd, 5th, and 6th blocks need to be considered. The influ-
ence
 of the blocks in lower order on the blocks in higher order is small because
aj  < 1. For example, the values of aM 5 and aM 5 obtained from the 2nd block or
5
9 10
M5 M5 M5
the values of aM 11 ; a12 ; a13 ; a14 from the 1st block have almost no effect on the
Downloaded by [University of California Davis] at 10:13 11 April 2014

blocks in high order in case M ¼ 8. The 1st block, with 8 elements, may be determ-
ined by solving the linear equations in 4 variables and an equation of 4th power since
it corresponds to the case of n ¼ 4; k ¼ 8 in Eq. (42). Thus, after solving the 1st
block, return to the 6th block and do the same iterative procedure till the variation
becomes small   enough. Finally, we have the values of Aj ; aj , which should satisfy the
condition aj  < 1.
From the above statement, we know the number of elements in each block in
(A4) does not have to be 4. If it is 6, in order to determine the fractional expres-
sions in each block, the linear equations in 3 variables instead of 2, and a cube
equation instead of the square one must be solved. And it does not matter to
use 4 elements in the 1st block in (A4). The selection depends on the required ac-
curacy, the interval of aj and the value of M. The term of (A4) must present in the
mapping function in case that there is a projection in the configuration to be ana-
lyzed. The index a depends on the angle of the projection. Accordingly, e.g.,
a ¼ 0:5 when the angle of the projection is 90 degrees. It is not necessary to com-
pute the values of Aj ; aj every time in the analysis once they have been obtained for
the particular angles.

APPENDIX 2: THE TEMPERATURE AND THE STRESS FUNCTIONS

FOR A HEAT SOURCE COUPLE
The temperature and the stress functions for an infinite plane without hole
subjected to a point heat source at za and a heat sink at infinity can be expressed
by [17]

M
Xa ðzÞ ¼
logðz
za Þ
2pk
a0 MGR
ua ðzÞ ¼ ðz
za Þflogðz
za Þ
1g ðA6Þ
4pk
a0 MGR
wa ðzÞ ¼

za logðz
za Þ
4pk

Conversely, the temperature and the stress functions for an infinite plane subjected
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 645

to a point heat sink at zb and a heat source at infinity are expressed by

M
Xb ðzÞ ¼ logðz
zb Þ
2pk
a0 MGR
ub ðzÞ ¼
ðz
zb Þflogðz
za Þ
1g ðA7Þ
4pk
a0 MGR
wb ðzÞ ¼
zb logðz
zb Þ
4pk

Consider the functions Fa ðza Þ and Fb ðzb Þ for the heat source at za and the heat sink
at zb , respectively. When the two points approach each other, we can obtain the lim-
iting function W ðza ; zb Þ as shown in Figure 10,
Downloaded by [University of California Davis] at 10:13 11 April 2014

W ðza ; zb Þ ¼ Fa ðza Þ þ Fb ðzb Þ ¼
fFa ðzb Þ
Fa ðza Þg ðA8Þ

Considering the following expression when the heat source points zb ! za

 
W ðza ; zb Þ Fa ðzb Þ
Fa ðza Þ dFa ðza Þ
lim ¼ lim
¼
ðA9Þ
zb !za zb
za zb !za zb
za dza

and the angle k between the heat source and the x-axis with ðzb
za Þ ¼ eeik ,
we have
dFa ðza Þ
W ðza ; zb Þ
ðzb
za Þ ¼
eeik Fa0 ðza Þ ðA10Þ
dza

Noting the limit Me ! P when e ! 0, the temperature and the stress functions can
be obtained in the following forms when using the definition of the intensity of heat
source couple P:

P 1
X0 ðzÞ ¼
2pk z
za
a0 GRP ik
uðzÞ ¼
e logðz
za Þ ðA11Þ
4pk  
a0 GRP
ik
za
wðzÞ ¼
e logðz
za Þ
eik
4pk z
za

By introducing the mapping function, the above equations can be expressed in the
f plane as

P 1
X1 ðfÞ ¼ 0
2pk x ðfÞðf
fa Þ
a0 GRP ik
/1 ðfÞ ¼
e logðf
fa Þ ðA12Þ
4pk ( )
a0 GRP
ik xðf a Þ
w1 ðfÞ ¼
e logðf
fa Þ
eik 0
4pk x ðfa Þðf
fa Þ
 646 N. HASEBE AND X. WANG

a0 GRP ik
where fa is the point of the heat source couple in the f-plane. Using e  C,
2k
the above expression can be rewritten as

C
/1 ðfÞ ¼
logðf
fa Þ
2p

ðA13Þ
C C xðfa Þ
w1 ðfÞ ¼
logðf
fa Þ þ
2p 2p x0 ðfa Þðf
fa Þ

The Derivation of the Temperature and the Stress

Functions Eq. (170)
The temperature function to be determined is expressed
Downloaded by [University of California Davis] at 10:13 11 April 2014

XðfÞ ¼ X1 ðfÞ þ X2 ðfÞ ðA14Þ

where X1 ðfÞ is given by the 1st equation in Eq. (A12). Substituting the above
equation into Eq. (22), multiplying by dr=½2piðr
fÞ and carrying out the Cauchy
integral on the unit circle, the temperature function is determined as the 1st equation
in Eq. (170).
The stress functions to be determined are

/ðfÞ ¼ /1 ðfÞ þ /2 ðfÞ þ A log f

ðA15Þ
wðfÞ ¼ w1 ðfÞ þ w2 ðfÞ þ B log f

where /1 ðfÞ; w1 ðfÞ are given by the 2nd and 3rd equations in Eq. (A12). The
constants A and B are determined from the single valuedness condition of the stres-
ses and the displacement by the 3rd equation in Eq. (170). Substituting the above
equation into the boundary condition (26), multiplying it by dr=½2piðr
fÞ and car-
rying out the Cauchy integral on the unit circle, the stress functions are determined
as the equation in Eq. (170).

APPENDIX 3: DERIVATIVE OF THE INTEGRAL [24,33]

The derivative of the following integral function can be expressed in the form
without integrals:

Z b 
d 1
vðfÞ dr ¼
yðfÞfg1 ðfÞ þ g2 ðfÞg ðA16Þ
df a vðrÞðr
fÞ

where
vðfÞ
yðfÞ ¼
Q
N
ðf
ak Þðf
bk Þ
k¼1 ðA17Þ
1 1
g1 ðfÞ ¼

yðaÞðf
aÞ yðbÞðf
bÞ
 COMPLEX VARIABLE METHOD FOR 2-D THERMAL STRESS PROBLEM 647

Here, when N ¼ 1,
g2 ðfÞ ¼ 0 ðA18Þ

when N ¼ 2,

g2 ðfÞ ¼ Ið0Þfð1
mÞðf
a1 Þðf
a2 Þ þ mðf
b1 Þðf
b2 Þ
ma1 a2

ð1
mÞb1 b2 g þ Ið1Þfmða1 þ a2 Þ þ ð1
mÞðb1 þ b2 Þg
Ið2Þ ðA19Þ

in which
Z b
rk
IðkÞ ¼ dr ðk ¼ 0; 1; 2Þ ðA20Þ
a vðrÞ
Downloaded by [University of California Davis] at 10:13 11 April 2014

The numerical quadrature is relatively simple in evaluating the integrals for higher
than N ¼ 3 because the analytical formulation becomes quite complex [26,32,33].

APPENDIX 4: NUMERICAL QUATRATURE FOR EQS. (176), (181),

AND STRESS INTENSITY FACTORS
Numerical evaluation of Eq. (176) Eq. (176) is obtained by using the follow-
ing variable replacement when the density of the heat source couple is given by
Eq. (175)

t ¼
f cos h ð
f  t  f ; 0  h  pÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðA21Þ
dt ¼ f 2
t2 dh

The Gauss–Chebyshev method is used to solve the integral equation. Dividing the
interval 0  h  p by equal M parts, and using the relation dh ¼ p=M, Eq. (176)
can be rewritten as follows:
XM
p   2
Gðtj Þ T sk ; tj
T s0 ; tj f
t2j ¼
½XB ðsk Þ
XB ðs0 Þ ðA22Þ
j¼1
M

where sk ; tj are
 
ð2j
1Þp
tj ¼ f cos ðj ¼ 1; 2; 3; . . . ; MÞ
2M
  ðA23Þ
kp
sk ¼ f cos ðk ¼ 1; 2; 3; . . . ; MÞ
M

The values of Gðtj Þ are obtained by solving M equations (A22). Using these values
Pðtj Þ ¼ Gðtj Þðf 2
t2j Þp=M ðj ¼ 1; 2; 3; . . . ; MÞ in C of Eq. (170), stress components,
NC ðsÞ þ iTC ðsÞ at the point s, can be calculated from stress function of Eq. (170).
Numerical evaluation of Eq. (181) The numerical evaluation of the integral
equation (181) can be carried out in the same way as for Eq. (176). The distribution
 648 N. HASEBE AND X. WANG

density of dislocation hn ðtÞ; hs ðtÞ is given by Eq. (180). Using the variable
replacement of (A21), we have

XM
p h i
Hn ðtj Þ þ iHs ðtj Þ ¼ 0
j¼1
M
XM
p n    o
Hn tj Nn sk ; tj þ iTn ðsk ; tj Þ þ Hs ðtÞ Ns ðsk ; tj Þ þ iTs ðsk ; tj Þ ðA24Þ
j¼1
M
¼
fNB ðsk Þ þ NC ðsk Þ þ i½TB ðsk Þ þ TC ðsk Þg

where sk ; tj are given by Eq. (A23). The values of Hn ðtj Þ; Hs ðtj Þ ðj ¼ 1; 2; 3; . . . ; MÞ

Downloaded by [University of California Davis] at 10:13 11 April 2014

are obtained from a system of 2M equations for the real and the imaginary parts
of the above equations for sk ðk ¼ 1; 2; 3; . . . ; MÞ.

Stress intensity factors The stress intensity factors for the crack tips A and
B (see Figure 8) can be expressed in terms of Hn ðtj Þ; Hs ðtj Þðj ¼ 1; 2; 3; . . . ; MÞ
obtained in the previous section

rffiffiffi( X
)
p 1 M jþ1   ð2j
1Þp
KA ¼
ð
1Þ Hn ðtj Þ þ iHs ðtj Þ cot
f M j¼1 4M
rffiffiffi (
) ðA25Þ
p 1 X M
jþM   ð2j
1Þp
KB ¼ ð
1Þ Hn ðtj Þ þ iHs ðtj Þ tan
f M j¼1 4M

The stress intensity factor for the crack tip C is obtained by the superposition of
problems B, C, and D:
Z f
KC ¼ UB þ ½cðtÞUC ðtÞ þ hðtÞUD ðtÞdt ðA26Þ

f

where

pffiffiffi /0m ðf0 Þ

Um ¼ 2 p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi ðm ¼ B; C; DÞ ðA27Þ
eik x00 ðf0 Þ

is obtained from Eq. (182). Here, k is the angle between the crack C and the x-axis,
k ¼ 0 in Figure 8; f0 is the point on the unit circle corresponding to the crack tip C.
Using the values of Gðtj Þ; Hn ðtj Þ; Hs ðtj Þðj ¼ 1; 2; 3; . . . ; MÞ, Eq. (A26) becomes

XM
p h 2 i
KC ¼ UB þ ðf
t2j ÞGðtj ÞUC ðtj Þ þ Hðtj ÞUD ðtj Þ ðA28Þ
j¼1
M