You are on page 1of 30

ON SOME FINITE DEFORMATIONS OF INHOMOGENEOUS

COMPRESSIBLE ELASTIC SOLIDS

By
U. Saravanan
Department of Civil Engineering, Structural Engineering Division
IIT-Madras, Chennai 600036, Tamil Nadu, India

and

K.R. Rajagopal∗
Department of Mechanical Engineering and Mathematics, Texas A&M University
College Station, TX 77843-3368, USA

[Accepted 25 September 2006. Published 4 April 2007.]

Abstract
We study several simple boundary value problems for a special class of inhomoge-
neous, isotropic, compressible elastic solids: inflation of a spherical shell; bending,
stretching and shearing of a rectangular block; and straightening, stretching and
shearing of a sector of a hollow cylinder. The purpose of the study is two-fold: to es-
tablish new exact solutions to boundary value problems concerning inhomogeneous,
isotropic, compressible elastic solids that have technological relevance (to the de-
formation of layered composites); and to reaffirm a thesis concerning an inherent
difficulty with regard to the homogenization techniques that are in vogue that ap-
peal to the stored energy in the homogenized body being the same as that for the
inhomogeneous body. In addition to establishing new exact solutions for a class of
inhomogeneous elastic solids, the study reinforces the thesis of Saravanan and Ra-
jagopal that great care ought to be exercised in approximating even a very mildly
inhomogeneous body by an equivalent homogeneous body when large deformations
are involved.

1. Introduction
All bodies are inhomogeneous. However, many of them, those whose inhomogeneities
are appropriately ‘mild’, are usually approximated as homogeneous bodies. When
the inhomogeneities are not mild, in that the properties of the body vary signifi-
cantly over the body, then such bodies cannot be approximated as homogeneous
bodies. When the homogeneities are very mild they can be, and usually are, ap-
proximated as a homogeneous body. Even in the case of homogeneous, compressible,
non-linear elastic solids, few exact solutions are available and such exact solutions
are even rarer when one considers inhomogeneous, compressible, non-linear elastic
bodies. In this paper, we find several new exact solutions for inhomogeneous, non-
linear elastic solids, even when the inhomogeneity is not mild. These solutions are

∗ Corresponding author, e-mail: krajagopal@mengr.tamu.edu

c Royal Irish Academy


Mathematical Proceedings of the Royal Irish Academy, 107A (1), 43–72 (2007)°
44 Mathematical Proceedings of the Royal Irish Academy

established for a variety of layered composites in which the inhomogeneity is not


necessarily mild. Approximating homogeneous bodies, even when the inhomogene-
ity appears ‘mild’, can lead to serious errors resulting in a flawed design of various
load-bearing bodies.
We illustrate the relevant issues by solving three classes of boundary value
problems—inflation of a spherical shell; bending, stretching and shearing of a rect-
angular block; and straightening, stretching and shearing of a sector of a hollow
cylinder—for a class of inhomogeneous Blatz–Ko bodies (see [2]). These studies go
hand in hand with previous studies by Saravanan and Rajagopal [20; 21; 22], and
together they show that in a large class of deformations, significant differences are
to be found between the solutions obtained for the inhomogeneous body and those
obtained for the homogeneous approximation. The deformations that are studied
in this current paper are a subclass of the universal deformations for homogeneous,
isotropic, incompressible simple materials that have been studied by Wineman [26],
by Carroll [5] and by Fosdick [14].
There are two aspects to the present work. The first concerns the appropri-
ateness of the homogenization procedure that is usually appealed to in non-linear
elasticity, namely that the stored energy in the inhomogeneous and the homoge-
nized models have to be the same. This issue has been discussed previously within
the context of a different set of deformations of a class of inhomogeneous bodies. It
is also important to bear in mind that in our earlier work we had shown that one
ought to be careful concerning homogenization by using a specific boundary value
problem. We want to document that our thesis is not a happenstance, and in order
to do so, we feel it is important to consider a different class of boundary value prob-
lems and thus show that our thesis is once again valid, hence lending more credence
to our thesis in general. The second aspect of the present paper concerns, as men-
tioned earlier, the establishment of new solutions to boundary value problems for a
class of inhomogeneous, compressible elastic solids that have technical relevance. It
will be apparent to any one who scans the literature that even within the context
of homogeneous isotropic bodies, very few exact solutions are available. And when
it comes to inhomogeneous, compressible elastic bodies, such exact solutions are
practically non-existent.
Thus, the exact solutions established in this paper, for inhomogeneous com-
pressible elastic bodies, takes on added significance. More importantly, our solution
procedure is valid for general inhomogeneous materials and not merely for mildly
inhomogeneous materials. The classes of problems that are studied have a lot of
relevance to layered inhomogeneous materials that are used commonly (many com-
posite solid bodies are such layered bodies), and the material functions that char-
acterize these layers could be very different. It is important to recognize that the
inhomogeneity need not be mild and that the different layers can have properties
that vary significantly. We feel that our solutions from this perspective would be a
valuable addition to the literature.
Two material points P1 and P2 are said to be materially uniform (see Truesdell
and Noll [25])—when attention is restricted to purely mechanical processes (i.e.,
we do not consider the thermal, electro-magnetic or other responses of the body)—
if there exist two placers κ1 and κ2 such that there exist neighborhoods NX1 of
Saravanan and Rajagopal—Deformation of inhomogeneous solids 45

X1 = κ1 (P1 ) and NX2 of X2 = κ2 (P2 ) that are indistinguishable with respect to


their mechanical response. A body, B, is said to be homogeneous if all the material
points are materially uniform with respect to a single placement. A body that is
not homogeneous is said to be inhomogeneous. In other words, a body is considered
to be inhomogeneous if there exists no single placer for the entire body with respect
to which the response at each material point is the same.
Ericksen [13] showed that the only deformations that can be maintained by
the application of boundary tractions alone in all homogeneous compressible elas-
tic bodies are homogeneous deformations. Hence, inhomogeneous deformations of
inhomogeneous compressible elastic bodies are possible only for specific forms of
the stored energy function. This has resulted in studies [6; 7; 24] concerned with
delineating constitutive relations that admit deformations of a given form. Also,
several dynamic problems involving isotropic, homogeneous, compressible elastic
solids have been studied (see [3; 4; 11; 12]).
Here we find that, for some forms of the deformation considered, the equilibrium
equations reduce to a second order ordinary differential equation of the form r,RR
= f (R, r, r,R ) for a variety of stored energy functions. We then observe that the
solution to the above second order ordinary differential equation can be represented
as a converging Taylor’s series if f ∈ C ∞ (Ω), where Ω = {(R, r, d)|Rinn ≤ R ≤
Rout , rinn ≤ r ≤ rout , dmin ≤ d ≤ dmax }. We illustrate our ideas using a stored
energy function proposed by Blatz and Ko [2] for polyurethane and study the
inflation of a spherical shell;, the bending, stretching and shearing of a rectangular
block; and the straightening, stretching and shearing of a sector of a hollow cylinder.
The inflation, extension, torsion and shearing of a right circular annular cylinder
studied by Saravanan and Rajagopal [22] can also be studied within the above
framework.
Even though the deformations studied here have been investigated in detail
in the case of compressible solids (see [1; 8; 9; 10; 17; 18; 23] and the references
therein), they seem to have not been investigated for the stored energy function
considered here. More importantly, it seems that these deformations have not been
studied within the context of inhomogeneous solids. As the exact solutions that are
established in this paper are given explicitly and depicted unambiguously in the
figures, we shall not comment on their structure.
Also, the purpose here is not to present a procedure for homogenization, nor
to obtain bounds for errors due to homogenization; the aim instead is to discuss
the appropriateness of the methodology behind certain homogenization procedures
that are based on the equivalence of the stored energy. This is achieved by compar-
ing the solutions obtained by solving identical boundary value problems for both
the inhomogeneous body and certain standard homogeneous approximations. It is
pertinent to point out that many of the available [15; 19] homogenization schemes
require uniform deformation of the boundary or uniform traction on the boundary
of the body. Such conditions do not apply for the boundary value problems studied
here.
From a practical standpoint we have to
• ensure that the actual inhomogeneous body will not fail, when subject to
46 Mathematical Proceedings of the Royal Irish Academy

the external loading at which the corresponding approximate homogeneous


model predicts that the body is intact (or vice versa), and
• obtain a good estimate of the force or traction required to enforce a dis-
placement boundary condition (or vice versa).
In keeping with the observations made by Saravanan and Rajagopal [20; 21;
22], we find that even the sense of stress cannot be captured correctly by models
obtained by using standard homogenized approximations. Thus, even if homoge-
nization might lead to reasonable results with regard to the global response of the
body, it is quite inappropriate when one is interested in determining the failure of
the body that depends on local quantities such as the stress. This should not be
surprising, because homogenization merely requires the stored energy to be approx-
imately the same, while the stress is the derivative of the stored energy with respect
to the deformation gradient. However, it is surprising that, given this obvious fact,
most studies concerned with homogenization appeal to energy equivalence. We fur-
ther find that even if the deformation fields found by solving identical boundary
value problems for the inhomogeneous body and its homogeneous approximation
are ‘close’, in some sense, their gradients are relatively far apart.
Let γ denote some generic material parameter, which varies over the body. Then,
we could define a mean value for γ as
R
γ(X)dV
γmean = V R . (1.1)
V
dV

We could also define the mean by introducing weight functions in the above integral.
We find that the above mean has little relevance to the constant value for the
property γexp in the homogeneous approximation, determined by correlation with
an experiment; this is completely in keeping with the observations made by Sara-
vanan and Rajagopal [20; 21; 22]. To elaborate, suppose we perform an experiment
in which we radially inflate a spherical shell that is in actuality inhomogeneous,
isotropic and compressible but we assume it to be homogeneous, and find the
Sp−inf
constant value γexp by corroborating the experiment, i.e., by correlating the
pressure required to engender a given radial inflation. We find that, in general, the
value for the material modulus is not close to that obtained by using (1.1). Also,
the constant value for the material modulus inferred from another experiment for
the same body will, in general, be different. For a given inhomogeneous body, de-
pending on the specific boundary value problem, we find that γexp could vary by as
much as 180% with regard to correlations from different experiments, for the defor-
mations considered here. This is an order of magnitude less than that reported by
Saravanan and Rajagopal [22] for the boundary value problems that they studied.
Even so, the variations are large and suggest that the bounds for these material
parameters cannot be tight. Here we have assumed that the stored energy associ-
ated with the homogeneous approximation belongs to the same class as that of the
inhomogeneous body. We could of course model an inhomogeneous body comprised
of homogeneous bodies of a certain class by a homogenized model of a different
class, i.e. a body comprised of different Blatz–Ko bodies could be approximated
by a homogenized body that is not a Blatz–Ko body. However, we should not ap-
Saravanan and Rajagopal—Deformation of inhomogeneous solids 47

proximate an inhomogeneous isotropic body by a homogeneous anisotropic body,


since material symmetry concerns response at a material point and inhomogeneity
is associated with response at different points; thus, the response of the two bodies
will not be equivalent in that the symmetry is different.
Sp−inf
Furthermore, γexp obtained by correlating the pressure required to engender
a given inflation depends on the amount of inflation. This suggests that the stored
energy for the homogeneous approximation cannot belong to the same class as that
of the inhomogeneous body. However, we find that except for the straightening
of a hollow cylinder, γexp tends towards γmean with increasing value of inflation
or bending. In other words, for certain boundary value problems the effect of the
inhomogeneity is predominant only for small displacements of the boundary!
We study the problems for different types of inhomogeneities reflected by dif-
ferent departures from the mean, i.e. we examine whether the results obtained are
significantly different for a sinusoidal variation versus a piecewise constant or lin-
ear variation, all having the same mean value. We find that, depending on the
inhomogeneity, for example whether γ(X) decreases or increases monotonically be-
tween the same extremum values, γexp can vary by as much as 560%, which clearly
emphasizes the need to know the actual form of the inhomogeneity.
Instead of correlating the pressure required to engender a given inflation, we can
seek the constant value of the parameter, γmth in the homogeneous approximation,
by requiring the total stored energy in both the inhomogeneous body and its homo-
geneous approximation to be same. Comparing γexp and γmth for a given boundary
value problem, we find that γmth could be 300% larger than γexp . Thus, the ho-
mogenization that appeals to the equivalence of stored energy will not, in general,
predict the boundary traction required to engender a given boundary displacement
correctly.
The above observations point to the pitfalls of modelling inhomogeneous bodies
as homogeneous bodies, especially when non-linear response is considered.
The remainder of this article is arranged as follows. In Section 2 we introduce
the basic kinematical quantities. In Section 3 we discuss the stored energy function
considered in this study. Section 4 introduces the various types of inhomogeneities
considered. Section 5 discusses the solution procedure for a generic, governing,
ordinary differential equation obtained in this study. In Section 6 we examine the
role of inhomogeneities in the inflation of a spherical shell. Section 7 is devoted
to the study of bending, stretching and shearing of an inhomogeneous rectangular
block; and in the final section we consider straightening, stretching and shearing of
an sector of an inhomogeneous hollow cylinder.

2. Kinematics
Let X denote a typical particle in the stress-free reference configuration κR (B) of
the body, and let x denote the position occupied by X at time t in the configuration
κt (B). The motion of the body is defined through the mapping χ that is one to one
for each t ∈ R:
x = χ(X, t). (2.1)
48 Mathematical Proceedings of the Royal Irish Academy

We shall assume that the motion is sufficiently smooth to render all the derivatives
that follow meaningful. The gradient of the deformation F is defined as

∂χ
F= , (2.2)
∂X
and the Cauchy–Green stretch tensors B and C are defined through

B = FFt , C = Ft F. (2.3)

The principal invariants of any second order tensor A are defined through
1
I1 = tr(A), I2 = [(tr(A))2 − tr(A2 )], I3 = det(A). (2.4)
2
We shall find it easier to express some of our results in terms of the invariants
I2 1/2
J1 = I1 , J2 = = tr(A−1 ), J3 = I3 . (2.5)
I3

3. Constitutive Relations
We consider elastic bodies whose stored energy, W, is given by

W := Ŵ (X, J1 , J2 , J3 ). (3.1)

Thus, the Cauchy stress is given by


2
T = W3 1 + [W1 B − W2 B−1 ], (3.2)
J3
where, W1 , W2 , W3 denote the derivative of the stored energy function with respect
to J1 , J2 , J3 , respectively. We find it convenient to decompose the Cauchy stress as

T = W3 1 + Te , (3.3)

where Te = J23 [W1 B − W2 B−1 ], in order to present our results.


The homogeneous version of the stored energy that we consider, for illustration,
was proposed by Blatz and Ko [2] for modelling solid polyurethane rubber. The
inhomogeneous version of this stored energy is given by
h i
W = 0.5µ1 (X) (J3−2µ3 − 1)/µ3 + J1 − 3 , (3.4)

and the stress obtained from (3.4) and (3.2) as

µ1 (X) h i
T= B − J3−2µ3 1 , (3.5)
J3

where µ1 (X) > 0 satisfies the E-inequalities (see Truesdell and Noll [25]) and µ3 is a
constant. If Poisson’s ratio, ν, is such that 0 < ν < 0.5, then µ3 (= ν/(1 − 2ν)) > 0.
Saravanan and Rajagopal—Deformation of inhomogeneous solids 49

Also, we note that Horgan [16] found that (3.4) satisfies strong ellipticity conditions
for all values of the stretch. In this study, we consider two values of µ3 , 6.25 and
0.5, corresponding to a Poisson’s ratio of 0.463 and 0.25 (see Blatz and Ko [2] and
Horgan [16]).
In the absence of body forces, the equation of equilibrium reduces to

div(T) = 0. (3.6)

Next, we record the non-dimensionalization procedure. We introduce dimen-


sionless prescriptions of the position, gradient and divergence through
1 ] = L ∗ grad(·), f
e=
x x, grad(·) div(·) = L ∗ div(·), (3.7)
L
] = ∂(·) and grad(·) = ∂(·) . We note that
where L is a relevant length scale, grad(·) ∂e
x ∂x
the deformation gradient F is already a dimensionless quantity.
We introduce a parameter µo with units of stress to render the Cauchy stress
dimensionless. Here, we identify the parameter µo with the mean value of µ1 . Thus,
Equation (3.2) becomes
h i
T f3 1 + 2 W
e =W f1 B − W
f2 B−1 , (3.8)
J3
fi = Wi /µo and T̃ = T/µo . Hence, Equation (3.6) becomes
where, as usual, W

f T)
div( e = 0. (3.9)

For the sake of convenience, we drop the tilde with the understanding that all
the quantities considered henceforth are dimensionless unless otherwise explicitly
stated.

4. Forms of inhomogeneities
Here, we assume that the material properties vary only along one direction, i.e.
µ1 (X) = µ1 (X1 ). We find it convenient to discuss the variation of the material
property in terms of the parameter X 1 , defined as

X1 − X1min
X1 = , (4.1)
X1max − X1min

and hence X 1 ranges between 0 and 1.


The functional forms chosen for µ1 (X1 ) are such that
R X1max Z
X1min
µ1 (X1 )dX1 1
(µ1 )mean = = µ1 (X 1 )dX 1 = 1. (4.2)
X1max − X1min 0

First, we consider cases where the material parameter varies monotonically. Here we
consider two types, one in which µ1 (X 1 ) increases from X1min to X1max and another
50 Mathematical Proceedings of the Royal Irish Academy

in which it decreases. While this can happen in a variety of ways, we choose the
following simple variations:

4.1. Linear Variation:

µ1 (X 1 ) = 2(1 − δ)X 1 + δ, (4.3)


where 0 < δ < 2. Thus,
½
dµ1 >0 , 0<δ<1
(4.4)
dX 1 < 0 , 1 < δ < 2.

4.2. Exponential Variation:


Here, we suppose that
δ
µ1 (X 1 ) = eδX 1 , (4.5)
eδ −1
where −∞ < δ < ∞. Thus,
½
dµ1 >0 , δ>0
(4.6)
dX 1 < 0 , δ < 0.

In the next three cases, the variation of µ1 is non-monotonic.

4.3. Piecewise constant (PWC) Variation:


We shall assume that
( Pk−1
δ + 2(1 − δ) n=0 (−1)n H(X 1 − nk ), k is even,
µ1 (X 1 ) = k
Pk−1 n n (4.7)
δ + 2(1 − δ) k+1 n=0 (−1) H(X 1 − k ), k is odd,

where,
½ n
n 0 if X1 < k,
H(X 1 − ) = n (4.8)
k 1 if X1 > k,
and δ and k determine the amplitude and frequency of the variation with the
restriction that 0 < δ < 2. (The restriction on δ arises from the requirement that
µ1 > 0.)

4.4. Sinusoidal Variation:


Here we assume that
µ1 (X 1 ) = 1 + δ sin(2kπX 1 ), (4.9)
where −1 < δ < 1 determines the amplitude of the variation and k the frequency.

4.5. Cosine Variation:


Finally, we study the case

µ1 (X 1 ) = 1 + δ cos(2kπX 1 ), (4.10)
Saravanan and Rajagopal—Deformation of inhomogeneous solids 51

with the δ and k having the same meaning as in the previous case. In all the above
cases, we require k to be an integer.

5. Solution Procedure
For all the deformations considered in this article, the balance of linear momentum
reduces to an equation of the form

d2 r
= f (R, r, d), (5.1)
dR2
with the boundary conditions
o
r(Rout ) = rout , g(Rout , r(Rout ), d(Rout )) = Trr , (5.2)
dr
where, d = dR and Rinn ≤ R ≤ Rout . We shall first solve Equation (5.2b) to obtain
d(Rout ) = do . We therefore seek a solution to (5.1) given by Taylor’s series

(R − Rout ) o (R − Rout )2
r(R) = rout + d + f (Rout , rout , do )
1! 2!
(R − Rout )3 d3 r (R − Rout )m dm r
+ |R=R + · · · + |R=Rout +em ,
3! dR3 out
m! dRm
where

d3 r ∂f ∂f ∂f
:= f1 (R, r, d) = +d +f ,
dR3 ∂R ∂r
· ∂d ¸
d4 r ∂2f ∂2f ∂2f ∂2f
:= f2 (R, r, d) = +2 d +f + fd
dR4 ∂R2 ∂r∂R ∂R∂d ∂r∂d
2 2
· ¸2
∂ f ∂ f ∂f ∂f ∂f ∂f ∂f ∂f
+d2 2 + f 2 2 + f +d + +f ,
∂r ∂d ∂d ∂r ∂d ∂d ∂R ∂r
dm+3 r ∂fm ∂fm ∂fn
m+3
:= fm+1 (R, r, d) = +d +f ,
dR ∂R ∂r ∂d
(R − Rout )m+1 dm+1 r
em = |R=ξ , (5.3)
(m + 1)! dRm+1

Rinn < ξ < Rout . (We follow the standard notation that d(·)dR denotes the total
∂(·)
derivative with respect to R and ∂R denotes the partial derivative with respect to
R.) For the above series to converge we need

lim em → 0. (5.4)
m→∞

While the above scheme is implementable in matlab or maple, it is still com-


putationally costly. Hence, for the simulations we use numerical computation. We
convert the second order ODE (5.1) to a system of two first order ODEs by a simple
52 Mathematical Proceedings of the Royal Irish Academy

change of variables:
dr
u = r, v= . (5.5)
dR
The differential equations relating these functions are
du dv
= v, = f (R, u, v), (5.6)
dR dR
with the condition
u(Rout ) = rout , v(Rout ) = do . (5.7)
This system of first order ODEs is integrated using ODE45 in matlab, in which
higher order terms in the Taylor’s series are estimated without requiring differenti-
ation of f .

5.1. Solution procedure for PWC variation:


In this case, the material parameter is piecewise continuous and hence the governing
Equation (5.1) has to be solved in each sub-domain in which the parameter varies
continuously. At the interface, the displacement field has to be continuous and tn
= −t−n , where t(·) denotes the traction vector in the current configuration and n is
the normal to the interface in the current configuration. In other words, we require
that there be no de-bonding at the interface.
Now, if R1 , R2 , . . ., Rn denote the locations where the material parameter is
discontinuous, then we begin by solving the governing Equation (5.1) with the
boundary conditions (5.2) over the domain R1 < R ≤ Rout , instead of over the
domain Rinn ≤ R ≤ Rout . (R = Rj , a constant, denotes a surface across which the
material parameter is discontinuous and R1 > R2 > Rn .) Now the value of r(R1+ )
= r1+ and dRdr
|R=R+ = d+ dr −
1 is known. Then, the value of dR |R=R1− = d1 is obtained
1
by solving
h1 (R1− , r1− , d− + + +
1 ) = h1 (R1 , r1 , d1 ), (5.8)
where the only unknown is d−
1, since r1−
= r1+ .
(For the assumed forms of deforma-
tion and hence the stress, the requirement tn = −t−n reduces to a scalar equation.)
Now, we solve the governing Equation (5.1) for the boundary condition r(R1− )
= r1− and dRdr
|R=R− = d−
1 over the domain R2 < R < R1 . This process is continued
1
until the other boundary of the body (i.e. R = Rinn ) is reached.
As before, the governing equation in each sub-part of the body is solved numer-
ically here.

6. Inflation of a spherical shell


In this section, we confine ourselves to a body, B, that is the annular region between
two concentric spheres:

B = {(R, Θ, Φ)|Rinn ≤ R ≤ Rout , 0 ≤ Θ ≤ 2π, 0 ≤ Φ ≤ π} . (6.1)

For this body, we shall use Rout for the non-dimensionalization of the length.
Saravanan and Rajagopal—Deformation of inhomogeneous solids 53

Here we seek a semi-inverse solution of the form

r = f (R), θ = Θ, φ = Φ, (6.2)

for the deformation in spherical polar coordinates, with (R, Θ, Φ) denoting the
coordinates of a typical material point in the reference configuration, and (r, θ, φ)
denoting the coordinates of a typical material point in the current configuration.
This deformation carries the region between two concentric spheres into a region
between two other concentric spheres.
For the assumed deformation (6.2), the matrix components of the deformation
gradient and left Cauchy–Green stretch tensor in spherical coordinates are given
by
 dr   ¡ ¢2 
dr
0 0 0 0
dR  dR ¡ r ¢2 
F= 0 R
r
0 , B= 0 R 0 . (6.3)
0 0 R r ¡ r
¢ 2
0 0 R

Substituting (6.3) in (3.5), we obtain the matrix components of Cauchy stress in


spherical coordinates to be
 ¡ ¢2 
dr 1
dR − 2µ 0 0
J3 3
µ1 (R) 
 ¡ r ¢2 1


T=  0 R − 2µ 0 , (6.4)
J3  J3 3 
¡ r ¢2 1
0 0 R − 2µ3
J3

dr
¡ r ¢2
where J3 = dR R , and we have restricted µ1 to be a function of R only.
For the special form of the deformation (6.2), the deformation gradient is a
function of only R and hence, if we assume that W = Ŵ (R, J1 , J2 , J3 ), the only
relevant equation in the equilibrium equation (3.6) is

dTrr 1
+ [2Trr − Tθθ − Tφφ ] = 0. (6.5)
dr r

We note that if W = Ŵ (Θ, Φ, J1 , J2 , J3 ), then a deformation of the form (6.2) is


not possible. On substituting (6.4) into (6.5) we obtain

d2 r
f1 + f2 = 0, (6.6)
dR2
where
"µ ¶2 #
R 2µ3 + 1 ³ r ´2
f1 = µ1 + 2µ3 +2 , (6.7)
r J3 R
"µ ¶2 # ·
dµ1 dr 1 1 dr R 1
f2 = − 2µ3 + 2µ1 −
dR dR J3 J3 dR r2 r
54 Mathematical Proceedings of the Royal Irish Academy
µ ¶#
2µ3 + 1 dr r dr r
+ 2µ3 +2 − . (6.8)
J3 dR R2 dR R

We solve (6.6) for the mixed boundary conditions

r(Rout ) = rout , Trr (rout ) = 0, (6.9)

by the method outlined in Section 5. From (6.9b) we obtain


µ ¶ µ2µ+1
3

o Rout 3
d = . (6.10)
rout

If µ1 >· 0 and µ3 > −0.5 (a condition


¸ that is required to ensure that f1 6= 0, i.e. µ3
1
¡ R ¢4 ³ r2 d ´2µ3 +1
6= − 2 r R2 + 1 when (R, r, d) ∈ Ω), then it immediately follows that
f1 > 0. Defining, f = −f2 /f1 , we note that f ∈ C ∞ (Ω), provided µ1 ∈ C ∞ (ΩR )
and Rinn > 0 where Ω = {(R, r, d)|Rinn ≤ R ≤ Rout , 0 < r ≤ rout , 0 < d < ∞} and
ΩR = {R|Rinn ≤ R ≤ Rout }. Thus, there exists a deformation of the form (6.2) for
the special class of Blatz–Ko constitutive relation studied here when Rinn > 0 and
µ1 ∈ C ∞ (ΩR ).
If the variation of µ1 (R) is given by (4.7), then at the interface (surfaces defined
by R = constant, at which the parameter µ1 is not continuous) we require Trr (rj− )
= Trr (rj+ ), which translates into finding (d−
j )∗ > 0 such that

y((d−
j )∗ ) = 0, (6.11)

where
 Ã !2µ3 +1 
− 2 − 2
−  −
(R j ) (R j )
y(d−
j ) = µ1 (Rj ) dj − 
(rj− )2 d−j (r − 2
j )
 Ã !2µ3 +1 
+ 2 + 2
(R j ) (R j )
−µ1 (Rj+ ) d+j − . (6.12)
(rj+ )2 d+j (r + 2
j )

In general, it is not possible to solve (6.11) analytically, and hence we seek a nu-
merical solution using the bisection algorithm. However, when µ3 = 0.5, one of the
cases studied here, we obtain

c 2b2 b
(d−
j )∗ = 2
+ 2
+ 2, (6.13)
6a 3a c 3a
where

√ p Rj−
c = [108a8 + 8b3 + 12 3 27a8 + 4b3 a4 ]1/3 , a= ,
rj−
Saravanan and Rajagopal—Deformation of inhomogeneous solids 55
 Ã !2µ3 +1 
µ1 (Rj+ ) + (Rj+ )2 (Rj+ )2
b= d − .
µ1 (Rj− ) j (rj+ )2 d+ + 2
j (rj )

Other cases for which analytical solutions are possible are when µ3 = −0.5 and µ3
= 0.
Since, y is a continuous function in d−
j and since µ3 > −0.5,

lim y(d−
j ) → −∞, and lim y(d−
j ) → ∞, (6.14)
d−
j →0 d−
j →∞

there exists (d− −


j )∗ ∈ (0, ∞) such that y((dj )∗ ) = 0. Further, since (6.12) is mono-
− −
tonic in dj for dj > 0 and µ3 > −0.5, (6.11) has a unique real valued solution,
(d−j )∗ .
Figure 1 shows the stress distribution when µ1 (R) varies linearly, as prescribed
in (4.3), and Figure 2 depicts the same for PWC variation of µ1 (R), given in (4.7).
It is apparent from these figures that the location of the maximum stress as well as
its magnitude is different for certain classes of inhomogeneous bodies in comparison
with the homogeneous body, for the stress component Tθθ . As seen from Figures
1b and 2b, even the sense of stresses is not preserved for certain classes of inhomo-
geneous bodies with respect to their homogeneous counterparts. From Figure 1 we
infer that these observations are valid even when the deformation is small!
dr
Figure 3 depicts the variation of r and dR when µ1 (R) varies linearly, and Figure
4 depicts the same when the variation of µ1 (R) is given by (4.7). It is apparent
from these figures that even if the deformation fields of inhomogeneous bodies and
homogeneous bodies are in some sense close, their gradients can be relatively far
apart.
The normal component of the radial stress at rinn , which we denote by −P, is
given by
Z rout
£ e e e
¤ dr
P = −Trr (rinn ) = −Trr (rout ) − 2Trr − Tθθ − Tφφ . (6.15)
rinn r
−3
x 10 0.025

δ = 0.1
0 Rinn = 0.5, µ3 = 0.5, δ = 0.1
δ = 0.5 0.02
homog
rout/Rout = 1.001 δ = 0.5
θθ

δ = 1.5 homog
Non dimensional stress T
Non dimensional stress Trr

δ = 1.9 δ = 1.5
0.015
δ = 1.9

0.01

−10
Rinn = 0.5, µ3 = 0.5, 0.005
rout/Rout = 1.001

−0.005
0 0.5 1 0 0.5 1
(R−Rinn)/(Rout − Rinn) (R−Rinn)/(Rout − Rinn)
(a) (b)

Fig. 1—Plot of the non-dimensional stress (a) Trr and (b) Tθθ for a spherical shell sub-
jected to inflation when µ1 (R) is given by (4.3) for various values of δ.
56 Mathematical Proceedings of the Royal Irish Academy

On requiring that Trr (rout ) = 0 and substituting (6.4) into the above equation, we
obtain
Z Rout " µ ¶2 #
R 2 ³ r ´2 dr
P= 2µ1 3 − dR. (6.16)
Rinn r R dR
Let (µ1 )exp denote the constant value for the material modulus, µ1 , for the homoge-
neous approximation of the same type. We determine the value of (µ1 )exp through
a correlation with an experiment, such that both the homogeneous approximation
and the inhomogeneous body require the same boundary traction to engender a
given inflation. Thus,
Z Rout " µ ¶2 #
R 2 ³ r ´2 dr
P(rout ) = µ1 3 − dR
Rinn r R dR
Z Rout 2 "³ ´2 µ ¶2 #
Sp−Inf R rh drh
= (µ1 )exp 3 − dR. (6.17)
Rinn rh R dR

The r(R) obtained for the inhomogeneous body will be different from that obtained
for the homogeneous body (rh (R)), in general, for compressible bodies (see Figures
3 and 4). Thus, while rout , the inflation due to a given pressure P, is same for both
the inhomogeneous body and its homogeneous counterpart, rinn would in general
be different, or vice versa. Now, from (6.17) we obtain
R Rout R2 h¡ r ¢2 ¡ dr ¢2 i
µ
Rinn 1 r 3 R − dR dR
(µ1 )Sp−Inf
exp = h
R Rout R2 ¡ rh ¢2 ¡ drh ¢2 i . (6.18)
Rinn r 3 R − dR dR
h

Instead of correlating the boundary traction with the inflation, we could require
that the total stored energy in the inhomogeneous and corresponding homogeneous
body be the same. Assuming that the stored energy function associated with the
homogeneous approximation for the inhomogeneous body belongs to the same class

0.1 4.5
Rinn = 0.5, µ3 = 6.25,
0 4 [δ = 1.9, k = 3]
r /R = 1.2
out out [δ = 1.3, k = 3]
−0.1 3.5 homog
Rinn = 0.5, µ3 = 6.25, [δ = 0.7, k = 3]
−0.2
θθ

[δ = 0.1, k = 3]
rr

r /R = 1.2 3
Non dimensional stress T

Non dimensional stress T

out out
−0.3
2.5
−0.4
2
−0.5
1.5
−0.6 [δ = 1.9, k = 3]
[δ = 1.3, k = 3]
1
−0.7 homog
[δ = 0.7, k = 3]
−0.8 0.5
[δ = 0.1, k = 3]

−0.9 0

−1 −0.5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(R−Rinn)/(Rout − Rinn) (a) (R−Rinn)/(Rout − Rinn) (b)

Fig. 2—Plot of the non-dimensional stress (a) Trr and (b) Tθθ for a spherical shell sub-
jected to inflation when µ1 (R) is given by (4.7) for various values of δ.
Saravanan and Rajagopal—Deformation of inhomogeneous solids 57
1
1
0.995

0.99
0.9 δ = 0.1
Rinn = 0.5, µ3 = 0.5, 0.985 δ = 0.5
rout/Rout = 1.001 homog
0.98 δ = 1.5
0.8 δ = 0.1 Rinn = 0.5, µ3 = 0.5, δ = 1.9
δ = 0.5
r

r,R
0.975 r /R = 1.001
homog out out

δ = 1.5 0.97
0.7
δ = 1.9
0.965

0.6 0.96

0.955

0.5 0.95
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(R−R )/(R −R ) (a) (R−Rinn)/(Rout − Rinn) (b)
inn out inn

dr
Fig. 3—Plot of (a) r and (b) dR for a spherical shell subjected to inflation when µ1 (R)
is given by (4.3) for various values of δ.

1.25 0.75

0.7
1.2
R = 0.5, µ = 6.25, r /R = 1.2 0.65
inn 3 out out R = 0.5, µ = 6.25,
inn 3
1.15 0.6 rout/Rout = 1.2

0.55
1.1
,R

0.5
r

1.05
0.45
[δ = 1.9, k = 3] [δ = 1.9, k = 3]
[δ = 1.3, k = 3] 0.4 [δ = 1.3, k = 3]
1
homog homog
[δ = 0.7, k = 3] 0.35 [δ = 0.7, k = 3]
[δ = 0.1, k = 3] [δ = 0.1, k = 3]
0.95
0.3

0.9 0.25
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(R−R )/(R −R ) (a) (R−R )/(R −R ) (b)
inn out inn inn out inn

dr
Fig. 4—Plot of (a) r and (b) dR for a spherical shell subjected to inflation when µ1 (R)
is given by (4.7) for various values of δ.

and (µ1 )mth denotes the corresponding material parameter, we obtain that
R Rout h i
−2µ3
R
µ 1 (R) (J 3 − 1)/µ 3 + J1 − 3 R2 dR
(µ1 )Sph
inn
mth = R Rout h i , (6.19)
−2µ3 2 dR
Rinn
(J 3h − 1)/µ 3 + J 1h − 3 R

where, J1h and J3h denote the invariants found from the solution of the identical
boundary value problem for the corresponding homogeneous body.
Figure 5 depicts the variation of (µ1 )Sp−inf
exp and (µ1 )sph
mth with changing thick-
ness of the spherical shell for a linear variation of µ1 (R). It can immediately be
observed that for this deformation, both the parameters (µ1 )Sp−inf exp and (µ1 )sph
mth
tend to (µ1 )mean as the thickness of the shell decreases. Also, it can be seen from the
figures that (µ1 )sph Sp−inf
mth is almost three times as great as (µ1 )exp for some inhomo-
geneous bodies with the same thickness. These observations hold qualitatively for
all the six variations of µ1 (R) studied here. Furthermore, we find that depending on
whether µ1 (R) increases or decreases monotonically between the same extreme val-
58 Mathematical Proceedings of the Royal Irish Academy

ues, (µ1 )Sp−inf


exp could vary from 0.3 to 1.7, i.e. a variation of nearly 600%, indicating
the importance of knowing the precise form of the inhomogeneity.
Figure 6 displays the variation of (µ1 )Sp−inf
exp and (µ1 )sph
mth with increasing in-
flation. Here, we observe that both the parameters tend towards their mean value
with increasing values of inflation. This also holds for all the six variations of µ1 (R)
considered here. Moreover, these observations are consistent with those made pre-
viously by Saravanan and Rajagopal [22] (for a different boundary value problem).

7. Bending, stretching and shearing of a rectangular block


Next, we study the deformation of a rectangular block:

B = {(X, Y, Z)|X1 ≤ X ≤ X2 , −Y1 ≤ Y ≤ Y1 , −Z1 ≤ Z ≤ Z1 } , (7.1)

with X1 > 0. For this body we shall use X2 for the non-dimensionalization of the
length.

1.8 1.6
δ = 0.1 δ = 0.1
δ = 0.5 δ = 0.5
1.6
δ = 1.5 δ = 1.5
1.4
δ = 1.9 δ = 1.9
1.4

1.2
1.2
Sp−inf

Sph
(µ1)mth
(µ1)exp

1 1

0.8
0.8

0.6 r /R = 1.001, µ = 0.5


rout/Rout = 1.001, µ3 = 0.5 out out 3
0.6
0.4

0.2 0.4
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Rinn/Rout R /R
(a) inn out (b)

Fig. 5—Plot of the variation of (a) (µ1 )Sp−Inf


exp and (b) (µ1 )sph
mth vs. Rinn when the variation
of µ1 (R) is given by (4.3).

1.35 1.6
[δ = 1.9, k = 3] [δ = 1.9, k = 3]
1.3 [δ = 1.3, k = 3] [δ = 1.3, k = 3]
[δ = 0.7, k = 3] 1.5 [δ = 0.7, k = 3]
1.25 [δ = 0.1, k = 3] [δ = 0.1, k = 3]

1.2 1.4

1.15 Rinn = 0.5, µ3 = 6.25 1.3


(µ )Sp−inf

Sph
(µ1)mth
1 exp

1.1
Rinn = 0.5, µ3 = 6.25
1.2
1.05

1 1.1

0.95
1
0.9

0.85 0.9
1 1.05 1.1 1.15 1.2 1 1.05 1.1 1.15 1.2
rout/Rout (a) rout/Rout (b)

Sp−Inf
Fig. 6—Plot of the variation of (a) (µ1 )exp and (b) (µ1 )sph
mth vs. rout when the variation
of µ1 (R) is given by (4.7).
Saravanan and Rajagopal—Deformation of inhomogeneous solids 59

Here, we seek a semi-inverse solution of the form

r = r(X), θ = βY, z = λZ + κY (7.2)

for the deformation, with (X, Y, Z) denoting the coordinates of a typical material
point in the reference configuration in a Cartesian coordinate system, and (r, θ, z)
denoting the coordinates of a typical material point in the current configuration in
cylindrical polar coordinates. This deformation carries a rectangular block into an
annular wedge bounded by cylinders r = rinn and r = rout and the planes θ = ±θo
and z = ±zo .
For the assumed deformation (7.2), the deformation gradient and the left Cauchy–
Green stretch tensor are given by

dr
F= er ⊗ Ex + βreθ ⊗ Ey + κez ⊗ Ey + λez ⊗ Ez , (7.3)
dX
µ ¶2
dr 2
B= er ⊗ er + (βr) eθ ⊗ eθ + rβκ(eθ ⊗ ez + ez ⊗ eθ )
dX
+(λ2 + κ2 )ez ⊗ ez , (7.4)

where [er , eθ , ez ] are cylindrical polar coordinate basis vectors and [Ex , Ey , Ez ] are
Cartesian coordinate basis vectors.
Substituting (7.4) into (3.5), the matrix components of the Cauchy stress in
cylindrical coordinates are
 ¡ ¢2 
dr 1
dX − 2µ 0 0
J3 3
µ1 (X) 
 1

,
T= 0 (rβ)2 − rκβ (7.5)
J3  

J3 3
1
0 rκβ λ 2 + κ2 − 2µ3
J3

dr
where J3 = rβλ dX and we have restricted µ1 to be a function of only X, i.e.
µ̂1 (X, Y, Z) = µ1 (X).
For the assumed form of the deformation, the deformation gradient is only a
function of r. Hence, if we assume that W = Ŵ (X, J1 , J2 , J3 ), the balance of linear
momentum (3.6) reduces to requiring

dTrr Trr − Tθθ


+ = 0. (7.6)
dr r

We note that if W = Ŵ (Y, Z, J1 , J2 , J3 ), then the deformation of the form (7.2) is


not possible.
Substituting (7.5) into (7.6) we obtain

d2 r
f1 + f2 = 0, (7.7)
dX 2
60 Mathematical Proceedings of the Royal Irish Academy

where
" #
1 2µ3 + 1
f1 = µ1 + rλβ 2µ3 +2 , and (7.8)
rλβ J3
" # " µ ¶2 #
dµ1 1 dr 1 (2µ3 + 1) dr β
f2 = − 2µ3 +1 + µ1 λβ − . (7.9)
dX rλβ dX J3 J32µ3 +2 dX λ

We solve (7.7) subject to the boundary conditions

r(X2 ) = rout , Trr (rout ) = 0, (7.10)

by the method outlined in Section 5. From (7.10b) we obtain

dr µ3
|X=X2 = do = (rout λβ)− µ3 +1 . (7.11)
dX
If µ1 £ > 0 and µ3 > −0.5 ¤ (a requirement that ensures that f1 6= 0, i.e. µ3 6=
−0.5 1 + (rλβ)2µ3 d2µ3 +2 when (X, r, d) ∈ Ω), then f1 > 0 and hence f (= − ff12 )
∈ C ∞ (Ω) provided rinn > 0 and µ1 ∈ C ∞ (ΩX ), where Ω = {(X, r, d)|X1 ≤ X ≤
X2 , 0 < r ≤ rout , 0 < d < ∞} and ΩX = {X|X1 ≤ X ≤ X2 }. Thus, the special
form of the Blatz–Ko constitutive relation considered here has a unique solution of
the form (7.2) provided rinn > 0 and µ1 ∈ C ∞ (ΩX ).
As before, if the variation of µ1 (X) is given by (4.7), then at the interface we
require that Trr (rj− ) = Trr (rj+ ), which translates into finding (d−
j )∗ > 0 such that
y((d−j )∗ ) = 0, where,
" #
d−
j ¡ ¢−(2µ3 +1)
y(d−
j ) = µ1 (Xj− ) − λβrj− d−
j
rj− λβ
" #
d+
j ¡ ¢−(2µ +1)
−µ1 (Xj+ ) − λβrj+ d+
3
j . (7.12)
rj+ λβ

As before, while it is not possible to obtain a general analytical solution for d−


j , for
the special case when µ3 = 0.5 we obtain

c 2b2 a3 ba
d−
j = + + , (7.13)
6a 3c 3
where
√ p
c = [(108 + 8b3 a4 + 12 3 27 + 4b3 a4 )a2 ]1/3 , a = λβrj− ,
" + #
µ1 (Xj+ ) dj ¡ ¢
+ + −(2µ3 +1)
b= − λβrj dj . (7.14)
µ1 (Xj− ) rj+ λβ

For the general case, since (7.12) is a continuous function of d−


j and since when µ3
Saravanan and Rajagopal—Deformation of inhomogeneous solids 61

> −0.5,
lim y(d−
j ) → −∞, and lim y(d−
j ) → ∞, (7.15)
d−
j →0 d−
j →∞

there exists a (d− −


j )∗ ∈ (0, ∞) such that y((dj )∗ ) = 0, which we find using the
bisection algorithm. Further, since (7.12) is monotonic in d− −
j for dj > 0 and µ3 >

−0.5, y((dj )∗ ) = 0 has a unique, real valued solution.
Figures 7 and 8 are plots for the stress distribution when µ1 (R) is given by (4.7).
We infer from these figures that the location and the magnitude of maximum stress
is different in the inhomogeneous body and its homogeneous counterpart belonging
to the same class. In fact, even the sense of the stress at a given location depends on
the type of inhomogeneous body being studied (see Figure 8). These observations
seem to be independent of the value of k, the frequency of variation.
dr
Figure 9 is a plot of r(X) and dX when µ1 (R) is given by (4.7). As before, it
follows from the figure that even though the deformation fields for the inhomoge-

0.05 4.5
[δ = 1.9, k = 10]
0 [δ = 1.5, k = 10]
4
homog
−0.05 [δ = 0.5, k = 10]
3.5 [δ = 0.1, k = 10]
rr

−0.1
θθ
Non dimensional stress T

Non dimensional stress T

X1 = 0.5, µ3 = 0.5, rout = 2 3


−0.15

−0.2 2.5

−0.25 2

−0.3
[δ = 1.9, k = 10] 1.5 X1 = 0.5, µ3 = 0.5, rout = 2
−0.35 [δ = 1.5, k = 10]
homog 1
−0.4 [δ = 0.5, k = 10]
[δ = 0.1, k = 10]
0.5
−0.45

−0.5 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(X−X )/(X − X ) (a) (X−X1)/(X2 − X1) (b)
1 2 1

Fig. 7—Plot of the non-dimensional stress (a) Trr and (b) Tθθ for bending of a rectangular
block when µ1 (X) is given by (4.7) for various values of δ and k = 10 with [λ, β, κ]
= [1, 1, 0].

0.5
X1 = 0.5, µ3 = 0.5, rout = 2
0.4

0.3
zz
Non dimensional stress T

0.2

0.1

0
[δ = 1.9, k = 10]
−0.1 [δ = 1.5, k = 10]
homog
[δ = 0.5, k = 10]
−0.2 [δ = 0.1, k = 10]

−0.3
0 0.2 0.4 0.6 0.8 1
(X−X )/(X − X )
1 2 1

Fig. 8—Plot of the non-dimensional stress Tzz for bending of a rectangular block when
µ1 (X) is given by (4.7) for various values of δ and k = 10 with [λ, β, κ] = [1, 1, 0].
62 Mathematical Proceedings of the Royal Irish Academy
2
0.9
1.95 X1 = 0.5, µ3 = 0.5, rout = 2
X1 = 0.5, µ3 = 0.5, rout = 2 0.8
1.9

0.7
1.85

0.6
1.8
r

r,X
[δ = 1.9, k = 10] 0.5
1.75 [δ = 1.5, k = 10]
homog [δ = 1.9, k = 10]
1.7 [δ = 0.5, k = 10] 0.4 [δ = 1.5, k = 10]
[δ = 0.1, k = 10] homog
[δ = 0.5, k = 10]
1.65 0.3
[δ = 0.1, k = 10]

1.6 0.2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(X−X1)/(X2 − X1) (X−X1)/(X2 − X1)
(a) (b)

dr
Fig. 9—Plot of (a) r and (b) dX for bending of a rectangular block when µ1 (X) is given
by (4.7) for various values of δ and k = 10 with [λ, β, κ] = [1, 1, 0].

neous and its homogeneous counterpart are close (within 1%), their gradients are
relatively far apart, i.e. a variation of 17%.
Denoting the normal component of the radial stress at rinn (= r(X1 )) by −P,
we obtain
Z rout
e e dr
P = −Trr (rinn ) = −Trr (rout ) − (Trr − Tθθ ) , (7.16)
rinn r
which for the present case reduces to
Z " µ ¶2 #
X2
µ1 2 dr
P= (rβ) − dX. (7.17)
X1 r2 βλ dX

Similarly, the axial load, L, defined as


Z rout
L = 2π Tzz rdr, (7.18)
rinn

reduces to
Z " #
X2
µ1 1
L = 2π λ2 + κ2 − 2µ3 dX. (7.19)
X1 λβ J3
Next, the expression for torque, T , is given by
Z rout Z X2
κ
T = 2π Tθz r2 dr = µ1 r2 dX. (7.20)
rinn X1 λ

Finally, the moment per unit length, M, in the current configuration is given by
Z rout Z X2 " #
µ1 2 1
M= Tθθ rdr = (rβ) − 2µ3 dX. (7.21)
rinn X1 λβ J3

Having obtained the above expressions, we illustrate the role of inhomogeneities


Saravanan and Rajagopal—Deformation of inhomogeneous solids 63

within the context of the special stored energy function introduced in Section 3
above and the various forms of inhomogeneities introduced in Section 4. As before,
let (µ1 )exp denote the constant value of the material parameter µ1 for the homog-
enized approximation of the inhomogeneous body of sub-bodies of the same type.
Then, we determine the value of (µ1 )exp through a correlation with the appropri-
ate boundary value problem, such that both the homogeneous and inhomogeneous
body require the same boundary traction to engender a given stretch or bending.
We propose to correlate with different boundary tractions to obtain (µ1 )exp and
to compare their values. Here, we present the details only for the case β = 1, λ =
1 and κ = 0.1. Then, we can correlate the radial traction required to engender a
given amount of bending to obtain

Z " µ ¶2 #
X2
µ1 (X) 2 dr
P(rout ) = (rβ) − dX
X1 r2 βλ dX
Z X2 " µ ¶2 #
T rr−ben 1 2 drh
= (µ1 )exp 2 (rh β) − dX,
X1 rh βλ dX
R X2 µ1 (X) h 2 ¡ dr ¢2 i
X1 r2 r − dX dX
(µ1 )Texp
rr−ben
= RX h ¡ ¢ i . (7.22)
2 1
r 2 − drh 2 dX
X1 r 2 h dX
h

We could instead correlate the axial load that has to be applied to maintain λ = 1
to obtain
R X2 h i
1
X1
µ1 (X) λ2 + κ2 − 2µ 3
dX
J3
(µ1 )Ax−ben
exp = · ¸ , (7.23)
R X2
λ 2 + κ2 − 1 dX
X1 2µ3
J3h

where J3h = rh drdX λβ. Similarly, we can correlate the moment per unit length in
h

the current configuration that has to be applied to get

R X2 h i
2 1
X1
µ 1 (X) (βr) − 2µ dX
J3 3
(µ1 )M om−ben
exp = · ¸ . (7.24)
R X2
(βr )2− 1 dX
X1 h 2µ3
J3h

Using an identical argument, we can correlate the torque required to maintain κ =


0.1 to obtain
R X2
µ1 (X)r2 dX
T rq−ben
(µ1 )exp = X1R X2 . (7.25)
X1 h
r2 dX

Instead, we could require the total stored energy in both the inhomogeneous
64 Mathematical Proceedings of the Royal Irish Academy

and its homogeneous counterpart to be the same and obtain that


R X2 h³ ´ ¡ dr ¢2 i
−2µ3 2 2 2
X1
µ 1 (X) J 3 − 1 /µ 3 + dX + (βr) + κ + λ − 3 dX
(µ1 )rec
mth = R X2 h³ −2µ3 ´ ¡ drh ¢2 i .
X1
J3h − 1 /µ3 + dX + (βrh )2 + λ2 + κ2 − 3 dX
(7.26)
While Figures 10 and 11 depict the variation of (µ1 )exp with X1 when µ1 (X)
varies in a piecewise constant manner, for various values of amplitude and frequency,
Figure 12 plots the variation of (µ1 )rec mth for the same cases. It follows from these
figures that as X1 tends to X2 , (µ1 )exp tends towards the mean value of µ1 (X).
Except for (µ1 )Ax−ben
exp , all other (µ1 )exp tend towards the mean value of µ1 (X) with
increasing frequency, k. We find that depending on the boundary traction that is
being correlated, (µ1 )exp could vary from 0.7 to 1.35 (i.e. by 200%) for a given
inhomogeneous body.
Figures 13 and 14 show the variation of (µ1 )exp with rout and Figure 15 por-
trays the variation of (µ1 )rec mth with rout , when µ1 (X) varies as given in (4.3), for

1.05 1.5
[δ = 1.9, k = 2]
[δ = 0.1, k = 2]
1.4 [δ = 1.9, k = 10]
1
[δ = 0.1, k = 10]

1.3
0.95 rout = 2, λ = 1, β = 1, κ = 0.1
Ax−ben
(µ )Trr−ben

(µ1)exp
1 exp

rout = 2, λ = 1, β = 1, κ = 0.1 1.2

0.9
1.1
[δ = 1.9, k = 2]
[δ = 0.1, k = 2]
0.85 [δ = 1.9, k = 10]
[δ = 0.1, k = 10] 1

0.8 0.9
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
X1/X2 (a) X /X (b)
1 2

Fig. 10—Plot of (a) (µ1 )Texp


rr−ben
and (b) (µ1 )Ax−ben
exp vs. X1 when the variation of µ1 (R)
is given by (4.7) and µ3 = 0.5.

1.4 1.2
[δ = 1.9, k = 2]
1.15
1.3 [δ = 0.1, k = 2]
[δ = 1.9, k = 10]
1.1
[δ = 0.1, k = 10]
1.2
1.05
1.1
(µ )Trq−ben

1
Mom−ben

1 exp

1 0.95 = 2, λ = 1, β = 1, κ = 0.1
(µ1)exp

r
out

0.9
0.9
0.85 [δ = 1.9, k = 2]
0.8 [δ = 0.1, k = 2]
r = 2, λ = 1, β = 1, κ = 0.1 0.8
out [δ = 1.9, k = 10]
0.7 [δ = 0.1, k = 10]
0.75

0.6 0.7
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
X1/X2 (a) X1/X2 (b)

Fig. 11—Plot of (a) (µ1 )M om−ben


exp and (b) (µ1 )Texp
rq−ben
vs. X1 when the variation of µ1 (R)
is given by (4.7) and µ3 = 0.5.
Saravanan and Rajagopal—Deformation of inhomogeneous solids 65
1.5

1.25

(µ1)mth
rec
1

rout = 2, λ = 1, β = 1, κ = 0.1

0.75 [δ = 1.9, k = 2]
[δ = 0.1, k = 2]
[δ = 1.9, k = 10]
[δ = 0.1, k = 10]

0.5
0 0.2 0.4 0.6 0.8 1
X1/X2

Fig. 12—Plot of (µ1 )rec


mth vs. X1 when the variation of µ1 (R) is given by (4.7) and µ3 =
0.5.

various values of δ. It can be seen from these figures that (µ1 )exp as well as (µ1 )rec
mth
tend towards the mean value of µ1 (X), with increasing values of rout except for
Ax−ben
(µ1 )exp . The significant variation of (µ1 )exp with rout suggests that, for at least
some types of inhomogeneities, one ought not to choose a homogenized body that
belongs to the same class as the constituents of the inhomogeneous body.

8. Straightening, stretching and shearing of a sector of a hollow cylinder


Finally, we study the straightening, stretching and shearing of a sector of a hollow
cylinder denoted by B, where

B = {(R, Θ, Z)|Rinn ≤ R ≤ Rout , −Θ1 ≤ Θ ≤ Θ1 , −Z1 ≤ Z ≤ Z1 } . (8.1)

For this body we shall use Rout for the non-dimensionalization of length.
Now we seek a semi-inverse solution of the form

x = x(R), y = βΘ, z = λZ + κΘ, (8.2)

1.2 1.15

δ = 0.1 δ = 0.1
δ = 0.5 δ = 0.5
δ = 1.5 1.1 δ = 1.5
1.1 δ = 1.9 δ = 1.9

1.05
(µ )Ax−ben

1
(µ1)Trr−ben

1 exp
exp

X1 = 0.5, λ = 1, β = 1, κ = 0.1
1
0.9

X1 = 0.5, λ = 1, β = 1, κ = 0.1
0.95

0.8

0.9
1.5 2 2.5 1.5 2 2.5
rout
(a) rout (b)

Fig. 13—Plot of (a) (µ1 )Texp


rr−ben
and (b) (µ1 )Ax−ben
exp vs. rout when the variation of µ1 (R)
is given by (4.3) and µ3 = 0.5.
66 Mathematical Proceedings of the Royal Irish Academy

δ = 0.1 δ = 0.1
δ = 0.5 δ = 0.5
δ = 1.5 δ = 1.5
1.2 1.1
δ = 1.9 δ = 1.9

1.1 1.05
Mom−ben

(µ )Trq−ben
1

1 exp
(µ1)exp

0.9 0.95

X = 0.5, λ = 1, β = 1, κ = 0.1
1
0.8 0.9

X = 0.5, λ = 1, β = 1, κ = 0.1
1
0.7 0.85
1.5 2 2.5 1.5 2 2.5
r (a) rout (b)
out

Fig. 14—Plot of (a) (µ1 )M om−ben


exp and (b) (µ1 )Texp
rq−ben
vs. rout when the variation of
µ1 (R) is given by (4.3) and µ3 = 0.5.

1.5

δ = 0.1
1.4
δ = 0.5
δ = 1.5
1.3 δ = 1.9

1.2

1.1
(µ1)mth
rec

0.9

0.8
X1 = 0.5, λ = 1, β = 1, κ = 0.1
0.7

0.6
1.5 2 2.5
rout

Fig. 15—Plot of (µ1 )rec


mth vs. rout when the variation of µ1 (R) is given by (4.3) and µ3 =
0.5.

for the deformation, with (R, Θ, Z) denoting the coordinates of a typical material
point in the reference configuration in a cylindrical-polar coordinate system, and
(x, y, z) denoting the coordinates of a typical material point in the current configu-
ration in a Cartesian coordinate system. This deformation straightens and stretches
the body bounded by cylinders R = Rinn and R = Rout and by the planes Θ =
±Θ1 and the planes Z = ±Z1 , and then shears the body in the y − z plane to carry
it into a parallelepiped.
For the assumed deformation (8.2), the deformation gradient and left Cauchy–
Green stretch tensor are given by:

dx β κ
F= ex ⊗ ER + ey ⊗ EΘ + ez ⊗ EΘ + λez ⊗ EZ , (8.3)
dR R R
µ ¶2 µ ¶2
dx β βκ
B= ex ⊗ ex + ey ⊗ ey + 2 [ey ⊗ ez + ez ⊗ ey ]
dR R R
· ³ κ ´2 ¸
+ λ2 + ez ⊗ ez , (8.4)
R
Saravanan and Rajagopal—Deformation of inhomogeneous solids 67

where (ER , EΘ , EZ ) and (ex , ey , ez ) are basis vectors for a cylindrical polar and
Cartesian coordinate system, respectively.
Substituting (8.4) into (3.5), we obtain the components of the Cauchy stress in
cartesian coordinates as
 ¡ dx ¢2 1

dR − 2µ 3
0 0
J3
µ1 (R) 
 ³ ´2
β 1 βκ


T=  0 R − 2µ R 2 , (8.5)
J3  J3 3
¡ ¢ 
βκ κ 2 1
0 R2 λ2 + R − 2µ 3
J3

where J3 = βλ dx
R dR and we have restricted µ1 to be only a function of R. We note that
if µ1 depends on Θ and Z, then the deformation of the form (8.2) is not possible.
Since T is a function of only x, the balance of linear momentum (3.6) requires
Txx = c, a constant, or equivalently

d2 x
f1 + f2 = 0, (8.6)
dR2
where
" #
R 2µ3 + 1 βλ
f1 = µ1 + 2µ3 +2 , and (8.7)
λβ J3 R
· ¸ " #
dµ1 R dx −(2µ3 +1) 1 dx 2µ3 + 1 βλ dx
f2 = − J3 + µ1 − 2µ3 +2 2 , (8.8)
dR βλ dR λβ dR J3 R dR

with the boundary conditions

x(Rinn ) = x1 , Txx (x1 ) = c. (8.9)

In general, we find it easier to solve the ODE than to solve the nonlinear equation
dx
resulting from the requirement that Txx = c for dR and then integrating to find x.
However, if c = 0 then
µ ¶ µµ+1
3 · 2µ3 +1
¸
dx R 3 µ3 + 1 µ3
− µ +1
2µ3 +1
µ3 +1
= , x= (λβ) 3 R µ3 +1
− Rinn + x1 . (8.10)
dR λβ 2µ3 + 1

Thus, the deformation field is independent of µ1 .


Next, we consider the case when µ3 = 0.5 but c 6= 0. For this case, we can solve
dx
the equation Txx = c for dR to obtain

dx k 2b2 b
= + + , (8.11)
dR 6a 3ak 3a
£ √ √ ¤1/3 R
where k = 108a4 + 8b3 + 12a2 3 27a4 + 4b3 , a = λβ , b = c/µ1 . Except for
the case when µ1 (R) is constant, analytical integration of (8.11) is not possible.
Hence, we have to numerically integrate (8.11), instead of which we can directly
68 Mathematical Proceedings of the Royal Irish Academy

integrate the ODE (8.6). However, we observe that for this case the solution, x,
depends on µ1 (R), unlike the previous case.
Using arguments similar to those outlined in the previous two sections, we can
show that there exists a solution of the form (8.2) for the special form of Blatz–Ko
stored energy function and variations of µ1 (R) studied here, provided Rinn > 0, µ3
> −0.5.
The stress distribution is plotted in Figure 16, when the variation of µ1 (R) is
given by (4.3) for the straightening of a cylindrical sector when β = λ = 1 and
κ = 0. As indicated before, we infer from the figure that the magnitude of the
maximum stress and its location change with the specific inhomogeneous body
dx
being studied. The deformation field x and its gradient dR are portrayed in Figure
17 for the same case as before. The figure provides ample evidence that even if
the deformation field for various classes of inhomogeneous bodies are close, their
gradients can be far apart.
Let us consider the case for which β = λ = 1 and κ = c = 0. Suitable normal
tractions Tyy and Tzz have to be applied on faces y = constant and z = constant,

4 0.5

0.45 δ = 0.1
3.5 δ = 0.1
δ = 0.5 δ = 0.5
0.4 homog
homog
3 δ = 1.5
δ = 1.5
Non dimensional stress Tyy

zz

0.35 δ = 1.9
Non dimensional stress T

δ = 1.9
2.5
0.3

2 0.25
Rinn = 0.5, µ3 = 0.5, c = 0.1
Rinn = 0.5, µ3 = 0.5, c = 0.1
0.2
1.5
0.15
1
0.1
0.5
0.05

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(R−R )/(R −R ) (R−Rinn)/(Rout − Rinn)
inn out inn (a) (b)

Fig. 16—Plot of the non-dimensional stress (a) Tyy and (b) Tzz for straightening of a
cylindrical sector when µ1 (R) is given by (4.3) for various values of δ.

1 2.2

δ = 0.1
2
0.9 δ = 0.5
Rinn = 0.5, µ3 = 0.5, c = 0.1 homog
1.8 δ = 1.5
0.8 δ = 1.9

1.6
x,R

0.7
x

δ = 0.1 1.4
δ = 0.5 Rinn = 0.5, µ3 = 0.5, c = 0.1
0.6 homog
δ = 1.5 1.2
δ = 1.9
0.5
1

0.4 0.8
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(R−R )/(R −R ) (a) (R−Rinn)/(Rout − Rinn) (b)
inn out inn

dx
Fig. 17—Plot of (a) x and (b) dR for straightening of a cylindrical sector when µ1 (R) is
given by (4.3) for various values of δ.
Saravanan and Rajagopal—Deformation of inhomogeneous solids 69

respectively, and the distribution of these normal tractions depends on the variation
of µ1 (R). Then, if we are interested in correlating the force per unit length in the
current configuration that has to be applied along ey , we obtain that
Z z2 Z x2 Z Rout à !
1 1
Fy = Tyy dxdz = (z2 − z1 ) µ1 (R)R − 2µ3 dR,
z1 x1 Rinn R2 J3
Z Rout µ ¶
Fy 1 2µ3
= µ1 (R)R 2
− R µ3 +1 dR
(z1 − z2 ) Rinn R
Z Rout µ ¶
1 2µ3
= (µ1 )F y−str
exp R − R µ3 +1
dR. (8.12)
Rinn R2

Thus,
R Rout ³ 3µ3 +1 ´
1
Rinn
µ1 (R) R − R µ3 +1 dR
(µ1 )F y−str
exp = ³ ´ · 4µ3 +2 4µ3 +2
¸. (8.13)
Rout µ3 +1 µ3 +1 µ3 +1
ln Rinn − 4µ3 +2 Rout − Rinn

Similarly,
Z Rout ³ 2µ3 ´
Fz
= µ1 (R)R 1 − R µ3 +1 dR
y2 − y1 Rinn
Z ³
3µ3 +1 ´
Rout
= (µ1 )F z−str
exp R − R µ3 +1 dR,
Rinn
R Rout ³ 2µ3 ´

Rinn 1
µ (R)R 1 − R µ3 +1 dR
(µ1 )F z−str
exp = · 4µ3 +2 4µ3 +2
¸. (8.14)
2 − R2 ) − µ3 +1 R µ3 +1 − R µ3 +1
0.5(Rout inn 4µ3 +2 out inn

Instead, we could require that the total stored energy in the inhomogeneous
body and its homogeneous counterpart be the same and obtain that
R Rout h³ 2µ3 ´ 2µ3 i
−2
R
µ 1 (R) R µ3 +1
− 1 /µ3 + R µ3 +1
+ R − 2 RdR
(µ1 )cyl
inn
mth = · 4µ3 +2 4µ3 +2
¸ ³ ´ .
(µ3 +1)2 µ3 +1 µ3 +1 Rout 2µ3 +1 2 − R2 ]
µ3 (4µ3 +2) R out − R inn + ln Rinn − 2µ3 [Rout inn

(8.15)
The variation of (µ1 )exp with Rinn when µ1 (R) is given by (4.7) is described in
Figure 18. As Rinn tends to Rout , (µ1 )exp instead of tending to (µ1 )mean tends to
some other constant value, which is in contrast with the above results and with the
results presented in Saravanan and Rajagopal [22]. The variation of (µ1 )cyl mth with
Rinn when µ1 (R) is given by (4.7) and is plotted in Figure 19. (µ1 )cylmth also tends
to some constant value other than the mean value as Rinn tends to Rout .
The variation of (µ1 )exp with c when µ1 (R) varies as given by (4.7) is shown
in Figure 20. The variation of (µ1 )cyl
mth with c for the same variation in µ1 (R) as
above is depicted in Figure 21. Increasing the values of c does not result in (µ1 )exp or
70 Mathematical Proceedings of the Royal Irish Academy

(µ1 )cyl
mth tending to some constant value. With increasing values of c, the magnitude

1.8 x = 2, λ = 1, β = 1, κ = 0, µ = 0.5
1.45 2 3

[δ = 0.9, k = 2]
1.7 [δ = 0.1, k = 2] 1.4
[δ = 0.9, k = 10]
1.6 [δ = 0.1, k = 10] 1.35

1.3
1.5
Fy−str

(µ )Fz−str
1.25 [δ = 0.9, k = 2]
(µ1)exp

1 exp
1.4 [δ = 0.1, k = 2]
x2 = 2, λ = 1, β = 1, κ = 0, µ3 = 0.5 1.2 [δ = 0.9, k = 10]
1.3 [δ = 0.1, k = 10]
1.15
1.2
1.1

1.1 1.05

1 1
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Rinn/Rout R /R (b)
(a) inn out

Fig. 18—Plot of (a) (µ1 )F y−str


exp and (b) (µ1 )F z−str
exp vs. Rinn when the variation of µ1 (R)
is given by (4.7) and µ3 = 0.5 for c = 0.

1.9
x2 = 2, λ = 1, β = 1, κ = 0, µ3 = 0.5
1.8

1.7

1.6
[δ = 0.9, k = 2]
1.5 [δ = 0.1, k = 2]
(µ1)mth
cyl

[δ = 0.9, k = 10]
1.4 [δ = 0.1, k = 10]

1.3

1.2

1.1

1
0 0.2 0.4 0.6 0.8 1
Rinn/Rout

Fig. 19—Plot of (µ1 )cyl


mth vs. Rinn when the variation of µ1 (R) is given by (4.7) and µ3
= 0.5 for c = 0.
1.5 1.5
Rinn = 0.5, λ = 1, β = 1, κ = 0
1.45
1.4
1.4 [δ = 0.9, k = 2]
[δ = 0.1, k = 2]
1.35 1.3 [δ = 0.9, k = 10]
R = 0.5, λ = 1, β = 1, κ = 0 [δ = 0.1, k = 10]
inn
1.3
1.2
Fz−str
(µ )Fy−str

(µ1)exp
1 exp

1.25 [δ = 0.9, k = 2]
[δ = 0.1, k = 2]
1.1
1.2 [δ = 0.9, k = 10]
[δ = 0.1, k = 10]
1.15 1

1.1
0.9
1.05

1 0.8
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
c (a) c (b)

F y−str
Fig. 20—Plot of (a) (µ1 )exp and (b) (µ1 )F z−str
exp vs. c when the variation of µ1 (R) is
given by (4.7) and µ3 = 0.5.
Saravanan and Rajagopal—Deformation of inhomogeneous solids 71
4.5
[δ = 0.9, k = 2]
[δ = 0.1, k = 2]
4 [δ = 0.9, k = 10]
[δ = 0.1, k = 10]

3.5

Rinn = 0.5, λ = 1, β = 1, κ = 0
3

(µ1)mth
cyl 2.5

1.5

1
0 0.1 0.2 0.3 0.4 0.5
c

Fig. 21—Plot of (µ1 )cyl


mth vs. c when the variation of µ1 (R) is given by (4.7) and µ3 = 0.5.

of the difference between (µ1 )exp and (µ1 )cyl


mth increases. The above results show that
if the inhomogeneous Blatz–Ko body is made up of layers of homogeneous Blatz–Ko
bodies with different material moduli (µ1 ), varying about a mean, then it can not
be modelled by a homogeneous Blatz–Ko body.

Acknowledgement
We thank National Science Foundation and National Institutes for Health for sup-
port received to undertake this work.

References
[1] M. Aron, Combined axial shearing, extension and straightening of elastic annular cylindrical
sectors, IMA Journal of Applied Mathematics 70 (2005), 53–63.
[2] P.J. Blatz and W.L. Ko, Application of finite elastic theory to the deformation of rubbery
materials, Transactions of the Society of Rheology VI (1962), 223–51.
[3] P. Boulanger and M. Hayes, Finite amplitude motions in some nonlinear elastic media, Pro-
ceedings of the Royal Irish Academy, Section A-Mathematical and Physical Sciences 89
(1989), 135–46.
[4] P. Boulanger, M. Hayes and C. Trimarco, Finite amplitude plane waves in deformed hadamard
elastic materials, Geophysical Journal International 118 (1994), 447–58.
[5] M.M. Carroll, Controllable deformations of incompressible simple materials, Journal of Elas-
ticity 5 (1967), 515–25.
[6] M.M. Carroll, Finite strain solutions in compressible isotropic elasticity, Journal of Elasticity
20 (1988), 65–92.
[7] M.M. Carroll, Some universal deformations for a class of compressible elastic solids, Interna-
tional Journal of Non-linear Mechanics 36 (2001), 443–6.
[8] M.M. Carroll and C.O. Horgan, Finite strain solutions for a compressible elastic solid, Quar-
terly of Applied Mathematics 48 (1990), 767–80.
[9] M.M. Carroll, J.G. Murphy, and F.J. Rooney, Plane stress problems for compressible mate-
rials, International Journal for Solids and Structures 31 (1994), 1597–607.
[10] D.T. Chung, C.O. Horgan and R. Abeyratane, The finite deformation of internally pressurized
hollow cylinders and spheres for a class of compressible elastic materials, International
Journal of Solids and Structures 22 (1986), 1557–70.
[11] M. Destrade and M. Hayes, Circularly polarized plane waves in a deformed Hadamard mate-
rial, Wave Motion 35 (2002), 289–309.
72 Mathematical Proceedings of the Royal Irish Academy

[12] M. Destrade and M. Hayes, Inhomogeneous ‘longitudinal’ plane waves in a deformed elastic
material, Journal of Elaticity 75 (2004), 147–65.
[13] J.L. Ericksen, Deformations possible in every compressible isotropic perfectly elastic material,
Journal of Mathematics and Physics 34 (1955), 126–8.
[14] R.L. Fosdick, Dynamically possible motions of incompressible isotropic simple materials,
Archive for Rational Mechanics and Analysis 29 (1968), 272–88.
[15] R. Hill, On constitutive macro-variables for heterogeneous solids at finite strain, Proceedings
of the Royal Society of London, series A 326 (1972), 131–47.
[16] C.O. Horgan, Remarks on ellipticity for the generalized blatz-ko constitutive model for a
compressible nonlinearly elastic solid, Journal of Elasticity 42 (1996), 165–75.
[17] E. Kirkinis and R.W. Ogden, On extension and torsion of a compressible elastic circular
cylinder, Mathematics and Mechanics of Solids 7 (2002), 373–92.
[18] E. Kirkinis, R.W. Ogden, and D.M. Haughton, Some solutions for a compressible isotropic
elastic material, Zeitschrift fur Angewandte Mathematik und Physik ZAMP 55 (2004),
136–58.
[19] O. Lopez-Pamies and P.P. Castaneda, Second order homogenization estimates incorporating
field fluctuations in finite elasticity, Mathematics and Mechanics of Solids 9 (2004),
243–70.
[20] U. Saravanan and K.R. Rajagopal, A comparison of the response of isotropic inhomogeneous
elastic cylindrical and spherical shells and their homogenized counterparts, Journal of
Elasticity 71 (2003), 205–33.
[21] U. Saravanan and K.R. Rajagopal, On the role of homogeneties in the deformation of elastic
bodies, Mathematics and Mechanics of Solids 8 (2003), 349–76.
[22] U. Saravanan and K.R. Rajagopal, Inflation, extension, torsion and shearing of compressible,
inhomogeneous annular cylinder, Mathematics and Mechanics of Solids 10 (2005), 603–
50.
[23] Y. Wang, M. Aron and C. Christopher, On the straightening of compressible, nonlinearly
elastic annular cylindrical sectors, Mathematics and Mechanics of Solids 3 (1998), 131–
45.
[24] P.K. Currie and M. Hayes, On non-universal elastic deformations, In D.E. Carlson and R.T.
Shield (eds), Proceedings of the IUTAM symposium on finite elasticity, 1981, pp 143–50.
[25] C. Truesdell and W. Noll, The nonlinear field theories, Handbuch der Physik, vol. III/3.
Springer–Verlag, Berlin, 1965.
[26] A.S. Wineman, Universal deformations of incompressible simple materials, University of
Michigan Technical Report, Ann Arbor, 1967.

You might also like