Chapter 2

General Anisotropic Elasticity

Abstract This Chapter is an introduction to general anisotropic elasticity, i.e. to the

elasticity of 3D anisotropic bodies. The main classical topics of the matter are treated
in detail: starting from the Hooke’s law for anisotropic bodies, the two principal
notations of Voigt and Kelvin are introduced and the reasons for the use of the
last one are argued. Then, after an explanation of the mechanical meaning of the
anisotropic elastic constants, the key topic of elastic symmetries is treated in detail.
The technical elastic constants are then introduced as well as the elastic bounds for the
components of the elastic tensor and for the technical constants. After an observation
on the decomposition of the strain energy for anisotropic bodies, the Chapter ends
with the determination of the symmetry planes, the curvilinear anisotropy and some
examples of typical anisotropic materials.

2.1 The Hooke’s Law for Anisotropic Bodies

Be Ω a body acted upon by body forces f and by surface tractions t on its frontier
∂Ω whose outward unit normal is n. We consider a small arbitrary variation δu of
the displacement field on Ω, compatible with the given boundary conditions and
satisfying the kinematical conditions (1.25). The total mechanical work dW done by
the applied forces can be transformed as follows
   
dW = f · δu dω + t · δu ds = f · δu dω + σ n · δu ds =
Ω ∂Ω Ω
 ∂Ω

f · δu dω + σ δu · n ds = [f · δu + div(σ δu)] dω = (2.1)

Ω ∂Ω Ω

[(f + divσ ) · δu + σ · ∇δu] dω = σ · δε dω
Ω Ω

© Springer Nature Singapore Pte Ltd. 2018 19

P. Vannucci, Anisotropic Elasticity, Lecture Notes in Applied and Computational
Mechanics 85, DOI 10.1007/978-981-10-5439-6_2
20 2 General Anisotropic Elasticity

In establishing Eq. (2.1) we have used successively the Cauchy’s stress theorem
(1.26), the symmetry of σ , the Gauss theorem, two standard results for tensor and
vector fields1 and the motion equation (1.27).
The quantity
δV = σ · δε = σi j δεi j , (2.2)

represents the variation of the internal energy of the body per unit of volume produced
by a small variation of the strain state.
Following the energetic approach of Green (1839), we define as elastic a body
for which the total variation V of the internal energy due to a finite transformation
from a state A to a state B is independent from the integration path. In particular,
V must then be null for any transformation where A=B:
 B
V = δV = VB − V A . (2.3)
A

Hence, for an elastic body δV must be the exact differential d V of a scalar function
V (ε), the strain energy density or elastic potential2 :

∂V
V = V (ε) : d V = dεi j . (2.4)
∂εi j

In such a case, Eq. (2.2) must be rewritten as

d V = σ · dε = σi j dεi j , (2.5)

so that we get the Green’s formula

∂V
σi j = . (2.6)
∂εi j

1 Namely, we have used the identity

div(L v) = divL · v + L · ∇v,

with L a second-rank tensor field and v a vector field, see (Gurtin 1981, p. 30), and the fact that,
∀L : L = L ,

∇v + ∇  v 1 1
L· = (L · ∇v + L · ∇v ) = (L · ∇v + L · ∇v ) = L · ∇v,
2 2 2
because of the property of tensor scalar product:

A · B = Ai j Bi j = A · B .

2. Theexistence of such a function can be established upon physical arguments, using the first law
of thermodynamics for adiabatic transformations or the second law for the isothermal ones, (Love
1944, p. 94).
2.1 The Hooke’s Law for Anisotropic Bodies 21

We now postulate that in the initial state, i.e. when the body is not acted upon by
forces, ε = O and σ = O, i.e. the body is unstrained and unstressed in its initial
state. Then, developing V (ε) in a Taylor series about ε = O we get
 
∂ V  1 ∂ 2 V 
V (ε) = V (ε = O) + εi j + εi j εkl + . . . (2.7)
∂εi j ε=O 2 ∂εi j ∂εkl ε=O

Choosing arbitrarily V (ε = O) = 0, always possible for a potential, and limiting

the development to the first non null term, which is correct for small strains, gives

1 ∂ 2 V 
V = εi j εkl ; (2.8)
2 ∂εi j ∂εkl ε=O

the second derivatives in the above equation are linear operators depending upon
four indexes; they are the components of a fourth-rank tensor E,

∂ 2 V 
E i jkl := , (2.9)
∂εi j ∂εkl ε=O

the (stiffness) elasticity tensor, so that

1 1
V = E i jkl εi j εkl = ε · Eε. (2.10)
2 2
Collecting all the parameters describing the elastic response of the material, E is
the operator that describes the elastic response of the continuum. It has 81 Cartesian
components, the elastic moduli E i jkl . Nevertheless, the number of independent elastic
moduli is far less than 81. In fact, first of all, by the Schwarz theorem we get

∂2V ∂2V
E i jkl = = = E kli j ; (2.11)
∂εkl ∂εi j ∂εi j ∂εkl

the above 15 relations are known as major symmetries and reduce the number of
independent Cartesian components of E from 81 to 66.
Now, if we apply the Green’s formula (2.6) to Eq. (2.10) we get

σi j = E i jkl εkl → σ = Eε. (2.12)

This is the Hooke’s law (1660), establishing a linear relation between stress and
strain.3 This linearity is a direct consequence of the quadratic structure of V and of
the Green’s formula; though initially formulated for isotropic bodies, it is the basic

3 An alternative, classical, approach to elasticity is to postulate the Hooke’s law and the existence of

V ; once obtained the Green’s formula, using the Schwarz theorem gives again the major symmetries,
while the minor ones are still given by the symmetry of ε and σ . Then, the expression of V is obtained
integrating d V :
22 2 General Anisotropic Elasticity

law of elasticity also for the more general case of anisotropic continua. Using now
the Hooke’s law, we obtain another reduction of the number of independent elastic
moduli due to the symmetry of σ and ε:

σi j = σ ji and εi j = ε ji ⇒ E i jkl = E jikl = E i jlk = E jilk ∀i, j, k, l ∈ {1, 2, 3}.

(2.13)

The above 45 relations among the components of E are called the minor symmetries
and reduce the number of independent elastic moduli to only 21. This is the highest
number of independent moduli that an elastic material can have. In such a case, the
material is completely anisotropic or triclinic. Further reductions of the number of
independent moduli can be obtained only if special conditions, not universal but
depending upon the material type, are introduced. Such conditions are called elastic
symmetries, and indicate the invariance of some elastic moduli under some geometric
transformations. Injecting the Hooke’s law into the expression (2.10) of V we get
also
1
V = σ · ε. (2.14)
2
Let us now consider the inverse of the Hooke’s law:

ε = Zσ, Z = E−1 , (2.15)

with Z the compliance elasticity tensor; introducing this last equation for ε into
(2.10) gives
1
V = σ · Zσ , (2.16)
2
an expression called stress energy density in the literature.
A last remark: in this section, the word symmetry has been used for denoting the
equivalence of the positions of an index for two or more components of the elasticity
tensor; to make the distinction with the concept of elastic symmetry, the expression
tensor or index symmetry could be used. Anyway, the reader should be aware of the
fact that the same word, symmetry, can have two rather different meanings in our
context.

2.1.1 The Voigt’s Notation

The general, tensorial, expression of the Hooke’s law needs the use of quantities with
four indexes,

(Footnote 3 continued)
1 1
d V = σ · dε = Eε · dε = E i jkl εkl dεi j → V = E i jkl εi j εkl = ε · Eε.
2 2
2.1 The Hooke’s Law for Anisotropic Bodies 23

σi j = E i jkl εkl , (2.17)

which can be somewhat cumbersome and heavy. That is why some simplified nota-
tions have been proposed. In particular, they allow for a matrix representation of
(2.17); these formalisms switch the algebra from that of a fourth-rank tensor to that
of a 6×6 square symmetric matrix.
The most well known of the matrix formalisms for anisotropic elasticity is that of
Voigt (1910): the stress and strain tensors are written as follows:
⎧ ⎫ ⎧ ⎫
⎪
⎪ σ1 = σ11 ⎪
⎪ ⎪
⎪ ε1 = ε11 ⎪
⎪
⎪
⎪ ⎪ ⎪ ⎪
⎪
⎪ σ2 = σ22 ⎪
⎪
⎪
⎪
⎪
⎪ ε2 = ε22 ⎪
⎪
⎪
⎨ ⎬ ⎨ ⎬
σ3 = σ33 ε3 = ε33
{σ } = , {ε} = . (2.18)
⎪
⎪ σ4 = σ23 ⎪
⎪ ⎪
⎪ ε4 = 2ε23 ⎪
⎪
⎪
⎪ ⎪ ⎪ ⎪
⎪
⎪ σ5 = σ31 ⎪
⎪
⎪
⎪
⎪
⎪ ε5 = 2ε31 ⎪
⎪
⎪
⎩ ⎭ ⎩ ⎭
σ6 = σ12 ε6 = 2ε12

Equation (2.18) shows the relations and order established for the indexes by the
Voigt’s notation:

11 → 1, 22 → 2, 33 → 3, 23 → 4, 31 → 5, 12 → 6. (2.19)

The introduction of the coefficient 2 for the terms ε4 , ε5 and ε6 is needed for taking
into account for the symmetry of σ and ε in the Hooke’s law. This fact imposes some
accuracy in the use of the Voigt’s notation, because the algebras for tensors σ and ε
are not completely the same, namely for their transformation upon axes rotation and
tensor inversion.
The Voigt’s notation transforms hence second rank symmetric tensors into column
vectors; correspondingly, the fourth-rank elasticity tensor is transformed into a 6×6
symmetric square matrix, the symmetry of such a matrix corresponding to the major
symmetries of E. According to the index transformation rule (2.19), the matrix form
of the Hooke’s law with the Voigt’s notation is
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ σ1 ⎪
⎪ C11 C12 C13 C14 C15 C16 ⎪ ⎪ ε1 ⎪⎪
⎪
⎪ ⎪
⎪ ⎢ ⎪ ⎪
⎪
⎪ σ2⎪ C C22 C23 C24 C25 C26 ⎥
⎥⎪
⎪ ε2⎪
⎪
⎨ ⎪ ⎬ ⎢ ⎪ ⎪
12
⎢ C13 ⎨ ⎬
σ3 ⎢ C23 C33 C34 C35 C36 ⎥
⎥ ε
{σ } = [C] {ε} → = 3
. (2.20)
⎪ ⎪ ⎢
⎪ σ4 ⎪ ⎢ C14 C24 C34 C44 C45 C46 ⎥
⎥⎪⎪ ε4 ⎪⎪
⎪
⎪ ⎪
⎪ ⎪ ⎪
⎪ σ
⎪ 5⎪ ⎪ ⎣ C15 C25 C35 C45 C55 C56 ⎦ ⎪
⎪ ε5 ⎪
⎪ ⎪
⎩ ⎭ ⎩ ⎪ ⎭
σ6 C16 C26 C36 C46 C56 C66 ε6

The name [C] is usually preferred to E to make a clear distinction between the
tensor and matrix representation (for the same reason, we will name differently the
compliance tensor and matrix). Thanks to the introduction of coefficients 2 in (2.18)2 ,
there is a perfect coincidence between the E i jkl and the C pq ; it is sufficient to remind
rule (2.19) to make correspond to each pair of indexes i j and kl in E i jkl the correct
p and q in C pq ; for instance, E 2312 = C46 , E 1322 = C52 and so on.
24 2 General Anisotropic Elasticity

Let us now consider the inverse of the Hooke’s law (2.15), that we will write in
the Voigt’s notation as
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ ε1 ⎪ S11 S12 S13 S14 S15 S16 ⎪ σ1 ⎪
⎪ ⎪
⎪ ⎪
⎪ ⎢ S12
⎪
⎪ ⎪
⎪
⎪
⎪ ε ⎪ S22 S23 S24 S25 S26 ⎥ ⎪
⎪ σ2⎪
⎪
⎨ ⎪ ⎬ ⎢ ⎥⎪⎨ ⎪
2
⎢ S13 ⎥ ⎬
ε3 S23 S33 S34 S35 S36 ⎥ σ3
{ε} = [S] {σ } → =⎢
⎢ S14 ⎥ . (2.21)
⎪
⎪ ε4 ⎪
⎪ ⎢ S24 S34 S44 S45 S46 ⎥ ⎪
⎪ σ4 ⎪
⎪
⎪
⎪ ⎪
⎪ ⎣ S15 ⎪
⎪ ⎪
⎪
⎪
⎪ ε5 ⎪
⎪ S25 S35 S45 S55 S56 ⎦ ⎪
⎪ σ5 ⎪
⎩ ⎭ ⎩ ⎪ ⎭
ε6 S16 S26 S36 S46 S56 S66 σ6

Unlike the case of the stiffness matrix [C], not all the components of [S] are equal
to the corresponding ones of Z. This is a consequence of the introduction of the
factors 2 in (2.18)2 and the correct transformation is
 
  Z ppqq 2Z ppr s
Si j = →
2Z ppr s 4Z pqr s
⎡ ⎤
S11 = Z 1111 S12 = Z 1122 S13 = Z 1133 S14 = 2Z 1123 S15 = 2Z 1131 S16 = 2Z 1112
⎢ S22 = Z 2222 S23 = Z 2233 S24 = 2Z 2223 S25 = 2Z 2231 S26 = 2Z 2212 ⎥
⎢ ⎥
⎢ S33 = Z 3333 S34 = 2Z 3323 S35 = 2Z 3331 S36 = 2Z 3312 ⎥
⎢ ⎥.
⎢ S44 = 4Z 2323 S45 = 4Z 2331 S46 = 4Z 2312 ⎥
⎢ ⎥
⎣ sym S55 = 4Z 3131 S56 = 4Z 3112 ⎦
S66 = 4Z 1212
(2.22)

The above equations show that passing to the Voigt’s notation implies a different
algebra for stiffness and compliance, as anticipated above. Mathematically, matrices
[C] and [S] do not represent some second order tensors in R6 . Practically, the use of
the Voigt’s notation needs some carefulness, not only in the differences between [C]
and [S], but also in the transformation of these matrices produced by axes rotation.

2.1.2 The Kelvin’s Notation

The Kelvin’s notation (by somebody named Mandel’s notation) was proposed by
W. Thomson, baron of Kelvin, as early as 1856 (Kelvin 1856, 1878), but, probably
because making use of irrational quantities, it has not known an as large success as
the Voigt’s notation. Nevertheless, rather recently a new attention has been brought
by scientists on it, mainly for its algebraic properties: the Kelvin’s notation has not
the drawbacks of the Voigt’s one, as it will be shown below.
The Kelvin’s notation is different from the Voigt’s one in that the coefficients
2 affecting ε, Eq. (2.18), are equally distributed over σ and ε, in such a way their
product still amounts to 2:
2.1 The Hooke’s Law for Anisotropic Bodies 25
⎧ ⎫ ⎧ ⎫
⎪
⎪ σ1 = σ11 ⎪⎪ ⎪
⎪ ε1 = ε11 ⎪
⎪
⎪
⎪ σ2 = σ22 ⎪⎪ ⎪
⎪ ε2 = ε22 ⎪
⎪
⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪
⎨ σ =σ ⎬ ⎨ ε =ε ⎬
3 √ 33 3 √ 33
{σ } = , {ε} = . (2.23)
⎪
⎪ σ4 = √2σ23 ⎪
⎪ ⎪
⎪ ε4 = √2ε23 ⎪
⎪
⎪
⎪ ⎪
⎪ ⎪
⎪ ⎪
⎪
⎪
⎪ σ = 2σ ⎪ ⎪ ε = 2ε ⎪
⎩ 5 √ 31 ⎪ ⎭ ⎪
⎩ 5 √ 31 ⎪ ⎭
σ6 = 2σ12 ε6 = 2ε12

In this way, Eqs. (2.20) and (2.21) still hold but there is no difference between σ and
ε in the transition from the tensor to the matrix representation; in particular, the way
the components of [C] and [S] are deduced from the corresponding ones of E and
Z, are exactly the same (no summation over dummy indexes):
 √   √ 
  E ppqq 2E ppr s   Z ppqq 2Z ppr s
Ci j = √ , Si j = √ . (2.24)
2E ppr s 2E pqr s 2Z ppr s 2Z pqr s

The above symbolic relations can be easily translated in the detailed expressions of
Ci j and Si j , applying a scheme quite similar to that detailed in Eq. (2.22).
Merhabadi and Cowin (1990), have shown that the Kelvin’s notation gives a
representation of elasticity by matrices, [C] and [S], representing second-rank sym-
metric tensors in R6 , which is not the case with the Voigt’s notation. Hence, the
Kelvin’s notation transfers the algebra of elasticity from fourth-rank tensors in R3 ,
to second-rank tensors in R6 . This fact has some advantages, for instance the rotation
of matrices [C] and [S] is made in the same way, unlike with the Voigt’s notation.
For these reasons, the Kelvin’s notation is preferred in this text.

2.1.3 The Mechanical Meaning of the Anisotropic Elastic

Constants

In the most general case of a triclinic material, the number of independent elastic
moduli is as great as 21; it is important to understand the mechanical meaning of
these parameters, because, unlike in the case of an isotropic material, some unusual,
strange mechanical effects can arise in anisotropy. To discover these effects and
connect them to particular elastic parameters, it is worth to use the compliances, i.e.
the components of [S] or Z.
Let us consider a cube of a triclinic material, submitted to the only traction σ1 , see
Fig. 2.1. In such a case, Eq. (2.21) gives εk = Sk1 σ1 ∀k = 1, . . . , 6 (or, equivalently,
εi j = Z i j11 σ11 ∀i, j = 1, 2, 3).
So, while in an isotropic solid only the terms ε1 , corresponding to the direct
stretching effect, and ε2 = ε3 , corresponding to the Poisson’s effect, i.e. the defor-
mation in a plane orthogonal to the direction of the normal stress, are not null for a
uniaxial traction, in a completely anisotropic body all the components of ε are not
null: a normal stress produces also shearing strains. The coupling effects are hence
26 2 General Anisotropic Elasticity

Fig. 2.1 Anisotropic

stretched cube

not restricted to the only Poisson’s effect, due to the terms Si j , i, j = 1, 2, 3, i = j:

in the anisotropic case, there is also a coupling between normal stress and shear
strain, due to terms Skl , k = 4, 5, 6, l = 1, 2, 3. In addition, generally speaking
S12 = S23 = S31 , so that the Poisson’s effect is different in the orthogonal directions,
i.e. ε2 = ε3 . In the same way, usually S4k = S5k = S6k , k = 1, 2, 3, so that also for
the shearing stresses it is ε4 = ε5 = ε6 . Finally, the anisotropic cube does not only
change its volume under the unique action of a traction, like in isotropic bodies, but
it changes also its form: it becomes a prism with no orthogonal faces.
Let us now consider the same cube submitted to a unique shear stress, say
σ5 ; Eq. (2.21) gives then εk = Sk5 σ5 ∀k = 1, . . . , 6 (or, equivalently, εi j =
Z i j31 σ31 ∀i, j = 1, 2, 3). This time, we can observe a coupling between shear
stresses and extension strains, due to the terms Slk , k = 4, 5, 6, l = 1, 2, 3 and also
a coupling between a shear stress and the shearing strains in orthogonal planes, due to
the terms Si j , i, j = 4, 5, 6, i = j. This last effect is called the Chentsov’s effect: it is
completely analogous to the Poisson’s effect, but it concerns shear stresses and strains
in the place of tractions and extensions. Also in this case, the couplings shear stress-
extensions and the Chentsov’s effect are not necessarily the same in all the planes,
because generally speaking Sl4 = Sl5 = Sl6 , l = 1, 2, 3 and S45 = S56 = S64 . It
is then apparent that, submitted to simple shear stress, the cube changes not only its
shape, but also its volume, unlike in the case of isotropic bodies.
Finally, the compliance matrix can be subdivided into parts in charge of a particular
effect, like in Fig. 2.2. It is immediately recognized that a similar partition is possible
also for the stiffness matrix [C].

direct effect of normal stresses

1 S11 S12 S13 S14 S15 S16 1 couplings extension strains-

2 S12 S22 S23 S24 S25 S26 2
shear stresses

3 S13 S23 S33 S34 S35 S36 3

= .
4 S14 S24 S34 S44 S45 S46 4
5 S15 S25 S35 S45 S55 S56 5
6 S16 S26 S36 S46 S56 S66 6

direct effect of shear stresses

Fig. 2.2 Partition of the compliance matrix by mechanical effects

2.2 Elastic Symmetries 27

2.2 Elastic Symmetries

2.2.1 Taking into Account for Elastic Symmetries

Recalling what said in Sect. 1.1, when some kind of symmetry in the behavior is
present, then some equivalent directions exist, whereon the behavior is the same.
The figure formed with these directions is a symmetrical figure allowing all the
covering operations of a certain group.
Just because the behavior is the same along equivalent directions, the forms of the
elasticity matrix [C]4 and of the strain energy are the same in two frames related by
a covering operations. This gives some relations among the components of [C], e.g.
some of them are null.5 Let us sketch the procedure for obtaining such relations:
• the expressions of the strain energy in two frames R and R
related by a covering
operation are6
1 1
V = {ε} [C]{ε}, V
= {ε
} [C]{ε
}; (2.25)
2 2

• the strain tensor {ε

} can be written in the frame R:

{ε
} = [R]{ε}, (2.26)

with [R] the orthogonal matrix corresponding to the covering operation, i.e. to the
symmetry of the material;
• injecting Eq. (2.26) into V
, Eq. (2.25), and putting V = V
, gives the equation

{ε} [C]{ε} = ([R]{ε}) [C][R]{ε} ∀{ε}; (2.27)

• this unique scalar equation gives all the relations that must be true for the compo-
nents of [C] exactly because it is independent from the strain state, i.e. because it
is true ∀{ε}.
Now, the question is: which is the orthogonal matrix [R] corresponding to a given
covering operation? This will be the matter of the following Sections.
Before, just a last remark: the procedure sketched above is not the only one; in
fact, in place of working with the strain energy, one could directly state that [C],

4 The use of [C] is here preferred to that of E because it facilitates calculations; of course, the results

found for components Ci j are immediately valid also for the E pqr s , see Eqs. (2.22) and (2.24).
5 The same is true for the stress energy; in such a case one can obtain relations among the components

of [S] that are exactly the same ones found for [C].
6 We denote here by a prime a component in B
or also, for the sake of shortness though with a
slight abuse of notation, a vector or tensor whose components are intended to be given in B
.
B = {e1 , e2 , e3 } and B
= {e1
, e2
, e3
} are two orthonormal bases of the vector space of translations
V associated with the ordinary euclidean space E , and they are associated with the frames R =
{o; B } and R
= {o
; B
} respectively, o, o
∈ E .
28 2 General Anisotropic Elasticity

or [S], does not change when passing from R to R

. This approach is practically
equivalent to the previous one, but it gives six scalar equations. For instance, for [C]
we have:

{σ } = [C]{ε}, and {σ
} = [C]{ε
} → [R]{σ } = [C][R]{ε} →
(2.28)
{σ } = [R] [C][R]{ε} ⇒ [C] = [R] [C][R].

2.2.2 Rotation of Axes

Let us consider two orthonormal bases B = {e1 , e2 , e3 } and B

= {e1
, e2
, e3
} and
let us suppose that these two bases are related by the orthogonal tensor U.7
We define U as the tensor such that

ei = Uei
⇒ ei
= U ei ; (2.29)

with this definition, it is easy to show that

⎡ ⎤
e1

U = ⎣ e2
⎦ , (2.30)
e3

i.e. the matrix representing U in the basis B

has for rows the Cartesian components
of the vectors of B
, components expressed in the base B. Algebraically, these
components are the director cosines of the angles between two corresponding axes
in B and B
.
Using the above equations, the components in B
of a tensor of any rank r can
be expressed as a linear combination of its components in B, the coefficients of the
combination being products of r components of U. In fact, considering that

ei = Uei
= U pq (e
p ⊗ eq
)ei
= U pq δqi e
p = U pi e
p , (2.31)

then, for a vector (r = 1) it is

w = wi ei = wi Uki ek
⇒ wk
= Uki wi , (2.32)

for a second-rank tensor (r = 2) it is

L = L i j ei ⊗ e j = L i j Umi em
⊗ Un j en
= Umi Un j L i j em
⊗ en
⇒
(2.33)
L
mn = Umi Un j L i j

7U is not necessarily a proper rotation, because reflections are possible too.

2.2 Elastic Symmetries 29

and finally for a fourth-rank tensor (r = 4) it is8

E = E i jkl ei ⊗ e j ⊗ ek ⊗ el = E i jkl Umi em

⊗ Un j en
⊗ U pk e
p ⊗ Uql eq
=
Umi Un j U pk Uql E i jkl em
⊗ en
⊗ e
p ⊗ eq
⇒ E mnpq

= Umi Un j U pk Uql E i jkl .

(2.34)
Given A, B ∈ Lin, the conjugation product A  B is the fourth-rank tensor defined
by the operation
(A  B)C := ACB ∀C ∈ Lin. (2.35)

It is worth to remark that Eq. (2.35) implies that for the vectors of a basis it is

(ei ⊗ e j )  (ek ⊗ el ) = ei ⊗ ek ⊗ e j ⊗ el , (2.36)

which gives
(A  B)i jkl = Aik B jl . (2.37)

Once defined the transpose A of a fourth-rank tensor A as the unique tensor such
that
L · (AM) = M · (A L) ∀L, M ∈ Lin, (2.38)

it is immediate to show that

(A ⊗ B) = B ⊗ A,
(2.39)
(A  B) = A  B .

Like for tensors in Lin, also a tensor A ∈ Lin is said to be symmetric ⇐⇒ A = A .

It is simple to check that

A = A ⇒ Aijkl = Akli j , (2.40)

i.e., the major symmetries of the elastic tensors E and Z actually coincide with the
definition of symmetric tensor in Lin.
For an orthogonal second-rank tensor U, we define its orthogonal conjugator U
as
U := U  U; (2.41)

it is not difficult to show that just as U preserves scalar products of elements in V ,

its associated orthogonal conjugator U preserves scalar products in Lin:

UA · UB = A · B ∀A, B ∈ Lin. (2.42)

8 ∀ A, B and L ∈Lin, A ⊗ B is the fourth-rank tensor defined by the operation (A ⊗ B)L := (B · L)A.

Applying this rule to the dyads of a basis, we get a fundamental result: (ei ⊗e j ⊗ek ⊗el )(e p ⊗eq ) =
(ek ⊗ el ) · (e p ⊗ eq )(ei ⊗ e j ) = δkp δlq (ei ⊗ e j ).
30 2 General Anisotropic Elasticity

In other words, U is an orthogonal tensor in Lin (Podio-Guidugli 2000, p. 55).

Introducing the identity of Lin,

I = I  I ⇒ I = Ii jkl (ei ⊗ e j ⊗ ek ⊗ el ) = δik δ jl (ei ⊗ e j ⊗ ek ⊗ el ), (2.43)

it is easy to recognize that also for rotations in Lin

UU = U U = I. (2.44)

Be n ∈ V a unit vector and let us suppose that n is orthogonal to a symmetry

plane. Then
U := I − 2n ⊗ n, U = U , U = U  U = U , (2.45)

is the orthogonal tensor describing the symmetry in the plane whose normal is n. In
fact,
Un = −n
(2.46)
Um = m ∀m ∈ V : m · n = 0, |m| = 1.

Thanks to these last definitions, it is possible to give a compact form to results

(2.32), (2.33) and (2.34):

w
= Uw,
L
= ULU = (U  U)L = UL, (2.47)
E
= (U  U)E(U  U) = UEU .

Using Eq. (2.23) and the result of Eq. (2.37), we can now obtain the matrix [R]
that corresponds, in the Kelvin’s notation, to tensor U; the calculations are rather
tedious and a little bit long, but the final result is (Mehrabadi and Cowin 1990),
⎡ 2
U11 2
U12 2
U13
√
2U12 U13
√
2U13 U11
√
2U11 U12
⎤
√ √ √
⎢ U212 2
U22 2
U23 2U22 U23 2U23 U21 2U21 U22 ⎥
⎢ √ √ √ ⎥
⎢ U312 2
U32 2
U33 2U32 U33 2U33 U31 2U31 U32 ⎥
[R] = ⎢ √ √ √ ⎥.
⎢ 2U21 U31 2U23 U33 U23 U32 + U22 U33 U33 U21 + U31 U23
2U22 U32 U31 U22 + U32 U21 ⎥
⎣ √2U U √
2U U
√
2U33 U13 U32 U13 + U33 U12 U31 U13 + U33 U11 U U +U U
⎦
31 11 32 12 31 12 32 11
√ √ √
2U11 U21 2U12 U22 2U13 U23 U12 U23 + U13 U22 U11 U23 + U13 U21 U11 U22 + U12 U21
(2.48)

The above matrix [R] allows for the change of basis of any second-rank tensor in the
Kelvin’s notation. In particular for σ and ε:

{σ
} = [R]{σ }, {ε
} = [R]{ε}. (2.49)

It can be checked that, when U is an orthogonal tensor, then

[R][R] = [R] [R] = [I ], (2.50)

2.2 Elastic Symmetries 31

i.e., [R] is an orthogonal matrix in R6 ; this is not the case with the Voigt’s notation.
Hence, [R] represents, in the given basis, an orthogonal tensor of Lin over a manifold
of dimension six.
It is impossible to put the result of Eq. (2.34) in matrix form, because also in the
Kelvin’s notation it depends upon four indexes; nevertheless, it is of course possible
to express all the components of such an operator, but actually in the most general
case these components have so extremely complicate and long expressions that it is
practically impossible to write down all of them, so they are omitted here.

2.2.3 A Tensorial Characterization of Elastic Symmetries

The results of the previous section give us the possibility of characterizing in an

elegant tensorial form the existence of elastic symmetries in a solid (Podio-Guidugli
2000, p. 56).
Let us suppose that a material has a given elastic symmetry and that the two
bases B and B
correspond to equivalent directions with respect to the symmetry of
concern. Physically, this means that it is not possible to detect the change from B to
B
by experiments measuring stresses, because the behavior is exactly the same in
the two cases: E = E
. Then, applying Eq. (2.47)2 to σ and ε,

σ
= Uσ , ε
= Uε, (2.51)

and the Hooke’s law, Eq. (2.12), we get, because E = E

,

σ
= Eε
→ Uσ = EUε → UEε = EUε ⇒ UE = EU. (2.52)

Hence, an orthogonal transformation U is in the elastic symmetry group of the mate-

rial if and only if E and U commute, U being the orthogonal conjugator of U.
The result of Eq. (2.52) constitutes also a way for determining the number and
type of independent elastic moduli, i.e. the distinct components of E; this is the way
sketched, with reference to matrix [C], in the last paragraph of Sect. 2.2.1, Eq. (2.28);
however, as said thereon, the energetic approach is preferred in this text.

2.2.4 Triclinic Bodies

A triclinic body has no material symmetries, so Eq. (2.27) cannot be written. As a

consequence, it is not possible to reduce the number of independent elastic compo-
nents, that remains fixed to 21. Matrix [C] appears hence as
32 2 General Anisotropic Elasticity
⎡ ⎤
C11 C12 C13 C14 C15 C16
⎢ C22 C23 C24 C25 C26 ⎥
⎢ ⎥
⎢ C33 C34 C35 C36 ⎥
[C] = ⎢
⎢
⎥. (2.53)
⎢ C44 C45 C46 ⎥
⎥
⎣ sym C55 C56 ⎦
C66

2.2.5 Monoclinic Bodies

The only symmetry of a monoclinic body is a reflection in a plane. Without loss in

generality, we can suppose x3 = 0 to be the symmetry plane. In such a case it is, see
Eqs. (2.45) and (2.48),
⎡ ⎤
1 0 0 0 0 0
⎡ ⎤ ⎢0 1 0 0 0 0⎥
1 0 0 ⎢ ⎥
⎢0 0 1 0 0 0⎥
⎣
U= 0 1 0 ⎦ ⇒ [R] = ⎢
⎢0
⎥, (2.54)
⎢ 0 0 −1 0 0 ⎥
⎥
0 0 −1 ⎣0 0 0 0 −1 0 ⎦
0 0 0 0 0 1

that applied to Eq. (2.27) gives the condition

C14 ε1 ε4 + C24 ε2 ε4 + C34 ε3 ε4 + C15 ε1 ε5 +

(2.55)
C25 ε2 ε5 + C35 ε3 ε5 + C46 ε4 ε6 + C56 ε5 ε6 = 0,

which is satisfied ∀ε ⇐⇒

C14 = C24 = C34 = C15 = C25 = C35 = C46 = C56 = 0. (2.56)

Hence, a monoclinic body depends upon only 13 distinct elastic moduli:

⎡ ⎤
C11 C12 C13 0 0 C16
⎢ C22 C23 0 0 C26 ⎥
⎢ ⎥
⎢ C 0 0 C36 ⎥
⎢
[C] = ⎢ 33 ⎥. (2.57)
⎢ C44 C45 0 ⎥ ⎥
⎣ sym C55 0 ⎦
C66
2.2 Elastic Symmetries 33

2.2.6 Orthotropic Bodies

Let us now add another plane of symmetry orthogonal to the previous one, say the
plane x2 = 0. Following the same procedure, we get successively:
⎡ ⎤
1 0 0 0 0 0
⎡ ⎤ ⎢0 1 0 0 0 0 ⎥
1 0 0 ⎢ ⎥
⎢0 0 1 0 0 0 ⎥
U= ⎣ 0 −1 0 ⎦ ⇒ [R] = ⎢
⎢0
⎥, (2.58)
⎢ 0 0 −1 0 0 ⎥
⎥
0 0 1 ⎣0 0 0 0 1 0 ⎦
0 0 0 0 0 −1

(C14 ε1 + C24 ε2 + C34 ε3 + C45 ε5 )ε4 +

(C16 ε1 + C26 ε2 + C36 ε3 + C56 ε5 )ε6 = 0 ∀ε ⇐⇒ (2.59)
C14 = C24 = C34 = C45 = C16 = C26 = C36 = C56 = 0.

So, the existence of the second plane of symmetry has added the four supplementary
conditions
C16 = C26 = C36 = C45 = 0 (2.60)

to the previous eight ones, reducing hence to only 9 the number of distinct elastic
moduli. Let us now suppose the existence of a third plane of symmetry, orthogonal
to the previous ones, the plane x1 = 0. With the same procedure, we get:
⎡ ⎤
1 0 0 0 0 0
⎡ ⎤ ⎢0 1 0 0 0 0 ⎥
−1 0 0 ⎢ ⎥
⎢0 0 1 0 0 0 ⎥
U = ⎣ 0 1 0 ⎦ ⇒ [R] = ⎢
⎢0
⎥, (2.61)
⎢ 0 0 1 0 0 ⎥
⎥
0 0 1 ⎣0 0 0 0 −1 0 ⎦
0 0 0 0 0 −1

(C15 ε1 + C25 ε2 + C35 ε3 + C45 ε4 )ε5 +

(C16 ε1 + C26 ε2 + C36 ε3 + C46 ε4 )ε6 = 0 ∀ε ⇐⇒ (2.62)
C15 = C25 = C35 = C45 = C16 = C26 = C36 = C46 = 0.

Rather surprisingly, this new symmetry condition does not give any supplementary
condition to those in (2.56) and (2.60). Because the procedure does not depend upon
the order of the symmetries, as it is immediately recognized, the only consequence
is that the existence of two orthogonal planes of elastic symmetry is physically
impossible: only the presence of one or three mutually orthogonal planes of symmetry
is admissible. A material having three planes of symmetry is called orthotropic. The
class of orthotropic materials is very important, because a lot of materials or structures
belong to it. An orthotropic material depends hence upon 9 distinct elastic moduli
and its matrix [C] looks like
34 2 General Anisotropic Elasticity
⎡ ⎤
C11 C12 C13 0 0 0
⎢ C C 0 0 0 ⎥
⎢ 22 23 ⎥
⎢ C 0 0 0 ⎥
[C] = ⎢
⎢
33 ⎥. (2.63)
⎢ C 44 0 0 ⎥⎥
⎣ sym C55 0 ⎦
C66

2.2.7 Axially Symmetric Bodies

We have seen in Sect. 1.4.2 that there are only four possible cases of axial symmetries
for crystals: the 2-, 3-, 4- and 6-fold axis of symmetry. In elasticity, there is another
possibility, that will be examined in the next Section. Let us then consider the above
four cases, in the order, taking as symmetry axis, without loss in generality, the axis
x3 .
Let us begin with a 2-fold axis of symmetry; the covering operation corresponds
hence to a rotation of π about x3 . In such a case, we have
⎡ ⎤
1 0 0 0 0 0
⎡ ⎤ ⎢0 1 0 0 0 0⎥
−1 0 0 ⎢ ⎥
⎢0 0 1 0 0 0⎥
U= ⎣ 0 −1 0 ⎦ ⇒ [R] = ⎢ ⎢ ⎥, (2.64)
⎢ 0 0 0 −1 0 0 ⎥ ⎥
0 0 1 ⎣ 0 0 0 0 −1 0 ⎦
0 0 0 0 0 1

and we can observe that the matrix [R] is the same of the monoclinic case, Eq. (2.54).
Hence, a 2-fold axis of symmetry coincides with a plane of symmetry.
For a 3-fold axis of symmetry, the covering operation corresponds to a rotation
of 2π/3 about x3 , which gives
⎡  ⎤
1 3
0 0 0 − 38
⎢ 4 4
 ⎥
⎢ ⎥
⎡ √ ⎤ ⎢ 3 1 3 ⎥
⎢ 4 0 0 0 8 ⎥
− 21 3
0 ⎢ 4
⎥
⎢ √
2
⎥ ⎢ 0 0 1 0 0√ 0 ⎥
U = ⎣− 3 − 21 0 ⎦ ⇒ [R] = ⎢
⎢
⎥;
⎥
2
⎢ 0 0 0 − 2 − 23
1
0 ⎥
0 0 1 ⎢ √ ⎥
⎢ 0 3
− 21 0 ⎥
⎢ 0 0 ⎥
⎣  2
⎦
3
8
− 83 0 0 0 − 21
(2.65)

in this case, condition (2.27) is very long and omitted here, but finally it gives 14
conditions on the components of [C]:
2.2 Elastic Symmetries 35

C16 = C26 = C34 = C35 = C36 = C45 = 0,

C22 = C11 , C55 = C44
√, C23 = C13 , √
C24 = −C14 , (2.66)
C25 = −C15 , C56 = 2C14 , C46 = 2C15 , C66 = C11 − C12 .

So, there are only 7 distinct elastic moduli:

⎡ ⎤
C11 C12 C13 C14 C15 0
⎢ C C −C −C 0 ⎥
⎢ 11 13 14 15 ⎥
⎢ C 0 0 ⎥
[C] = ⎢
33 √0 ⎥. (2.67)
⎢ C44 0 −√ 2C15 ⎥
⎢ ⎥
⎣ sym C44 2C14 ⎦
C11 − C12

For a 4-fold axis of symmetry, the covering operation corresponds to a rotation

of π/2 about x3 , which gives
⎡ ⎤
0 1 0 0 0 0
⎡ ⎤ ⎢1 0 0 0 0 0 ⎥
0 1 0 ⎢ ⎥
⎢0 0 1 0 0 0 ⎥
U = ⎣ −1 0 0 ⎦ ⇒ [R] = ⎢
⎢0
⎥; (2.68)
⎢ 0 0 0 −1 0 ⎥
⎥
0 0 1 ⎣0 0 0 1 0 0 ⎦
0 0 0 0 0 −1

we omit also in this case the long expression of Eq. (2.27), but the final result are 14
conditions different from the (2.66):

C14 = C24 = C34 = C15 = C25 = C35 = C45 = C36 = C46 = C56 = 0,
(2.69)
C22 = C11 , C55 = C44 , C23 = C13 , C26 = −C16 ,

leaving an elastic matrix [C] still depending upon only 7 distinct moduli, but different
from the previous case, Eq. (2.67):
⎡ ⎤
C11 C12 C13 0 0 C16
⎢ C11 C13 0 0 −C16 ⎥
⎢ ⎥
⎢ C33 0 0 0 ⎥
[C] = ⎢
⎢
⎥. (2.70)
⎢ C44 0 0 ⎥⎥
⎣ sym C44 0 ⎦
C66

The last case is that of a 6-fold axis of symmetry, with as covering operation a
rotation of π/3 about x3 , which gives
36 2 General Anisotropic Elasticity
⎡  ⎤
1 3 3
0 0 0
⎢ 4 4 8 ⎥
⎡ ⎤ ⎢ ⎥
√ ⎢ 3 1
0 0 0 − 3
⎥
1 3
0 ⎢ 4 4 8 ⎥
⎢ 2√ 2 ⎥ ⎢ 0 0 1 0 0√ 0 ⎥
U = ⎣− 3 1
0 ⎦ ⇒ [R] = ⎢
⎢
⎥;
⎥
2 2
⎢ 0 0 0 1
− 23 0 ⎥
0 0 1 ⎢ 2
√ ⎥
⎢ 0 3 1 ⎥
⎣  0 0 2 2
0 ⎦
− 83 3
8
0 0 0 − 21
(2.71)

condition (2.27), omitted because too long, gives in this case 16 conditions:

C14 = C24 = C34 = C15 = C25 = C35 =

C45 = C16 = C26 = C36 = C46 = C56 = 0, (2.72)
C22 = C11 , C55 = C44 , C23 = C13 , C66 = C11 − C12 ,

for a final elastic matrix [C] depending upon only 5 moduli:

⎡ ⎤
C11 C12 C13 0 0 0
⎢ C C 0 0 0 ⎥
⎢ 11 13 ⎥
⎢ C 0 0 0 ⎥
[C] = ⎢
⎢
33 ⎥.
⎥ (2.73)
⎢ C 44 0 0 ⎥
⎣ sym C44 0 ⎦
C11 − C12

2.2.8 Transversely Isotropic Bodies

Let us now consider the case of a material with an axis of cylindrical symmetry, i.e.
an axis of symmetry where the covering operation is a rotation by any angle θ ; such a
material is called transversely isotropic, and many materials belong to this class, like
for instance wood, fiber reinforced composites, laminated steel and so on. Also in
this case we can proceed in the usual way. Denoting, for shortening the expressions,
c = cos θ, s = sin θ , we get:
⎡ √ ⎤
c2 s2 0 0 0 √2cs
⎡ ⎤ ⎢ s2 c2 0 0 0 − 2cs ⎥
c s 0 ⎢ ⎥
⎢ 0 0 1 0 0 0 ⎥
U = ⎣ −s c 0 ⎦ ⇒ [R] = ⎢
⎢ 0
⎥ . (2.74)
⎢ 0 0 c −s 0 ⎥ ⎥
0 0 1 ⎣ 0
√ √ 0 0 s c 0 ⎦
− 2cs 2cs 0 0 0 c2 − s 2

In this case we obtain exactly the same 16 conditions (2.72); this means that, elas-
tically, the 6-fold axis of symmetry is strictly identical to an axis of cylindrical
2.2 Elastic Symmetries 37

symmetry. Hence, two such materials cannot be distinguished using only the results
of tests on stress or strain. This must not surprise, because this fact is in perfect accor-
dance with the Neumann’s principle, as the 6-fold axis of symmetry is contained in
the more general case of cylindrical symmetry.
Finally, Eq. (2.73) represents also the elastic matrix of a transversely isotropic
material, who has 5 distinct elastic moduli.

2.2.9 Isotropic Bodies

Iostropy is the complete symmetry: all the directions are equivalent. The conditions
of isotropy could be find following the usual procedure, imposing that Eq. (2.27) is
valid for any orthogonal transformation [R]. However, this general approach, that
can be followed using for instance the Euler angles for expressing a generic [R],
results to be very cumbersome and computationally heavy.
A more direct approach is the following one: for a transversely isotropic body,
all the directions orthogonal to the axis of symmetry, say x3 , are equivalent. In other
words, fixing the axes of x1 and x2 is completely arbitrary. Let us then suppose that,
besides the equivalence of all the directions in the plane perpendicular to x3 , also
x1 and x3 are equivalent. We then impose to a material described by a transversely
isotropic elastic matrix, Eq. (2.73), this further equivalence, which is described by
⎡ ⎤
0 0 1 0 0 0
⎡ ⎤ ⎢0 1 0 0 0 0⎥
0 0 1 ⎢ ⎥
⎢1 0 0 0 0 0⎥
⎣
U= 0 1 0 ⎦ ⇒ [R] = ⎢ ⎢ ⎥. (2.75)
⎢ 0 0 0 0 0 1⎥ ⎥
1 0 0 ⎣0 0 0 0 1 0⎦
0 0 0 1 0 0

This gives three new conditions:

C13 = C12 , C33 = C11 , C44 = C66 , (2.76)

which reduce the number of distinct elastic constants from 5 to only 2:

⎡ ⎤
C11 C12 C12 0 0 0
⎢ C11 C12 0 0 0 ⎥
⎢ ⎥
⎢ C11 0 0 0 ⎥
[C] = ⎢⎢
⎥.
⎥ (2.77)
⎢ C11 − C12 0 0 ⎥
⎣ sym C11 − C12 0 ⎦
C11 − C12

Because x1 is any direction, all the directions of the space are equivalent; this can
be proved showing that the elastic matrix (2.77) is insensitive to any change of basis
leaving x2 unchanged, i.e.
38 2 General Anisotropic Elasticity
⎡ √ ⎤
c2 0 s2 0 2cs 0
⎡ ⎤ ⎢ 0 1 0 0 0 ⎥
c 0 s ⎢ √0 ⎥
⎢ s2 0 c2 0 − 2cs 0 ⎥
U= ⎣ 0 1 0 ⎦ ⇒ [R] = ⎢
⎢ 0
⎥ , (2.78)
⎢ √ 0 √0 c 0 −s ⎥
⎥
−s 0 c ⎣ − 2cs 0 2cs 0 c2 − s 2 0 ⎦
0 0 0 s 0 c

which gives as only condition C44 = C11 − C12 , a condition already contained in the
previous ones, Eq. (2.72) and (2.76). This proves that nothing is added to the previous
conditions and hence that all the directions in any meridian plane are equivalent, i.e.
that the body is isotropic.
There is another, more elegant and direct way to prove that an isotropic body
depends upon only two distinct moduli: because of isotropy, the elastic response is
insensitive to a change of frame, so the elastic moduli of an isotropic material cannot
be frame-dependent. This means that for an isotropic material, V cannot depend
upon the εi j , that are frame-dependent quantities, but rather on the invariants of ε.9
As a consequence, being V a quadratic function of the εi j , its general expression is
of the type
1
V = c1 I12 + c2 I2 , (2.79)
2

with10
tr 2 ε − trε 2 εii εii − εi j ε ji
I1 = trε = εii , I2 = = . (2.80)
2 2
The third order invariant of ε, i.e. det ε, cannot enter in the expression of V ,
because it is a cubic function of the εi j , while V must be a quadratic function of the
εi j . Then,
1
V = [(c1 + c2 )εii εii − c2 εi j ε ji ], (2.81)
2

so that11
∂V
σii = = (c1 + c2 )εii − c2 εii ,
∂εii
(2.82)
∂V
σi j = = −c2 ε ji = −c2 εi j .
∂εi j

9 The elastic potential V is, as any other quantity derived by a scalar product, an invariant, i.e. it is not
frame-dependent. Hence, because [C] for an isotropic material is frame independent, the expression
of V cannot depend upon frame-dependent quantities, the εi j , but only upon frame-independent
functions of the εi j : the invariants of ε.
10 ε 2 = εε = ε e ⊗ e ε e ⊗ e = ε ε
i j hk e j · eh (ei ⊗ ek ) = εi j εhk δ j h (ei ⊗ ek ) → trε =
2
ij i j hk h k
εi j εhk δ j h tr(ei ⊗ ek ) = εi j εhk δ j h δik = εi j ε ji .
11 Following a common practice, when an index is underlined, it is not a dummy index: no summation

over it.
2.2 Elastic Symmetries 39

For instance:
∂V
σ11 = = (c1 + c2 )(ε11 + ε22 + ε33 ) − c2 ε11 ,
∂ε11
(2.83)
∂V
σ12 = = −c2 ε12 etc.
∂ε12

We see hence that in the case of isotropic materials, only two constants are sufficient
to characterize the elastic behavior.

2.2.10 Some Remarks About Elastic Symmetries

Some remarks can be done about the results found in the previous Sections. First of
all, the results given, in all the cases, for [C] are completely valid also for [S]; this is
not the case with the Voigt’s notation, where for some symmetries, not each Si j has
the same expression of the corresponding Ci j .
A mechanically important remark is the fact that typically some coupling compo-
nents disappear in a symmetry basis. The case of orthotropic bodies is emblematic:
in the orthotropic frame, the skyline of [C] is exactly the same of an isotropic body
and the only coupling is the Poisson’s effect. Nevertheless, this is no longer true
in any other basis: in a generic basis, all the anisotropic materials, regardless of
their symmetries, behave like a triclinic body, i.e. they have all the coupling terms
(generally speaking, their elastic matrix is complete, none of its terms vanishes).
The only exception to this fact is isotropy; in fact, for an isotropic body the matrices
[C] and [S] are completely invariant, i.e. their only two distinct moduli are tensor
invariants and the only possible coupling is the Poisson’s effect. This is the obvious
consequence of the fact that all the directions of the space are equivalent. Physically,
the fact that the least number of independent elastic constants is two means that in
a stressed elastic body there are, in general, at least two distinct and independent
deformation effects.

2.2.11 Elasticity of Crystals and Elastic Syngonies

Crystals have an elastic behavior that belongs to one of the cases above or is a
combination of these cases. Examining their cases, allows us for entirely defining the
ten elastic syngonies introduced in Sect. 1.4.2. In particular, referring to the Voigt’s
classification, Table 1.1, it is12 :

12 We recall that the following classification is based upon the definition of elastic syngony as a
class of materials sharing the same number and type of independent elastic moduli, see Sect. 1.4.2.
40 2 General Anisotropic Elasticity

1. classes 1 and 2 belong to the triclinic case, with 21 constants; their matrix [C] is
like in Eq. (2.53) and this crystal syngony corresponds with the triclinic elastic
syngony;
2. classes 3, 4 and 5 belong to the monoclinic case, with 13 constants; their matrix
[C] is like in Eq. (2.57) and this crystal syngony corresponds with the monoclinic
elastic syngony;
3. classes 6, 7 and 8 of the orthorhombic syngony belong to the orthotropic case,
with 9 constants; their matrix [C] is like in Eq. (2.63) and the orthorhombic
syngony corresponds hence entirely with the orthotropic elastic syngony;
4. classes 12 and 13 of the trigonal syngony belong to the 3-fold rotational symmetry
case, with 7 constants; they have a matrix [C] as in Eq. (2.67) and they constitute
the trigonal elastic syngony with 7 constants;
5. classes 17, 18 and 20 of the tetragonal syngony belong to the 4-fold rotational
symmetry case, with 7 constants; their matrix is like in Eq. (2.70) and they
constitute the tetragonal elastic syngony with 7 constants;
6. classes 9, 10 and 11 of the trigonal syngony are a combination of the 3-fold rota-
tional symmetry and the monoclinic symmetry cases: if the plane of symmetry
is the plane x1 = 0, then the usual procedure applied to the matrix (2.67) gives
C15 = 0, and matrix (2.67) becomes
⎡ ⎤
C11 C12 C13 C14 0 0
⎢ C C −C 0 0 ⎥
⎢ 11 13 14 ⎥
⎢ C 0 0 0 ⎥
[C] = ⎢ ⎥;
33
⎢ ⎥ (2.84)
⎢ C 44 0 √0 ⎥
⎣ sym C44 2C14 ⎦
C11 − C12

if it is x2 = 0 the plane of symmetry, then it is C14 = 0 and matrix (2.67)

becomes ⎡ ⎤
C11 C12 C13 0 C15 0
⎢ C11 C13 0 −C15 0 ⎥
⎢ ⎥
⎢ C33 0 0 0 ⎥
[C] = ⎢⎢
√ ⎥; (2.85)
⎢ C44 0 − 2C15 ⎥ ⎥
⎣ sym C44 0 ⎦
C11 − C12

these cases constitute the trigonal elastic syngony with 6 constants;

7. classes 14, 15, 16 and 19 of the tetragonal syngony are a particular case of the
orthotropic symmetry: they have identical elastic properties along the axis x1
and x2 , which gives the three supplementary conditions C22 = C11 , C23 =
C13 , C55 = C44 , so reducing matrix (2.63) to
2.2 Elastic Symmetries 41
⎡ ⎤
C11 C12 C13 0 0 0
⎢ C C 0 0 0 ⎥
⎢ 11 13 ⎥
⎢ C 0 0 0 ⎥
[C] = ⎢
⎢
33 ⎥; (2.86)
⎢ C 44 0 0 ⎥⎥
⎣ sym C44 0 ⎦
C66

these cases constitute the tetragonal elastic syngony with 6 constants;

8. classes of the hexagonal syngony, from the 21 to the 27, belong to the 6-fold
rotational symmetry, with 5 constants; together with transversely isotropic mate-
rials, that do not exist as crystals, they form the axe-symmetric elastic syngony,
with [C] as in Eq. (2.73);
9. classes of the cubic syngony, from the 28 to the 32, are a particular case of
the orthotropic symmetry: they have identical properties along the three axes,
which gives the six supplementary conditions C33 = C22 = C11 , C23 = C13 =
C12 , C66 = C55 = C44 , so reducing matrix (2.63) to
⎡ ⎤
C11 C12 C12 0 0 0
⎢ C C 0 0 0 ⎥
⎢ 11 12 ⎥
⎢ C 0 0 0 ⎥
[C] = ⎢
⎢
11 ⎥; (2.87)
⎢ C 44 0 0 ⎥⎥
⎣ sym C44 0 ⎦
C44

the cubic crystal syngony corresponds entirely with the cubic elastic syngony;
10. the last elastic syngony is the isotropic elastic syngony; of course, no crystal
syngonies belong to this case; nevertheless, a huge number of materials have
an isotropic behavior. Though the texts on crystals and anisotropy usually for-
get to consider the isotropic case, this one actually exists and for the sake of
completeness we prefer here to consider it as an elastic syngony; the isotropic
matrix (2.77) can be obtained as a particular case of the cubic one, (2.87), when
C44 = C11 − C12 .

2.3 The Technical Constants of Elasticity

In practical applications, engineers usually prefer to replace the use of the elastic stiff-
ness matrix components by the so-called technical elasticity constants or engineer
moduli.
Technical constants quantify an effect, a direct or a coupling one, whose mechan-
ical meaning is immediate and that can be easily identified and measured in simple
laboratory tests, like for instance unidirectional tensile tests.
Of course, the set of technical constants must be equivalent to the set of indepen-
dent elastic moduli, which means, on one side, that the number of technical constants
42 2 General Anisotropic Elasticity

and distinct elastic moduli must be the same, i.e. 21, and that the technical constants
must represent all the mechanical effects in a stressed body.
Though replacing the components of the stiffness elastic matrix [C], the technical
constants are defined as functions of the Si j . Unlike the elastic moduli, only 6 tech-
nical constants are moduli: they measure a direct effect i.e. they correspond to terms
on the diagonal of [C], and are homogenous to a stress. The remaining 15 technical
constants are coefficients: they are dimensionless quantities because constructed as
the ratio between two strain components and they measure a coupling effect, i.e. they
correspond to terms out of the diagonal of [C]. Let us introduce all of them.

2.3.1 The Young’s Moduli

The three Young’s moduli generalize to anisotropy the analogous isotropic modulus
and are defined in a similar way:
σi
E i := , i = 1, 2, 3, σi = 0, σ j = 0 for j = i, j = 1, . . . , 6. (2.88)
εi

As a consequence, from Eq. (2.21) we get the relations (no summation over dummy
indexes)
1
Sii = Z iiii = , i = 1, 2, 3. (2.89)
Ei

The mechanical meaning of the Young’s moduli is self-evident: each one of them
measures the extension stiffness along the direction of one of the frame axes, i.e.
the stress to be applied in the direction xi to stretch the same direction with a strain
equal to unity. Generally speaking, the three Young’s moduli are different, i.e. in
anisotropy the directions of the space have different stiffnesses.

2.3.2 Shear Moduli

Also in this case, the three shear moduli generalize to anisotropy the isotropic concept
of shear modulus13 :
σk
G i j := , i, j = 1, 2, 3, i = j, k = 4, 5, 6, σk = 0, σh = 0 for h = k, h = 1, . . . , 6.
2εk
(2.90)

13 The reader should consider that the definition of the shear moduli normally found in the literature

is σi j
G i j := ,
γi j
where γi j is the so-called technical shear strain, γi j := 2εi j . The above equation in the Kelvin’s
notation just corresponds to Eq. (2.90).
2.3 The Technical Constants of Elasticity 43

To be remarked that in the literature there is a strange discrepancy in the nomenclature

of the G i j s: in fact, the Kelvin notation is used for σk and εk but in G i j the indexes are
those indicating the directions. The correspondences between k and i j are of course
those indicated by Eq. (2.19). As a consequence, from Eqs. (2.21) and (2.24) we get
the relations (no summation over dummy indexes)

1
2Skk = 4Z i ji j = , i, j = 1, 2, 3, i = j, k = 4, 5, 6. (2.91)
Gi j

The mechanical meaning of the shear moduli is completely analogous to that of the
Young’s moduli, but it concerns shear stress and strain, and the same remarks can be
done.

2.3.3 Poisson’s Coefficients

The definition of the Poisson’s coefficients or ratios in anisotropy is quite similar to

that given for isotropic bodies:
εj
νi j := − , i, j = 1, 2, 3, σi = 0, σh = 0 for h = i, h = 1, . . . , 6. (2.92)
εi

Like for shear moduli, also in this case the nomenclature makes use, in the same
formula, of the Kelvin’s notation along with the classical tensorial one.
From the Young’s moduli definition, Eq. (2.88), we get
σi
ε j = −νi j εi = −νi j , i, j = 1, 2, 3. (2.93)
Ei

Through Eq. (2.21) this gives (no summation over dummy indexes)

νi j S ji
S ji = Z j jii = − ⇒ νi j = − , i, j = 1, 2, 3. (2.94)
Ei Sii

Finally, the symmetry of matrix [S], consequence of the major symmetries of Z,

gives the reciprocity relations
νi j ν ji
= , i, j = 1, 2, 3, (2.95)
Ei Ej

which reduce the number of distinct Poisson’s coefficients from 6 to only 3.

Some remarks about the Poisson’s coefficients: they measure the Poisson’s effect,
i.e. the deformation in a direction transversal to that of the normal stress. Because,
generally speaking, the three Poisson’s coefficients are different, the Poisson’s effect
is different in the different directions. Also, due to the dependence upon the frame
44 2 General Anisotropic Elasticity

orientation, Eq. (2.34), it is possible that in some directions νi j ≤ 0, i.e. directions

with null or negative Poisson’s coefficients are quite possible in anisotropic elasticity.
To end, it is worth to remark that some authors exchange the place of suffixes i and
j in the definition of νi j .

2.3.4 Chentsov’s Coefficients

The Chentsov’s coefficients μi j,kl play the same role of the Poisson’s coefficients
with respect to shear stress and strain. They are defined as follows:
εi j
μi j,kl := , i, j, k, l = 1, 2, 3, i = j, k = l,
εkl (2.96)
σkl = 0, σ pq = 0 for pq = kl, p, q = 1, 2, 3.

Hence, coefficient μi j,kl measures the Chentsov’s effect in the plane i j due to the
shear stress σkl , i.e. the ratio between the shear strain components εi j and εkl . By the
definition of the G i j s, Eq. (2.90), it follows that (no summation over dummy indexes)

σkl
εi j = μi j,kl εkl = μi j,kl i, j, k, l = 1, 2, 3, (2.97)
2G kl

and through Eqs. (2.21) and (2.24) we get

μi j,kl S pq
2S pq = 4Z i jkl = ⇒ μi j,kl = , i, j, k, l = 1, 2, 3, p, q = 4, 5, 6,
G kl Sqq
(2.98)
with p that corresponds to the couple i j and q to kl according to the scheme (2.19).
The symmetry of [S] gives the reciprocity relations
μi j,kl μkl,i j
= , (2.99)
G kl Gi j

that, along with the minor symmetries of σ and ε reduce to only three the number
of distinct Chentsov’s coefficients. Finally, the remarks done for the νi j s can be
rephrased verbatim for the μi j,kl s.

2.3.5 Coefficients of Mutual Influence of the First Type

These coefficients characterize the normal strain εii due to the shear σ jk (no summa-
tion over dummy indexes):
2.3 The Technical Constants of Elasticity 45

εii
ηi, jk := i, j, k = 1, 2, 3, j = k, σ jk = 0, σ pq = 0 for pq = jk, p, q = 1, 2, 3.
2ε jk
(2.100)
By the definition of the G i j s, Eq. (2.90), it follows that

σ jk
εii = 2ηi, jk ε jk = ηi, jk , (2.101)
G jk

and through Eqs. (2.21) and (2.24) we get

√ ηi, jk Si p
2Si p = 2Z ii jk = ⇒ ηi, jk = √ , i, j, k = 1, 2, 3, p = 4, 5, 6,
G jk 2S pp
(2.102)
with p that corresponds to the couple jk according to the scheme (2.19). For the
symmetry of σ and ε, the exchange of suffixes j and k has no effects, so the number
of distinct coefficients is only 9.

2.3.6 Coefficients of Mutual Influence of the Second Type

These coefficients characterize the shear strain εi j due to the normal stress σkk (no
summation over dummy indexes):

2εi j
ηi j,k := i, j, k = 1, 2, 3, i = j, σkk = 0, σ pq = 0 for pq = kk, p, q = 1, 2, 3.
εkk
(2.103)
By the definition of the E i s, Eq. (2.88), it follows that
σkk
2εi j = ηi j,k εkk = ηi j,k , (2.104)
Ek

and through Eqs. (2.21) and (2.24) we get

√ ηi j,k √ S pk
2S pk = 2Z i jkk = ⇒ ηi j,k = 2 , i, j, k = 1, 2, 3, p = 4, 5, 6,
Ek Skk
(2.105)
with p that corresponds to the couple i j according to the scheme (2.19). Like for the
coefficients of the first type, the symmetries of σ and ε reduce the number of distinct
coefficients of the second type to only 9.
The coefficients of the second type are not independent from those of the first
type; in fact, the symmetry of [S] gives immediately the reciprocity relations
ηi j,k ηk,i j
= , i, j, k = 1, 2, 3. (2.106)
Ek Gi j
46 2 General Anisotropic Elasticity

So the use of the coefficients of the first or of the second type is arbitrary and
equally valid. Also for the coefficients of the first and second type can be repeated
almost verbatim the remarks done about the other coefficients.

2.3.7 Some Remarks About the Technical Constants

The relations between a technical constant and the corresponding component of Z,

given in the previous Sections, are valid regardless of the notation used, i.e. they
are the same also with the Voigt’s notation. On the contrary, the relations with the
components Si j depends upon the notation, and those found above are not completely
identical with the Voigt’s notation, see for instance (Jones 1999, p. 79).
It is possible, of course, to express also the components of [C] as functions of
the technical constants; this necessitates the inversion of [S] and in the most general
case it gives so complicate and long expressions that it is impossible to write them.
Nevertheless, in the important case of orthotropic materials the transformation
is rather simple. In fact, in the orthotropic frame, the inverse of matrix [S], which
is perfectly analogous to matrix (2.63), is given by (no summation on the dummy
indexes)

S j j Skk − S 2jk 1 − ν jk νk j
Cii = = E i , i, j, k = 1, 2, 3, i = j = k,
S 
Sik Sk j − Si j Skk νi j + νik νk j
Ci j = = E j , i, j, k = 1, 2, 3, i = j = k,
S 
1 1 1 (2.107)
C44 = = 2G 23 , C55 = = 2G 31 , C66 = = 2G 12 , with
S44 S55 S66
S = S11 S22 S33 − S11 S23 − S22 S13 − S33 S12 + 2S12 S23 S13 ,
2 2 2

 = 1 − ν12 ν21 − ν23 ν32 − ν31 ν13 − 2ν32 ν21 ν13 .

In matrix form we have

⎡ ⎤
S22 S33 −S23
2
S13 S32 −S12 S33 S12 S23 −S13 S22
0 0 0
⎢ S S 2
S11 S33 −S13
S ⎥
⎢ S21 S13 −S23 S11
0 0 0 ⎥
⎢ S S 2 ⎥
⎢ S11 S22 −S12 ⎥
[C] = ⎢
⎢ S
0
1
0 0 ⎥,
⎥ (2.108)
⎢ 0 0 ⎥
⎢ S44
⎥
⎣ sym 1
S55
0 ⎦
1
S66

and with the technical constants

2.3 The Technical Constants of Elasticity 47
⎡ 1−ν23 ν32 ν12 +ν13 ν32 ν13 +ν12 ν23 ⎤

E1 
E2 
E3 0 0 0
⎢ 1−ν13 ν31
E ν23 +ν21 ν13
E 0 0 0 ⎥
⎢  2  3 ⎥
⎢ 1−ν12 ν21
E 0 0 0 ⎥
[C] = ⎢
⎢
 3 ⎥ (2.109)
⎢ 2G 23 0 0 ⎥ ⎥
⎣ sym 2G 13 0 ⎦
2G 12

and conversely
⎡ ⎤
1
E1
− νE212 − νE313 0 0 0
⎢ 1
− νE323 0 0 0 ⎥
⎢ E2 ⎥
⎢ 1 ⎥
⎢ 0 0 0 ⎥
[S] = ⎢ E3
⎥. (2.110)
⎢ 1
0 0 ⎥
⎢ 2G 23 ⎥
⎣ sym 1
2G 13
0 ⎦
1
2G 12

It is also worth to specify these results also for the isotropic case
⎡ (1−ν)E νE νE ⎤
(1−2ν)(1+ν) (1−2ν)(1+ν) (1−2ν)(1+ν)
0 0 0
⎢ (1−ν)E νE ⎥
⎢ (1−2ν)(1+ν) (1−2ν)(1+ν)
0 0 0 ⎥
⎢ (1−ν)E ⎥
⎢ 0 0 0 ⎥
[C] = ⎢ (1−2ν)(1+ν) ⎥, (2.111)
⎢ E
0 0 ⎥
⎢ 1+ν ⎥
⎣ sym E
0 ⎦
1+ν
E
1+ν

and conversely
⎡ ⎤
1
E
− Eν − Eν 0 0 0
⎢ 1
− Eν 0 0 0 ⎥
⎢ E ⎥
⎢ 1
0 0 0 ⎥
[S] = ⎢
⎢
E
1+ν
⎥.
⎥ (2.112)
⎢ E
0 0 ⎥
⎣ sym 1+ν
0 ⎦
E
1+ν
E

To remark that the Voigt’s notation can be obtained simply dividing by a factor 2
the components C44 , C55 and C66 , and multiplying by 2 the components S44 , S55 and
S66 .
48 2 General Anisotropic Elasticity

2.4 Bounds on the Elastic Constants

2.4.1 General Conditions and Results

Elastic constants cannot take any value: they are bounded because of the physical fact
that the deformation of an elastic body Ω cannot produce energy, i.e. the overall work
Lext done by the applied forces must be positive. From the Clapeyron’s Theorem
  
1
Lext = 2Vtot = 2 σ · ε dΩ , (2.113)
2 Ω

we get hence the condition that the total strain energy Vtot must be positive. Assuming
the strain as independent field over Ω, then the overall condition is

1
Vtot = σ · ε dΩ > 0 ∀ε = O. (2.114)
2 Ω

The above constraint on the deformation of an elastic body is a strong condition. By

a procedure of limit towards small volumes, it is easy to see that it must be true also
locally, i.e. ∀ p ∈ Ω; it is just the local form of (2.114) that gives the bounds on the
elastic constants of a material. In fact, getting the local form of (2.114) and injecting
the Hooke’s law (2.12) gives

1 1
V = σ · ε = ε · Eε > 0 ∀ε = O. (2.115)
2 2
Equation (2.115) is the mathematical condition corresponding to the thermody-
namical fact that no energy can be produced deforming an elastic body: the elasticity
stiffness tensor E must be positive definite.
If the σ is taken as independent field over Ω in place of ε, we get a similar
restriction on the stress energy and finally the condition that the elasticity compliance
tensor Z must be positive definite. Of course, the two approaches give in the end the
same results for the elastic constants.

2.4.2 Mathematical Conditions for the Elastic Matrices

It is easier to obtain practical results for the components of matrices [C] and [S] than
for tensors E and Z, so let us rewrite condition (2.115) in its equivalent matrix form:

1 
{ε} [C]{ε} > 0 ∀{ε} = {O}. (2.116)
2
2.4 Bounds on the Elastic Constants 49

Of course, the above condition states the positive definiteness of matrix [C], and
an analogous condition can be written for matrix [S].
Mathematically, the problem is hence clear: being [C] symmetric, so, by the
Spectral Theorem, with real eigenvalues λi , condition (2.116) corresponds to impose
that all the eigenvalues are positive:

1 
{ε} [C]{ε} > 0 ∀{ε} = {O} ⇐⇒ λi > 0 ∀i = 1, . . . , 6. (2.117)
2
The above result is almost useless, because the Laplace’s equation of [C] is an
algebraic equation of degree 6. Hence, generally speaking, it is not possible to get
an analytic expression of the roots of this equation, the eigenvalues λi , for obtaining
the searched bounds on the Ci j .
Nevertheless, a first qualitative result is that the number of conditions on the Ci j s
is 6. As distinct components are, in the most general case, 21, the conditions on
the Ci j s are not necessarily simple bounds but at least some of them are necessarily
relations among some of the components. Also, for the hexagonal, cubic and isotropic
syngonies the number of conditions is redundant with respect to the distinct elastic
constants, so some of them have lower and upper bounds and/or some of the bounds
are redundant (this, anyway, can be true also for other syngonies).
Though the approach by eigenvalues is practically impossible, there is another
mathematical approach which is completely general and feasible. To this purpose,
let us introduce the following definitions and theorems of matrix algebra.
A principal minor of a matrix [A] is the determinant of the sub-matrix extracted
from [A] removing an equal number of rows and columns having the same indexes,
i.e. preserving the leading diagonal.
A leading principal minor of order r is the determinant of a principal r × r sub-
matrix whose rows and columns are the first r rows and columns of [A]. Hence, a
n × n matrix has n leading principal minors.
Theorem 1 (Necessary condition for a symmetric matrix to be positive definite) All
the principal minors of a positive definite n × n symmetric matrix [A] are positive.

Proof By the hypothesis,

{x} [A]{x} > 0 ∀{x} = {0}. (2.118)

Then, for a principal r × r sub-matrix [Ar ] extracted from [A] deleting n − r rows
and columns with the same indexes, we have

{x r } [Ar ]{x r } = {x} [A]{x} > 0 ∀{x}, {x r } = {0} (2.119)

where {x} is any vector whose components corresponding to the removed rows of
[A] are null and with at least one of the other components different from zero,
while {x r } is the r −dimensional vector obtained removing from {x} the components
corresponding to the removed rows of [A].
50 2 General Anisotropic Elasticity

Hence, [Ar ] is positive definite and because it is symmetric, its eigenvalues are
positive, see Eq. (2.117). Then, its determinant, which is the product of its eigenval-
ues, is positive too. This proves that det[Ar ] > 0 and hence the theorem.

Theorem 2 (Necessary and sufficient condition for a symmetric matrix to be positive

definite) For a n × n symmetric matrix [A] to be positive definite it is necessary and
sufficient that its n leading principal minors are all positive.

The proof of this theorem is non trivial and the reader is referred to (Hohn 1958, p.
340).
The six principal minors of [C] are
 
   C11 C12 C13 
 C11 C12   
M1 = C11 , M2 =   , M3 =  C12 C22 C23  ,
  
C12 C22  C13 C23 C33 
 
   C11 C12 C13 C14 C15 
 C11 C12 C13 C14  
   C12 C22 C23 C24 C25 
 C12 C22 C23 C24  
M4 =   , M5 =  C13 C23 C33 C34
  C35  , (2.120)
 C13 C23 C33 C34   C14 C24 C34 C44 C45 
 C14 C24 C34 C44  
 C15 C25 C35 C45 C55 
M6 = det[C].

Contrarily to the eigenvalues, it is always possible to explicit the above expressions

and hence the 6 conditions

Mi > 0, i = 1, . . . , 6. (2.121)

That is why the use of Theorem 2 is more interesting than condition (2.117), though
to write down the 6 conditions in the most general case of a triclinic material gives
so long expressions that they are omitted here.
We can, however, consider the different elastic syngonies and because they have a
simpler form of [C] than in the triclinic case, also conditions (2.121) will be simpler.
In particular, let us consider here some cases whose results are particularly simple
(the bounds are written for matrix [C], but similar results can be written for [S] too;
redundant bounds have been omitted):
• orthotropic elastic syngony, Eq. (2.63):

Cii > 0, i = 1, 4, 5, 6,
C11 C22 − C12
2
> 0, (2.122)
C11 C22 C33 − 2
C33 C12 − 2
C11 C23 − 2
C22 C13 + 2C12 C13 C23 > 0;

• tetragonal elastic syngony with 6 constants, Eq. (2.86):

2.4 Bounds on the Elastic Constants 51

C44 > 0,
C66 > 0,
(2.123)
2
C11 − C12
2
> 0,
 
(C11 − C12 ) C33 (C11 + C12 ) − 2C13
2
> 0;

• axially symmetric elastic syngony, Eq. (2.73):

C44 > 0,
2
C11 − C12
2
> 0, (2.124)
 
(C11 − C12 ) C33 (C11 + C12 ) − 2C13
2
> 0;

• cubic elastic syngony, Eq. (2.87)

C44 > 0,
C11 − C12 > 0, (2.125)
C11 + 2C12 > 0;

• isotropic elastic syngony, Eq. (2.77):

C11 − C12 > 0,

(2.126)
C11 + 2C12 > 0.

2.4.3 A Mechanical Approach

The bounds on the elastic constants can be found also by a direct mechanical
approach, based upon the fact that the strain energy must be positive for each pos-
sible choice of the strain field ε. This allows for choosing particularly simple strain
fields, giving some direct, simple results. Let us see how (no summation over dummy
indexes): choose a field {ε} with only one component εi = 0. Then,

V > 0 ⇐⇒ Cii > 0, i = 1, . . . , 6; (2.127)

we get hence six conditions. Unfortunately, each one of them is only a necessary
condition for the strain energy be positive, so the (2.127) do not constitute a set
of necessary and sufficient conditions for the positiveness of V . Nevertheless, they
give us a precious information: all the moduli responsible for the direct effects are
strictly positive. Using the stress energy instead of the strain energy, it is immediately
recognized that it is also:
Sii > 0 ∀i = 1, . . . , 6. (2.128)
52 2 General Anisotropic Elasticity

2.4.4 Bounds on the Technical Constants

The results of Eqs. (2.89), (2.91) and (2.128) give immediately

E i > 0, G i j > 0 ∀i, j = 1, 2, 3 : (2.129)

all the Young’s and shear moduli are strictly positive quantities, result that is valid
for any kind of elastic syngony.
To these necessary conditions some other relations for the technical constants can
be added. First of all, let us consider a spherical state of stress; it is then easy to see
that
{σ } = σ {I } ⇒ {σ } [S]{σ } > 0 ⇐⇒
(2.130)
S11 + S22 + S33 + 2(S13 + S32 + S21 ) > 0.

Replacing in the above result the expressions of the Si j s from Eqs. (2.89) and (2.94)
gives the condition

1 − 2ν12 1 − 2ν23 1 − 2ν31

+ + > 0. (2.131)
E1 E2 E3

This result is valid regardless of the elastic syngony; for the cubic and isotropic
syngonies it becomes the well known bound ν < 1/2 on the Poisson’s coefficient.
A simpler but rougher estimation can be obtained from bound (2.131), (Lekhnitskii
1950, p. 85):

3 − 2(ν12 + ν23 + ν31 ) 1 − 2ν12 1 − 2ν23 1 − 2ν31

> + + >0 ⇒
min{E 1 , E 2 , E 3 } E1 E2 E3
(2.132)
3
ν12 + ν23 + ν31 < .
2
Some other necessary conditions can be given expressing the Cii in terms of the
technical parameters. This is impossible in the most general case of the triclinic
syngony, for the calculations are too complicate. However, this can be done for the
orthotropic syngony; the supplementary bounds can be found expressing the (2.127)
as functions of the technical constants through Eq. (2.107) and taking into account
the positivity of the Young’s moduli, Eq. (2.129):

1 − νi j ν ji > 0 ∀i, j = 1, 2, 3;
(2.133)
 = 1 − ν12 ν21 − ν23 ν32 − ν31 ν13 − 2ν32 ν21 ν13 > 0.

Condition (2.133)2 can be transformed to

 
1 2 E2 2 E1 2 E3 1
ν32 ν21 ν13 < 1 − ν32 − ν21 − ν13 < . (2.134)
2 E3 E2 E1 2
2.4 Bounds on the Elastic Constants 53

Through the reciprocity conditions on the Poisson’s coefficients, Eq. (2.95), condi-
tions (2.133)1 can be written also as

Ei
|νi j | < ∀i, j = 1, 2, 3, (2.135)
Ej

or equivalently 
|Si j | < Sii S j j ∀i, j = 1, 2, 3. (2.136)

Some remarks to end this part; first of all, the bounds concern frame dependent
quantities, and of course they are more easily written in a frame composed by sym-
metry directions. Then, the only, general, necessary and sufficient conditions are
the (2.121), that can always be written and used in numerical applications, e.g. for
checking the validity of the results of experimental tests.
In the case of orthotropic materials, a set of conditions on the technical constants
can be easily written, but it is still questionable whether or not it constitutes a set
of necessary and sufficient conditions for the positivity of the strain energy, a point
never treated in the literature. Finally, bounds on the Chentsov’s and mutual influence
coefficients are apparently unknown in the literature.
In the case of isotropic materials, the conditions of positivity of the strain energy
reduce to the well known three bounds on E and ν
1
E > 0, −1 < ν < , (2.137)
2
but when the bounds are written for the two distinct components of [C], C11 and C12 ,
then rather surprisingly the bounds are only two, see Eq. (2.126):

C11 − C12 > 0, C11 + 2C12 > 0. (2.138)

Also when the isotropic constitutive law is written under the form of the Lamé’s
equations
σ = 2με + λtrε I, (2.139)

it is easy to show that the only two bounds on the Lamé’s constants λ and μ are

μ > 0, 2μ + 3λ > 0, (2.140)

that corresponds exactly to bounds (2.138). This fact shows that the number of
necessary and sufficient conditions for the strain energy to be positive depends upon
the choice of the elastic constants and that, anyway, it is quite hard to establish a
priori its value, whose maximum remains however 6.
A last remark: all the bounds and conditions written in this Section are written on
frame dependent quantities, apart those written for the isotropic case, Eq. (2.126),
of course. In particular, conditions (2.122) to (2.125) are valid exclusively in the
symmetry frame where the respective matrices [C] have been written. In the plane
54 2 General Anisotropic Elasticity

case, we will see that it is possible, with the polar formalism, to give completely
invariant bounds, i.e. bounds established on tensor invariants, which are not yet
known for the general 3D case.

2.5 An Observation About the Decomposition

of the Strain Energy

Let us consider a point which is true at least for isotropic materials but often thought as
generally true also for other elastic syngonies: is it possible to decompose the strain,
or stress, energy into spherical and deviatoric parts? In other words, we ponder
whether or not it is always possible to write

V = Vsph + Vdev , (2.141)

where Vsph , the spherical part of V is produced exclusively by the spherical part of
ε and by its corresponding part of σ , i.e.

1
Vsph = ε sph · Eε sph , (2.142)
2
and Vdev , the deviatoric part of V is produced exclusively by the deviatoric part of ε
and by its corresponding part of σ , i.e.

1
Vdev = ε dev · Eε dev . (2.143)
2
Mechanically, such a decomposition means that V can be considered as the sum of
two parts: one, Vsph , due to volume changes not accompanied by shape changes, the
other one, Vdev , produced by isochoric shape changes. This decomposition is, for
instance, at the basis of the Hüber–Hencky–von Mises criterion, where the only Vdev
is considered to be responsible of yielding.
We recall that it is always possible to decompose σ and ε into a spherical and a
deviatoric part

1
σ = σ sph + σ dev , σ sph = trσ I, σ dev = σ − σ sph ,
3 (2.144)
1
ε = εsph + ε dev , ε sph = trε I, ε dev = ε − ε sph ,
3
and that any spherical part is orthogonal to any deviatoric part:
2.5 An Observation About the Decomposition of the Strain Energy 55
 
1 1 1 1
σ sph · εdev = trσ I · ε − trε I = trε trσ − trε trσ = 0,
3 3 3 3
  (2.145)
1 1 1 1
σ dev · ε sph = σ − trσ I · trε I = trε trσ − trε trσ = 0.
3 3 3 3

Using decomposition (2.144) we have

1 1
V = ε · Eε = (ε sph + ε dev ) · E(ε sph + ε dev ) =
2 2 (2.146)
1 1 1 1
ε sph · Eε sph + εdev · Eε dev + ε sph · Eε dev + ε dev · Eε sph .
2 2 2 2
For the decomposition (2.141) to be true, it is necessary and sufficient that
 
εsph · Eε dev = 0 ⇒ tr ε 
sph (Eε dev ) = 0 ∀ε. (2.147)

In fact, whenever Eq. (2.147) is satisfied, for definition (2.38) it is

ε dev · Eε sph = E ε dev · εsph = ε sph · Eε dev , (2.148)

because of the symmetry of E, i.e. for its major symmetries. This result shows that
the two mixed terms in (2.146) are identical.
Through (2.144), condition (2.147) can be written as
 
1
tr trε I(Eεdev ) = 0 ∀ε ⇐⇒ tr(Eεdev ) = 0. (2.149)
3

The components of E must satisfy Eq. (2.149) for the decomposition (2.141) to be
possible. It can be rewritten as
  
1
tr E ε − trε I = 0 ⇒ 3tr(Eε) − trε tr(EI) = 0 ∀ε. (2.150)
3

Actually, it is easy to check that condition (2.147) corresponds to impose that

σ dev = Eε dev , σ sph = Eε sph . (2.151)

Condition (2.150) can be written by components:

E hhkk εii − 3E j j pq ε pq = 0 ∀εmn , i, j, h, k, p, q, m, n = 1, 2, 3. (2.152)

Generally speaking, this quantity does not vanish for any possible choice of ε. As a
consequence, for a generic anisotropic material decomposition of the strain energy
into a spherical and deviatoric part is not possible. Nevertheless, it can be checked
that for the cubic syngony Eq. (2.152) is always satisfied. In fact, for an orthotropic
56 2 General Anisotropic Elasticity

material condition (2.152) becomes

1
[E 1111 (2ε11 − ε22 − ε33 ) + E 2222 (2ε22 − ε11 − ε33 ) +
3
E 3333 (2ε33 − ε22 − ε11 )]+
(2.153)
2
[E 1122 (ε11 + ε22 − 2ε33 ) + E 1133 (ε11 + ε33 − 2ε22 ) +
3
E 2233 (ε22 + ε33 − 2ε11 )] = 0,

condition which is not yet satisfied, generally speaking, but which is always satisfied
when
E 1111 = E 2222 = E 3333 , E 1122 = E 2233 = E 1133 , (2.154)

i.e. by cubic materials. Hence, for materials of the cubic syngony, decomposition
(2.141) is always possible and, a fortiori, by isotropic materials too, as well known,
because they can be considered as a special case of the cubic syngony, see Sect. 2.2.11,
point 10.

2.6 Determination of Symmetry Planes

The classification in elastic syngonies presupposes that, for a given material, the
existing equivalent directions are known, so as to write E, or equivalently [C], in
a symmetry frame, which makes some of the E i jkl s, and the corresponding Ci j s,
vanish.
But when a material is completely unknown, the independent measures to be done
in experimental tests to characterize the material are as much as 21; practically, it
is very complicate to do all of these tests. Nevertheless, the existence of possible
symmetry planes remains unknown also once all the Ci j s are known, if [C] is a full
matrix.
The problem is hence the following one: given a general matrix [C], is it possible
to determine if some planes of symmetry exist and which they are? We will see in
Sect. 4.1 that in the planar case it is very simple to determine whether or not the
material has some kind of elastic symmetry and the symmetry directions using the
polar formalism. In the three dimensional case, the problem is much more complicate;
it has been solved by Cowin and Mehrabadi in two works, (Cowin and Mehrabadi
1987; Cowin 1989), successively completed by Ting (1996). We give here a brief
account of these results.
Be n and m two vectors such that ||n|| = ||m|| = 1, m · n = 0, with n orthogonal
to a symmetry plane for a material whose elastic tensor is E. Consider the following
second-rank symmetric tensors: V = EI, W the acoustic14 tensor relative to the basis

14 Theacoustic or Green-Christoffel tensor Au relative to the direction u is the unique tensor such
that Au w = E(w ⊗ u)u ∀w ∈ V .
2.6 Determination of Symmetry Planes 57

direction e p , X and Y the acoustic tensors relative to n and m, respectively.15 We can

now state the following:
Theorem 3 The following statements are equivalent (λi ∈ R, i = 1, . . . , 6):
1. the material has a plane of symmetry whose normal is n;
2. Vn = λ1 Yn = λ2 n;
3. Wn = λ3 Yn = λ4 n;
4. Xn = λ5 Yn = λ6 n.
Proof Without loss of generality, let us suppose that n = e1 and m = cos θ e2 +
sin θ e3 . When n is an eigenvector of V, W, X or Y then

Vn = λv n → E i1qq ei = λv e1 ,
Wn = λw n → E i p1 p ei = λw e1 ,
Xn = λx n → E i111 ei = λx e1 , (2.155)
Yn = λ y n →
 
E i212 cos2 θ + E i313 sin2 θ + (E i213 + E i312 ) sin θ cos θ ei = λ y e1 ∀θ.

For i = 1, the above results give the values of the respective eigenvalues, but for
i = 2, 3 we get, respectively,

E 21qq = E 31qq = 0,
E 2 p1 p = E 3 p1 p = 0,
E 2111 = E 3111 = 0, (2.156)
E 2212 cos2 θ + E 2313 sin2 θ + (E 2213 + E 2312 ) sin θ cos θ =
E 3212 cos2 θ + E 3313 sin2 θ + (E 3213 + E 3312 ) sin θ cos θ = 0 ∀θ.

Passing to the Ci j s for the sake of convenience, and writing down in the order all the
above relations, we get

C15 + C25 + C35 = C16 + C26 + C36 = 0,

C46 C45
C15 + C35 + √ = C16 + C26 + √ = 0,
2 2 (2.157)
C15 = C16 = 0,
C25 = C26 = C35 = C36 = C45 = C46 = 0.

15 It is simple to verify that

V = EI = E ikqq ei ⊗ ek ,
W = E i pkp ei ⊗ ek ,
X = E ilkm nl n m ei ⊗ ek ,
Y = E i jkh m j m h ei ⊗ ek .
.
58 2 General Anisotropic Elasticity

If the material has x1 = 0 as unique plane of symmetry, it belongs to the monoclinic

syngony and its matrix [C] is given by, see Sect. 2.2.5,
⎡ ⎤
C11 C12 C13 C14 0 0
⎢ C22 C23 C24 0 0 ⎥
⎢ ⎥
⎢ C33 C34 0 0 ⎥
[C] = ⎢
⎢
⎥, (2.158)
⎢ C44 0 0 ⎥⎥
⎣ sym C55 C56 ⎦
C66

that is:

C15 = C16 = C25 = C26 = C35 = C36 = C45 = C46 = 0. (2.159)

It is then clear that conditions (2.157)1,4 , (2.157)2,4 or (2.157)3,4 imply (2.159) and
vice-versa, which proves the theorem.
This theorem states that the material has a plane of symmetry whose normal is n
if and only if n is the eigenvector of Y and of at least another tensor among V, W or
X.

2.6.1 Physical Interpretations

A physical interpretation of Theorem 3 is possible in the frame of the acoustics

theory, see (Ting 1996, p. 61): tensor X is the acoustic tensor for the elastic waves
that propagate in the direction of n. An elastic wave is a longitudinal wave whenever
n is an eigenvector of X; in such a case, n is called a specific direction of X. It has
been proved by Kolodner that in an anisotropic material there exist always at least
three different specific directions (Kolodner 1966). When n is an eigenvector of Y,
then the wave is transversal, m is the direction of the wave propagation and n is
called the specific axis.
Then conditions (2.157)3,4 , i.e. when n is an eigenvector of X and Y, are equivalent
to say that n is at the same time a specific direction and a specific axis, i.e. it is
simultaneously the direction of propagation of longitudinal waves and the transversal
direction of transversal waves propagating along the direction of m orthogonal to n.
A statical interpretation has also been given by Hayes and Norris. It traduces
the above acoustics conditions into equivalent statical conditions. They have been
resumed in the following
Theorem 4 A material has a plane of symmetry if and only if at least two orthogonal
planes of pure shear exist, sharing a common shear direction which is the normal to
the plane of symmetry.
For the proof of this Theorem, rather articulated, we address the reader to the
original paper (Hayes and Norris 1991).
2.7 Curvilinear Anisotropy 59

2.7 Curvilinear Anisotropy

When in a body there are directions that are not parallel but mechanically equivalent,
then the body possesses a curvilinear anisotropy. It is still possible to write the
Hooke’s law in a rectangular coordinate system, as done until now. However, in
doing so, the components of [C] or [S] are no more constants, but vary with the
position according to the variation of the coordinate directions with respect to the
equivalent directions.
Be {ξ, η, ζ } the coordinate directions of the curvilinear coordinates that coin-
cide with the mechanically equivalent directions. With self-evident meaning of the
symbols, the Hooke’s law can be written in the curvilinear coordinate system as
(Lekhnitskii 1950, p. 64),
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ σξ ξ ⎪ ⎪ C11 C12 C13 C14 C15 C16 ⎪ ⎪ εξ ξ ⎪⎪
⎪
⎪ ⎪
⎪ ⎢ ⎪
⎪ εηη ⎪
⎪
⎪ σ ηη ⎪
⎪ C22 C23 C24 C25 C26 ⎥
⎥⎪
⎪
⎪
⎨ σ ⎬ ⎢⎢
⎪
⎨ ε ⎪
⎬
√ ζ ζ C33 C34 C35 C36 ⎥
⎥ √ ζ ζ
=⎢
⎢ , (2.160)
⎪
⎪ 2σ ⎪
⎪ C44 C45 C46 ⎥
⎥⎪ √2εηζ ⎪
⎪ ⎢ ⎪ ⎪
√ ηζ
⎪
⎪ ⎪ ⎪
⎪
⎪ 2σζ ξ ⎪⎪
⎪
⎣ sym C55 C56 ⎦ ⎪
⎪
⎪ 2ε ⎪ ⎪
⎩ √ ⎭ ⎩ √ ζξ ⎪ ⎭
2σξ η C66 2εξ η

where the Ci j s are constants. In some cases of non homogenous bodies, the Ci j s can
depend upon the coordinates {ξ, η, ζ }. Of course, if some type of elastic symmetry
is present in the body, then some of the Ci j s can be null, as in the ordinary cases of
the elastic syngonies.
A special case of curvilinear anisotropy is that of cylindrical anisotropy: the body
has an axis of symmetry, not necessarily inside the body itself, all the directions
crossing this axis at right angles are equivalent, as well as all the directions parallel
to the axis and the directions orthogonal to the first two directions.
Using a customary set of cylindrical coordinates {r, θ, z}, with z the axis of sym-
metry, then the Hooke’s law can be written as
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ σrr ⎪ ⎪ C11 C12 C13 C14 C15 C16 ⎪ ⎪ εrr ⎪ ⎪
⎪
⎪ ⎪
⎪ ⎢ ⎪
⎪ εθθ ⎪
⎪
⎪ σ θθ ⎪
⎪ C22 C23 C24 C25 C26 ⎥
⎥⎪
⎪
⎪
⎨ σ ⎬ ⎢⎢
⎪
⎨ ε ⎪
⎬
√ C33 C34 C35 C36 ⎥
⎥ √
=⎢
zz zz
⎢ . (2.161)
⎪
⎪ √ θz ⎪
2σ ⎪ ⎢ C44 C45 C46 ⎥
⎥⎪⎪ √2εθ z ⎪
⎪
⎪
⎪ ⎪
⎪ ⎪ ⎪
⎪
⎪ 2σ ⎪
⎪
⎣ sym C55 C56 ⎦ ⎪
⎪ √ 2εzr ⎪
⎪ ⎪
⎪
⎩√ zr
⎭ ⎩ ⎭
2σr θ C66 2εr θ

A special case of cylindrical anisotropy is that of cylindrical orthotropy: each

plane which is radial, tangential or orthogonal to the symmetry axis is a plane of
symmetry. In such a case matrix [C] in Eq. (2.161) is simplified:
60 2 General Anisotropic Elasticity
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ σrr ⎪ ⎪ C11 C12 C13 0 0 0 ⎪
⎪ εrr ⎪ ⎪
⎪
⎪ σθθ ⎪ ⎪ ⎢ ⎪ εθθ ⎪
⎪
⎪ ⎪
⎪ ⎢ C22 C23 0 0 0 ⎥⎥
⎪
⎪
⎪
⎪
⎪
⎪
⎨ σ ⎬ ⎢ ⎥ ⎨ εzz ⎬
√ zz ⎢ C 0 0 0 ⎥ √
=⎢ 33
. (2.162)
⎪
⎪ 2σθ z ⎪
⎪ C44 0 0 ⎥⎥⎪ √2εθ z ⎪
⎪√
⎪ ⎪ ⎢
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ 2σ ⎪ ⎣ sym C55 0 ⎦ ⎪ 2ε ⎪
⎩ √ zr ⎪ ⎭ ⎪
⎩ √ zr ⎪ ⎭
2σr θ C66 2εr θ

It is worth noting that cylindrical orthotropy is not equivalent to transverse isotropy

(that in fact depends upon only 5 constants, not upon 9). Actually, transverse isotropy
is a special case of cylindrical orthotropy, because not only the radial and tangential
directions are equivalent, but all the directions lying in a plane orthogonal to the
symmetry axis are equivalent directions.
Some examples of cylindrical anisotropy are a block of wood with regular yearly
cylindrical layers, or metallic pipes, for their manufacturing process, or a circular
reinforced concrete slab with steel bars disposed radially and circumferentially, a
bicycle wheel, when homogenized, a circular stone arch and so on.
Another, less common, type of curvilinear anisotropy is that of spherical
anisotropy: there is a center of symmetry, not necessarily belonging to the body,
and all the rays emanating from it are equivalent directions. Also, the tangents to
the meridians and to the parallels are equivalent directions too. Using a standard
spherical coordinate systems {ρ, θ, ϕ}, where the directions of the coordinate axes
coincide with the equivalent directions, Eq. (2.160) becomes
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ σρρ ⎪ ⎪ C11 C12 C13 C14 C15 C16 ⎪ ⎪ ερρ ⎪⎪
⎪
⎪ ⎪
⎪ ⎢ ⎪
⎪ εθθ ⎪
⎪
⎪ σ θθ ⎪
⎪ C22 C23 C24 C25 C26 ⎥
⎥⎪
⎪
⎪
⎨ σ ⎬ ⎢⎢
⎪
⎨ ε ⎪
⎬
√ ϕϕ C33 C34 C35 C36 ⎥
⎥ √ ϕϕ
=⎢
⎢ . (2.163)
⎪
⎪ 2σ ⎪
⎪ C44 C45 C46 ⎥
⎥⎪ √2εθϕ ⎪
⎪ ⎢ ⎪ ⎪
√ θϕ
⎪
⎪ ⎪ ⎪
⎪
⎪ 2σϕρ ⎪⎪
⎪
⎣ sym C55 C56 ⎦ ⎪
⎪
⎪ 2ε ⎪ ⎪
⎩ √ ⎭ ⎩ √ ϕρ ⎪⎭
2σρθ C66 2ερθ

The case of spherical orthotropy is get when each meridian and tangential plane
is a plane of symmetry as well as each plane orthogonal to these two planes. Then,
Eq. (2.163) becomes
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎪
⎪ σρρ ⎪⎪ C11 C12 C13 0 0 0 ⎪
⎪ ερρ ⎪⎪
⎪
⎪ σθθ ⎪⎪ ⎢ ⎪ εθθ ⎪
⎪
⎪ ⎪
⎪ ⎢ C22 C23 0 0 0 ⎥ ⎥
⎪
⎪
⎪
⎪
⎪
⎪
⎨ σ ⎬ ⎢ ⎥ ⎨ ε ⎬
√ ϕϕ ⎢ C33 0 0 0 ⎥ √ ϕϕ
=⎢ . (2.164)
⎪
⎪ √2σθϕ ⎪
⎪ ⎢ C44 0 0 ⎥ ⎥⎪⎪ √2εθϕ ⎪
⎪
⎪
⎪ ⎪ ⎪ ⎪
⎪
⎪ 2σ ⎪ ⎪ ⎣ sym C55 0 ⎦ ⎪ ⎪ 2ε ⎪ ⎪
⎩ √ ϕρ ⎪⎭ ⎪
⎩ √ ϕρ ⎪⎭
2σρθ C 66 2ερθ

To remark the difference between isotropy and spherical orthotropy: isotropy is

a special case of spherical orthotropy, because all the directions are equivalent, not
only those emanating from the centre of symmetry. This reduces the number of
independent elastic constants from 9 to only 2.
2.8 Some Examples of Anisotropic Materials 61

Fig. 2.3 Scheme of the x3

frame rotation for tracing the
elastic constants 3D-graphics
x3 x1

x2
0
x2

x1

2.8 Some Examples of Anisotropic Materials

To end this Chapter, we give in this Section some examples of anisotropic materials,
showing the matrix [C] (in GPa) and the 3D-directional diagrams of some of the
technical constants. These last have been obtained as the value get by the constant
on the axis of x1
of a frame {x1
, x2
, x3
} rotated with respect to the frame {x1 , x2 , x3 }
where the matrix [C] is known, see Fig. 2.3. The rotation matrix [R] is obtained
according to Eq. (2.48), with a rotation tensor U that is
⎡ ⎤
sin ϕ cos θ sin ϕ sin θ cos ϕ
U = ⎣ − sin θ cos θ 0 ⎦. (2.165)
− cos ϕ cos θ − cos ϕ sin θ sin ϕ

So, with this choice axis x2 lies always in the horizontal plane.
The compliance matrix [S
] in the rotated frame can be obtained using the inverse
of relations (2.49):

{ε} = [S]{σ } → [R] {ε

} = [S][R] {σ
} →
(2.166)
{ε
} = [R][S][R] {σ
} ⇒ [S
] = [R][S][R] .

This result can be applied to [C] too, and it is the matrix corresponding of Eq. (2.34).
Once the Si j s known, the technical constants can be easily calculated using the results
of Sect. 2.3.
Through Eqs. (2.165) and (2.166) it can be shown that for the materials of the
hexagonal elastic syngony it is always

S14 = S16 = S24 = S26 = S34 = S36 = S45 = S56 = 0. (2.167)

For these materials, the only Chentsov’s and mutual influence coefficients that are
not identically null are μ23,12 , η1,31 , η2,31 , η3,31 , η31,1 , η31,2 and η31,3 .
Different cases are considered below; for each one of them, the directional 3D-
graphics of E 1 , G 12 , ν12 , μ23,12 , η1,31 and η31,1 are traced. For the four last constants,
62 2 General Anisotropic Elasticity

when needed a blue translucent sphere is also traced: it represents the surface where-
on the property traced in the graphics vanishes. So, the part of the 3D-graphics
inside the sphere corresponds to negative values of the property itself. The graphics
show clearly that the Poisson’s, Chentsov’s and mutual influence coefficients can get
negative values. The values of the Ci j s, E 1 and G 12 are in GPa. To remember that
[C] is given in the Kelvin’s notation, Eq. (2.24).
From the graphics below, one can appreciate the extreme variety of forms of
the technical constants. It can be remarked how anisotropy properties change very
quickly for small changes of direction (Fig. 2.9, 2.10, 2.11, 2.12, 2.13 and 2.14).
The case of the hexagonal syngony is very articulated, and it can be shown that
there are as much as 8 possible different profiles of E 1 (ϕ), (Vannucci 2015) (of course
E 1 , like all the other properties, does not depend upon θ , because the hexagonal elastic
syngony is equivalent to transverse isotropy).

• Anorthite (CaAl2 Si2 O8 )

Crystal syngony: Monoclinic, N = 13, plane of symmetry: x2 = 0.
Source: Evans and Grove (2004)

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.4 Directional 3D-graphics of Anorthite

2.8 Some Examples of Anisotropic Materials 63
⎡ ⎤
124 66 50 0 −26.9 0
⎢ 205 42 0 −9.9 0 ⎥
⎢ ⎥
⎢ 156 0 −25.4 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 48 0 −2 ⎥
⎥
⎣ sym 80 0 ⎦
84

• Perovskite (CaTiO3 )
Crystal syngony: Orthorhombic, N = 9.
Source: Evans and Grove (2004)
⎡ ⎤
515 117 117 0 0 0
⎢ 525 139 0 0 0 ⎥
⎢ ⎥
⎢ 435 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 48 0 0 ⎥⎥
⎣ sym 404 0 ⎦
350

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.5 Directional 3D-graphics of Perovskite

64 2 General Anisotropic Elasticity

• Dolomite (CaMg(CO3 )2 ) (*estimated)

Crystal syngony: Trigonal, N = 7.
Source: Bakri and Zaoui (2011)

⎡ ⎤
196.6 64.4 54.7 31.7 25.3∗ 0
⎢ 196.6 54.7 −31.7 −25.3∗ 0 ⎥
⎢ ⎥
⎢ 110 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 83.2 0 −35.84 ⎥
⎥
⎣ sym 83.2 44.8 ⎦
132.2

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.6 Directional 3D-graphics of Dolomite

2.8 Some Examples of Anisotropic Materials 65

• Calcium Tungstate (CaWO4 )

Crystal syngony: Tetragonal, N = 7.
Source: Landolt and Börnstein (1992)

⎡ ⎤
141 61 41 0 0 1.9
⎢ 141 41 0 0 −1.9 ⎥
⎢ ⎥
⎢ 125 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 67.4 0 0 ⎥⎥
⎣ sym 67.4 0 ⎦
81.4

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.7 Directional 3D-graphics of Calcium Tungstate
66 2 General Anisotropic Elasticity

• Quartz (SiO2 )
Crystal syngony: Trigonal, N = 6.
Source: Landolt and Börnstein (1992)

⎡ ⎤
86.8 7.1 14.4 24.3 0 0
⎢ 86.8 14.4 −24.3 0 0 ⎥
⎢ ⎥
⎢ 107.5 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 116.4 0 0 ⎥ ⎥
⎣ sym 116.4 34.4 ⎦
79.7

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.8 Directional 3D-graphics of Quartz

2.8 Some Examples of Anisotropic Materials 67

• Zircon (ZrSiO4 )
Crystal syngony: Tetragonal, N = 6.
Source: Evans and Grove (2004)

⎡ ⎤
424 70 149 0 0 0
⎢ 424 149 0 0 0 ⎥
⎢ ⎥
⎢ 489 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 262 0 0 ⎥⎥
⎣ sym 262 0 ⎦
96

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.9 Directional 3D-graphics of Zircon

68 2 General Anisotropic Elasticity

• Ice (H2 O)
Crystal syngony: Hexagonal, N = 5.
Source: Evans and Grove (2004)

⎡ ⎤
13.5 6.5 6 0 0 0
⎢ 13.56 0 0 0⎥
⎢ ⎥
⎢ 15 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 6 0 0⎥⎥
⎣ sym 6 0⎦
7

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.10 Directional 3D-graphics of Ice

2.8 Some Examples of Anisotropic Materials 69

• Titanium Boride (TiB2 )

Crystal syngony: Hexagonal, N = 5.
Source: Landolt and Börnstein (1992)

⎡ ⎤
648.3 404.2 317.7 0 0 0
⎢ 648.3 317.7 0 0 0 ⎥
⎢ ⎥
⎢ 439.3 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 500 0 0 ⎥ ⎥
⎣ sym 500 0 ⎦
244.1

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.11 Directional 3D-graphics of Titanium Boride

70 2 General Anisotropic Elasticity

• Pine Wood
Transversely isotropic, N = 5.
Source: Lekhnitskii (1950)

⎡ ⎤
0.45 0.11 0.13 0 0 0
⎢ 0.45 0.13 0 0 0 ⎥
⎢ ⎥
⎢ 10.1 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 1.5 0 0 ⎥ ⎥
⎣ sym 1.5 0 ⎦
0.34

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.12 Directional 3D-graphics of Pine Wood

2.8 Some Examples of Anisotropic Materials 71

• Gold (Au)
Crystal syngony: Cubic, N = 3.
Source: Evans and Grove (2004)

⎡ ⎤
191 162 162 0 0 0
⎢ 191 162 0 0 0 ⎥
⎢ ⎥
⎢ 191 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 84 0 0 ⎥⎥
⎣ sym 84 0 ⎦
84

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.13 Directional 3D-graphics of Gold

72 2 General Anisotropic Elasticity

• Diamond (C)
Crystal syngony: Cubic, N = 3.
Source: Evans and Grove (2004)

⎡ ⎤
1079 124 124 0 0 0
⎢ 1079 124 0 0 0 ⎥
⎢ ⎥
⎢ 1079 0 0 0 ⎥
[C] = ⎢
⎢
⎥
⎢ 1156 0 0 ⎥ ⎥
⎣ sym 1156 0 ⎦
1156

(a) E1 (b) G12 (c) ν12

(d) μ 23,12 (e) η 1,31 (f) η 31,1

Fig. 2.14 Directional 3D-graphics of Diamond

References 73

References

Z. Bakri, A. Zaoui, Structural and mechanical properties of dolomite rock under high pressure
conditions: a first-principles study. Phys. Status Solidi 248, 1894–1900 (2011)
S.C. Cowin, Properties of the anisotropic elasticity tensor. Q. J. Mech. Appl. Math. 42, 249–266
(1989)
S.C. Cowin, M.M. Mehrabadi, On the identification of material symmetry for anisotropic elastic
materials. Q. J. Mech. Appl. Math. 40, 451–476 (1987)
J. Evans, T. Grove, Structure of Earth Materials - MIT Open CourseWare (MIT, Boston,
2004), http://ocw.mit.edu/courses/earth-atmospheric-and-planetary-sciences/12-108-structure-
of-earth-materials-fall-2004/lecture-notes/lec20.pdf
G. Green, On the laws of reflexion and refraction of light at the common surface of two non-
crystallized media. Camb. Philos. Soc. Trans. 7 (1839)
M.E. Gurtin, An Introduction to Continuum Mechanics (Academic Press Inc., New York, 1981)
M.A. Hayes, A.N. Norris, Static implications of the existence of a plane of symmetry in an
anisotropic elastic solid. Q. J. Mech. Appl. Math. 45, 141–147 (1991)
F.E. Hohn, Elementary Matrix Algebra (MacMillan, New York, 1958)
R.M. Jones, Mechanics of Composite Materials, 2nd edn. (Taylor & Francis, Philadelphia, 1999)
I.I. Kolodner, Existence of longitudinal waves in anisotropic media. J. Acoust. Soc. Am. 40, 730–731
(1966)
H.H. Landolt, R. Börnstein, Second and Higher Order Elastic Constants, Numerical Data and
Functional Relationships in Science and Technology III, vol. 29/a (Springer, Berlin, 1992)
S.G. Lekhnitskii, in Theory of Elasticity of an Anisotropic Elastic Body. English translation by P.
Fern (Holden-Day, San Francisco, 1963), p. 1950
A.E.H. Love, A Treatise on the Mathematical Theory of Elasticity (Dover, New York, 1944)
M.M. Mehrabadi, S.C. Cowin, Eigentensors of linear anisotropic elastic materials. Q. J. Mech.
Appl. Math. 43, 15–41 (1990)
P. Podio-Guidugli, A primer in elasticity. J. Elast. 58, 1–104 (2000)
W. Thomson Lord Kelvin, Elements of a mathematical theory of elasticity. Philos. Trans. R. Soc.
146, 481–498 (1856)
W. Thomson Lord Kelvin, Mathematical theory of elasticity. Encycl. Br. 7, 819–825 (1878)
T.C.T. Ting, Anisotropic Elasticity (Oxford University Press, Oxford, 1996)
P. Vannucci, A note on the computation of the extrema of Young’s modulus for hexagonal materials:
an approach by planar tensor invariants. Appl. Math. Comput. (2015)
W. Voigt, Lehrbuch der Kristallphysik (B.G. Teubner, Leipzig, 1910)
http://www.springer.com/978-981-10-5438-9