CONTINUUM MECHANICS
Page
Chapter 3 DEFORMATION AND STRAIN
3.1 Particles and Points ................................................ 77
3.2 Coritinuum Configuration . Deformation and Flow Concepts . . . . . . . . . . . . . . 77
3.3 Position Vector . Displacement Vector .................................. 77
3.4 Lagrangian and Eulerian Descriptions .................................. 79
3.5 Deformation Gradients. Displacement Gradients ........................ 80
3.6 Deformation Tensors . Finite Strain Tensors ............................ 81
3.7 Small Deformation Theory . Infinitesimal Strain Tensors ................ 82
3.8 Relative Displacements . Linear Rotation Tensor . Rotation Veztor ........ 83
3.9 Interpretation of the Linear Strain Tensors ............................ 84
3.10 Stretch Ratio . Finite Strain Interpretation ............................ 86
3.11 Stretch Tensors . Rotation Tensor ...................................... 87
3.12 Transformation Properties of Strain Tensors ............................ 88
3.13 Principal Strains. Strain Invariants . Cubical Dilatation ................ 89
3.14 Spherical and Deviator Strain Tensors .................................. 91
3.15 Plane Strain . Mohr's Circles f o r Strain ................................ 91
3.16 Compatibility Equations for Linear Strains .............................. 92
Page
6.7 Airy's Stress Function ................................................ 147
6.8 TwoDimensional Elastostatic Problems in Polar Coordinates .............. 148
6.9 Hyperelasticity . Hypoelasticity ........................................ 149
6.10 Linear Thermoelasticity ................................................ 149
Chapter 7 FLUIDS
7.1 .
Fluid Pressure . Viscous Stress Tensor Barotropic Flow ................ 160
7.2 . .
Constitutive Equations Stokesian Fluids Newtonian Fluids ............ 161
7.3 .
Basic Equations f o r Newtonian Fluids NavierStokesDuhem Equations .... 162
7.4 . .
Steady Flow Hydrostatics Irrotational Flow .......................... 163
7.5 .
Perfect Fluids Bernoulli Equation . Circulation ........................ 164
7.6 .
Potential Flow Plane Potential Flow .................................. 165
Chapter 8 PLASTICITY
8.1 Basic Concepts and Definitions ........................................ 175
8.2 Idealized Plastic Behavior .............................................. 176
8.3 Yield Conditions . Tresca and von Mises Criteria ........................ 177
8.4 . .
Stress Space The 1I.Plane Yield Surfaces ............................ 173
8.5 .
PostYield Behavior Isotropic and Kinematic Hardening ................ 180
8.6 .
Plastic StressStrain Equations Plastic Potential Theory ................ 181
8.7 .
Equivalent Stress Equivalent Plastic Strain Increment .................. 182
8.8 .
Plastic Work Strain Hardening Hypotheses ............................ 182
$8.9 Total Deformation Theory .............................................. 183
8.10 Elastoplastic Problems ................................................ 183
8.11 Elementary Slip Line Theory f o r Plane Plastic Strain .................... 184
Chapter 9 VISCOELASTICITY
9.1 Linear Viscoelastic Behavior ............................................ 196
9.2 Siniple Viscoelastic Models ............................................ 196
9.3 .
Generalized Models Linear Differential Operator Equation .............. 198
9.4 Creep and Relaxation .................................................. 199
9.5 Creep Function . Relaxation Function . Hereditary lntegrals . . . . . . . . . . . . . . 200
9.6 Complex Moduli and Compliances ...................................... 202
9.7 Three Dimensional Theory .............................................. 203
9.8 .
Viscoelastic Stress Analysis Correspondence Principle .................. 204
Chapter 1
Mathematical Foundations
Fig. 11
Fig. 12
CHAP. 11 MATHEMATICAL FOUNDATIONS 3
Multiplication of a vector by a scalar produces in general a new vector having the same
direction a s the original but a different length. Exceptions are multiplication by zero to
produce the nu11 vector, and multiplication by unity which does not change a vector. Multi
plication of the vector b by the scalar m results in one of the three possible cases shown ip
Fig. 13, depending upon the numerical value of m.
O<m<l
Fig. 13
In the important case of a vector multiplied by the reciproca1 of its magnitude, the
result is a unit vector in the direction of the original vector. This relationship is expressed
by the equation A
b = blb (1.7)
Fig. 14
in which O is the angle less than 180" between the vectors a and b, and ê is a unit vector
perpendicular to their plane such t h a t a righthanded rotation about ê through the angle
0 carries a into b. The magnitude of v is equal to the area of the parallelogram having
a and b as adjacent sides, shown shade'd in Fig. 14(b). The cross product is not commutative.
4 MATHEMATICAL FOUNDATIONS [CHAP. 1
The scalar triple product is a dot product of two vectors, one of which is a cross product.
a.(bxc) = ( a x b ) * c= a.(bxc)= h (1.10)
As indicated by (1.10) the dot and cross operation may be interchanged in this product.
Also, since the cross operation must be carried out first, the parentheses are unnecessary
and may be deleted as shown. This product is sometimes written [abc] and called the box
product. The magnitude h of the scalar triple product is equal to the volume of the
parallelepiped having a, b, c as coterminous edges.
The vector triple product is a cross product of two vectors, one of which is itself a
cross product. The following identity is frequently useful in expressing the product of a
crossed into b x c.
a x (b x c) = (a c)b  (a b)c = w (1.11)
From (1.11), the product vector w is observed to lie in the plane of b and c.
which is, however, never unique. In symbolic notation, dyadics are denoted by boldfaced
sansserif letters as above.
If in each dyad of (1.12) the antecedents and consequents are interchanged, the resulting
dyadic is called the conjugate dyadic of D and is written
Dc = blal + bzaz + +b ~ a ~ (1 .I 3)
If each dyad of D in (1.12) is replaced by the dot product of the two vectors, the result is a
scalar known as the scalar o f t h e dyadic D and is written
Ds = a i  b l + azbz + +a ~ * b ~ (1.l4)
If each dyad of D in (1.12) is replaced by the cross product of the two vectors, the result is
called the vector o f t h e dyadic D and is written
Dv = a 1 x b l + a z x b z + +a,vXb~ (1.15)
I t can be shown that Dc, D, and Dv are independent of the representation (1.12).
The indeterminate vector product obeys the distributive laws
+
a(b c) = ab + ac (1.16)
(a + b)c = ac + bc (1.l7)
(a + b)(c + d) = ac + ad + bc + bd (1.18)
and if h and p are any scalars,
(h + p)ab = Aab + pab (1.19)
(ha)b = a(hb) = hab (1 .20)
CHAP. 11 MATHEMATICAL FOUNDATIONS 5
If v is any vector, the dot products v D and D v are the vectors defined respectively by
v ' D = (v.a1)bl+(v.az)bz+ + ( v * ~ N ) ~ N= i.i (1.21)
In (1.21) D is called the postfactor, and in (1.22) it is called the prefactor. Two dyadics D
and E are equal if and only if for every vector v, either
v * D = v.E or D.v = Ev (1.23)
The unit dyadic, or idemfactor I, is the dyadic which can be represented as
+
I = êl& i%, êS3 + (124)
where 21, ê2, ê3 constitute any orthonormal basis for threedimensional Euclidean space
(see Section 1.7). The dyadic I is characterized by the property
1.v = v  I = v (1.25)
for a11 vectors v.
The cross products v x D and D x v are the dyadics defined respectively by
+
v x D = (v X a1)bl (vX az)bz + +
(vX a ~ ) =b ~F (1.26)
DXv = al(b1 X v) + az(bz X v) + +a ~ ( Xbv)~ = G (1.27)
From these definitions, double dot and cross products of dyadics may be readily developed.
Also, some authors use the double dot product defined by
a b  . c d = (b*c)(a.d) = A, ascalar (1.35)
A dyadic D is said to be selfconjugate, or symmetric, if
D = D,
and antiselfconjugate, or antisymmetric, if
Every dyadic may be expressed uniquely as the sum of a symmetric and antisymmetric
dyadic. For the arbitrary dyadic D the decomposition is
D = ?j(D + Dc) + ?j(D Dc) = G + H (1.38)
6 MATHEMATICAL FOUNDATIONS [CHAP. 1
set of unit vectors i, j, k along the coordinate axes as shown in Fig. 15. These base vectors
constitute a righthanded unit vector triad, for which
$ $ A A A C A A h
i x j = k, j x k = i, kxi = j (1.45)
A A A A A A
and i.; j  j = k.k = 1
a*b = (a,?+a,~+a,k)*(b,i+b,j+b~)
+ +
= axbx ayby a,b, (1.49) Fig. 16
+
a x b = (ayb, a,by)i (a,b,  a,b,) j
A
+ (a,b,  aybx)k
A
in which the elements a r e treated as ordinary numbers. The triple scalar product may also
be represented in component form by the determinant
a, a, a,
[abc] = bx b~ b~
C, Cy c,
I n Cartesian component form, the dyad ab is given by
c
+
ab = (a,i +a,; a,k)(b,? b,; b,k) + +
A A
= a,b,i i+ axbyij + a,b,i k
A A
9"
+ ayb,S$ + a , b y +~ ayb,$k
~
+ a,b,kT + a,bykj + a,b,kk
A A
(1.53)
Because of the nine terms involved, (1.53) is known a s the nonion form of the dyad ab.
It is possible to put any dyadic into nonion form. The nonion form of the idemfactor in
A A A
terms of the unit triad i, j , k is given by
A A A A A A
I = ii+jj+kk (1.54)
In addition to the rectangular Cartesian coordinate system already discussed, curvi
linear coordinate systems such as the cylindrical (R, O, z ) and spherical ( r ,O, systems C$)
shown in Fig. 17 below a r e also widely used. Unit triads ( $ R , ês, ê,) and (ê,, 60,ê+) of base
vectors illustrated in the figure a r e associated with these systems. However, the base
vectors here do not all have fixed directions and a r e therefore, in general, functions of
position.
8 MATHEMATICAL FOUNDATIONS [CHAP. 1
+ +
A A
where D = u i v j wk. This demonstrates that any linear vector function f may be
expressed a s a dyadicvector product. In (1.61) the dyadic D serves a s a linear vector
operator which operates on the argument vector b to produce the image vector a.
In the "mixed" form, where both subscripts and superscripts appear, the dot shows that j
is the second index.
Under the rules of indicial notation, a letter index may occur either once or twice in a
given term. When an index occurs unrepeated in a term, that index is understood to take
on the values 1,2,. . .,N where N is a specified integer that determines the range of the
index. Unrepeated indices are known as free indices. The tensorial rank of a given term
is equal to the number of free indices appearing in that term. Also, correctly written
tensor equations have the same letters as free indices in every term.
When an index appears twice in a term, that index is understood to take on a11 the
values of its range, and the resulting terms summed. In this socalled summation conven
tion, repeated indices are often referred to as dummy indices, since their replacement by
any other letter not appearing as a free index does not change the meaning of the term in
which they occur. In general, no index occurs more than twice in a properly written term.
If it is absolutely necessary to use some index more than twice to satisfactorily express a
certain quantity, the summation convention must be suspended.
The number and location of the free indices reveal directly the exact tensorial character
of the quantity expressed in the indicial notation. Tensors of first order are denoted by
kernel letters bearing one free index. Thus the arbitrary vector a is represented by a symbol
having a single subscript or superscript, i.e. in one or the other of the two forms,
The following terms, having only one free index, are also recognized as firstorder tensor
quantities:
aijbj, Fikk, rijkUjZ)k
Secondorder tensors are denoted by symbols having two free indices. Thus the arbitrary
dyadic D will appear in one of the three possible f o m s
In the "mixed" form, the dot shows that j is the second index. Secondorder tensor
quantities may also appear in various f o m s as, for example,
Aijip, B?.jk, 8ifikvk
By a logical continuation of the above scheme, thirdorder tensors are expressed by symbols
with three free indices. Also, a symbol such as X which has no indices attached, represents
a scalar, or tensor of zero order.
In ordinary physical space a basis is composed of three, noncoplanar vectors, and so
any vector in this space is completely specified by its three components. Therefore the
range on the index of ai, which represents a vector in physical threespace, is 1,2,3.
Accordingly the symbol aiis understood to represent the three components a1,az,a3. Also,
ai is sometimes interpreted to represent the ith component of the vector or indeed to rep
resent the vector itself. For a range of three on both indices, the symbol Aij represents
nine components (of the secondorder tensor (dyadic) A). The tensor Aij is often presented
explicitly by giving the nine components in a square array enclosed by large parentheses as
Ai1 A12 A13
Aij = ( 1.62)
10 MATHEMATICAL FOUNDATIONS [CHAP. 1
I n the same way, the components of a firstorder tensor (vector) in threespace may be
displayed explicitly by a row or column arrangement of the form
assign to any point (xl, x2,x3) in the xi system a new set of coordinates (O1, 02,B3) in the Oi
system. The functions Bi relating the two sets of variables (coordinates) are assumed to
be singlevalued, continuous, differentiable functions. The determinant
is called the Jacobian of the transformation. If the Jacobian does not vanish, (1.72)
possesses a unique inverse set of the form
The coordinate systems represented by xi and Bi in (1.72) and (1.75) are completely general
and may be any curvilinear or Cartesian systems.
From (l.72), the differential vector dOi is given by
This equation is a prototype of the equation which defines the class of tensors known as
contravariant vectors. I n general, a set of quantities bi associated with a point P are said
to be the components of a c o n t r a v a ~ a n ttensor of order one if they transform, under a
coordinate transformation, according to the equation
where the partia1 derivatives are evaluated at P. In (1.77), bj are the components of the
tensor in the xj coordinate system, while bti are the components in the Oi system. In general
12 MATHEMATICAL FOUNDATIONS [CHAP. 1
tensor theory, contravariant tensors are recognized by the use of superscripts as indices.
I t is for this reason that the coordinates are labeled xi here rather than Xi, but it must be
noted that it is only the differentials dxi, and not the coordinates themselves, which have
tensor character.
By a logical extension of the tensor concept expressed in (l.77), the definition of con
travariant tensors of order two requires the tensor components to obey the transformation
law
Contravariant tensors of third, fourth and higher orders are defined in a similar manner.
The word contravariant is used above to distinguish that class of tensors from the
class known as covariant tensors. In general tensor theory, covariant tensors are recognized
by the use of subscripts as indices. The prototype of the covariant vector is the partia1
derivative of a scalar function of the coordinates. Thus if + = +(xl, x2,x3) is such a function,
In (1.80), b: are the covariant components in the ei system, bi the components in the xi
system. Secondorder covariant tensors obey the transformation law
where the secondorder tensor g,, = (dxi/dOp)(dxi/d$q) is called the metric tensor, or funda
mental tensor of the space. If Oi represents a rectangular Cartesian system, say the x"
system, then
where S, is the Kronecker delta (see Section 1.13) defined by S,, = O if p # q and S,, = 1
if p = q .
Any system of coordinates for which the squared differential element of distance takes
the form of (1.83) is called a system of homogeneous coordinates. Coordinate transforma
tions between systems of homogeneous coordinates are orthogonal transformations, and
when attention is restricted to such transformations, the tensors so defined are called
Cartesian tensors. In particular, this is the case for transformation laws between systems
of rectangular Cartesian coordinates with a common origin. For Cartesian tensors there
is no distinction between contravariant and covariant components apd therefore it is cus
tomary to use subscripts exclusively in expressions representing Cartesian tensors. As
will be shown next, in the transformation laws defining Cartesian tensors, the partia1
derivatives appearing in general tensor definitions, such as (1.80) and (1.81), are replaced
by constants.
Fig. 19
From this definition of aij, the unit vector along the x: axis is given according to (1.48)
and the summation convention by
An obvious generalization of this equation gives the arbitrary unit base vector as
In component form, the arbitrary vector v shown in Fig. 19 may be expresse(; in the
unprimed system by the equation
v = ,ê,.
2). (1.90)
and in the primed system by
V = vi ei (1.91)
Comparing (1.92) with (1.90) reveals that the vector components in the primed and unprimed
systems are related by the equations
v 3. 
 a,.
%,Vi
I
(1.93)
The expression (1.93) is the transformation law for firstorder Cartesian tensors, and as
such is seen to be a special case of the general form of firstorder tensor transformations,
expressed by (1.80) and (1.77). By interchanging the roles of the primed and unprimed
base vectors in the above development, the inverse of (1.93) is found to be
I t is important to note that in (1.93) the free index on aij appears as the second index. In
(1.94), however, the free index appears as the first index.
By an appropriate choice of dummy indices, (1.93) and (1.94) may be combined to pro
duce the equation
v j = a..
tjazkVk
. (1.95)
Since the vector v is arbitrary, (1.95) must reduce to the identity vj = vj. Therefore the
coefficient aijaik, whose value depends upon the subscripts j and k , must equal 1 or O
according to whether the numerical values of j and Ic are the same or different. The
Kronecker delta. defined b'r
may be used to represent quantities such as aijaik. Thus with the help of the Kronecker delta
the conditions on the coefficient in (1.95) may be written
In expanded form, (1.97) consists of nine equations which are known as the orthogonality
or orthonormality conditions on the direction cosines aij. Finally, (1.93) and (1.94) may also
be combined to produce vi = &ijakjvk from which the orthogonality conditions appear in the
alternative form
aijakj = s i k (1.98)
A linear transformation such as (1.93) or (1.94), whose coefficients satisfy (1.97) or (1.98),
is said to be an orthogonal transformation. Coordinate axes rotations and reflections of
the axes in a coordinate plane both Iead to orthogonal transformations.
CHAP. 11 MATHEMATICAL FOUNDATIONS 15
The Kronecker delta is sometimes called the substitution operator, since, for example,
8ijbj = 8iibi + 8i2b2 f 8i3b3 = bi
and, likewise,
8ijFik = 8ijFik +8 2 j F 2 k $. 83jF3k = F j k
It is clear from this property that the Kronecker delta is the indicial counterpart to the
symbolic idemfactor I, which is given by (1.54).
According to the transformation law (1.94), the dyad UiVj has components in the primed
coordinate system given by
U I V ~ = ( U i p ~ ) ( ~ j q=~ qUipUjqUpVq
) (I.I 01)
In an obvious generalization of (1.101), any secondorder Cartesian tensor Tij obeys the
transf ormation Iaw
TC = aipajqTpq
With the help of the orthogonality conditions i t is a simple calculation to invert (1.102),
thereby giving the transformation rule from primed components to unprimed components:
The transformation laws for first and secondorder Cartesian tensors generalize for an
Nth order Cartesian tensor to
T~&.. . = aipajqalcm. . . Tpqm... (IJ04)
The sum is a tensor of the same order as those added. Note that like indices appear in the
same sequence in each term.
Multiplication of every component of a tensor by a given scalar produces a new tensor of
the same order. For the scalar multiplier A, typical examples written in both indicial
and symbolic notation are
bi = xai or b = Xa (1.106)
As indicated by the above examples, outer products are formed by simply setting down the
factor tensors in juxtaposition. (Note that a dyad is formed from two vectors by this very
procedure.)
16 MATHEMATICAL FOUNDATIONS [CHAP. 1
Contraction of a tensor with respect to two free indices is the operation of assigning
to both indices the same letter subscript, thereby changing these indices to dummy indices.
Contraction produces a tensor having an order two less than the original. Typical examples
of contraction are the following.
( a ) Contractions of Tij and uivj
+ +
Tii = Til TZZ T33
UiVi = UiVi+ + U2V2 U3V3
( b ) Contractions of Eijak
E..
va3. = bi
Eijai = cj
Eiiak = dk
(C) Contractions of EijFkm
EijFim = Gim EijFkk = Pij
EijFki = Hik EijFjm = Qim
EiiFkm = Kkm EijFkj = Rik
An Znner product of two tensors is the result of a contraction, involving one index from
each tensor, performed on the outer product of the two tensors. Severa1 inner products
important to continuum mechanics are listed here for reference, in both the indicial and
symbolic notations.
Outer Product Inner Product
Indicial Notation Symbolic Notation
1. aibj aibi a*b
2. aiEjk aiEik = fk a.E = f
aiEji = hj Ea = h
3. Eij Fkm EijFjm = Gim E.F = G
4. Eij Ekm Eij Ejm = Bim E . E = (E)2
Multiple contractions of fourthorder and higher tensors are sometimes useful. Two
such examples are
1. Eij Fkm contracted to Eij Fij, or E : F
2. EijEkmE,, contracted to Eij Ejm Emq, or ( E ) 3
Since the same box product is given in the form of a determinant by (1.52), it is not sur
prising that the permutation symbol is frequently used to express the value of a 3 X 3
determinant.
I t is worthwhile to note that cijk obeys the tensor transformation law for third order
Cartesian tensors as long as the transformation is a proper one (det aij = 1) such as arises
from a rotation of axes. If the transformation is improper (det aij = I), e.g. a reflection
in one of the coordinate planes whereby a righthanded coordinate system is transformed
into a lefthanded one, a minus sign must be inserted into the transformation law for cijk.
Such tensors are called pseudotensors.
The dual vector of an arbitrary secondorder Cartesian tensor Tij is defined by
2  c <. . ~ T.
v.  k ik (1.110)
which is observed to be the indicial equivalent of T,, the "vector of the dyadic Tu, as defined
by (1.15).
where 4 is the identity matrix, having ones on the principal diagonal and zeros elsewhere,
and so named because of the property
4 c A = CAJ = (1.116)
I t is clear, of course, that 4 is the matrix representation of 6,,, the Kronecker delta, and of I,
the unit dyadic. Any matrix cA for which the condition cAT = cAI is satisfied is called
an orthogonal matrix. Accordingly, if O4 is orthogonal,
As suggested by the fact that any dyadic may be expressed in the nonion form (1.53),
and, equivalently, since the components of a secondorder tensor may be displayed in the
square array (1.62), it proves extremely useful to represent secondorder tensors (dyadics)
by square, 3 x 3 matrices. A firstorder tensor (vector) may be represented by either a
1 x 3 row matrix, or by a 3 x 1 column matrix. Although every Cartesian tensor of order
two or less (dyadics, vectors, scalars) may be represented by a matrix, not every matrix
represents a tensor.
If both matrices in the product 4 % = C are 3 x 3 matrices representing secondorder
tensors, the multiplication is equivalent to the inner product expressed in indicial notation by
where the range is three. Expansion of (1.I18) duplicates the "row by column" multiplica
tion of matrices wherein the elements of the ith row of the prefactor matrix are multiplied
in turn by the elements of the kth column of the postfactor matrix, and these products
summed to give the element in the .ith row and kth column of the product matrix. Severa1
such products occur repeatedly in continuum mechanics and are recorded here in the various
notations for reference and comparison.
where parentheses around the indices denote the symmetric part of Dii, and square brackets
on the indices denote the antisymmetric part.
Since the interchange of indices of a secondorder tensor is equivalent to the interchange
of rows and columns in its matrix representation, a square matrix cA is symmetric if it is
equal to its transpose CAT. Consequently a symmetric 3 x 3 matrix has only six independent
components as illustrated by
Ali Ai2
(1.126)
A23 A33
An antisymmetric matrix is one that equals the negative of its transpose. Consequently
a 3 x 3 antisymmetric matrix CB has zeros on the main diagonal, and therefore only three
independent components as illustrated by
20 MATHEMATICAL FOUNDATIONS [CHAP. 1
Symmetry properties may be extended to tensors of higher order than two. In general,
an arbitrary tensor is said to be symmetric with respect to a pair of indices if the value of
the typical component is unchanged by interchanging these two indices. A tensor is anti
symmetric in a pair of indices if an interchange of these indices leads to a change of sign
without a change of absolute value in the component. Examples of symmetry properties
in higherorder tensors are
(a) Rijkm = Rikjm (symmetric in k and j )
( b ) cijk = c kii (antisymmetric in k and i)
(c) Gijkm = Gjimk (symmetric in i and j; k and m)
(d) Pijk = Pikj = P k j i = pjik (syrnmetric in a11 indices)
Here Tij may be envisioned as a linear vector operator which produces the vector vi conjugate
to the direction ni. If the direction is one for which vi is parallel to ni, the inner product
may be expressed as a scalar multiple of ni. For this case,
and the direction ni is called a principal direction, or principal axis of Tij. With the help
of the identity ni = Sijnj, (1.129) can be put in the form
which represents a system of three equations for the four unknowns, ni and h, associated
with each principal direction. In expanded form, the system to be solved is
Note first that for every A, the trivial solution ni = O satisfies the equations. The purpose
here, however, is to obtain nontrivial solutions. Also, from the homogeneity of the system
(1.131) it follows that no loss of generality is incurred by restricting attention to solutions
for which nini = 1, and this condition is imposed from now on.
For (1.130) or, equivalently, (1.131) to have a nontrivial solution, the determinant of
coefficients must be zero, that is,
]Tij ASijl = O (1.132)
Expansion of this determinant leads to a cubic polynomial in h, namely,
A3  ITX' + IITA IIIT = O (I .133)
which is known as the characteristic equation of Tij, and for which the scalar coefficients,
CHAP. I] MATHEMATICAL FOUNDATIONS 21
. are called the first, second and third invariants, respectively, of Tij. The three roots of the
cubic (1.133), labeled X(I),h ( ~A(,),
, are called the principal values of Tij. For a symmetric
tensor with real components, the principal values are real; and if these values are distinct,
the three principal directions are mutually orthogonal. When referred to principal axes,
both the tensor array and its matrix appear in diagonal form. Thus
If X(I) = h(z), the tensor has a diagonal form which is independent of the choice of X ( I )
and htz) axes, once the principal axis associated with XG) has been established. If a11
principal values are equal, any direction is a principal direction. If the principal values are
ordered, it is customary to write thern as h ( , ) ,h(II),h(,,,) and to display the ordering as in
Ao, > A,,,, > ~ ( I I I , .
For principal axes labeled OxTx2x8, the transformation from Oxixzx3 axes is given by
the elements of the table
$1 x2 x3
2;
aZ1= n(2) aZ2= n(2) (L*, = n(32)
in which nlj) are the direction cosines of the jth principal direction.
A comparison of (1.135) and (1.137) indicates that Tij and a11 its integer powers have the
same principal axes.
Since each of the principal values satisfies (1.133), and because of the diagonal matrix
form of T ngiven by (1.138), the tensor itself will satisfy (1.133). Thus
in which ,4 is the identity rnatrix. This equation is called the HamiltonCayley equation.
Matrix multiplication of each term in (1.139) by I produces the equation,
22 MATHEMATICAL FOUNDATIONS [CHAP. 1
V = êi a
aXi
= eidi
A
(1.I46)
Frequently, partia1 differentiation with respect to the variable xi is represented by the
commasubscript convention as illustrated by the following examples.
avi aTij
(b) 
axi = vi,i (e) ã2* = Ti!, k
avi a2Tij 
(c) = vi,j
ax ( f ) a s r d z , km
From these examples i t is seen t h a t the operator ai produces a tensor of order one higher
if i remains a free index ((a)and (c) above), and a tensor of order one lower if i becomes
a dummy index ((b) above) in the derivative.
Severa1 important differential operators appear often in continuum mechanics and are
given here f o r reference.
taken along the curve from A to B is known as the line .integral of F aIong C. In the indiciaI
notation, (1.151) becomes (z~)s
S,Fidxi J
(zi)* Fidxi (1.I 52)
Stokes' theorem says that the line integral of F taken around a closed'reducible curve
C, as pictured in Fig. 111, may be expressed in terms of an integral over any twosided
surface S which has C as its boundary. Explicitly,
in which 6 is the unit normal on the positive side of S , and dS is the differential element of
surface as shown by the figure. In the indicial notation, (1.153) is written
where 2 is the outward unit normal to the bounding surface S , of the volume V in which
the vector field is defined. . I n the indicial notation, (1.155) is written
i vi,i dV = L vini dS (1.156)
Solved Problems
ALGEBRA OF VECTORS AND DYADICS (Sec. 1.11.8)
Determine in rectangular Cartesian form the
unit vector which is (a) parallel to the vector
+
v = 2? 3 3  6g, ( b ) along the line joining
points P(1,0,3) and Q(O,2 , l ) .
+
( a ) Ivl = v = 1 / ( 2 ) ~ (3)2 (6)2 = 7 +
A
v = VIV = (217)i
A
+ (317)j A
 (617)k
A
u u
[(xzxI)~+ (v2y1)j
A
+( 2 ,  2,) k ] [a? +
b3 cc] +
U
1 +
=  ( a s 2 by2+ cz2 axl  byl  cz,) = X X
= O
U
Z
1.3. +
If r = x? + y 3 z c is the vector extending from the origin to the arbitrary point
+ +
P ( x , y, z) and d = a? b j c k is a constant vector, show that (r  d) r = O is the
vector equation of a sphere.
Expanding the indicated dot product,
A A A A A A
(rd)*r = [(xa)i+(yb)j+(zc)k]*[xi+yj+zk]
= x2+y2+z2axbycz = O
Adding d2/4 = (a2 + b2 + c2)/4 to each side of this equation gives the desired equation
( x  a/2)2 + ( y  b/2)2 + ( z  c/2)2 = (d/2)2
which is the equation of the sphere centered a t d/2 with radius d/2.
1.5. Show that if the vectors a, b and c are linearly dependent, a b x c = O . Check the
linear dependence or independence of the basis
A A A
u = 3i+j2k
A A A
v = 4 i  j k
A A A
w = i 2j +k
The vectors a, b and c are linearly dependent if there exist constants h, p and v, not a11 zero,
such that ha pb + + vc = O. The component scalar equations of this vector equation are
1.6. Show that any dyadic of N terms may be reduced to a dyadic of three terms in a
form having the base vectors êl, ê2,ê3 as (a) antecedents, (b) consequents.
Let D = albl + a2b2+    + aNbN= aibi (i = 1,2, . . .,N ) .
( a ) In terms of base vectors, ai = aliel + azie2 + a3ie3 = a j i e j
A A A A
3 1 3 =
and so D = a2.b. 3 t t = $.c.
~ ê.(a..b.)
3 1 1
with j = 1,2,3.
( b ) Likewise setting bi = bjiêj i t follows that D = aibjiêj = (bjiai)êj= g j ê j where j = 1 , 2 , 3 .
1.7. For the arbitrary dyadic D and vector v, show that D  v = v . D,.
Let D = albl + a2b2+ .   + aNbN. Then
D v = al(b, v) + a,(b2 v ) + . . . + a N ( b N v )
= (V. b,)a, + ( v .b2)a2+ . . + ( v . b N ) a N = v . D,
1.10.
A A
+
If D = a ii b j j + c k k and r is the position vector r = x i
AA AA A
+ y j + xk,
A A
show
that r D r = 1 represents the ellipsoid ax2 by2 cx2 = 1. + +
r  ~ . r= (s?+ y3+z k ) (a??+ b33+ ckk) (s?+ zk)
A A A A A A
= ( z i + y j + z k ) * ( a x i + b y j + c z k ) = as2+by2+cz2 = 1
A A AA AA AA A A
1.11. Forthedyadics D = 3 ? ? + 2 j j  j k + s k k and F = 4 i k  t  6 3 3  3 k j + k k , com
pute and compare the double dot products D :F and D F. ..
From the definition ab : cd = (a c)(b d) i t is seen that D : F = 12 + 5 = 17. Also, from
ab cd = (b e)(a d) i t follows that D F = 12 3 5 = 20. + +
1.12. Determine the dyadics G = D . F and H = F * D if D and F are the dyadics given in
Problem 1.11.
From the definition ab cd = (b c)ad, .
O = (3?? + 2;:  3 k + 5 k k ) . (4;: + :6: 3k3 +kk)
AA AA A A A A AA A A
= 12ik+12j j + 3 j j  jk 15kj+5kk
AA AA AA AA AA AA AA
Similarly, H = (4ik+6jj3kj+kk).(3ii+2jjjk+5kk)
AA
= 20ik+123!63k.6k3+8kk
1.13. Show A A
directly from
A
theAnonion form
A
ofA the dyadic D that D = (Do?)? + (D?)? +
( D 0 k ) k and also i w D . i = Dz,, i  D  j = D,,, etc.
Writing D in nonion form and regrouping terms,
II = (D,,i + D,,j + D,,k) i + (D,, i + D,,; + Dryk)j + (Dxzi + D,,? + D,$);
A A A A A A A A
= d l i + d 2 j +d,k = (o?)?+ ( D * 3 ) 3 + ( ~  2 ) k
A A A
Also now
A A e A
i  d l = i * ( ~  i =) i . ( D , , i + D Y , ~ +D,,k)
A 6 A
= D,,
A A
6
j * d l = ; . ~ . i = D,,, j * d 2 = 3 . ~ . j = D,,, etc.
1.14. For an antisymmetric dyadic A and the arbitrary vector b, show that 2 b . A = A, x b.
From Problem 1.6(a), A = &cl + &c2 + $,c3; and because it is antisymmetric, 2A = (A  A,)
or
( elel + e2c2+ e3c3  elel  c2e2  c3e3)
A A A A A A
2A =
= (êlcl  elel
A
+ e2c2 c2e2 + esc3  c3e3)
A A A A
and so .
2b A = [(b $,)cl  (b c,) ê,] . + [(b.ê2)c2 (b . ê2]+ [(b ê3)c3 C,)  (b c3)$31
= [ ( ê , x c 1 ) x b + ( ê z X c 2 ) X b+ ( ê 3 X ~ 3 ) X b ]= (AvXb)
1.15.
that D . ( u x v ) = ( D x u ) . ~ .
A A A A A
Since u X v = ( 2 i + k ) X 5 j = 1 0 k  5 i ,
AA AA A A A A
D * ( U X V=
) (6ii+3ij+4kk)(5i+lOk) = 30?+40k
A A AA AA AA AA
Next, D ~ =U ( 6 ? ? + 3 ? ; + 4 k k ) ~ ( 2 i + k ) = 6ik+8kj6ij+3ii
AA AA AA AA A A A
snd (DXu)*v = ( 3 i i  6 i j  6 i k + 8 k j ) . 5 j = 30i+40k
CHAP. 11 MATHEMATICAL FOUNDATIONS 27
Fig. 114
1.17. Determine the dyadic D which serves as a 'inea? vec^r operator for the vector
function a = f(b) = b + b x r where r = x i y j xk and b is a constant vector. + +
In accordance with (1.59) and (1.60), construct the vectors
u = f(i) = i
A A
+ iXr
A A
= i zj+yk
A A
v = f(j) = j
A A
+ j Xr
A
= xi
A
+j A
xk
A
A A A A A A
w = f(k) = k + k X r =  y i + x j + k
A A A A
Then ü = ui+vj
A A
+ wk A A
= (i zj+yk)i+(xi
A A A A A A
t j xk)j +(yi+xj+k)k
A
AS a check the same result may be obtained by direct expansion of the vector function,
a = b +bXr A
= b,i ib,3 + b,k + (b,z  b,y) i + (b,x  bxz) j + (bxy b,x)k
A A A A
A
e, = ( sin e) i + (cos e) j
A A
1 k
A
e+ x ê, = cos g cos e cos g sin e  sin @
 sin e cos e O
= (sin @ cos e ) i
A
+ (sin g sin e) j
A
+ [(cosze + sin2 e) cos g] k A
= 2,.
1.19. Resolve the dyadic D = 3;: + 4:: + 6 3 1+ 7;: + 10 ko;f 2 k3 into its symmetric
and antisymmetric parts.
Let D = E +F where E = E, and F = F,. Then
E +
= (1/2)(D D,) = (1/2)(6 i i
A A
+ 4 i k + 4 k i + 6 j i + 6 i j + 14 j j
A A A A A A A A A A
+ ioct + i o t k + 2 2 3 + 2;;)
A A A A A A A A A A A A A A A A
= 3ii+3ij+7ik+3ji+7jj+jk+7ki+kj = E,
A A A A A A A A A A A A A A A A
F = (1/2)(~o,) = ( 1 / 2 ) ( 4 i k  4 k i + 6 j i  6 i j + 1 0 k i  10ik+2kj2jk)
A A A A A A A A A A A A
= 3i j  3 i k + 3 j i  jk+3ki+kj = F,
28 MATHEMATICAL FOUNDATIONS [CHAP. 1
1.20. With respect to the set of base vectors ai, a2, a3 (not necessarily unit vectors), the set
a', a2, a3 is said to be a reciprocal basis if ai a3 = aij. Determine the necessary rela
tionships for constructing the reciprocal base vectors and carry out the calculations
for the basis
bs = ? + $ + %
A A A
bl = 3 i + 4 3 , bz =  i + 2 ? + 2 k ,
By definition, al a1 = 1, a, a1 = O, a3 a1 = O. Hence a1 is perpendicular to both a, and
a,. Therefore i t is parallel to a, X a,, i.e. a1 = X(a2 X a3). Since al a1 = 1, al Xa2 X a, = 1 and
h = l / ( a l a, X a,) = l/[ala2a3]. Thus, in general,
122. Evaluate the following expressions involving the Kronecker delta aij for a range of
three on the indices.
( a ) Sii = S l l + Sz2 + = 3
(b) SijSij = S l j S l j + 62jS2j + S3jS3j = 3
(C) S i j S i k S j k = S l j s l k 8 j k + 82j62k8jk + 83j83k8jk = 3
( d ) Gijsjk = s i l S l k + Si2S2k + 8i383k = Sik
(e) 6ifiik = S1#11, + 82jA2k + 83jA3k = Ajk
1.23. For the permutation symbol Cijk S ~ O Wby direct expansion that (a) Cijk'kij = 6,
( b ) ' i j k a j a k = 0.
(a) First sum on i, i ~ k i..
€.. j
= €ljkCklj + ' 2 j k e k 2 j + '3jkek3j
CHAP. 11 MATHEMATICAL FOUNDATIONS 29
1.24. Determine the component f2 for the vector expressions given below.
fi 'ijkTjk
(C) fi = Bijf;
1.25. Expand and simplify where possible the expression Dijxixj for (a) Dij = Dji,
(b) D i j =  D j i .
Expanding, Dijxixj = Dljx1xj + D2jx2xj+ Dsjx3xj
= D11x1x1 + D 1 2 x 1 x 2 + 0 1 3 x 1 5 3 + D 2 1 ~ 2 ~+1D 2 2 ~ 2 ~ 2
+ D23x2x3 + D31x3x1 + D32x3x2 + D33x3x3
(a) Dijxixj = D11(x1)2 + D22(x2)2 + D33(x3)2 + 2D12x1x2+ 2D23x2x3+ 2D13x1x3
( b ) Dijxixj = O since Dll = Dll, D12 = D2,, etc.
+
~pqibi = & ~ ~ q i ~ i j k=~ j&k( a p j a q k  SpkSqj)Bjk = & ( B p q B q p ) = &(Bpq B p q ) = B p q
Fig. 116
Hence
ax1
  ax1  8x1 
 sin O , cos 8 ,   e , cos 0 , cos e3   e, sin e, sin e,
ao1 as2 863
8x2
  8x2  8x2 
 sin e2 sin e,   0, cos O , sin e,   6, sin e2 cos e,
861 302 as3
8x3
  8x3
 el sin e,  = o
as2 863
f r o m which g,,
axi axi
= 
as, asl

 sinz e, cos2 8 , + sin2 o, sin2 e, + cos2 e, = 1
axi axi
g33 =
 
ao3 ae,  eSsin2 e, sin2 e, + eSsin2 6 , cos2 e , = eSsin2 O ,
1.31. If the angle between the line elements represented by (doi, O, O) and (O, don,O) is
812
denoted by p,,, show that cosp,, =
66.
Let ds, = 6de, be the length of line element represented by (del, O , 0 ) and dsz = G d 6 2
axi
be that of ( O , de,, O). Write (1.85) a s dxi = dek, and since (A),= cos plz h,dsz,
aek
ax, a%,
  de, de,
ae, ae,
+ as,

ax,
ae, ae2
ax, ax3
 de, de2 i  de1 de2
ae, as,
= giz de1 dez
51 22 x3
X; cos e sin e O
xó sin e cos 6 O
x3 o o 1
cos 6 sin 6 O
A = sin e cos 6 O
Fig. 118
v j = a3jvj = V,
The unit vector along the xí axis is given by the first row of the table a s $I
= (315)ê1 (415)ê 2 .
Also from the table 2; = ê3. For a righthanded primed system 23 = Xêi, or Ae t, =
[ ( 3 / 5 ) ê 1 (4/5)$2]x 2, = (3/5)&  (416)$~and the third row is 1 1
x: 415 1 3/5 1 O I.
1.34. Let the angles between the primed and
unprimed coordinate directions be given
by the table shown on the right. Deter
mine the transformation coefficients aij
and show that the orthogonality condi
tions are satisfied.
The coefficients aij are direction cosines and
may be calculated directly from the table. Thus
 1 112 112
aij  o llfi
 1 112 112
The orthogonality conditions aiiaik = S i , require:
1.35. Show that the sum x A +~pBij ~ represents the components of a secondorder tensor
if Aij and Bij are known secondorder tensors.
By (1.108) and the statement of the problem, Aij = a p i a q j ~ p and
q Bii = apiaqjBiq. Hence
XAij + @Bii = ~ ( ~ ~ ~ ~ ~ = k a p~i a ,)i ( ~ ~+
+ ay(apia,iBg,) k qyBP,)
which demonstrates that the sum transforms a s a secondorder Cartesian tensor.
1.36. S ~ O that
W (Pijk + Pjki + Pjik) XiXjXk = 3Pijk xixjxk.
Since a11 indices are dummy indices and the order of the variables xi is unimportant, each
tem of the sum is equivalent to the others. This may be readily shown by introducing new dummy
variables. Thus replacing i, j, k in the second and third terms by p, q , r , the sum becomes
Pijkxixjxk + Pqrpxpxqxr + Pqprxpxqx,
Now change dummy indices in these same terms again so that the form is
Pijkxixixk + Pijk + Pijkxjxixk = 3Pijkxixjxk
1.38. Show t h a t the quadratic form Dijxixjis unchanged if Dij is replaced by its symmetric
part D(ij).
Resolving Dij into its symmetric and antisymmetric parts,
Dij = D(;j) + DLijl = S ( D i j+ Dji) + &Dij Dji)
This result may also be obtained by direct expansion of the determinant. An equivalent expression
for the determinant is eijkAilAj2Ak3.
1.41. If the vector vi is given in terms of base vectors a, b, c by vi = aiai + pbi + yCi,
34 MATHEMATICAL FOUNDATIONS [CHAP. 1
By Cramer's rule, a =
I v3 b3 c3 I and by (1.52) and (1.I 0 9 ) , a = eiikvibjck
a1 bl c1 'pqrapbq~r'
a2 b2 c2
EijkaivjCk cijkaibjvk
Likewise /3 =
epqrapbqcr = ~pqrapbqcr'
.
Let D b = w; then
O 5 6 10
1.43. Determine the principal directions and principal values of the secondorder Cartesian
tensor T whose matrix representation is
3 1 o
[Tij] = 1 3 O
O 0 1
From (1.132), for principal values h,
1.44. Show that the principal axes determined in Problem 1.43 form a righthanded set of
orthogonal axes.
Orthogonality requires that the conditions aijaik= 6 j k be satisfied. Since the condition
nini = 1 was used in determining the aii, orthogonality is automatically satisfied for j = k.
Multiplying the corresponding elements of á n y row (column) by those of any other row (column)
+
and adding the products demonstrates that the conditions for j k are satisfied by the solution
in Problem 1.43.
Finally for the system to be righthanded,
~ ( 3 =
A
n(2) ) n ( l ) . ~h~~
A A A
e1 e2 e3
i i o = (4+ 4)ê3 A
= e3
116 1 o
As indicated by the plus and minus values of
aij in Problem 1.43, there are two sets of prin
cipal axes, x: and x:*. As shown by the sketch
both sets a r e along the principal directions
with x: being a righthanded system, x;" a
lefthanded system. Fig. 119
1.45. Show that the matrix of the tensor Tij of Problem 1.43 may be put into diagonal
(principal) form by the transformation law T: = a;,ajqTPq, (or in matrix symbols
T* = cATcAT).
36 MATHEMATICAL FOUNDATIONS [CHAP. 1
Since Tij is symmetric, the dummy indices i and j may be interchanged on the lefthand side of
the second of these equations and that equation subtracted from the first to yield
(A,,,  ~ ( ~ , ) n : ~ ' n=: ~O'
Since h(,, Z h(,,, their difference is not zero. Hence n:2'n:3' = 0 , the condition for the two
directions to be perpendicular.
1.47. Compute the principal values of ( T ) 2 of Problem 1.43 and verify that its principal
axes coincide with those of T.
$.
2
9n;" = O
= n21' = 0' n(l)
3
=21
6n(2)  Cn(2) = O
1 2
For A ( 2 , = 4, Gn?' + 6nk2' = 0 = ni2'= k l f i , nF) = 0
Sn'2' 
3  0
For = 16,
15ni3' = O
which are the same a s the principal directions of 1.
1.48. Use the fact that (T)2 has the same principal directions as the symmetrical tensor T
to obtain fiwhen
First, the principal values and principal directions of T are determined. Following the procedure
of Problem 1.43, the diagonal form of T is given by
CHAP. 11 MATHEMATICAL FOUNDATIONS 37
[aijl = [ llfi
0
216
lldz
llfi
llh
11,h
Therefore fi = ;" ó 3 o o
and using [aij; to relate this to the original axes by the
transformation fi = ~ A,~ f i
the matrix equation is
Consider  =
ax axi
A.. X . + A,.%.a%,
 Since
axi
z Sik, i t i~ seen that 
ax = A k j z j+ Aikzi =
axk %3axk 3 a> 'axk ' Jxk Jxk
(A + A jk)xj. a2A 8%
Continuing the differentiation,  ( A k j + A . ) 

 A k P + A P k .If A 23. . = A3.i 9
ax, axk lk axp
 = 2Akjxj and  2APk.
axk as, axk
1.52. If is a secondorder Cartesian tensor, show that its derivative with respect to
Aij
xk, namely Aij,k, is a thirdorder Cartesian tensor.
F o r the Cartesian coordinate systems xi and xl, xi = ajixi and âxildxi = ajt. Hence
i hbini d S = i rijkhvk,jni d s
MISCELLANEOUS PROBLEMS
1.56. For the arbitrary vectors a and b, show that
A = ( a x b) (a x b) + (a.b)2 = ( ~ b ) ~
Interchange the dot and cross in the first term. Then
h = a  b x (axb) + (a.b)(a0b)
= a [(b. b)a  (b.a)b] + ( a . b)(a0b)
= ( a . a ) ( b  b )  ( b . a ) ( a W b+) (a.b)(a.b)
= (ab)2
d
1.57. If ii = o x u and V = w x v, show that 
dt
(u X v) = ca X (u x v).
(a) In symbolic notation,
d
z(uxv) = h X v + ux; = (mXu)Xv+ uX(mXv)
= (V. m)u  (V U)OI + (U v)a  (U @)V
d
( b ) I n indicial notation, let (u X V) = W. Then
wi =
d
( E ~ ~ ~=u ~eijkGjwk
v ~ ) + eijkuj;k
Let the determinant of Aij be given by det A = AZl Az2 A23 . An interchange of rows
or columns causes a sign change. Thus
A31 A32 A33
or column changes,
A,, A,, A I S

42, 42, 4%  epqS det A
A,, 43, A,,
Hence f o r a n arbitrary row and column interchange sequence,
1.59. Use the results of Problem 1.58 to prove (a) cpqscsnr = S P n S ~ r Spraqn, ( b ) ~ P , S ~ S Q=~  2 6 ~ ~ .
Expanding the determinant of the identity in Problem 1 . 5 8 ,

'pqs'mnr  S m p ( S n q S r s  s n s a r q ) +
Omq(SnsSrp  Onpsrs) + Sms(SnpSrq  SnqSrp)
( a ) Identifying s with m yields,
'pqs'snr = Ssp(8nqsrs  SnsSrq) f S s q ( 8 n s s r p  SnpSrs) + S s s ( S n p ~ r q SnqSrp)


srpsnq  SpnSrq + SqnSrp  SnpSqr + 3SnpSrq  3Snqsrp

 S n p a r q  SnqSrp
( b ) Identifying q with n in ( a ) ,

~pqs'sqr  Sqparq  aqqsrp = S p r  3 S p r = 2Spr
By the HamiltonCayley theorem the tensor satisfies its own characteristic equation. Hence
+
 2 ( ~ ) 6~8 91 = 0, and multiplying this equation by B yields = 2(B)3 6 ( 8 ) 2  9 8 01 +
+
(B)4 = 1 0 ( B ) z 38  181. Hence
CHAP. 11 MATHEMATICAL FOUNDATIONS 41
(23)4 =
O 26
1.62. Prove that (a) Aii, (b) AijAij, ( c ) c i j k ~ k j p A t pare invariant under the coordinate trans
formation represented by (1.103), i.e. show t h a t Aii = ~ : i ,etc.
(a) By (1.103), Aij = aPiaqjAP,. Hence Aii = a,,aqiA;, = S,,A:, = A:, = Aii.
(b) AijAij = a p i a q j A ~ q a m i a n j =
A ~ nB ~ ~ s ~ ~ A= PApqApq
~ A & =~ A!.A!.
21 13
1.63. Show that the dual vector of the arbitrary tensor Tij depends only upon T ~ i j ibut that
the product TijSijof Tij with the symmetric tensor Sij is independent of Triji.
) eijk T L j k l since
By (1.110) the dual vector of Tij is vi = cijkTjk, or vi = eijk(TCjk)T c j k l = +
= O (cijk is antisymmetric in j and k, T ( j k ,symmetric in j and k).
cijkT(jk)
For the product T i j S i j= TCii>Sij TLij1 +
Si? Here TLii,Sii = O and T,Sij = TCij,S,.
1.65. Use the indicial notation to prove the vector identity V x (a x b) = b Va  b(V a) +
a ( V * b ) a  V b .
Let V X (a x b ) = v; then v, = ~ ~ , ~ c ~or~ ~ a ~ a ~ b ~
Vp = 'pqi'ijk (ajbk),q = epqi'ijk(aj,qbk ajbk,q)
= (Spjsqk  Spksqj)(aj,q bk + ~ j ~ k , q=) bq  aq,q bp +  aqbp,q
Hence v = b V a  b ( V a)  + a(V b)  a Vb.
1.67. For the reflection of axes shown in Fig. 121 show that the transformation is
orthogonal.
From the figure the transformation matrix is
i: :l[ui 01
L 0 0 1 ] L 0 0 1 J
r: : 01
L 0 0 1 J Fig. 121
Supplementary Problems
1.69.
A
+ A A A A
Show t h a t u = i j  k and v = i  j a r e perpendicular to one another.
h
Determine w so
t h a t u, v, & forms a righthanded triad. Ans. w = (  l / f i ) ( ? + 2c) 3+
1.70. Determine the transformation rnatrix between the u,v,W axes of Problem 1.69 and the coordinate
directions.
1.71. Use indicial notation to prove ( a ) V x = 3, (b) V X x = O, (c) a . V x = a where x is the position
vector and a is a constant vector.
1.72. Determine the principal values of the symmetric p a r t of the tensor Tij =
Ans. h ( , , = 15, h ( Z )= 5 , h,,) = 10 3 18
CHAP. 11 MATHEMATICAL FOUNDATIONS 43
1.73. F o r the symmetric tensor Tij = determine the principal values and the directions
of the principal axes.
Ans. h c l , = 2 , A ( z , = 7, A,,, = 1 2 ,
1.74. Given the arbitrary vector v and a n y unit vector ê, show t h a t v may be resolved into a component
parallel and a component perpendicular to 2, i.e. v = ( v * ê ) ê ê X ( v X 3 ) . +
1.75. If .
V v = O, V x v = 6 and V X w = c, show t h a t V2v = y.
1.76. Check the result of Problem 1.48 by direct rnultiplication to show t h a t fifi = T.
*(&+I) #&I) 0
1.79. Verify t h a t (a) O3,,vp= v3, ( b ) 8siAji = Aj3, (C) Sijeijk = 0 , ( d ) ai2aj3Aij = A23.
h
1.81. Show t h a t the volume V enclosed by the surface S may be given a s V = Q
x ia the ponition vector and n the unit normal to the surfaee. Hint: Write V = (116)
and use (1.157). '
1 S
(xixi),inidS
Analysis of Stress
Adofion
 
of t_hecontinuum viewpoint
 
material behavior means that field
as the
 basis for the mathematical
__ _ _ such as stress
_ quantities and
 . . 
displacement
description of
are expressed
 __r _ 
functions of the space coordinates and time.
The average force per unit area on AS is given by Af,lAS. The Cauchy stress principie
asserts that this ratio Afi/As tends to a definite limit d f i / d S as AS approaches zero a t the
point P, .while a t the same time the moment of afiabout the point P vanishes
  inthgimiting
__h
process. The resulting vector d f i l d S (force per unit area) is called the stress vector ti"'
and i s shown in Fig. 23. If the moment a t P were not to vanish in the limiting process,
a couplestress vector, shown by the doubleheaded arrow in Fig. 23, would also be defined
a t the point. One branch of the theory of elasticity considers such couple stresses but
they are not considered in this text.
46 ANALYSIS O F STRESS [CHAP. 2
V upon the material outside is the vector t:"'. Thus by Newton's law of action and
reaction,
=  t' 
 "'
h
(2.5)
The stress v e c t o r is very often referred to as the t r a c t i o n v e c t o r .
vector t:"' with each unit normal vector ni, representing the orientation of a n infinitesimal
surface element having P a s a n interior point. This is illustrated in Fig. 23. The
totality of a11 possible pairs of such vectors tiE;'and ni a t P defines the s t a t e of stress a t that
point. Fortunately it is not necessary to specify every pair of stress and normal vectors
to completely describe the state of stress a t a given point. This may be accomplished by
giving the stress vector on each of three mutually perpendicular planes a t P. Coordinate
transformation equations then serve to relate the stress vector on any other plane a t the
point to the given three.
Adopting planes perpendicular to the coordinate axes f o r the purpose of specifying
the state of stress a t a point, the appropriate stress and normal vectors are shown in
Fig. 24.
Fig. 24
For convenience, the three separate diagrams in Fig. 24 are often combined into a
single schematic representation a s shown in Fig. 25 below.
Each of the three coordinateplane stress vectors may be written according to (1.69) in
terms of its Cartesian components as
CHAP. 21 ANALYSIS O F STRESS 47
are the components of a secondorder Cartesian tensor known as the stress tensor. The
equivalent stress dyadic is designated by Z, so that explicit component and matrix rep
resentations of the stress tensor, respectively, take the forms
Pictorially, the stress tensor components may be displayed with reference to the
coordinate planes as shown in Fig. 26. The components perpendicular to the planes
(all,a2,, a,,) are called normal stresses. Those acting in (tangent to) the planes (al2,a,,, a,,,
a,,, os?,o,,) are called shear stresses. A stress component is positive when it acts in the
pgsitive direction of the coordinate axes, and on a plane whose outer normal points in one
of the positive coordinate directions. The component uii acts in the direction of the jth
coordinate axis and on the plane whose outward normal is parallel to the ith coordinate
axis. The stress components shown in Fig. 26 are a11 positive.
The average traction vectors t:'&' on the faces and t:':' on the base, together with the
average body forces (including inertia forces, if present), acting on the tetrahedron are
shown in the figure. Equilibrium of forces on the tetrahedron requires that
tf(.) d S  tf'"' d s l  tf'"' d s 2  tf'"' d s 3 + pbf d V = 0 (2.10)
If now the linear dimensions of the tetrahedron are reduced in a constant ratio to one
another, the body forces, being an order higher in the small dimensions, tend to zero more
rapidly than the surface forces. At the same time, the average stress vectors approach
the specific values appropriate to the designated directions a t P. Therefore by this limiting
process and the substitution (2.9), equation (2.10) reduces to
Cancelling the common factor d S and using the identity tje" =.u j i , (2.11) becomes
Since the volume V is arbitrary, the integrand in (2.17) must vanish, so that
+ pbi =
ajijj O or V I + pb = O (2.18)
which are called the equilibrium equations.
In the absence of distributed moments or couplestresses, the equilibrium of moments
about the origin requires that
in which xi is the position vector of the elements of surface and volume. Again, making
h
the substitution tin' = ujini, applying the theorem of Gauss and using the result expressed
in (2.18), the integrais of (2.19) are combined and reduced to
which shows that the stress tensor is szjmmetric. In view of (2.22), the equilibrium equations
(2.18) are often written
+ ~ O
o ~ ~p b i, =
Fig. 29
50 ANALYSIS OF STRESS [CHAP. 2
According to the transformation law for Cartesian tensors of order one (1.93), the
A
components of the stress vector tln)referred to the unprimed axes are related to the primed
A
Likewise, by the transformation law (1.102) for secondorder Cartesian tensors, the stress
tensor components in the two systems are related by
O; = aiPajqap, or 1' = A . X A, (2.27)
and
is called the stress quadric of Cauchy. From this definition i t follows that the magnitude
O , of the normal stress component on the surface element dS perpendicular to the position
In the three equations (2.36), there are four unknowns, namely, the three direction cosines
ni and the principal stress value O.
For solutions of (2.36) other than the trivial one nj = 0, the determinant of coefficients,
vanish. Explicitly,
luij  Sijal, must
lu..
11
 S..
7 1
I ~
= 0 or a~~ u22  u23 = 0 (2.,?7)
as 1 a32 O33 
11, = +(aiiujj
 a.21.c..)
li
are known respectively as the first, second and third stress invariants.
The three roots of (2.38), a,,,, a,,,, a(,) are the three principal stress values. Associated
with each principal stress a,,,, there is a principal stress direction for which the direction
cosines n,'k'are solutions of the equations
CHAP. 21 ANALYSIS O F STRESS 53
The maximum and minimum values of U, may be obtained from (2.50) by the method of
Lagrangian rnultipliers. The procedure is to construct the function
F = U:  Anini (2.51)
in which the scalar A is called a Lagrangian multiplier. Equation (2.51) is clearly a function
of the direction cosines ni, so that the conditions for stationary (maximum or minimum)
values of F are given by JFldni = O. Setting these partials equal to zero yields the
equations
+ +
n2{uTI  2uI1(uIn; ~ , , n ; ~,,,n;) + A} = O (2.523)
n3{uSII 2 ~ , , , ( ~ , n+; ~ , , n +
; ~,,,n;)+ A} = O (2.52~)
which, together with the condition nini = 1, may be solved for A and the direction cosines
nl, ne, na, conjugate to the extremum values of shear stress.
One set of solutions to (2.52), and the associated shear stresses from (2.50), are
n1=&1, nz=O, n3=0; forwhich~,=O (2.53~)
nl = O, na = 11, n3 = 0; for which a, =O (2.533)
ni = O, nn = O, na = ? I ; for which U, =O (2.53~)
The shear stress values in (2.53) are obviously minimum values. Furthermore, since (2.35)
indicates that shear components vanish on principal planes, the directions given by (2.53)
are recognized as principal stress directions.
A second set of solutions to (2.52) may be verified to be given by
nl = 0, n2 = 
+ I , n3 = +l/fi;
 for which U, = (a,,  uIII)/2 (2.54a)
ni = *l/fi, n z = 0, n3 = *1/fi; for which a, = (oIII uI)/2 (2.54b)
nl = 11/fi, nz = e l l f i , n3 = 0; for which U, = (aI  uII)/2 (2.54c)
Equation (2.54b) gives the maximum shear stress value, which is equal to half the difference
of the largest and smallest principal stresses. Also from (2.543), the maximum shear stress
component acts in the plane which bisects the right angle between the directions of the
xhaximum and minimum principal stresses.
54 ANALYSIS O F STRESS [CHAP. 2
Combining these two expressions with the identity nini = 1 and solving for the direction
cosines ni. results in the eauations
These equations serve as the basis for Mohr's stress circles, shown in the "stress plane" of
Fig. 215, for which the U , axis is the abscissa, and the U , axis is the ordinate.
In (2.58a), since U ,  u,, > O and U ,  u I I 1> O from (2.55), and since ( n ~ )is
' nonnegative,
the numerator of the righthand side satisfies the relationship
which represents stress points in the (V,, os) plane that are on or exterior to the circle
Fig. 215
56 ANALYSIS OF STRESS [CHAP. 2
represents the components and O, of the stress vector tin) on the plane having the normal
direction ni a t Q in Fig. 216.
Fig. 217
=O
Fig. 218
CHAP. 21 ANALYSIS OF STRESS 57
If the principal stresses are not ordered and the direction of the zero principal stress
is taken as the X Q direction, the state of stress is termed plane stress parallel to the xlxz
plane. For arbitrary choice of orientation of the orthogonal axes X I and xz in this case,
the stress matrix has the form
ull @12 0
(2.65)
The stress quadric for this plane stress is a cylinder with its base lying in the XIXZ plane
and having the equation
u,,xS + +
2 u l z ~ , ~ ,u,,x~ = Ck2 (2.66)
Frequently in elementary books on Strength of Materials a state of plane stress is rep
resented by a single Mohr's circle. As seen from Fig. 218 this representation is necessarily
incomplete since a11 three circles are required to show the complete stress picture. In
particular, the maximum shear stress value a t a point will not be given if the single circle
presented happens to be one of the inner circles of Fig. 218. A single circle Mohr's diagram
is able, however, to display the stress points for a11 those planes a t the point P which include
the zero principal stress axis. For such planes, if the coordinate axes are chosen in
accordance with the stress representation given in (2.65), the single plane stress Mohr's
circle has the equation
The essential features in the construction of this circle are illustrated in Fig. 219. The
circle is drawn by locating the center C a t = (all + O,,)/:! and using the radius
+
R = j/[(u1,  uzz)/2]2 given in (2.67). Point A on the circle represents the stress
state on the surface element whose normal is nl (the righthand face of the rectangular
parallelepiped shown in Fig. 219). Point B on the circle represents the stress state on
the top surface of the parallelepiped with normal nz. Principal stress points U , and a,, are
so labeled, and points E and D on the circle are points of maximum shear stress value.
Fig. 219
where U , =  p = ukk/3 is the mean normal stress, and the second (the deviator stress tensor)
has the form
I t is easily shown that the first invariant of the deviator stress tensor Ir, is identically
zero, which accounts for its absence in (2.72).
Solved Problems
STATE OF STRESS AT A POINT. STRESS VECTOR.
STRESS TENSOR (Sec. 2.12.6)
2.1. At the point P the stress vectors tin)
A
t:^.*'n, =
A
2.2. The stress tensor values a t a point P are given by the array
7 o 2
2 o 4
Determine the traction (stress) vector on the plane a t P whose unit normal is
íi = (213)êi  (213)$2(113)$3. +
From (2.12), t';' = n. 1. The multiplication is best carried out in the matrix form of (2.13):
7
7 o 2

A A A
2.3. For the traction vector of Problem 2.2, determine (a) the component perpendicular
to the plane, (b) the magnitude of t i n ) , ( C ) the angle between tin)and G.
, + Qê3) =
A
+ 10019
A
( b ) jtin'( = d 1 6 = 5.2
( c ) Since tia' 6 = t : ; ' [ cos O , cos O = ( 4 4 / 9 ) / 5 . 2 = 0.94 and e = 20'.
, , A
2.4. The stress vectors acting on the three coordinate planes are given by t l e l ' , tlez' and t i ê 3 ) .
Show that the sum of the squares of the magnitudes of these vectors is independent
of the orientation of the coordinate planes.
Let S be the sum in question. Then
s A A A
= t;ei)t:ei)+ t(.edt!ez)+ t!edt(ed
t t
A
a
A
i
A
where a, b, c are constants and u is some stress value. Determine the constants a, b
and c so that the stress vector on the octahedral plane (2 = (11fi)êl ( 1 l f i ) ê ~+ +
(11fl )ê3)vanishes.
A
I n matrix forrn, tin' = uijnj must be zero for the given stress tensor and normal vector.
0/2 012
60 ANALYSIS O F STRESS [CHAP. 2
t : =
(1 5

3
i)
1
7 3 Fig. 221
From (2.14),the stress vector may be determined by matrix multiplication, 
Thus
2.7. The state of stress throughout a continuum is given with respect to the Cartesian
axes 0 ~ 1 ~ 2 by
x 3 the array
3xix2 5x;
I: =
2x3
Determine the stress vector acting a t the point P ( 2 , 1 , fi)of the plane that is tangent
+
to the cylindrical surface x ; x3 = 4 a t P.
A t P the stress components a r e given by
substitute t:" = oiini in the surface integral and convert the result to a volume integral by
(1.157):
~ ) ~ dV
( E ~ ~ d~ s x =~ o ~(eijkxjopk),p
~
Carrying out the indicated differentiation in this volume integral and combining with the first
volume integral gives
k ~ j ( ~ p k+
E i i k ( ~ j , p o p+ , p pbk)} dV = O
\
STRESS TRANSFORMATIONS (Sec. 2.8)
2.10. The state of stress a t a point is given with respect to the Cartesian axes 0 ~ ~ byx 2 ~
the array
Determine the stress tensor Z' for the rotated axes OxíxSxl related to the unprimed
axes by the transformation tensor
llfi llfi
1 112
The stress transformation law is given by (2.27) a s ali = aiPaiqopq O r 1' = A A,. The de
[aqi] given in
tailed calculation is best carried out by the matrix multiplication [ali] = [aip][opq]
(2.29). Thus
l l f i 1lfi
[o$] =
1 112 112
o
62 ANALYSIS O F STRESS [CHAP. 2
2.11. Show that the stress transformation law may be derived from (2.33), the equation
uN = u i j n i n j expressing the normal stress value on an arbitrary plane having the unit
normal vector ni.
Since U N is a zeroorder tensor, i t is given with respect to an arbitrary set of primed or
unprimed axes in the same form as
ohr = o!.n!nj
a3 1 = uijnin j
and since by (1.94) nE = aijnj,
U!.n!n! = o!.a. n a . n =
E3 Z 3 21 IP P 54 q "N = "PqnPnq
where new dummy indices have been introduced in the last term. Therefore
("!.a. a. 
23 IP 3 9 O
and since the directions of the unprimed axes are arbitrary,
Uij = (i ; :) o o
Using the orthogonality equations aijaik = Si,, the transformation matrix is determined by com
puting the missing entries in the table. I t is left as an exercise for the student to show that
CHAP. 21 ANALYSIS O F STRESS 63
l/,h l l f i 1lfi
1 1 6 1 1 6 llfi
The result obtained here is not surprising when one considers the Mohr's circles for the state of
stress having three equal principal stress values.
[i E I!]![
Thus using the matrix form,
Hence the quadric surface for uniform tension is the sphere 5: + 522 + c3 2 = kk2lcr.
Hence the quadric surface for uniaxial tension i s a circular cylinder along the tension axis.
and so the quadric surface for simple shear is a hyperbolic cylinder parallel to the 5, axis.
(d) [Si?52,531 [ []
O 0 0
= 05; + 2 ~ ( ~ {+; 05: = k 2
and so the quadric surface for plane stress is a general conic cylinder parallel to the zero principal
axis.
ai i
2.14. Show that the Cauchy stress quadric for a state of stress represented by
is an ellipsoid (the stress ellipsoid) when a, b and c are a11 of the same sign.
The equation of the quadric is given by
53 
Therefore the quadric surface is the ellipsoid bcg +g
a c +
ab 
+k2
aL
bc '
64 ANALYSIS O F STRESS [CHAP. 2
Determine the principal stress values and the principal stress directions represented
by the axes Oxtx,*x3.
From (9.37) the principal stress values o are given by
30 1 1
1 a 2 = O or, upon expansion, (a + 2)(0 4 ) ( o 1 ) = O
1 2 a
The roots are the principal stress values = 2, = 1, = 4. Let the x: axis be the
direction of o ( l ) , and let n;" be the direction cosines of this axis. Then from (2.42),
(3 + 2 ) n i 1 )+ n k l ) + n"'
3 = 0
216. show that the transformation tensor of direction cosines determined in Problem 2.15
transforms the original stress tensor into the diagonal principal axes stress tensor
According to (2.29), [Gj]= [ a i p ] [ a P q ] [ a Qwhich
j], for the ~ r o b l e ma t hand becomeS
o ,h 2 o o
O 0 4
CHAP. 21 ANALYSIS O F STRESS 65
2.17. Determine the principal stress values and principal directions for the stress tensor
From (2.37),
Hence = O, <rc2)
~
= 0,
7
7 L
7a
~
o(3)
a
T = O or (7  a)[27a
which together with nini = 1 a r e insufficient to determine uniquely the first and second principal
directions. Thus any pair of axes perpendicular to the n:3' direction and perpendicular to each
other may serve a s principal axes. For example, consider the axes determined in Problem 2.12, for
which the transformation matrix is
llfi llfi llfi
[aiil
According to the transformation law (2.29), the principal stress matrix [a*,] is given by
[aijl =
11,h 11,h llfi
I
1lfi 11d5
1 1 fi16
116 11a lld3
O
As before,
2.19. Show that the principal stresses and the stress components oij for an arbitrary
set of axes referred to the principal directions through the transformation coefficients
aij are related by oij = 5 ap,apjy.
p=1
From the transformation law for stress (2.27), oij = apiaqjop,;* but since o:, are principal
stresses, there are only three terms on the right side of this equation, and in each p = q. There
3
fore the right hand side may be written in form oij = 2
p=1
apiapjop.
= (aipair)(ajqajn)(aksakm)~pq"rs"mn
= aprsqnssmopqarsamn
= (sprupq)(aqnamn)(8smurs)
 arqoqmarm

 aijoikokj
2.21. Evaluate directly the invariants IE,111, 1111for the stress tensor
Determine the principal stress values for this state of stress and show that the
diagonal form of the stress tensor yields the same values for the stress invariants.
From (2.39), IE = oii = 6 + 6 f 8 = 20.
From (2.40), IIE = ( 1 / 2 ) ( ~~u~ ~0 ~ ~u ~ ~ )
= 01ia22 + O22O33 + 033011  412a12  023423  031a31
= 36+ 48 + 48  9 = 123.
From (2.41), IIIz = (oij(= 6(48) + 3(24) = 216.
The principal stress values of aij are o1 = 3, oI1 = 8, a111= 9. In terms of principal values,
1, = a1 + o11 + = 3 + 8 + 9 = 20
0111
IIE = o1011 + + = 24 + 72 + 27
~11~111 ~ 1 1 1 ~ 1 = 123
1111 = OIO~IOIII = (24)9 = 216
CHAP. 21 ANALYSIS O F STRESS 67
2.22. The octahedral plane is the plane which makes equal angles with the principal stress
directions. Show that the shear stress on this plane, the socalled octahedral shear
stress, is given by
UOCT = 3d(uI
 ~11)' I (011  ~ 1 1 1 ) ' + ( ~ 1 1 1 VI)'
1
= 
6
+ 011ê2 + ~ ~ ~ $ 3 )
2.23. The stress tensor a t a point is given by uij = . Determine the maxi
mum shear stress a t the point and show that it acts in the plane which bisects the
maximum and minimum stress planes.
From (2.38) the reader should verify that the principal stresses are o, = 10, o,, = 5,
= 15. From (2.51b) the maximum shear value is os = (oIII 4 1 2 = 12.5. The principal
axes 0 x ~ x are
~ x related
~ to the axes of maximum shear 0 x ; x ~ xby~ the transformation table
below and are situated as shown in Fig. 226.
Fig. 226
68 ANALYSIS O F STRESS [CHAP. 2
1 o llfi
[a&] =
12.5 O 2.5
The results here may be further clarified by showing the stresses on infinitesimal cubes a t the
point whose sides are perpendicular to the coordinate axes (see Fig. 227).
Fig. 227
from which a diagram of the relative orientation of the axes is made a s shown in Fig. 229. The
x1 and 2; axes are coincident. x2 and x3 are in the plane of x: x; as shown. From the angles
a = 36.8O and fl = 53.2O shown, the points A (  6 , 1 2 ) on the plane L to x2 and B ( 1 , 1 2 ) on the
plane I to x , are located. Point C(5,O) represents the stress state on the plane L to x,.
Determine analytically the stress vector components on the plane whose unit normal
+ +
is íi = (2/3)ê1 (1/3)ê2 (2/3)&. Check the results by the Mohr's diagram for
this problem.
From (2.13) and the symmetry property of the stress tensor, the stress vector on the plane of
A
n is given by the matrix product
Thus t':) = 10ê1/3  1 0 ê 2  10ê3/3; and from @.33), UN = t ( : ) hn = 7019. From (2.47),
a,= 70.719.
F o r oij the principal stress values a r e o, = 10, oI1 = 5, CIII = 15; and the principal axes
a r e related to 0 x l x 2 x 3by the table
tlie angles of Fig. 216 a r e given by 0 = ,8 = cosI 213 = 48.2' and $ = cos1113 = 70.5O, and
the Mohr's diagram comparable to Fig. 217 is a s shown in Fig. 230.
Fig. 230
70 ANALYSIS OF STRESS [CHAP. 2
2.26. Sketch the Mohr's circles for the three cases of plane stress depicted by the stresses
on the small cube oriented along the coordinate axes shown in Fig. 231. Determine
the maximum shear stress in each case.
Fig. 231
(b)
Fig. 232
uij = uMSij + ~ i j =
o 2 5
and sii = 4 + 1 .5 = 0.
2.28. Show that the deviator stress tensor is equivalent to the superposition of five simple
shear states.
The decomposition is
CHAP. 21 ANALYSIS OF STRESS 71
where the last two tensors are seen to be equivalent to simple shear states by comparison of cases
(a) and (b) in Problem 2.26. Also note that since sii = 0, sll  s~~= s ~ ~ .
2.29. Determine the principal deviator stress values for the stress tensor
10 6 O
0 0 1
Thus sI = 9, sII = 3, sIII = 6. The same result is obtained by first calculating the principal
stress values of oij and then using (2.71). For qj,a s the reader should show, a1 = 16, o11 = 4,
oII1 = 1 and hence sI = 16  7 = 9, sI1 = 4  7 = 3, sII1= 1 7 = 6.
2.30. Show that the second invariant of the stress deviator is given in terms of its principal
+ +
stress values by 111, = (sIsI1 SIISIII SIIISI), or by the alternative form IIn, =
+ + sh).
sS1
In terms of the principal deviator stresses the characteristic equation of the deviator stress
tensor is given by the determinant
SIs o o
o SI,s o = (SI  s)(sII s)(sIII s) = o
O 0 SI11  s
+ + 2811181
IInD = *(~SISII 2~11~111  (SI + + 12) = *(*I + + s : ~ ~ )
811 8111 SII
MISCELLANEOUS PROBLEMS
2.31. Prove that for any symmetric tensor such as the stress tensor uij, the transformed
tensor in any other coordinate system is also symmetric.
From (2.27), o; = aipajqo,,, = ajqaipaqp= aie
72 ANALYSIS O F STRESS [CHAP. 2
Substituting n l = n, and ni  1 = nf  n; = 2nS into this equation and solving for n,, the direc
tion cosines are found to be nl = 1 / 2 f i , n2 = f i / 2 , n3 = 1 / 2 f i . The reader should apply these
2.33. Show that the stress tensor uij may be decomposed into a spherical and deviatoric
part in one and only one way.
Assume two decompositions, <rij = Sijh t sij = Si$* +
sTi with sii = O and sTi = O. Then
aii = 3X = 3h*, so
X = h*; and from ASij + +
sij = hSij s$ i t foilows that sij = s*,.
Prove that the principal stress values are real if 1 is real and symmetric.
For real values of the stress components the stress invariants are real and hence the coefficients
in (2.38) are a11 real. Thus by the theory of equations one root (principal stress) is real. Let a ( , )
be this root and consider a set of primed axes a: of which xi is in the direction of ( r c 3 ) . With respect to
aí1. cri2 o
such axes the characteristic equation i s given by the determinant o O = O o& a&
0 a(,)  0 o
12
or ( a c 3 )  o) [(ai1 o)(aS2 a )  a12] = O. Since the discriminant of the quadratic in square brackets
+ +
D = (ail a i 2 ) 2  4[a;1a;12 ( o i 2 ) 2 ] = (aíl  o i 2 ) 2 4aii > 0, the remaining roots must be real.
Use the method of Lagrangian multipliers to show that the extrema1 values (maximum
and minimum) of the normal stress U , correspond to principal values.
From (2.33), U N = aijninj with qni = 1. Thus in analogy with (2.51) construct the function
H = oN  A n i n i for which aHlani = O. Then
  aijni,pnj+ aijninj,p 2hni,pni
8 % ~
= aijSipnj + oijniS,  2XSipni

 apjnj + a,ni  2XSiPni = 2(api XSip)ni = O
which is equivalent t o equation (2.36) defining principal stress directions.
2.36. Assume that the stress components uij are derivable from a symmetric tensor field
+ij by the relationship uij = Q ~ ~ ~ c Show
~ ~ +that
~ in
~ the
, ~ absence
~ . of body forces the
equilibrium equations (2.23) are satisfied.
Using the results of Problem 1.58, the stress components are given by
oij = Sij($qq,pp  $qp,qp) $. $ p i , p j + +jp,pi  $pp, ji  @ji,pp
or explicitly 011 = $33.22 + #22,33 012 = a21 = @33,21

a22 = $11,33+ @33,11 O23 = O32  @11,23

O33 = $22.11 + $11.22 a31 = "13  @22,13
CHAP. 21 ANALYSIS O F STRESS 73
2.37. Show that, as is asserted in Section 2.9, the normal to the Cauchy stress quadric
a t the point whose position vector is r is parallel to the stress vector tln).
Let the quadric surface be given in the form $ = oijlirj2 k2 = O. The normal a t any point is
then V g or &$/ali =
A
Hence $, = u ~ ~
oijriSjp +
S = 2uPiTi.
~ ~ li = rni, this becomes
~ Now~ since
2uPimi or 2 9  ( ~=~ 2 ~t ' n~) . )
From (2.6) and the identity tte" = uij (2.7), the traction vectors on the unprimed coordinate
+
axes are t ' ; l ) = 15ê1  10ê,, t ( ê z ) = 10 ê1 5ê2, t ' ; 3 ) = 20 ê3 which correspond to the rows
of the stress tensor. Projecting these vectors onto the primed axes by the vector form of (2.12),
A
t(G) = nlt(el) + A
n2t'ez) + A
n 3 t ( e 3 ) , gives
Likewise
and
so that
74 ANALYSIS O F STRESS [CHAP. 2
2.39. Slrow that the second invariant of the deviator stress tensor IIp, is related to the
octahedral shear stress by the equation a,,, = \/311r,.
From Problem 2.22, <rOCT + +
= $ d ( o I  a I I ) 2 (o11  o I I I ) 2 ( o I I I a I ) 2 , and because o1 = U M + SI,
OII = 6111 +
s I 1 , etc.,
~ O C T= SI  ~11)' + ( S I I  SIII)' + ( S I I I  SI)^
Hence
A A
The normal to the sphere xixi = (9)2 a t P i s ni = $,i with $ = xixi  81, or n = 49 e 1 .
+
$êz $ê3. In the matrix form of (2.14) the stress vector a t P is
(d) The Mohr's circles are shown in Fig. 233. Fig. 233
CHAP. 21 ANALYSIS O F STRESS 75
2.41.
Supplementary Problems
plane a t P parallel to plane ( a ) BGE, ( b ) BGFC of the small parallelepiped shown in Fig. 234.
Fig. 234
2.42. Determine the normal and shear stress components on the plane BGFC of Problem 2.41.
Ans. U N = 6315, os = 37.715
2.43. The principal stresses a t point P are o, = 12, o,, = 3, o,,, = 6. Determine the stress vector and
its normal component on the octahedral plane a t P.
Ans. t':' = (12êl+3ê26ê3)/fi, oN = 3
(a) oij =
(i 1)
1 0 1
Ans. ( a ) a1 = 2, o l ~
= olI1= 1, ( b ) o, = 4 , 0 1 , = o,,, = 1
3 10
4 Decompose the stress tensor oij = O 3:) into its spherical and deviator par. and
30 27
determine the principal deviator stresses. Ans. S I = 31, SI, = 8, 8111 = 39
2.46. Show that the normal component of the stress vector on the octahedral plane is equal to one third
the first invariant of the stress tensor.
so that the stress vector on some plane a t the point will be zero. Give the unit normal for this
tractionfree plane.
Ans. oz2 = 1, f; = ( S I  2 ê 2 + ê 3 ) / 6 .
76 ANALYSIS O F STRESS [CHAP.2
2.48. Sketch the Mohr's circles and determine the maximum shear stress for each of the following
stress states:
(a) uij
Ans. ( a ) U , = 7, ( b ) os = 3712
= i:: )
0 0
(b) 0. =
o
7 :
0 27
2.49. Use the result given in Problem 1.58, page 39, together with the stress transformation law (2.27),
page 50, to show that ~ ~ ~ is an~invariant.
~ ~ ~ ~ u ~ ~ u ~
o
Determine ( a ) the body force distribution if the equilibrium equations are to be satisfied throughout
the field, ( b ) the principal stress values a t the point P(a, 0,2&), (c) the maximum shear stress a t
P , (d) the principal deviator stresses a t P .
Ans. ( a ) b, = 4x,, ( b ) a, a, 8a, ( c ) 24.5~3,( d )  1 1 ~ 1 3 ,5a/3,16a/3
Deformation and Strain
_ _
L
.___I___
CI
Fig. 31
78 DEFORMATION AND STRAIN [CHAP. 3
with respect to the rectangular Cartesian axes OXlX2X3. Uppercase letters are used as
indices in (3.1) and will appear as such in severa1 equations that follow, but their use as
sixmmation indices is restricted to this section. In the remainder of the book uppercase
subscripts or superscripts serve as labels only. Their use here is to emphasize the connection
of certain expressions with the coordinates (XiXaX3), which are called the material coor
dinates. In the deformed configuration the particle originally a t PO is located a t the point
P and has the position vector
when referred to the rectangular Cartesian axes oxixnx3. Here lowercase letters are used
as subscripts to identify with the coordinates (x1x2x3)which give the current position of
the particle and are frequently called the spatial coordinates.
The relative orientation of the material axes OXlX2X3 and the spatial axes 0 x 1 ~ 2 ~is3
specified through direction cosines a,, and a,,, which are defined by the dot products of
unit vectors as
A A A A 
e, I, = I; e, = a,,  a,, (3.3)
No summ:,tion is implied by the indices in these expressions since k and K are distinct
A h
indices. lnasmuch as Kronecker deltas are designated by the equations I K  I P= SKP and
A A
ekDep= 8kp, the orthoganality conditions between spatial and material axes take the form
Tn Fig. 31 the vector u joining the points Po and P (the initial and final' positions,
respectively, of the particle), is known as the displacement vector. This vector may be
expressed as A
u = ukek (3.5)
A
or alternatively as U = U,I,
in which the components UK and uk are interrelated through the direction cosines A
a,,.
From (1.89) the unit vector êk is expressed in terms of the material base vectors IK as
A
A
ek = ~ILKIK
Therefore substituting (3.7) into ( 3 4 ,
Since the direction cosines a,, are constants, the components of the displacement vector are
observed from (3.9) to obey the law of transformation of firstorder Cartesian tensors, as
they should.
The vector b in Fig. 31 serves to locate the origin o with respect to O. From the
geometry of the figure,
u = b+xX (3.1O)
CHAP. 31 DEFORMATION AND STRAIN 79
However, for superimposed axes the unit triads of base vectors for the two systems are
identical, which results in the direction cosine symbols a,, becoming Kronecker deltas.
Accordingly, (3.12) reduces to
U, = X ,  X,$ (3.13)
in which only lowercase subscripts appear. In the remainder of this book, unless specifi
cally stated otherwise, the material and spatial axes are assumed superimposed and hence
only lowercase indices will be used.
which give the present location xi of the particle that occupied the point (XIX2X3) a t time
t = O. Also, (3.14) may be interpreted as a mapping of the initial configuration into the
current configuration. I t is assumed that such a mapping is onetoone and continuous,
with continuous partial derivatives to whatever order is required. The description of
motion or deformation expressed by (3.14) is known as the Lagrangian formulation.
If, on the other hand, the motion or deformation is given through equations of the form
in which the independent variables are the coordinates xi and t, the description is known
as the Eulerian formulation. This description may be viewed as one which provides a
tracing to its original position of the particle that now occupies the location ( x l ,X Z , ~ 3 ) . If
(3.15) is a continuous onetoone mapping with continuous partial derivatives, as was also
assumed for (3.14), the two mappings are the unique inverses of one another. A necessary
and sufficient condition for the inverse functions to exist is that the Jacobian
a
in which the differential operator V, = $idXi is applied from the right (as shown
explicitly in the equation). The matrix form of F serves to further clarify this property
of the operator V, when it appears as the consequent of a dyad. Thus
Partial differentiation of (3.15) with respect to x j produces the tensor dXi/dxj which is
called the spatial d e f o r m a t i o n gradient. This tensor is represented by the dyadic
The material and spatial deformation tensors are interrelated through the wellknown
chain rule for partia1 differentiation,
and
CHAP. 31 DEFORMATION AND STRAIN
81
Fig. 32
In the deformed configuration, the sqiare of the differential element of length between
P and Q is
The difference (dx)'  (dX)' for two neighboring particles of a continuum is used as
the measure of deformation that occurs in the neighborhood of the particles between the
initial and final configurations. If this difference is identically zero for a11 neighboring
particles of a continuum, a rigid displacement is said to occur. Using (3.34) and (3.28),
this difference may be expressed in the form
01 ( d ~ ) . .
' (dx)' = dX ( F , F  I) dX = dX 2Lc dX (3.36)
L..   or LG = &(Fc F  I)
/
is called the Lagrangian (or Green's) finite strain tensor.
*' /
Using (3.32) and (3.30), the same difference may be expressed in the form
In the same manner, if dXi/dxj from (3.25) is substituted into (3.39), the result is the
Eulerian finite strain tensor in the form
The matrix representations of (3.40) and (3.41) may be written directly from (3.26) and
(3.27) respectively.
In (3.40), if the displacement gradient components duildXj are each small compared to
unity, the product terms are negligible and rnay be dropped. The resulting tensor is the
Lagrangian infinitesimal strain tensor, which is denoted by
If both the displacement gradients and the displacements themselves are small, there is
very little difference in the material and spatial coordinates of a continuum particle.
Accordingly the material gradient components duildXj and spatial gradient components
dui/dxj are very nearly equal, so that the Eulerian and Lagrangian infinitesimal strain
tensors rnay be taken as equal. Thus
l i . = r i . or L = E (3.44)
if both the displacements and displacement gradients are sufficiently small.
v
dui = u i < Q ~ )  U ! P ~ ) or
1
du = ~ ( Q o ) u ( P o ) dx
Here the parentheses on the partia1 derivatives are to emphasize the requirement that the
derivatives are to be evaluated a t point Po. These derivatives are actually the components
of the material displacement gradient. Equation (3.46) is the Lagrangian form of the
relative displacement vector.
It is also useful to define the u n i t relative displacement vector d ~ l d Xin which d X is
the magnitude of the differential distance vector dXi. Accordingly if vi is a unit vector in
the direction of d X i so that d X i = V , d X , then
Since the material displacement gradient duildXj rnay be decomposed uniquely into a
symmetric and an antisymmetric part, the relative displacement vector dui rnay be written a s
84 DEFORMATION AND STRAIN [CHAP. 3
The first term in the square brackets in (3.48) is recognized as the linear Lagrangian strain
tensor lij. The second term is known as the linear Lagrangian rotation tensor and is
denoted by
In a displacement for which the strain tensor lij is identically zero in the vicinity of point
Po,the relative displacement a t that point will be an infinitesimal rigid body rotation.
This infinitesimal rotation may be represented by the rotation vector
The development of the Lagrangian description of the relative displacement vector, the
linear rotation tensor and the linear rotation vector is paralleled completely by an analogous
development for the Eulerian counterparts of these quantities. Accordingly the Eulerian
description of the relative displacement vector is given by
The first term in the square brackets of (3.54) is the Eulerian linear strain tensor eij. The
second term is the linear Eulerian rotation tensor and is denoted by
+
(dx)'  (dX)' = ( d x  d X ) ( d x d X ) = 2Ljd X i d X j
(dx)'  ( d X ) 2 = ( d ~d X ) ( d +
~ dX) = d x 21. d x (3.58)
Since d x  d X for small deformations, this equation may be put into the form
Fig. 34
Thus the normal strain for an element originally along the X z axis is seen to be the com
ponent 122. Likewise for elements originally situated along the X i and X 3 axes, (3.59) yields
normal strain values l1l and 133 respectively. I n general, therefore, the diagonal terms of
the linear strain tensor represent normal strains in the coordinate directions.
Fig. 35
The physical interpretation of the offdiagonal terms of lij may be obtained by a con
sideration of the line elements originally located along two of the coordinate axes. In
Fig. 35 the line elements POQOand POMOoriginally along the X2 and Xs axes, respectively,
become after deformation the line elements P Q and P M with respect to the parallel set of
local axes with origin a t P. The original right angle between the line elements becomes
the angle O. From (3.46) and the assumption of small deformation theory, a first order
approximation gives the unit vector a t P in the direction of Q as
86 DEFORMATION AND STRAIN [CHAP. 3
Therefore
Similarly, from (3.30) the reciproca1 of the squared stretch for the line element a t P along
the unit vector íi= dxldx is given by
For an element originally along the local X2 axis shown in Fig. 34, = êz and
therefore dXildX = dXJdX = O, dXzldX = 1 so that (3.66) yields for such an element
For a n element parallel to the xz axis after deformation, (3.67) yields the result
The stretch ratio provides a basis for interpretation of the finite strain tensors. Thus
the change of length per unit of original length is
and f o r the element POQOalong the X2 axis (of Fig. 34), the unit extension is therefore
This result may also be derived directly from (3.36). For small deformation theory, (3.71)
reduces to (3.60). Also, the unit extensions L ( 1 )and L(,) a r e given by analogous equations
in terms of Ltt and L33 respectively.
F o r the two differential line elements shown in Fig. 35, the change in angle y,, = ~ 1
26
is given in terms of A , ; ~ ,and A , ; ~ ,by
in which R is the orthogonal rotation tensor, and S and T are positive symmetric tensors
known a s the right stretch tensor and left stretch tensor respectively.
From these expressions the deformation of dXi into dxi a s illustrated in Fig. 32 may be
given either of two physical interpretations. In the first form of the right hand side of
(3.74), the deformation consists of a sequential stretching (by S ) and rotation to be followed
by a rigid body displacement to the point P. In the second form, a rigid body translation
to P is followed by a rotation and finally the stretching (by T). The translation, of course,
does not alter the vector components relative to the axes Xi and X i .
88 DEFORMATION AND STRAIN [CHAP. 3
and l!.
11  b.I P b.34 G,4 or lf=B*L.B, (3.76)
(b)
Fig. 36
Likewise, for the rotated axes xi' having the transformation matrix [aij] in Fig. 36(b),
the components of E; and are given by
By analogy with the stress quadric described in Section 2.9, page 50, the Lagrangian
and Eulerian linear strain quadrics may be given with reference to local Cartesian coor
dinates q i and (Si a t the points Po and P respectively as shown in Fig. 37. Thus the
equation of the Lagrangian strain quadric is given by
CHAP. 31 DEFORMATION AND STRAIN 89
Two important properties of the Lagrangian {Eulerian) linear strain quadric are:
1. The normal strain with respect to the original {final) length of a line element is
inversely proportional to the distance squared from the origin of the quadric PO
{P) to a point on its surface.
2. The relative displacement of the nedghboring particle located a t QO {Q) per unit
original {final) length is parallel to the normal of the quadric surface a t the point
of intersection with the line through POQO{PQ).
Additional insight into the nature of local deformations in the neighborhood of POis
provided by defining the strain ellipsoid a t that point. Thus for the undeformed continuum,
the equation of the bounding surface of an infinitesimal sphere of radius R is given in
terms of local material coordinates by (3.28) as
After deformation, the equation of the surface of the same material particles is given by
(3.30) as
(dx)' = Cijdxidxj = R2 or ( d X ) 2 = d x . C . d x = R2 (3.82)
which describes an ellipsoid, known as the material strain ellipsoid. Therefore a spherical
volume of the continuum in the undeformed state is changed into an ellipsoid a t Po by the
deformation. By comparison, an infinitesimal spherical volume a t P in the deformed
continuum began as an ellipsoidal volume element in the undeformed state. For a sphere
of radius r a t P, the equations for these surfaces in terms of local coordinates are given
by (3.32) for the sphere as
The ellipsoid of (3.84) is called the spatial strain ellipsoid. Such strain ellipsoids as
described here are frequently known as Cauchy strain ellipsoids.
n
Calling 1;"' the normal strain in the direction of the unit vector ni, (3.85) yields for pure
strain (Wij 0) the relation A A
which together with the condition nin+= 1 on the unit vectors ni provide the necessary
equations for determining the principal strain value 1 and its direction cosines ni. Nontrivial
solutions of (3.88) exist if and only if the determinant of coefficients vanishes. Therefore
which upon expansion yields the characteristic equation of lij, the cubic
P  IL12 + IIJ  IIIL = O (3.90)
The first invariant of the Lagrangian strain tensor may be expressed in t e m s of the
principal strains as
IL= Zii = Z(1) Z(2) Z(3) + + (3.92)
and has an important physical interpretation. To see this, consider a differential rec
tangular parallelepiped whose edges are parallel to the principal strain directions a s
shown in Fig. 38. The change in volume per unit original volume of this element is called
the cubical dilatation and is given by
For small strain theory, the firstorder approximation of this ratio is the sum
Do = l(l>+ 1<2>+ Z(3) = IL (3.94)
Fig. 38
CHAP. 31 DEFORMATION AND STRAIN 91
With regard to the Eulerian strain tensor eij and its associated unit relative displacement
vector E:"', the principal directions and principal strain values C,,,,E,,,, C,,, are determined
in exactly the same way as their Lagrangian counterparts. The Eulerian strain invariants
may be expressed in terms of the principal strains as
IE = E(i, + + E<3,
IIE = '(1)'<2) '(2)'(3) + '(3)'(1)
IIIE = c(1)E(2, E ( 3 )
(i: i: ;)
linear strain tensor is given by
t,
E..

 or [c] = [; :i %] (3.99)
When xi and xz are also principal directions, the strain tensor has the form
Ul = u1(x1,xz)
Uz = uz(x1,xz)
u3 = C (a constant, usually taken as zero)
Inserting these expressions into the definition of eij given by (3.43) produces the plane
strain tensor in the same form shown in (3.99).
Here the yij (with i # j ) are the socalled "engineering" shear strain components, which are
twice the tensorial shear strain components.
The state of strain a t an unloaded point on the ~ 1 2
'1
There are eightyone equations in a11 i n (3.103) but only six are distinct. These six written
in explicit and symbolic form appear as
CHAP. 31 DEFORMATION AND STRAIN
5. 

8x3 axi
d (dc23 
6.   +   '€31 
~ X S8x1 ax2 axs 8x1 8x2
Compatibility equations in terms of the Lagrangian linear strain tensor lij may also be
written down by an obvious correspondence to the Eulerian form given above. For plane
strain parallel to the ~ 1 x 2plane, the six equations in (3.104) reduce to the single equation
Solved Problems
DISPLACEMENT AND DEFORMATION (Sec. 3.13.5)
3.1. With respect to superposed material axes Xi and spatial axes xi, the displacement
+
field of a continuum body is given by xl = Xi, 2 2 = X2 AX3, x s = X3 AX2 where +
A is a constant. Determine the displacement vector components in both the material
and spatial forms.
From (3.13) directly, the displacement components in material form are ul = xl  X l = 0,
u2 = x2  X2 = AX3, u3 = x3  X3 = AX,. Inverting the given displacement relations to obtain
X, = x,, X2 = (x,  Ax3)l(1 AZ), X3 = (2,  Ax2)l(1 AZ), the spatial components of u are
U, = O, u, = A(x3 Ax2)l(l A2), ~3 = A(x2  Ax3)l(1 A2).
From these results i t is noted that the originally straight line of material particles expressed
+ +
by X, = O, X2 X3 = lI(1+ A ) occupies the location xl = 0, x2 x3 = 1 after displacement.
Likewise the particle line X1 = O, X2 = X3 becomes after displacement x1 = 0, x2 = x3. (Inter
pret the physical meaning of this.)
For the displacement field of Problem 3.1 determine the displaced location of the
material particles which originally comprise (a) the plane circular surface XI = 0,
+
XE X : = 1/(1 A2), (b) the infinitesimal cube with edges along the coordinate axes
of length d X i = d X . Sketch the displaced configurations for (a) and (b) if A = .S.
(a) By the direct substitutions X2 = (x,  Ax3)l(l  A2) and X3 = (x3  Ax2)l(1 A2), the circular
surface becomes the elliptical surface (1+ A2)xl  4Ax2x3 + (1+ A2)xi = (1  A2). For
A = 4, this is bounded by the ellipse 5x22  8x223 + 5x3 = 3 which when referred to its
prinzipal axes x i (at 45O with xi, i = 2,3) has the equation xi2 +9 ~ =
; 3.
~ Fig. 310 below
shows this displacement pattern.
94 DEFORMATION AND STRAIN [CHAP. 3
( b ) From Problem 3.1, the displacements of the edges of the cube are readily calculated. For the
edge X , = X,, X 2 = X 3 = O, u , = u , = u3 = O. For the edge X 1 = O = X2, X 3 = X3,
u, = u3 = O , u2 = A X 3 and the particles on this edge are displaced in the X , direction propor
tionally to their distance from the origin. For the edge X , = X 3 = O, X , = X,, u, = u, = 0 ,
u3 = AX,. The initial and displaced positions of the cube are shown in Fig. 311.
33. For superposed material and spatial axes, the displacement vector of a body is given
+ +
by u = 4x:ê1 X 2 X i ê 2 Xixiê3. Determine the displaced location of the particle
originally a t ( 1 , 0 , 2 ) .
The original position vector of the particle is X = ê, + 2êg. Its displacement is u = 4ê1 4 ê 3 +
and since x = X + u, its final position vector is x = 5ê1 6ê3. +
3.4. With respect to rectangular Cartesian ma
terial coordinates Xi, a displacement field is
given by U l =  A X 2 X 3 , U2 = A X l X 3 , U 3 = 0
where A is a constant. Determine the dis
placement components for cylindrical spatial
coordinates xi if the two systems have a com
mon origin.
From the geometry of the a h s (Fig. 312) the
A
transformation tensor apK = e p IK is
cosx, sin x , O
aPK (7x2 0)
and from the inverse form of (3.9) up = aPKUK.Thus
since Cartesian and cylindrical coordinates are related
through the equations X , = xl cos x,, X 2 = x1 sin x,,
X 3 = x,, equation (3.9) gives Fig. 312
+
ul = (cos x2)AX,X3 (sin x 2 ) A X l X 3
+
= (cos x2)Ax3xl sin 2 , (sin x2)Ax3x1cos x2 = O
u, = (sin x,)AX,X3 + (cos x,)AXlX3
+
= ( s i n ~ x 2 ) A x l x 3 (cos2 x2)Ax1x3 = A x 1 x 3
Ug = o
This displacement is that of a circular shaft in torsion.
CHAP. 31 DEFORMATION AND STRAIN 95
+
Since x = u X, the displacement field may also be described by equations xl = X1(l + x:),
+
x2 = X2(1 x:), x3 = X3(l x:)+from which F is readily found to be
Direct substitution of the calculated tensors F and J into (3.26) verifies that the equation is satisfied.
3.8. For the displacement field of Problem 3.7 determine the displaced position of the
position vector of particle C ( 2 , 6 , 3 ) which is parallel to the vector joining particles
A and B. Show that the two vectors remain parallel after deformation.
By the analysis of Problem 3.7 the position vector of C becomes U = 82, 7ê2 + + 25ê3 which
is clearly parallel to V. This is an example of socalled homogeneous deformation.
+
According to (3.16) the inverse equations Xi = (aij B i j ) x j exist provided the determinant
+
Iaij Aijl does not vanish. Assuming this is the case, the material plane piXi cu = O becomes +
+ +
&(aij B i j ) x j cu = O which may be written in standard form a s the plane h j x j a. = O +
where the coefficients = &(aij Bij). +
( b ) A straight line may be considered a s the intersection of two planes. I n the deformed geometry,
planes remain plane a s proven and hence the intersection of two planes remains a straight line.
so t h a t Gij] = [i 1l
o
f 2
o
yAj
Therefore from (3.37), L~ = S ( G  I) =
2
O 2A A2
=
[: l:Aj[il
1 +A2
2A
3.13. Calculate the change in squared length of the line elements of Problem 3.12 and check
the result by use of (3.36) and the strain tensor LG found in Problem 3.11.
Directly from the results of Problem 3.12, t h e changes are:
( a ) f o r OC, ( d ~ ) (dX)2
~ = ,(I+A Z ) ( ~ X ; dx;) + + 4 A dX, dX3  ( d x ; + d x ; )
= A Z ( ~ X : dx;) + + 4 A dX, dX3
( b ) for OB, (dx)2  (dX)2 = ( 1 + A ' ) d x ;  dx; = A2 d x ;
By equation (3.36), f o r OC
3.14. For the displacement field of Problem 3.11 calculate the material displacement
gradient J and use this tensor to determine the Lagrangian finite strain tensor LÇ.
Compare with resillt of Problem 3.11.
F r o m Problem 3.11 the displacement vector components a r e ul = O , u, = A X 3 , z(3 =AX, so
that
O O A2
Thus f r o m (3.40)
~ L G
O A O O A O O O A2 O 2A A2
+
3.15. A displacement field is given by X I = X I A X 2 , x2 = X Z AX3, x s = X3 A X l where + +
A is a constant. Calculate the Lagrangian linear strain tensor L and the Eulerian
linear strain tensor E. Compare L and E for the case when A is very small.
F r o m (3.42),
/ 0 A O\ 10 O A \ /o A A\
i2)
A2 1 A A2  A
+ K,)
:,)
A
2E = (K =
1 +A3
+ t2
h(:
When A i s very small, A? and higher powers may be neglected with the result t h a t E reduces to L.
98 DEFORMATION AND STRAIN [CHAP. 3
~ a u i l ~ x i=
3.17. For the displacement field of Problem 3.16 determine the unit relative displacement
vector a t P ( 1 , 2 , 1) in the direction of Q(4,2,3).
The unit vector a t P in the direction of Q is $ = 3ê1/5 + 4ê3/5, so that from (3.47) and the
matrix of J a s calculated in Problem 3.16,
3.18. Under the restriction of small deformation theory, L = E. Accordingly for a dis
placement field given by u = (xl  ~ 3 ) ~ ê (x2
l + +
~ ~ ) ~ x1x2ê3,
ê 2 determine the linear
strain tensor, the linear rotation tensor and the rotation vector at the point P(O,2, 1).
Here the displacement gradient is given in matrix form by
[auilaxj] =
2(x1 x3)
O
0
2(x2+ x3)
a21  x3)
2(x2 + 23)
1
which a t the point P becomes
r2
Decomposing this matrix into its symmetric and antisymmetric components gives
[eijI + [wijI =
2 o 1 o
Therefore from (3.56) the rotation vector wi has components wl = 1, w2 = w3 = 0.
CHAP. 31 DEFORMATION AND STRAIN 99
3.19. For the displacement field of Problem 3.18 determine the change in length per unit
+
length (normal strain) in the direction of Y = (8êi  ê2 4ê3)/9 a t point P(O,2, 1).
From (3.59) and the strain tensor a t P a s computed in Problem 3.18, the normal strain a t P in
the direction of Y is the matrix product
3.20. Show that the change in the right angle between two orthogonal unit vectors G and í;
in the undeformed configuration is given by G.2L.fi for small deformation theory.
Assuming small displacement gradients, the unit vectors in the deformed directions of $ and $
+ +
a r e given by (3.47) a s ($ J * c ) and (li J .fi) respectively. (The student should check equations
(3.61) and (3.62) by this method.) Writing J in the equivalent form $ * J c and dotting the two
displaced unit vectors gives (as i n (3.63)), cos e = sin ( ~ 1 2 . e) = sin y,, = y,, or y,, = [$ $ * J,] +
[ f i + J . C ] = $ . f i + G  ( ~ + ~ , ) . f i + Y . ~ , . ~ . f i . Here J,.J is of higher order f o r small displacement
gradients and since Y Ifi, 4. fi = O so t h a t finally by (3.42), y,, = $ 21. c.
3.21. Use the results of Problem 3.20 to compute the change in the right angle between
+ +
G = (8ê1 êz 4ê3)/9 and i; = (4ê1 4ê2 7ê3)/9 a t the point P(O,2, 1) f o r the dis
placement field of Problem 3.18.
Since L = E for small deformation theory, the strain tensor eij = lij and so a t P
Thus A:;, +
= (1  A2)2/(1 A)2 = (1 A)2 which is identical with A(m, calculated for OC. The
2
diagonal element OC does not change direction under the given shear deformation.
r1 O O 1
(3.35), G = F, .
F or here [Gij] = 1 0 1 + A2 2A I. The principal axes of G a r e given by
L0 2A i t ~ 2 J
a 4 5 O rotation about X1 with the tensor in principal form [G~;] =
o o
Therefore [Sij] = [ G ] =
I n this example the deformation gradient F is its own stretch tensor S and R = i. This is the
result of the coincidence of the principal axes of LG and EA f o r the given shear deformation.
[Gijl =
O 0 1
From ( 3 . 3 7 ) [Lij] = [! i ]: o
which f o r principaI axes given by the transformation
matrix [aij = [:
o 1lfi
11.
o
11fi
becomes [L: = [i ;fi
1
o
.
r 0 O O 1
is converted into the principalaxes form [E:] = . The student should
verify these calculations.
o l+fi
3.27. Using the definition (3.37), show that the Lagrangian finite strain tensor Lij trans
forms a s a second order Cartesian tensor under the coordinate transformations
xi = bjixf and X( = bijXj.
By (3.37), Lij = 
2
1
íax,ax,
axk axk
 Sij which by the stated transformation becomes
3.28. A certain homogeneous deformation field results in the finite strain tensor
deformation.
Being a symmetric second order Cartesian tensor, the principal strains a r e the roots of
102 DEFORMATION AND STRAIN [CHAP. 3
[aijl =
1 1 216
= E[ o
11,h llfi
From the deformation equations, the stretch tensor S = 6 i s given (calculation is similar to
t h a t in Problem 3.24) a s
O 1
with principal stretches A:~, = 3, A:,, = 6, A&) = 2. Note also t h a t the principal stretches may
be calculated directly from (3.66) using [aij] above.
3.30. For the deformation of Problem 3.29, determine the spatial strain ellipsoid and show
that i t is of the form A:,,x? +
A?,,X; t A:,,X~ = 1.
By (3.84) the sphere xixi = 1 resulted from the ellipsoid X G X = 1, o r
This ellipsoid is put into the principalaxes form 3 ~ : 6x22 + +2 : 1 by the transformation
~ =
CHAP. 31 DEFORMATION AND STRAIN 103
3.31. Verify by direct expansion that the second invariant IILof the strain tensor may be
expressed by
IIL =
I 111
1.1
112
122 1 /+ 111
131
113
133 1 1 122
132
123
133 1
Expansion of the given determinants results in IIL = ll1l2, 1,,13, + + 133111  ( 1 f 2 + l i 3 + l i l ) .
I n comparison, direct expansion of the second equation of ( 3 . 9 1 ) yields
IIL = +[(111 + lZ2 + 133)ljj  ( l l j l l j + 12j12j + 13j13j)]
= &[(I11 + l22 + 133)(111 + l22 + l33)  ( l l l l l l + 112112 + 113113
+ l2l12l + l22l22 + l23l23 + 131131 + l32l32 + 133133)l
= 111122 + l22l33 f 133111  (1S2 + 123 + 1321)
3.32. For the finite homogeneous deformation given by ui = AijXj where Aij a r e constants,
determine an expression for the change of volume per unit original volume. If the
Aij a r e very small, show that the result reduces to the cubical dilatation.
Consider a rectangular element of volume having original dimensions d X l , d X 2 , d X 3 along the
+
coordinate axes. F o r the given deformation, xi = ( A i j S i j ) X j . Thus by ( 3 . 8 3 ) the original volume
d V o becomes a skewed parallelepiped having edge lengths d x i = ( A i c n ) Si(,)) d X ( , , , n = 1,2,3. From+
+
( 1 . 1 0 9 ) this deformed element has the volume d V = e i j k ( A i l S i 1 ) ( A j 2 4 S j 2 ) ( A k 3 Sk3) d X 1 d X 2 d X 3 . +
Then
F o r linear theory the cubical dilatation lii = auilaXi, which f o r ui = A i j X j is lii = A,, + A,, + A33,
t h a t e ( , ) = c(,)  e k k / 3 .
2 1
i)
=
e$ =
(1
O 1 u).
= 5 and so the
Note
11. ,./I O] [:
5 x 104 e,,
7 .:Op4 ][';I
llfi
= 4 x 1Op4
Therefore
12 x 104 + 2 ~ , ,= 4 x 104 or ,€
, = 2 x 104.
2
Next a rotation of 45O about the x: axis (90° in the Mohr diagram) results in the xl axes and
the associated strain tensor e / j given by
the first two rows of which represent the state of strain specified by point F in Fig. 316. Note
t h a t a rotation of 45O about xg would correspond to point E in Fig. 316.
CHAP. 31 DEFORMATION AND STRAIN 105
MISCELLANEOUS PROBLEMS
3.37. Derive the indicial form of the Lagrangian finite strain tensor L, of (3.40) from its
definition (3.37).
From (3.24), dx,/dXi = S i j + aui/dXi. Thus (3.37) may be written
Li$ = [(Ski f 
"k)
ax, ( Skj + axj
a?1i) 
= Ldui
2 ax,++
au,
axi ax, axj
auk auk
I
3.38. A displacement field is defined by xl = X I  CX2 BX3, xz = C X i + Xz  AX3, +
+
x3 =  B X l + AX2 X3. Show that this displacement represents a rigid body rotation
only if the constants A, B, C a r e very small. Determine the rotation vector w f o r the
infinitesimal rigid body rotation.
1 C B
F o r the given displacements, F = and from (3.37),
If products of the constants a r e neglected, this strain tensor i s zero and the displacement reduces
to a rigid body rotation. From (3.50), the rotation vector is
A A A
e1 e2 e3
W = 
2
â/dXl a/ax,
alax, = Aê, + sê, + cê3
CXz + BX3 C X l  AX3 BXl + AX,
3.40. For a state o£ plane strain parallel to the ~ 2 x 3axes, determine expressions for the
normal strain r;, and the shear strain r;, when the primed and unprimed axes are
oriented as shown in Fig. 319.
By equation (3.59),
"42 = ~o,cosesinel[~
o
€J[:;:]

 E, cos2 e + 2e23sin e cos O i sin2 o

 '22 + €33 +  2 eos 2e +
'22 €83
sin 20
2
4 =
=
[o, cos e, sin e,
E23
€23 C O S ~0
€7
E33
[si!
 €23
COS
e]
0
sin2 e + 9, sin e cos e
"22  "33
= €23 COS20  sin 20
2
What is the change in the 90" angle ADC depicted by the small tetrahedron OABC
in Fig. 320 if OA = OB = OC, and D is the midpoint of A B ?
The unit vectors Y and a t D a r e given by f; = (êl  ê2)/\/S and 2 = (2 ê3 ê2êl)/& . From
the result of Problem 3.20,
CHAP. 31 DEFORMATION A N D STRAIN 107
3.42. At a point the strain tensor is given by rij  i) and in principal form
+
3.43. For the displacement field xl = X I AX3, x z = X z  AX3, x3 = X3  A X l AX2, +
determine the finite strain tensor LG. Show that if A is very small the displacement
represents a rigid body rotation.
Since u , = AX3, u, = AX3, u3 = AXl + AX,, by (3.40),
+
3.44. Show that the displacement field ul = A x l 3x2, u2 = 3x1  Bxz, u3 = 5 gives a state
of plane strain and determine the relationship between A and B for which the
deformation is isochoric (constant volume deformation).
Solving simultaneously f o r e12 and ez2 yields e l z = ( b  c ) / 6 and E,Q = (a + 2b + 2c)/3.
108 DEFORMATION AND STRAIN [CHAP. 3
3.46. Derive equation (3.72) expressing the change in angle between the coordinate direc
tions Xp and X3 under a finite deformation. Show that (3.72) reduces to (3.65) when
displacement gradients are small.
Let y , ~= ~ / 2 8 be the angle change a s shown in Fig. 35. Then sin yz3 = cos (r12  8) =
A A
n, n,, or by (3.33) and (3.34)
sin Yzn =
dx2 .
dx3 

dX, F, F dX3 . .
Idx21 ldxzl d d ~ ,G dX2 d d ~ ,G dX3
Now dividing t h e numerator and denominator of this equation by IdX21 and IdX31 and using (3.353
and (3.66) gives
sin yz3 =
A
A
e, G e, 

A
e, G ê, .
,I ,I A(:,) A(&]
A
Next from (3.37), e,. G e3= e,A
(21, + I) *e3= ê2.2~~ e3 + e,
A A
l e3
A A A
= 2Lz3 since e2 e3 = 0.
Also from (3.68), A(;2) = d m , etc., and so
or 7mE + +
4~ m2m3 3mi = 3. Also mi +
rn; = 1, and solving simultaneously m2 = ffi12,
m, = 7112, o r m2 = O, m3 = f 1. Thus there is zero strain along the X3 axis and f o r the element
a t 60' to the X3 axis.
A
The student should verify this result by using the relation & 2LG m = O derived from (3.36).
Supplementary Problems
3.48. F o r the shear displacement of Problem 3.47, determine the equation of the ellipse into which the
circle X: +X: = 1 is deformed. Ans. x: 9%; = 3+
3.49. Determine the shear angle y,, f o r the deformation of Problem 3.47
(Fig. 322). Ans. y,, = sin12/$?
3.50. +
Given the displacement field x, = X, 2X3, x2 = Xz  2X,, x, =
+
X3  2X1 2X2, determine the Lagrangian and Eulerian finite strain
tensors L, and E*.
2 2 o 2 2 o
x,
Fig. 322
CHAP. 31 DEFORMATION AND STRAIN 109
3.51. Determine the principalaxes form of the two tensors of Problem 3.50.
Ans. 1; = ( c 4).
4 o
E: = (: :)
419 o
419
' i),
3.52. For the displacement field of Problem 3.50 determine the deformation gradient i, and by a polar
3.53. Show that the first invariant of 1, may be written in terms of the principal stretches as
ILG

  1 ) (h:ê2) + +
 1 ) (h2,,  1 ) ] / 2 . Hint: See equation (3.68).
(e31
and fl = ê1/2 +
+
$212 $,/\/S. A n s . e ( AV ) = 6 , y,, = 0.
3.55. Determine the principalaxes form of eij given in Problem 3.54 and note that and fl of that
problem are principal directions (hence y,, = 0).
Ans. eij 
o 2
3.56. Draw the Mohr's circle for the state of strain giv'en in Problem 3.54 and determine the maximum
shear strain value. Verify this result analytically. Ans. y, =4
3.58. For eij of Problem 3.54, determine the deviator tensor eij and calculate its principal values.
A=
2112 2x3
The necessary and sufficient condition for the inverse functions (4.2) to exist is that the
Jacobian determinant
J = laxiIaXil (4.3)
should not vanish. Physically, the Lagrangian description fixes attention on specific
particles of the continuum, whereas the Eulerian description concerns itself with a particu
lar region of the space occupied by the continuum.
Since (4.1) and (4.2) are the inverses of one another, any physical property of the con
tinuum that is expressed with respect to a specific particle (Lagrangian, or material descrip
tion) may also be expressed with respect to the particular location in space occupied by the
particle (Eulerian, or spatial description). For example, if the material description of the
density p is given by
p = P(Xi,t) or p = p(X,t)
the spatial description is obtained by replacing X in this equation by the function given
in (4.2). Thus the spatial description of the density is
P = p(Xi(x,t), t) = p*(~i,
t) ~r p = p(X(x, t), t) = p*(x>t) (4.5)
where the symbol p* is used here to emphasize that the functional form of the Eulerian
description is not necessarily the same as the Lagrangian form.
The time rate of change of any property of a continuum with respect to specific particles
of the moving continuum is called the material derivative of that property. The material
derivative (also known as the substantial, or comoving, or convective derivative) may be
thought of as the time rate of change that would be measured by an observer traveling
with the specific particles under study. The instantaneous position xi of a particle is itself
a property of the particle. The material derivative of the particle's position is the instan
taneous velocity of the particle. Therefore adopting the symbol dldt or the superpositioned
dot as representing the operation of material differentiation (some books use DIDt), the
( O )
where the second term on the right arises because the specific particles are changing position
in space. The first term on the right of (4.10) gives the rate of change a t a particular
[dP'.ir9
Iocation and is accordingly called the local rate of change. This term is sometimes written
t)] to emphasize that x is held constant in this differentiation. The second term
X
on the right in (4.10) is called the convective rate of change since it expresses the contribu
tion due to the motion of the particles in the variable field of the property.
which is used in taking the material derivati,ves of quantities expressed in spatial coordinates.
If, on the other hand, the displacement is in the Eulerian form zci = u ~ ( xt),
, then
In (6.15) the velocity is given implicitly since i t appears a s a factor of the second term on
the right. The function
vi = vi(x,t ) or V = V(X, t) (4.16 )
is said to specify the instantaneous velocity field.
The material derivative of the velocity is the acceleration. If the velocity is given in
the Lagrangian form (4.14), then
is called the rate of deformation tensor. Many other names are used for this tensor; among
them rate of strain, stretching, strain rate and velocity strain tensor. The skewsymmetric
tensor
The rate of deformation tensor is easily shown to be the material derivative o'f the
Eulerian linear strain tensor. Thus if in the equation
the differentiations with respect to the coordinates and time are interchanged, i.e. if
d t axj
(e)
is replaced by
dxj dt
, the equation takes the form
By the same procedure the vorticity tensor may be shown to be equal to the material
derivative of the Eulerian linear rotation tensor. The result is expressed by the equation
and the motion in the neighborhood of P is a rigid body rotation. For this reason a velocity
field is said to be irrotational if the vorticity tensor vanishes everywhere within the field.
that formed dX1 now form the differential line segment dx,'" = (dxilaXi)dX1. Similarly
dX2 becomes d ~ : = ~ (axildxn)
' dX2 and dX3 becomes dx13' = (axiIdX3)dXa. Therefore the
differential volume element dV is a skewed parallelepiped having edges dx?', dx:", dxi3)and
a volume given by the box product
dV = dxC1)x dxC2) dxC3) = Éijk dxi(l)dxj2' dxk' (4.37)
But (4.37) is seen to be equal to
dV = c..
axi
 dX d X 2d X 3 = J dVo
ax, ax, ax3 (4.38)
where J = laxiIdXjl is the Jacobian defined by (4.3).
Using (4.38), it is now possible to obtain the material derivative of dV as
d
since dVo is time independent, so that  (dVo)= O. The material derivative of the Jacobian
dt
J rnay be shown to be (see Problem 4.28)
For the initial configuration of a continuum, a differential element of area having the
magnitude dSOrnay be represented in terms of its unit normal vector ni by the expression
dSoni. For the current configuration of the continuum in motion, the particles initially
making up the area d S O nnow fill an area element represented by the vector dSnt or dSi. It
rnay be shown that
ax.
dSi= J L d X j or dS= JdX*Xvx (4.42)
axi
from which the material derivative of the element of area is
The material derivative of the squared length of the differential line element dxi rnay
be calculated as follows,
where V is the volume that the considered part of the continuum occupies a t time t. The
material derivative of Pij.. .(t) is
and since the differentiation is with respect to a definite portion of the continuum (i.e. a
specific mass system), the operations of differentiation and integration may be interchanged.
Therefore
which, upon carrying out the differentiation and using (4.411, results in
Since the material derivative operator is given by (4.12) as dldt = d/& + v~aldxp, (4.52)
may be put into the form
dt P;. . .(x, t) dV =
V
+ ~(aX vPPt
P . . (x, t))] dV
By using Gauss' theorem (1.157), the second term of the righthand integral of (4.53) may
be converted to a surface integral, and the material derivative then given by
P: ( x , t ) d v = . '(xft, dV
at
+ 1 v,[P:. . .(x, t)] dSp (4.54)
This equation states that the rate of increase of the property Pij...(t) in that portion of the
continuum instantaneously occupying V is equal to the sum of the amount of the property
created within V plus the flux V,[P;. . .(x, t)] through the bounding surface S of V.
The procedure for determining the material derivatives of surface and line integrals is
essentially the same as that used above for the volume integral. Thus for any tensorial
property of a continuum represented by the surface integral
where S is the surface occupied by the considered part of the continuum a t time t, then, as
bef ore,
& 1Q: . . .(x, t) dSp = i& [Q: . . .(x, t) dSp] (4.56)
CHAP. 41 MOTION AND FLOW 117
R 13.. .( t ) = IR; C
...(~ , t ) d x , (4.58)
the material derivative is given by
Differentiating the right hand integral as indicated in (4.59), and making use of (4.45),
results in the material derivative
d
[R
dt
ij...(t)] = J C
d[R:. . .(x, t)]
dt dx, + 12 [ ~ ~ . . ( x , dt x) ,] (4.60)
Solved Problems
9
4.4. Integrate the velocity equations of Problem 4.3 to obtain the displacement relations
xi = xi(X, t ) and from these determine the acceleration components in Lagrangian
form for the motion.
+ +
By (4.13), v , = d x l / d t = x l l ( l t ) ; separating variables, dxllxl = d t l ( 1 t ) which upon inte
+ +
gration gives ln x , = ln ( 1 t ) ln C where C is a constant of integration. Since xl = X , when
+ +
t = O, C = X , and so x1 = X l ( l t). Similar integrations yield z2 = X 2 ( l t ) 2 and x, = X3(1 t)3. + '
+
Thus from (4.14) and (4.17), v , = X 1 , v , = 2X2(1 t ) , v , = 3X3(1 + t)2 and a, = O, a, = 2X,,
+
a , = 6X3(1 t).
4.5. +
The motion of a continuum is given by xl = A (eBAIX)sinA(A w t ) , x2 = B  +
+
(ecBVX)cos X(A d ) , x3 = X3. Show that the particle paths are circles and that the
velocity magnitude is constant. Also determine the relationship between X l and X Z
and the constants A and B.
By writing x ,  A = (eBVk) sin x ( A ~ t ) x, , + +
B = (eBV/h) cos X(A ~ t ) then
, +
squaring
and adding, t is eliminated and the path lines are the circles ( x l A)2 +
( x , B)2 = ezBh/X. +
+ +
From (4.6), v l = oeBh cos h ( A wt), v , = weBA sin h ( A wt), v , = O and v2 = v: v; vi = + +
+
~2ezBA. Finally, when t = O, xi = Xi and so X , = A (eBh/X) sin x A , X , = B  ( e  B ~ x cos
) xA.
4.6. A velocity field is specified by the vector v = xStêl x2t2ê2 x1x3tê3. Determine+ +
the velocity and acceleration of the particle a t P(1,3,2) when t = 1.
By direct substitution, v, = e,
A
+ 3 ê2+ 2 ê3. Using the vector form of (4.18) the accelera
tion field is given by
a = xSê, + 2x2tê2 + x1x3ê3 + + x2t2ê2 + x1x3tê3)
(xStê,
+ x3tê,ê3 + t2ê$, + x1tê3ê3)
(2x1tê$,
or a = (xS + 2x1t2)ê, + (2x2t + %,e)ê2+ (51x3 + 2x x3t2)ê3
Thus ap = 3 ê l + 9 ê 2 + 6 ê , .
4.7. For the velocity field of Problem 4.3 determine the streamlines and path lines of the
flow and show that they coincide.
A t every point on a streamline the tangent is in the direction of the velocity. Hence for the
differential tangent vector d x along the streamline, v X d x = O and accordingly the differential
equations of the streamlines become d x l l v l = dx2/vz = dx3/v,. For the given flow these equations
are d x l l x l = dx,/2x2 = dx3/3x3. Integrating and using the conditions xi = Xi when t = 0, the
equations of the streamlines are ( x l / X l ) z= x,/X,, (x1/X1)3= x3/X3, (x2/X2)3= (x3/X3)2.
Integration of the velocity expressions dxi/dt = vi a s was carried out in Problem 4.4 yields
+ + +
the displacement equations x l = X l ( l t ) , x2 = X z ( l t)*, x3 = X 3 ( l t)3. Eliminating t from
these equations gives the path lines which are identical with the streamlines presented above.
C H A P . 41 MOTION AND FLOW 119
"
from it compute the rate of deformation tensor D and the spin tensor V for the point
P(1,0,3) when t = 0.
Likewise, decomposition of the displacement gradient gives aui/dxj = cij + uij. Thus
120 MOTION AND FLOW [CHAP. 4
!
e% 3e3t/2
ds,ldt = e2t O = Dij
3e3tI2
O O o
The student should show that dwijldt = V i j .
4.12. A vortex line is one whose tangent a t every point in a moving continuum is in the
direction of the vorticity vector q. Show that the equations for vortex lines are
dxilqi = dx21q2 = dx31q3.
Let d x be a differential distance vector in the direction of q. Then q X d x E O , or
(9, dx3  q3 dx,) 2, 3 (93 dxl  d ~ 3 $2
) (91 dxz  9, d x i ) ê 3 E O
from which dxllql = dx21q2 = dx31q3.
4.13. Show that for the velocity field v = ( A x s  Bx2)êl (Bxl  Cx3)êz + + ( C X Z Axl)ê3
the vortex lines are straight lines and determine their equations.
+ +
From (4.29), q = V , X V = 2 ( C ê l A 3, sê3),and by Problem 4.12, the d.e. for the vortex
lines are A dx3 = B dx,, B d x , = C dx,, C d x z = A dxl. Integrating these in turn yields the equa
+ +
tions of the vortex lines x3 = Bx21A K1, x 1 = Cx31B K,, x z = A x J C K , where the Ki are +
constants of integration.
4.14. Show that the velocity field of Problem 4.13 represents a rigid body rotation by
showing that D = 0.
Calculating the velocity gradient avilaxj, i t is found to be antisymmetric. Thus avildxj =
(; r )
O B A
f = V i j and Dij = O.
4.15. For the rigid body rotation v = 3x3ê1 4x3& + (4x2 3xl)ê3, determine the rate of
rotation vector O and show that v = O X x.
From (4.30), 2iZ = q, or i2 = 4 ê 1 + 3ê2. This vector is along the axis of rotation. Thus
( 4 ê l f 3 ê 2 )X ( ~ ~ ê ~ f ~ =
~ ê3x3ê1
~ +  4x3ê2 +
~ ~ ê ( 4~x 2) 3 x 1 ) ê 3 = V
4.16. A steady velocity field is given by v = ( x ;  xix:)êl (x: $2 xz)ê2. Determine the + +
unit relative velocity with respect to P(1,1,3) of the particles a t Q1(1,0,3),Q2(1,3/4,3),
Q3(1,7/8,3) and show that these values approach the relative velocity given by (4.26).
By direct calculation v ,  v n , = e1 2 e,,
h
+ h
4(vp  vQ2)= 7ê1/4 + 2ê2 and 8 ( v p V Q3) =
15$,/8 +
22,. The velocity gradient matrix is
3x1  xp 2xix2
CHAP. 41 MOTION AND FLOW 121
Thus ( d v l d x ) ~= 22, + 2ê2 which i s the value approached by the relative unit velocities
vp  Vgi. e2
4.18. For the motion of Problem 4.17 determine the rate of shear a t P between the orthog
onal directions 2 = (3êi  4&)/5 and fi = (4&+ 3ê3)/5.
In analogy with the results of Problem 3.20 the shear rate Y, is given by Y, = c *28 2, or
in matrix form
Y, = [4/5,O, 3/51
4.19. A steady velocity field is given byv1 = 2x3, v2 = 2x3, 213 = O. Determine the principal
E : r]
directions and principal values (rates of extension) of the rate of deformation tensor
for this motion.
Here avi/axiI =
0
O
0
0
0
2
0
=
O
+ [ ;]
1
o o
o o
1
1
and for principal values
A of D,,
4.22. Determine the material derivative pidSi of the flux of the vector property pi
dt
through the surface S.
BY (4.5%
4.23. Show that the transport theorem derived in Problem 4.22 may be written in symbolic
notation as
$A P * & ~ s = [$ + v(T7.p) + V x ( p x v ) ] . & d s
By a direct transcription into symbolic notation of the result in Problem 4.22,
CHAP. 41 MOTION AND FLOW 123
4.24. Express Reynold's transport theorem a s given by equations (4.53) and (4.54) in
symbolic notation.
Let P*(x, t ) be any tensor function of the Eulerian coordinates and time. Then (4.53) is
4.25. If the function P*(x, t ) in Problem 4.24 is the scalar 1, the integral on the left is
simply the instantaneous volume of a portion of the continuum. Determine the
material derivative of this volume.
Using the vector form of (4.58) a s given in Problem 4.24, $i i dV = V . v dV. Here
V v d V represents the rate of change of dV, and so V v is known a s the cubical rate of dilata
tion. This relationship may also be established by a direct differentiation of (4.38). See Problem 4.43.
MISCELLANEOUS PROBLEMS
4.26. From the definition of the vorticity vector (4.29), q = curl v, show that qi = e i j k V k j
and that 2Vii = ejikqk.
By (4.29), qi = ~ ~= ~ ~ ~ ~ +
~ v (and ~, ~ ~ , =
~v since ~ O ~(see, for example, Problem
= (GriSsk  S r k S s i ) V k j = 2Vsr.
1.50), qi = qjkV[k,jl = eiikVkj. From this result q,,qi = eirsYjkVkj
av
4.27. Show that the acceleration a may be written a s a = 
dt
+ q x v + +Vv2.
avi avi
From (4.18), ai = 
at + v , axk
 and so
which, a s the student should confirm, is the indicial form of the required equation.
4.29. Show that for steady motion (avilat = 0) of a continuum the streamlines and
pathlines coincide.
As shown in Problem 4.7, a t a given instant t streamlines are the solutions of the differential
equations d x l l v l = dx21v2 = dx31v3. Pathlines are solutions of the differential equations dxildt =
vi(x, t). If vi = vi(x), these equations become d t = dx,lv, = dxZ/v2= dx3/v3 which coincide with
the streamline differential equations.
4.30. +
For the steady velocity field v1 = x l x z x i , v2 = x;  ~ 1 x 2 ,v3 = O determine
expressions for the principal values of the rate of deformation tensor D a t an arbitrary
point P(x1, X Z , x3).
4.31. Prove equation (4.43) by taking the material derivative of d S i in its cross product
form d S i = cijkd x j 2 )d x k 3 ) .
axi axi a%, axk
Using (S.&?), dSi = ~ ~ ~ ~ ( a xdXz(axklaX3)
~ / a X ~ ) d X 3 and axi  dSi = qjk dX2dX3 = ax,
ax ax. dSi = 6, dSi = d S p = 
8x1
J d X z dX3. Thus 2 J d X z d X 3 and by Problem 4.28,
axP ax, a x ~
4.32. Use the results of Problems 4.27 and 4.23 to show that the material rate of change
of the vorticity flux 
d"t L
q *6 dS equals the flux of the curl of the acceleration a.
i$ d~ = Sv
[Eijkak,i (.,srn~rn~r),sl dv
Supplementary Problems
A continuum motion is given by x 1 = Xlet + X3(et 1 ) , x2 = X 2 + X,(et  et), x3 = X3. Show
that J does not vanish for this motion and obtain the velocity components.
+
Ans. v , = ( X , X3)et, v 2 = X,(et +
et), v3 = O or v , = x1  x3, v2 = x3(et et),
i v3 = O
Show that for the flow vi = xil(l +t) the streamlines and path lines coincide.
The electrical field strength in a region containing a fluid flow is given by h = ( A cos 3t)lr where
r2 = x; +
x i and A is a constant. The velocity field of the fluid is v , = 2 2 , v , = x;  XSX, +
xixi, v , = O. Determine dXldt a t P(x,, x,, x,). Ans. dhldt = (3A sin 3t)lr
Show that for the velocity field v , = x l x 2 + 22, v , = x;  xixi, v3 = O the streamlines are
circular.
i
+ +
For the continuum motion x , = X,, x, = et(X2 X3)/2 et(X2  X,)/2, x3 = et(X2 X,)/2  +
c~~(X ,X3)12, show that Dij = deijldt a t t = O. Compare these tensors a t t = 0.5.
For the velocity field v , = s;x, + si, v , = (xt + x,x;), v3 = 0, determine the principal axes
and principal values of ü a t P ( l , 2,3).
Am. D: = (:5 0
o
o
o);
5
a = ( 31
1
O
3
11
o
o
For the velocity field of Problem 4.41 determine the rate of extension in the direction
V  (3, 2 3 ,
A
h
+
2 ê 3 ) / 3 a t P ( l , 2,3). What is the maximum shear rate a t P?
Ans. d"' = 2419, Ymax = 5
Show that d(axilaXj)ldt = v~,,x,,~ and use this to derive (4.41) of the text directly from (4.38).
Prove the identity ~pq,(v,vr,,),q = qPvq,, +vqqP,,  qqvP,, where vi is the velocity and qi the
vorticity. Also show that v,, jvi,i = Dij Dij  qiqi/2.
Fundamental Laws
of Continuum Mechanics
Since this equation holds for an arbitrary volume V , the integrand must vanish, or
This equation is called the continuity equation; using the material derivative operator it
may be put into the alternative form
where the integrals are taken over the same particles, i.e. V is the volume now occupied by
the material which occupied V0 a t time t = O . Using (4.1)and (4.38),the right hand integral
in (5.7) may be converted so that
CHAP. 51 FUNDAMENTAL LAWS OF CONTINUUM MECHANICS 127
Since this relationship must hold for any volume VO,it follows that
Po = PJ (5.9)
which implies that the product pJ is independent of time since V is arbitrary, or that
d
~ ( P J=
) 0 (5.10)
Equation (5.10) is the Lagrangian differentialform of the continuity equation.
Fig. 51
Based upon Newton's second law, the principle o f linear momentum states that the time
rate of change of an arbitrary portion, of a continuum is equal to the resultant force acting
upon the considered portion. Therefore if the interna1 forces between particles of the con
tinuum in Fig. 51 obey Newton's third law of action and reaction, the momentum principle
for this mass system is expressed by
Upon substituting tin' = ajini into the first integral of (5.12) and converting the result
ing surface integral by the divergence theorem of Gauss, (5.12) becomes
In calculating the material derivative in (5.13), the continuity equation in the form given
by (5.10) may be used. Thus
Replacing the right hand side of (5.13) by the right hand side of (5.14) and collecting terms
results in the linear momentum principle in integral form,
128 FUNDAMENTAL LAWS OF CONTINUUM MECHANICS [CHAP. 5
Since the volume V is arbitrary, the integrand of (5.15) must vanish. The resulting
equations,
in which xj is the position vector of the volume element dV. The moment of momentum
principle states that the time rate of change of the angular momentum of any portion of a
continuum with respect to an arbitrary point is equal to the resultant moment (with respect
to that point) of the body and surface forces acting on the considered portion of the con
tinuum. Accordingly, for the continuum of Fig. 51, the moment of momentum principle
is expressed in integral form by
Equation (5.19) is valid for those continua in which the forces between particles are equal,
opposite and collinear, and in which distributed moments are absent.
The moment of momentum principle does not furnish any new differential equation of
h
motion. If the substitution t p ' = uPknpis made in (5.19), and the symmetry of the stress
tensor assumed, the equation is satisfied identically by using the relationship given in (5.16).
If stress symmetry is not assumed, such symmetry may be shown to follow directly from
h
since Vijuji O. Finally, converting the first integral on the right hand side of (5.24) to a
surface integral by the divergence theorem of Gauss, and making use of the identity
ti"' = ~31. . n3 . ,the energy equation for a continuum appears in the form
This equation relates the time rate of change of total mechanical energy of the continuum
on the left side to the rate of work done by the surface and body forces on the right hand
side of the equation. The integral on the left side is known as the time rate of change of
internal mechanical energy, and written dUldt. Therefore (5.25) may be written briefly a s
where d W l d t represents the rate of work, and the special symbol d is used to indicate that
this quantity is not an exact differential.
If both mechanical and nonmechanical energies are to be considered, the principle of
conservation of energy in its most general form must be used. In this form the conservation
principle states that the time rate o f change of t h e kinetic plus the internal energy i s equal
t o t h e s u m o f the rate o f work plus a11 other energies supplied to, or removed f r o m the
continuum per unit time. Such energies supplied may include thermal energy, chemical
energy, or electromagnetic energy. In the following, only mechanical and thermal energies
are considered, and the energy principle takes on the form of the wellknown first law of
thermodynamics.
For a thermomechanical continuum it is customary to express the time rate of change
of internal energy by the integral expression
where u is called the specijic internal energy. (The symbol u for specific energy is so well
established in the literature that it is used in the energy equations of this chapter since
there appears to be only a negligible chance that it will be mistaken in what follows for the
magnitude of the displacement vector ui.) Also, if the vector s is defined as the heat fEux
per unit area per unit time by conduction, and z is taken a s the radiant heat constant per
unit mass per unit time, the rate of increase of total heat into the continuum is given by
Converting the surface integrals in (5.30) to volume integrals by the divergence theorem of
Gauss, and again using the fact that V is arbitrary, leads to the local form of the energy
equation:
or
Within the arbitrarily small volume element for which the local energy equation (5.31)
is valid, the balance of momentum given by (5.16) must also hold. Therefore by taking the
scalar product between (5.16) and the velocity p'Uivi = vioji,j+ pvibi and, after some simple
manipulations, subtracting this product from (5.31),, the result is the reduced, but h i g h l ~
useful form of the local energy equation,
This equation expresses the rate of change of internal energy as the sum of the stress power
plus the heat added to the continuum.
where d s is the increase in specific entropy, d s c e )is the increase due to interaction with the
exterior, and ds'" is the internal increase. The change dsCi)is never negative. I t is zero
for a reversible process, and positive for an irreversible process. Therefore
ddi) > O (irreversible process) (5.34)
ds(i) = O (reversible process) (5.35)
In a reversible process, if d q ( ~denotes
) the heat supplied per unit mass to the system, the
change dsCe)is given by
dS(e) = dq(Ri (reversible process) (5.36)
T
where e is the local entropy source per unit mass. The equality in (5.37) holds for reversible
processes; the inequality applies to irreversible processes.
The ClausiusDuhem inequality is valid for arbitrary choice of volume V so that trans
forming the surface integral in (5.37) by the divergence theorem of Gauss, the local form
of the internal entropy production rate y, per unit mass, is given by
This inequality must be satisfied for every process and for any assignment of state variables.
For this reason it plays an important role in imposing restrictions upon the socalled
constitutive equations discussed in the following section.
In much of continuum mechanics, it is often assumed (based upon statistical mechanics
of irreversible processes) that the stress tensor may be split into two parts according to
the scheme,
a,. = a!?) + u 'iDj ' (5.39)
where u::) is a conservative stress tensor, and is a dissipative stress tensor. With this
assumption the energy equation (5.32) may be written with the use of (4.25) a s
1
In this equation,  o;?) iij is the rate of energy dissipated per unit mass by the stress, and
0
d q l d t is the rate Af heat influx per unit mass into the continuum. If the continuum under
goes a reversible process, there will be no energy dissipation, and furthermore, dqldt =
d q ( ~ ) l d t so
, that (5.40) and (5.36)may be combined to yield
Therefore in the irreversible process described by (5.40),the entropy production rate may
be expressed by inserting (5.41). Thus
132 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS [CHAP. 5
The scalar ;ij is called the dissipation function. For an irreversible, adiabatic process
(dq = O), dsldt > O by the second law, so from (5.42) it follows that the dissipation function
is positive definite, since both p and T are always positive.
Assuming that body forces bi and the distributed heat sources z are prescribed, (5.43),
(5.44) and (5.45) consist of five independent equations involving fourteen unknown functions
of time and position. The unknowns are the density p, the three velocity components vi,
(or, alternatively, the displacement components ui), the six independent stress components
oij, the three components of the heat flux vector ci, and the specific interna1 energy u. In
addition, the ClausiusDuhem inequality (5.38)
which governs entropy production, must hold. This introduces two additional unknowns:
the entropy density s, and T, the absolute temperature. Therefore eleven additional equa
tions must be supplied to make the system determinate. Of these, six will be in the form
known as constitutive equations, which characterize the particular physical properties of
the continuum under study. Of the remaining five, three will be in the form of temperature
heat conduction relations, and two will appear as thermodynamic equations of state; for
example, perhaps as the caloric equation of state and the entropic equation of state. Specific
formulation of the thermomechanical continuum problem is given in a subsequent chapter.
I t should be pointed out that the function of the constitutive equations is to establish
a mathematical relationship among the statical, kinematical and thermal variables, which
will describe the behavior of the material when subjected to applied mechanical or thermal
forces. Since real materials respond in an extremely complicated fashion under various
loadings, constitutive equations do not attempt to encompass a11 the observed phenomena
related to a particular material, but, rather, to define certain ideal materials, such as the
ideal elastic solid or the ideal viscous fluid. Such idealizations or material models as they
are sometimes called, are very useful in that they portray reasonably well over a definite
range of loads and temperatures the behavior of real substances.
CHAP. 51 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS 133
Solved Problems
CONTINUITY EQUATION (Sec. 5.1)
5.1.' An irrotational motion of a continuum is described in Chapter 4 as one for which the
vorticity vanishes identically. Determine the form of the continuity equation for
such motions.

B y (1.29),curl v = O when q 0, and so v becomes the gradient of a scalar field +(xi, t ) (see
+
Problem 1.50). Thus vi = +,i and (5.3)is now dpldt p+,kk = O or dpldt pV2+ = 0 . +
5.2. If P:?, .(x, t) represents any scalar, vector or tensor property per unit mass of a con
tinuum so that P:. . .(x, t) = pp". .(x, t) show that
5.3. Show that the material form d(pJ)ldt = O of the continuity equation and the spatial
+
form dpldt p v , , , = O are equivalent.
Differentiating, d ( p J ) l d t = ( d p l d t ) J + p dJ/dt = O and from Problem 4.28, d J l d t = J v k , , so
+
that d ( p J ) l d t = J ( d p l d t P V , , ~ )= 0 .
5.4. Show that the velocity field vi = Axil?, where xixi = r2 and A is an arbitrary con
stant, satisfies the continuity equation for an incompressible flow.
From (5.5) v k , k = O for incompressible flow. Here
5.5. For the velocity field vi = x i l ( l + t), show that pxlx,x3 = p,,X1X2X3.
+
Here vk,k = 3 / ( 1 t ) and integrating (5.3) yields ln p = 1n (1 t)3 ln C where C is a + +
constant of integration. Since p = po when t = 0, this equation becomes p = pd(l t)3. Next +
by integrating the velocity field dxi/xi = d t / ( l + t ) (no sum on 9, xi = X i / ( l t ) and hence +
Pxlx2x3 =~ 0 ~ 1 ~ 2 ~ 3 .
5.7. If distributed body moments m i per unit volume act throughout a continuum, show
that the equations of motion (5.16) remain valid but the stress tensor can no longer
be assumed symmetric.
Since (5.16) is derived on the basis of force equilibrium, i t is not affected. Now, however, (5.19)
acquires a n additional term so that
which reduces to (see Problem 2.9) ( Y j k a j k+mil dV = O, and because V i s arbitrary, eijkajk +
mi = O for this case.
5.8. The momentum principle in differential form (the socalled local or "in the small"
form) is expressed by the equation d ( p v , ) l d t = pbi + ( u i j  ~ V ~ V ~ Show
) , ~ . that the equa
tion of motion (5.16) follows from this equation.
Carrying out the indicated differentiation and rearranging the terms in the resulting equation
yields
~,(a~/+
a t p,jvj + pvj, +j) p(avi/Jt + vivi, j) pbi aij, j
The first term on the left is zero by (5.4) and the second term is pai. Thus pai = pbi + aij,j which
is (5.16).
Substituting oPknpfor tp' in (5.19) and applying the divergence theorem (1.157) to the resulting
surface integral gives
Using the results of Problem 5.2, the indicated differentiations here lead to
5.10. For a rigid body rotation about a point, vi = cijkojxk. Show that for this velocity
(5.19) reduces to the wellknown momentum principle of rigid body dynamics.
The left hand side of (5.19) is the total moment Mi of a11 surface and body forces relative to
the origin. Thus for vi = eijkwjXk,
=
d
[O. J"~ ( 4 P q x q
where Zip = p(8ipzqzq 44 d V is the moment of inertia tensor.
From (5.2?)),
K = i p Y d v = i peiikUj%k€ipq~pZq
dv
= ii ~ ~ p w j ( 8 j p~6ki q~S k p ) x k z q dV
 wiw~
2 i p(6ipzqzq  z P z $ d V wiwpzip

2
o.l.0,
In symbolic notation note that K =  2
5.12. At a certain point in a continuum the rate of deformation and stress tensors are
given by
4 2 5
and
5.13. If uij = pSij where p is a positive constant, show that the stress power may be
.
p dp
expressed by the equation D i j u i j = 
P dt
By (4.19), Dij = v i B j Vij; and since V i j u i j= O, i t follows that Dijoij = = pvi,i.
From the continuity equation (5.3), vi,i = (llp)(dpldt) and so D i j q j = (plp)(dpldt) for oij = paij.
5.14. Determine the form of the energy equation if uij = (p + h* Dkk)8,, 4 2 , ~D,,
* and the
heat conduction obeys the Fourier law ci = kTai.
From (5.92),
du
p d t
 (p+ X*Dkk)SijDij 2p*DijDij

+ kTSii + +z
 P
 do + +
(h* 2 / ~ * ) ( 1 , )
P dt
~ 4/~*11, kT,ii + +z
where I, and 11, are the first and second invariants respectively of the rate of deformation tensor.
136 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS [CHAP. 5
5.15. If uij =  p s i j , determine an equation for the rate of change of specific entropy during
a reversible thermodynamic process.
Here oij '=
::o
dt
ds
and (5.41) gives T  = 
dt
du
+ 
p dp
p2 dt
upon use of the result in Problem 5.13.
5.16. For the stress having = /3DikDkj, determine the dissipation function in terms of
the invariants of the rate of deformation tensor D.
Here by (4.25), o:;) iij = PDsDkjDij which is the trace of D~ (see page 16) and may be evaluated
by using the principal axis values D ( , , , D ( 2 ) ,DC3). Thus by (1.138) the trace
DiiD,Dki = D:,) i D:Z) + ~ ( 3 )
5.20. Show that the constitutive equation of Problem 5.19 may be split into the equivalent
equations aii = (3A* + 2p*)Dii and sij = 2p*DG where sij and DC are the deviator ten
sors of stress and rate of deformation, respectively.
+ +
Substituting U i j = S i j Sijukk/3 and Dij = D;j SijDkk/3 int0 uij = h*SijDkk 2@*Dij 0f +
+ +
Problem 5.19 resuits in the equation sij aijukk/3 = h*SijDkk 28*(Dii SijDkk/3). From this +
when i f j, sij = 2@*D& and hence ukk = ('&h* 2@*)Dkk. +
MISCELLANEOUS PROBLEMS
581. Show that +
= ( ~ , , a ~ , ~ qjvi,j)lpwhere p is the density, ai the acceleration and
qi the vorticity vector.
By direct differentiation
dd, (f) = 4 

P p2'
But 4 = rijkak,i + qivi,i  qivjj (see Problem
4.32); and by the continuity equation (5.4, e = pvci. Thus
5.26. If the continuum of Problem 5.25 is considered incompressible, show that the diver
gente of the vorticity vanisbes and give the form of the equations of motion for
this case.
By (4.29), qi = e i j k v k , and div q = eijkvk,ji = O since eijk is antisymmetric and vk,ji is sym
metric in i and j . Thus for V v = O the equations of motion become p'Ui = pbi  p,i p*vi,jj +
or in Gibbs notation p; = p b  V p p*VZv. +
5.27. Determine the material rate of change of the kinetic energy of the continuum which
occupies the volume V and give the meaning of the resulting integrals.
,.
By (5.23). dK/dt = J pvi'UidV. Also the total stress power of the surface forces is
. L viti'' d S which may be written
V
viuijni d S and by the divergence theorem (1.157) and the
This sum of integrals represents the rate of work done by the body forces, the interna1 stresses
and the surface stresses, respectively.
5.29. For a continuum with uij =  p 8 . .23'. the specific e n t h a l p y h = u p l p . Show that the +
+
energy equation may be written h = +Ip Ti using this definition for the enthalpy.
From (5.41), = pSijDijlp .+ T S for the given stress; and by the result of Problem 5.13 and
the definition of h, ii = a  +Ip  p;/pZ = p;/pZ + T B . Canceling and rearranging, i= ;IP + TS.
5.30. If the continuum of Problem 5.25 undergoes an incompressible flow, determine the
equation of motion in terms of the vorticity q in the absence of body forces and
assuming constant density.
For incompressible flow, V v = 0 ; and if b 0, the equation of motion in Problem 5.25
reduces to pCi = pZi +
P * v i , j j . Taking the cross product V X with this equation for p = constant
+
gives c ~ ~ =~ B~ ~~ , ~ ~ ~( Pp* / P, ) ~~p q~i v i ,l j jBut
q~. ~ ~ = O ~ and ~by (4.29)
p the
, result
~ ~is 4, =
( ~ * / p ) q , ~In~ .symbolic notation, dqldt = (p*lp)V2q.
CHAP. 51 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS 139
Supplementary Problems
5.31. Show that for the rate of rotation vector a, P
5.32. Show that the flow represented by v , =  ~ X , X ~ ~ /v T = , (x:  x~)x,l+, v , = %,/r2 where
, ~
r2= x; +
xg, satisfies conditions for an incompressible flow. 1s this motion irrotational?
5.34. +
Show that the flow v, = (1 rz) cos elrz, v e = (1 r,) sin #/r,, v , = O satisfies the continuity
equation in cylindrical coordinates when the density p is a constant.
5.36. I f a continuum i s subjected to a body momeht per unit mass h in addition to body force b, and a
couple stress g'"^ in addition to the stress t'"', the angular momentum balance may be written
where m i s distributed angular momentum per unit mass. If G = g'L', show that the local form
of this relation is p dmldt = h V G Z,.+ +
5.37. If a continuum has the constitutive equation oii = pGij + PDij + aDikDkj, show that uii =
3(p  2 ~ ~ 1 1 ~ 1 3Assume
). incompressibility, Dii = 0.
5.38. For a continuum having uij = pSij, show that du = T ds  p dv where in this problem v = l I p ,
the specific volume.
5.39. I f T dsldt = vi,ilp and the specific free energy is defined by * = u  Ts, show that the energy
+
equation may be written p dqldt ps dTldt = oiiDij.
L = E = ~ ( u v+
, Vxu) = . &(uVx+ Vxu) = 4(uV + Vu)
In the following it is further assumed that the deformation processes are adiabati; (no heat
loss or gain) and isothermal (constant temperature) unless specifically stated otherwise.
The constitutive equations for a linear elastic solid relate the stress and strain tensors
through the expression 
uii = Cijkmekm or I: = C : E (62)
which is known a s the generalized Hooke's law. In (6.2) the tensor of elastic constants
Cijkm has 81 components. However, due to the symmetry of both the stress and strain
tensors, there are a t most 36 distinct elastic constants. For the purpose of writing Hooke's
law in terms of these 36 components, the double indexed system of stress and strain com
ponents is often replaced by a single indexed system having a range of 6. Thus in the
notation   
'=H  '=I u23  u32  O4
and

€33  €3 2c12 = 2~~~= E6
The interna1 energy in this case is purely mechanical and is called the strain energy (per
unit mass). From (6.6),
1
du =  u..d,,, (6.7)
P 23
and if u is considered a function of the nine strain components, u = u(c,,), its differential
is given by
au
du = de..
a,,, 3'
Furthermore, the zero state of strain energy may be chosen arbitrarily; and since the stress
must vanish with the strains, the simplest form of strain energy function that leads to a
linear stressstrain relation is the quadratic form
U* = &Cijkmcijckm (6.12)
From (6.2), this equation may be written
U* = +uijeij or U* = &Z: E (6.18)
In the single indexed system of symbols, (6.12) becomes
U* = &CKMcKeM (6.14)
in which CKM= CMK. Because of this symmetry on CKM,the number of independent elastic
constants is a t most 21 if a strain energy function exists.
When a strain energy function exists for the body, CKM= CMK, and the 36 constants in
(6.15) are reduced to 21.
142 LINEAR ELASTICITY [CHAP. 6
Inserting the values of (6.17) into the transformation laws for the linear stress and strain
tensors, (2.27) and (3.78) respectively, the elastic matrix for a material having ~ 1 x 2as a
plane of symmetry is
C11 C12 c13 0 O
C21 C22 C23 0 0
The 20 constants in (6.18) are reduced to 13 when a strain energy function exists.
If a material possesses three mutually perpendicular planes of elastic symmetry, the
material is called orthotropic and its elastic matrix is of the form
For isotropy, the number of independent elastic constants reduces to 2, and the elastic
matrix is symmetric regardless of the existence of a strain energy function. Choosing as
the two independent constants the wellknown Lamé constants, A and p, the matrix (6.19)
reduces to the isotropic elastic form
where C = c,, = IE. This equation may be readily inverted to express the strains in terms
of the stresses as
C.. =
A
+
2 ~ ( 3 A 2 ~S i j )u k k
+ 2P
1
oij ~r E =
A
2 ~ ( 3 +2P)
h
I@+E
1
2P
where o = u,, = I,, the symbol traditionally used in elasticity for the first stress ihvariant.
For a simple uniaxial state of stress in the X I direction, engineering constants E and V
may be introduced through the relationships u,, = Ec,, and c,, = C, = vc,,. The constant
E is known as Young's modulus, and V is called Poisson's ratio. In terms of these elastic
constants Hooke's law for isotropic bodies becomes
E
u..
13
= 
l E
+ v ( ) or E =  1( + V E +:ic)1  2~
or, when inverted,
 l + v E = 
l + v x Li@
E 001"
c..  E U i j Lf3 i j ukk E E
From a consideration of a uniform hydrostatic pressure state of stress, it is possible to
define the bulk modulus,
which relates the pressure to the cubical dilatation of a bpdy so loaded. For a socalled state
of pure shear, the shear modulus G relates the shear components of stress and strain. G
is actually equal to p and the expression
 E
p  G = 
2(1+ v)
may be proven without difficulty.
( b ) Hooke's law,
uij = X a i j c k k + 2peij or 2 = hlc + ~,.LE (6.28)
which are called the NavierCauchy equations. The solution of this type of problem is
therefore given in the form of the displacement vector Ui, satisfying (6.31) throughout the
continuum and fulfilling (6.30) on the boundary.
For those problems in which surface tractions are prescribed everywhere on the
boundary by equations of the form
the equations of compatibility (3.104) may be combined with Hooke's law (6.24) and the
equilibrium equation (6.27) to produce the governing equations,
which are called the BeltramiMichell equations of compatibility. The solution for this
type of problem is given by specifying the stress tensor which satisfies (6.33) throughout
the continuum and fulfills (6.32) on the boundary.
For those problems having "mixed" boundary conditions, the system of equations (6.27),
(6.28) and (6.29) must be solved. The solution gives the stress and displacement fields
throughout the continuum. The stress components must satisfy (6.32) over some portion
of the boundary, while the displacements satisfy (6.30) over the remainder of the boundary.
In the formulation of elastodynamics problems, the equilibrium equations (6.27) must
be replaced by the equations of motion (5.16)
CHAP. 61 LINEAR ELASTICITY 145
and initial conditions as well as boundary conditions must be specified. In terms of the
displacement field ui, the governing equation here, analogous to (6.31) in the elastostatic
case is ..
p ~ ~ , ~ ~ + ( h + p ) ~ ~ , ~0~ 1+ p~V b 2~ ~=+ p( A
~ +
~ p ) V V * ~ + p b = p Ü(6.35)
Solutions of (6.35) appear in the form ui = u~(x,t) and must satisfy not only initial condi
tions on the motion, usually expressed by equations such a s
ui = gi(x, t) or u = g(x, t)
o r on the surface tractions,
ui = u,'" +ui2) represent a solution to the system for body forces bi = bil' + b'2).
i
('J)
Fig. 62
For the plane stress problem of Fig. 62(a) the stress components a,,, a,,, a,, are taken
a s zero everywhere, and the remaining components are taken as functions of xl and 2 2 only,
na8 = (xl>xZ) (a,P =1~2) (6.39)
in which
a a
êl + $2 and
V = 8x1 ax,
(6.43)
Due to the particular form of the strain tensor in the plane stress case, the six compatibility
equations (3.104) may be reduced with reasonable accuracy for very thin plates to the
single equation
In terms of the displacement components u,, the field equations may be combined to give
the governing equation
where V 2 = 
d2
dx:
+ 8
d2
~ ;
'
For the plane strain problem of Fig. 62(b) the displacement component us is taken as
zero, and the remaining components considered as functions of X I and X n only,
CHAP. 61 LINEAR ELASTICITY 147
(a) u,,,,+pb, = O or ~ . I : + p b= 0
As in the case of plane stress, the compatibility equations for plane strain reduce to the
single equation (6.44).
If the forces applied to the edge of the plate in Fig. 62(a) are not uniform across the .
thickness, but are symmetrical with respect to the middle plane of the plate, a state of
generalized plane stress is said to exist. In formulating problems for this case, the field
variables a,,, r,, and u, must be replaced by stress, strain and displacement variables
averaged across the thickness of the plate. In terms of such averaged field variables, the
generalized plane stress formulation is essentially the same as the plane strain case if X is
replaced by
A' = 2XP   
VE (6.52)
X 2/.L +
1  v2
A case of generalized plane strain is sometimes mentioned in elasticity books when r,,
is taken as a constant other than zero in (6.50).
For plane elastostatic problems in the absence of body forces, the equilibrium equations
reduce to
ua4.~
 O or 82:= O (6.53)
and the compatibility equation (6.44) may be expressed in terms of stress components as
The stress components are now given as partia1 derivatives of the Airy stress function
= +(xl, XP) in accordance with the equations
   (6.55)
ull  4,229 u12  4,129 u22  +,I1
148 LINEAR ELASTICITY [CHAP. 6
The equilibrium equations (6.53) are satisfied identically, and the compatibility condition
(6.54) becomes the biharmonic equation
Functions which satisfy (6.56) are called biharmonic functions. By considering biharmonic
functions with singlevalued second partia1 derivatives, numerous solutions to plane elasto
static problems may be constructed, which satisfy automatically both equilibrium and
compatibility. Of course these solutions must be tailored to fit whatever boundary
conditions are prescribed.
in which R and Q represent body forces per unit volume in the directions shown.
'c'
Fig. 63
Taking the Airy stress function now as @ = @(r,O), the stress components are given by
in which Dij is the rate of deformation tensor. In a second classification, a material is said
to be hypoelastic if the stress rate is a homogeneous linear function of the rate of deforma
tion. In this case the constitutive equation is written
= KijkrnDkm
Heat conduction in an isotropic elastic solid is governed by the wellknown Fourier law
of heat conduction,
ci = kTai (6.71)
where the scalar k, the thermal conductivity of the body, must be positive to assure a
positive rate of entropy production. If now the specific heat a t constant deformation c(")is
introduced through the equation
 c ~ ,=
~ p~(")T (6.72)
and the interna1 energy is assumed to be a function of the strain components cij and the
temperature T, the energy equation (5.45) may be expressed in the form
150 LINEAR ELASTICITY [CHAP. 6
( a ) equations of motion ..
~ , , ~ + p b ~ = oru ~ V o Z + p b = Ü
I: = Xlr + ~ P E (3A + 2 ~ ) a l ( T
 To)
( c ) straindisplacement relations
This system must be solved for the stress, displacement and temperature fields, subject to
appropriate initial and boundary conditions. In addition, the compatibility equations must
be satisfied.
There is a large collection of problems in which both the inertia and coupling effects
may be neglected. For these cases the general thermoelastic problem decomposes into two
separate problems which must be solved consecutively, but independently. Thus for the
uncoupled, quasistatic, thermoelastic problem the basic equations are the
( b ) equilibrium equations
a i j , j + p b i =O ~r V0E+pb= O
Solved Problems
HOOKE'S LAW. STRAIN ENERGY. ISOTROPY (Sec. 6.16.3)
6.1. Show that the strain energy density u* for an isotropic Hookean solid may be
expressed in terms of the strain tensor by u* = X(tr E)V2 /.LE: E, and in terms of +
+
the stress tensor by u* = [(I v)Z: I:  ~ ( tZ)2]/2E.
r
Inserting (6.21) into (6.13), u* = ( X S i j e k k + 2peij)e i j / = Xeiiyjl2 + peijyj which in symbolic
notation is u* = X(tr E)2/2 pE : E.+
+
Inserting (6.24) into (6.13), u* = o i j [ ( l v)oij  vSijukk]/2E= [(I + v)oijoij voiiojj]12Ewhich in
+
symbolic notation is u* = [ ( 1 v)L : L  ~ ( tL)2]/2E.
r
6.2. Separating the stress and strain tensors into their spherical and deviator components,
express the strain energy density v,* as the sum of a dilatation energy density u&)
and distortion energy density u&).
Inserting (3.98) and (2.70) into (6.13),
+ +
u* = &(sij okksij/3)(eij e,,Sij/3) + +
= +(sijeij oiiejj/3 siieii/3 + u,ejj/3)
+
and since eii = sii = O this reduceS to U* = u'lS> uTD,= oiiejj/6 sijeij/2. +
6.3. Assuming a state of uniform compressive stress uij = pai,, develop the formulas
for the bulk modulus (ratio of pressure to volume change) given in (6.25).
With uij = pSii, (6.24) becomes eij = [ ( I V)(pSii) + +
vSij(3p)]lE and so yi = [  3 p ( l + V) +
9pvIIE. Thus K = p/eii = E / 3 ( 1  2v). Likewise from (6.21), oii = ( 3 1 2p) eii = 3p +
so that
+
K = (3h 2p)/3.
6.4. Express u:,, and u&, of Problem 6.2 in terms of the engineering constants K and G
and the strain components.
From a result in Problem 6.3, uii = 3Keii and so
u : ~ )= uiiejj/6 = Keiiejj/2 = K(IE)2/2
From (6.21) and (2.70), oij = ASij ekk + 2peii = sij + okkSij/3 and since oii = (3X + 2p)eii i t foiiows
that sij = 2p(eij  ekk Sijl3). Thus
u : ~ )= zP(eij  ~ ~ ~ / 3 ) (=~ p ~
~ ~ 8  ( ~ i~j ~
i i eiieji/3)
Note that the dilatation energy density u:,) appears a s a function of K only, whereas the distortion
energy u'lD, is in terms of p (or G), the shear modu?us.
Fig. 64
and from (2.27) and (J.78), o;( = UK, = cK for K = 1,2,3,6 whereas UL = UK, eg = cK
for K = 4,5. Thus, for example, from c; = CIMcM,
These two expressions for U; = o, are equal only if C14 = C15 = O. Likewise, from o; = o,,
03 = u3,U; = v4, U; = u5, U; = o, i t is found that C% = C25 = = C35 = C64 = C65 C41 =
= c43 = = = C53 = C5@= 0.
If Z2x3 (or XZX:) is a second plane of elastic symmetry such that a; = CKMcM, the trans
formation array is
I1 
and now from (2.27) and (5.78), 02 = UK, cK  €K for K = 1,2,3,4 whereas O; = UK,
)I 
c~  cK for K = 5,6. Now C16 = C2, = C3, = C45 = C54 = = Cs2 = Cs3 = O and the elastic
coefficient matrix attains the form (6.19). The student should verify that elastic symmetry with
respect to the (third) xlx3 plane is identically satisfied by this array.
6.7. Give the details of the reduction of the orthotropic elastic matrix (6.19) to the
isotropic matrix (6.20).
For isotropy, elastic properties are the same with respect to a11 Cartesian coordinate axes.
In particular, for the rotated zl axes shown in Fig. 65, the method of Problem 6.6 results in the
matrix (6.19) being further simplified by the conditions Cll = CZ2= C33, C44 = C55 = and
C12 = = Cl3 = = Cz3 = c32.
Finally, for the axes x:'obtained by a 45O rotation about x3 a s in Fig. 66 above, the transformation
matrix is
[aijl =
so that o[ = (o,  u1)/2 = (Cll  Clz)(az 4 1 2 and e; = e,  e,. But o[ = C44~[ and so
2C4, = C,,  C,,. Thus defining p = C44 and X = C1,, (6.20) is obtained.
6.9. Express the engineering constants V and E in terms of the Lamé constants h and p.
From (6.25), E l ( 1  2 ~ =
) 3X +
2p; and from (6.26), E l ( l + v ) = 2p. Thus (3X + 2p)(1 2v) =
2p(l + +
v ) from which v = X/2(X p). Now by (6.26), E = 2p(l + v) = p(3X f 2p)/(X + p).
6.10. Determine the elastic coefficient matrix for a continuum having an axis of elastic
symmetry of order N = 4. Assume CKM= CMK.
Let x3 be the axis of elastic symmetry. A rota
tion e = 2a/4 = r12 of the axes about x3 produces
equivalent elastic directions for N = 4. The trans
formation matrix is
[aiil =
o o
and by (2.27) and (3.78), o; = o,, o; = o,, 04 = o,,
o: = 05, o; = u4, u; = and e: = e2, E; = cl, = '3,
1 
e4  e5, 4 = e4, e: = eg. Thus, for example, from
o3 = 83, C34= C35= = 0, C3, = C3,. Likewise, from
the remaining five stress relations, the elastic matrix
becomes Fig. 67
[CKMI =
6.13. If body forces may be neglected, show that (6.35) is satisfied by ui = +,i +c ~ ~ ~ $
provided + and $i each satisfies the familiar three dimensional wave equation.
Substituting the assumed ui into ( 6 . 9 5 ) yields
+
~ ( @ , i k k eijk$'k,jqq) + ( A + ~ ) ( @ , j j i + €jpq$'q.pji)
.. + eijk$'k,j)..
= P( @,i
Since ypq\l.q,pji= 0, this equation may be written
( ( A f 2 1 ~ ) @ , k k P ? ) , i + 'ijk(~$'k,qq
..
 ~ $ ' k ) j, = O
= p?/(A
which is satisfied when V 2 @ +2 ~ and
) V2$'k = pYk/fl.
6.14. Writing c2V2+= where c2 = ( A + 2 p ) / p for the wave equation derived in Problem
6.13, show that + = g(r
ct) r h(r  ct) is a solution with g and h arbitrary func
+ +
6.15. Derive the BeltramiMichell equations (6.33) and determine the form they take when
body forces are conservative, i.e. when p b i = +,i.
Substituting ( 6 . 2 4 ) into ( 3 . 1 0 3 ) yields
where 8 = I= = uii. Only six of the eightyone equations represented here are independent. Thus
setting m = k and using ( 6 . 2 7 ) gives
from which = (i + ~ ) p b ~ , ~/ (v l) . inserting this expression for O , k k into the previous
equation leads to ( 6 . 9 3 ) .
If p b i = @ , i , then p ( b i , j + bj,i) = 2@,ij and p b k , , = @ , k k = V 2 @ so that ( 6 . 3 3 ) becomes
6.17. F o r plane strain parallel to x1x2,develop the stressstrain relations in terms of V and
becomes
inverting,

E. Show that these equations correspond to those given a s (6.48).
Here u3 O so that eg3 = O and (6.24) gives 033 = v(oll o,,) = Ao,,/2(A
+ +
+
+
= (1 v)o,plE  v ( l v ) S a p o Y y l E from which e,, = ( 1 v ) ( l  2v)o,,lE.
+
p). Thus (6.24)
Finally,
6.18. Develop the Navier equation for plane stress (6.45) and show that i t is equivalent to
the corresponding equation f o r plane strain (6.51) if A' = 2 A p / ( A 2 p ) is substituted +
f o r A.
Inverting (6.41) and using (6.42) leads to a,p = E ( U , , ~ up,,)/2(1 + + v) + 2vES,puy,y/2(1  v2).
Differentiating with respect to x p and substituting into (6.40) gives
Eu,, pp/2(1+ v) + Eup,p,/Z(l v) + pb, = h V 2 u a + p(3A + 2 p ) ~ ~ , p , / (+h 2p) + pb, = O
Thus since p(3X + 2p)/(A+ 2p) = (2Ap/(A+ 2p) + p) = (A' + p), (6.45) and (6.51) have the same form
for the given substitution.
Fig. 68
6.21. In Problem 2.36 it was shown that the equilibrium equations were satisfied in the
absence of body forces by uij = c ~ ~ ~ c Show
~ ~ ~ that+ Airy's
~ ~ ,stress
~ ~ function
. is
represented by the case +,,
= +(xl,x,) with = = = = +,, +,, +,, +,, +,,
0.
Since +,,is the only nonvanishing component, oij = eipqejmn+qn,pm becomes q j = eip3ejrn3+33,pm
which may be written oap = E , ~ ~ E ~ y t~. ~ Thus
+ ~ ~ ,since +33 = +, ",B = ( S a ~ S y ( ; =( ;
SYB)@,~
6,p+,yy  +,,p . The stress components are therefore 0 1 1 = +,li +
+,22  +,li = +,22, 0 1 2 = +,12,
" 2 2 = 9.11 + +,22  +,22 = +,
11.
156 LINEAR ELASTICITY [CHAP. 6
6.24. Develop the thermoelastic energy equation (6.73) by use of the free energy f = u  Ts.
Assuming the free energy to be a fynction of the strains and temperature, f = f(e,, T ) and
substituting into (5.41) p& +
aijS pTs where dots indicate time derivatives, the result is
+
(oij pôf/dsij)iij  p(s d f l d T ) T = O. Since the terms in parentheses are independent of strain
~ ~ s = dfldT.
and temperature rates, i t follows that uij = p d f l a ~ and From (5.38) for a reversible
isothermal process, =p ~ =
(a:j
S pT  i.. dTas
+ ">
 T . A t constant deformation, Zij = 0 and
comparing this equation with (6.72) gives c(") = T(ds/dT) or from above, since ds/dT = a2flaT2,
c(") =  d Z f l d T 2 . = dSjldT and so combining ( 5 3 8 ) with (6.71),
Also, from above, p(d2fld~ijdT)
MISCELLANEOUS PROBLEMS
6.26. Show that the distortion energy density u:,, may be expressed in terrns of princigal
stress values by the equation uy,, = [ ( a ,  (a2  u J 2 +
( a 3 ~ ~ ) ~ ] / 1 2 G . +
\
From Problem 6.2, u:,) = sijeij/2 = siisij/4G which in terms of stress components becomes
u f D , = (uij8ijukk/3)(aij 8ijop,/3)/4G = (oijoijoiioji/3)/4G
6.27. Use the results of Problem 6.1 to show that for an elastic material d u * l d ~ =
, ~ uij and
~ u * I ~=u É~i j . ,
+
From Problem 6.1, u* = Xeiiejjl2 prijsij and so
+ +
Ju*larPq = X/2[eii(aejjla~pq)ejj(âciilòepq)] Z p ~ ~ ~ ( a e ~ ~ l a $ , )
= X/2[eii8jp8jq + ~ ~ ~+ 82 / ~J ~~ ~ 8=~ 8X/2[€iispq
~ ~ ~~ 8] $
~ ~ € j j s p q ]f 2,Urpq

 heiiSpq + 2pePq = upq
Likewise from Problem 6.1, u* = [(I + v)uijuii  ~ u ~ ~ u and
~ ~so] / ~ E
au"la,, = [2(1+ u)uijSipSjq v(uiiôpq + ujjôW]/2E = [ ( I + v)uPq u8pquii]lE = epq
6.28. Express the strain energy density u* as a function of the strain invariants.
From Problem 6.1, u* = Xeiiqj/2 + priirij; and since by comparison with (3.91), I, = cii and
11, = (riirjj 4 j Y j ) / 2 , i t follows that
6.30. Show that for a continuum having an axis of elastic symmetry of order N = 2, the
elastic properties (Hooke's law and strain energy density) are of the same form as a
continuum having one plane of elastic symmetry.
Here a rotation of axes e = 2 ~ 1 N= 2 ~ 1 2= n produces equivalent elastic directions. But this
is precisely the same situation a s the reflection about a plane of elastic symmetry.
cos 0 sin 0 0
ai2 = (sin e cos e)ull + (cos2 e  sin2 e)o12 + (sin 6 cos @)oz2
or i n single index notation,
= (sin e cos e)ul + (cos2 8  sin2 e)a6 + (sin e cos e)u2
Likewise from (3.78)and (6.4),
= (2 sin e cos e)€,
e; +
(cos2 e  sin2 @)e6 ( 2 sin e cos @)e2 +
B u t a; = C44€; for a n isotropic body and so here u2  ai = 2C44(€2 cl). Finally f r o m (6.19) w i t h
the given conditions, ul = Cllel i +
C12(e2 e l ) and q = CtlY C12(el e,) +
and so o2  ul = +
(Cll  C12)(e2 el). Therefore (Cll  C12)= 2C44 and w i t h C* = p, G12 = h, Cll = h 2p as +
given i n (6.20).
A
6.32. For an elastic body in equilibrium under body forces bi and surface forces ti"', show
that the total strain energy is equal to onehalf the work done by the externa1 forces
acting through their displacements ui.
I t is required t o show t h a t
A
i pbiui dV +
Jv
t:"ui d S = 2 u* dV. Consider first the surface
integral w i t h ti("' = ujiuj and convert b y Gauss' theorem. T h u s
L uijuinjd S = l, (cijui), dV =
L (aij,jui + uijui, j) dV
4 t$'ui d S Sv
=  pbiui dV +2 S aiieij/2 dV
6.33. Use the result of Problem 6.32 to establish uniqueness of the elastostatic solution of a
linear elastic body by assuming two solutions O : ) , u:') and O : ; ) , ui2).
For linear elasticity superposition holds, so uij = b:;)  a:;', ui = uil'  ~ 1 would
~ ) also be
a solution for which bi = O. T h u s for this "difference" solution
LA t:")uidS = 2
Problem 6.32. SinceAthe t w o assumed solutions satisfy boundary conditions, t h e l e f t hand integral
Jv
u* dV from
6.34. The Navier equations (6.31) may be p u t in the form pui,s + uj.ji + pbi = O
which for the incompressible case ( V = 4) are clearly indeterminate. Use the equilib
rium equations for this situation to S ~ O Wthat pUi,jj @,i13 pbi = 0. + +
From equation (6.24), eii = ( 1  2v)aiilE; and for v = Q, eii uiBi= O. T h u s f r o m (6.24),
2cij, = u 1.33
. .. u 3,13 + +
. .. = 2(1 v)aij,jlE  2v8ijukk,j/E
B u t uj,ji= O and E = 3G when v = 3, so t h a t ui,j j = pbilG  ukkJ3G or pV2ui + e j i / 3+ pbi = 0.
Supplementary Problems
 6.35. Prove t h a t the principal axes o f the stress and strain tensors coincide for a homogeneous isotropic
elastic body.
6.36. Develop the expression for the strain energy density u* for an orthotropic elastic medium. Use
equations (6.14) and (6.19).
+ +
U* = ( C l l ~fl 2C12e2 2 C 1 3 ~ 5 ) ~f1 / 2(C22e2f 2c23e4)"2/2 C3363 f C44'42 Cs5e: f C66ei. +
CHAP. 61 LINEAR ELASTICITY 159
6.37. Determine the form of the strain energy density for the case of ( a ) plane stress, ( b ) plane strain.
Ans. ( a ) u* = [o;, + + +
 2volla22 2(1 v ) o y 2 ] 1 2 ~
( b ) u* = (a + ~/2)(e?1+egz) + Xe11'22 + 2aef2
2a
6.38. Determine the value of c for which ul = A sin ( x l c t ) , u 2 = u3 = O is a solution of equation
I
( 6 3 5 ) when body forces are zero. Ans. c = d w
6.39. Show that the distortion energy density uyD,= (uijoii oiiojj/3)/4G and the dilatation energy density *
urS, = oiiajjl18K.
6.40. Show that 1 / ( 1 + V ) = 2(X + & ( 3 ~+ 2p) and vl(1 V) = X/(h + 2p).
6.41. For plane strain parallel to x,x,, show that b, O and that b, and b2 are functions of x , and
x, only.
6.42. Use the transformation laws for stress and strain to show that the elastic constants Cijk,,, are the
components of a fourth order Cartesian tensor so that Ci;km = aipaiqakTa,,,sCpqrs.
6.43. +
Show that the Airy stress function + = 2%; 12x;x;  6x2 satisfies the biharmonic equation
V4+ = O and determine the stress components assuming plane strain.
2:  3x2 2x,x,
Ans. <rij = 24
o 2v(x;  x ; )
6.44. Determine the strains associated with the stresses of Problem 6.43 and show that the compatibility
equation (6.44) is satisfied.
6.45. For an elastic body having an axis of elastic symmetry of order N = 6, show that C2 = C,,,
C,, = CG, Ce6 = 2(Cll  C I 2 ) and that C13 and C33 are the only remaining nonzero coefficients.
6.46. Show that for an elastic continuum with conservative body forces such that pba = V+ = $,,, the
compatibility condition (6.44) may be written V20aa = Q2+/(1  V ) for plane strain, or
820aa = (1 +
v ) V 2 + for plane stress.
6.47. If V4Fi = 0, show t h a t ui = 2(1 v)V2Fi/G  FjBji/G is a solution of the Navier equation (6.31)
when bi O (see Problem 6.12). If F = B ( x 2 ê l  x l ê 2 ) / r where r 2 = xixi, determine the stress
components.
Ans. o,, = oz2 = 6QGx1x2/r5, = 0, u12 = 3QG(x;  4 ) / ~ 5 ,u13 = oz3 = 3QGx2x3/r5, where
Q = 4 B ( 1  v)lG.
6.48. In polar coordinates an Airy stress function is given by @ = Cr2(cos 2e  cos 2a) where C and a
are constants. Determine C if use = O, ora = T when e = a, and = O , a,, = T when
e = a. Ans. C = r / ( 2 sin 2a)
Fluids
where rii is called the viscous stress tensor and p is the pressure.
A11 real fluids are both compressible and viscous. However, these characteristics vary
widely in different fluids so that it is often possible to neglect their effects in certain situa
tions without significant loss of accuracy in calculations based upon such assumptions.
Accordingly, an inviscid, or socalled perfect fluid is one for which rij is taken identically
zero even when motion is present. VZscous fluids on the other hand are those for which
rij must be considered. For a compressible fluid, the pressure p is essentially the same as
the pressure associated with classical thermodynamics. From (7.3), the mean normal stress
is given by
For a fluid a t rest, r , vanishes and p reduces to p, which in this case is equal to the negative
of the mean normal stress. For an incompressible fluid, the thermodynamic pressure is
not defined separately from the mechanical conditions so that p must be considered as an
independent mechanical variable in such fluids.
In a compressible fluid, the pressure p, the density p and the absolute temperature T are
related through a kinetic equation of state having the form
where R is the gas constant. If the changes of state of a fluid obey an equation of state
that does not contain the temperature, i.e. p = P ( ~ ) such
, changes are termed barotropic.
An isothermal process for a perfect gas is an example of a special case which obeys the
barotropic assumption.
the fluid is called a Stokesian fluid. When the function is a linear one of the form
where the constants Ki,, are called viscosity coeficients, the fluid is known as a Newtonian
fluid. Some authors classify fluids simply as Newtonian and nonNewtonian.
Following a procedure very much the same as that carried out for the generalized
Hooke's law of an elastic media in Chapter 6, the constitutive equations for an isotropic
homogeneous Newtonian fluid may be determined from (7.7) and (7.3). The final form is
where A* and p* are viscosity coefficients of the fluid. From (7.9)' the mean normal stress
is given by
+
where K* = Q(3A* 2p*) is called the coeficient of bulk viscosity. The condition that
is known as Stokes' condition, and guarantees that the pressure p is defined as the average
of the normal stresses for a compressible fluid a t rest. In this way the thermodynamic
pressure is defined in terms of the mechanical stresses.
In terms of the deviator components sij = aii  6ijokk/3and D i = Dii  6iiD,k/3, equa
tion (7.9) above may be rewritten in the form
Therefore in view of the relationship (7.10),equation (7.12) may be expressed by the pair
of equations
sij = 2p*D,; or S = 2p*Df (7.13)
the first of which relates the shear effects in the fluid and the second gives the volumetric
relationship.
162 FLUIDS [CHAP. 7
When the flow is incompressible ( v , = O), (7.22) reduce to the NavierStokes equations
for incompressible flow,
If Stokes condition is assumed (A* = pp*), (7.22) reduce to the NavierStokes equations
for compressible flow
PVi = Pbi  P*i + *!J*vj,ji+ p*vcjj
or (7.24)
+
PV = pb  V p +p.*v(v V ) p*V2v +
CHAP. 71 FLUIDS 163
The NavierStokes equations (7.23), together with the continuity equation (7.15) form
a complete set of four equations in four unknowns: the pressure p and the three velocity
components vi. In any given problem, the solutions of this set of equations must satisfy
boundary and initial conditions on traction and velocity components. For a viscous fluid,
the appropriate boundary conditions a t a fixed surface require both the normal and
tangentiaI components of velocity to vanish. This condition results from the experimentally
established fact that a fluid adheres to and obtains the velocity of the boundary. For an
inviscid fluid, only the normal velocity component is required to vanish on a fixed surface.
If the NavierStokes equations are put into dimensionless form, severa1 ratios of the
normalizing parameters appear. One of the most significant and commonly used ratios is
the Reynolds number N,,, which expresses the ratio of inertia to viscous forces. Thus if
a flow is characterized by a certain length L, velocity V and density p, the Reynolds number
is
N,,, = VL/v (7.25)
where V = p*lPis called the kinematic viscosity. For very large Reynolds numbers, the
,
viscous contribution to the shear stress terms of the momentum equations may be neglected.
In turbulent flow, the apparent stresses act on the time mean flow in a manner similar to
the viscous stress effects in a laminar flow. If turbulence is not present, inertia effects
outweigh viscous effects and the fluid behaves as though it were inviscid. The ability of
a flow to support turbulent motions is related to the Reynolds number. It is only in the case
of laminar flow that the constitutive relations (7.18) apply to real fluids.
may be defined. Furthermore, if the body force may be prescribed by a potential function
which is known as the Euler equation of motion. For a barotropic fluid with conservative
body forces, (7.29) and (7.30) may be introduced so that (7.36) becomes
;i = (n+P), or V = v(a+P) (7.37)
For steady flow (7.37) may be written
v3. vt i 3. . = (n+P),i or v.Vv = V(n+P) (7.38)
If the Euler equation (7.37) is integrated along a streamline, the result is the well
known Bernoulli equation in the form (see Problem 7.17)
For steady motion, avilat = O and C(t) becomes the Bernoulli constant C which is, in general,
different along different streamlines. If the flow is irrotational as well, a single constant
C holds everywhere in the field of flow.
When the only body force present is gravity, the potential n = gh where g is the
gravitational constant and h is the elevation above some reference level. Thus with h, = Plg
defined as the pressure head, and v2/2g= hv defined as the velocity head, Bernoulli's equa
tion requires the total head along any streamline to be constant. For incompressible fluids
(liquids), the equation takes the form
h + h, + h" = h + plpg + v2/2g = constant (7.40)
From Stokes theorem (1.153) or (1.154), page 23, the line integral (7.41) may be converted
to the surface integral
where íiis the unit normal to the surface S enclosed by the path. If the flow is irrotational,
V x v = O and the circulation is zero. In this case the integrand of (7.42) is the perfect
differential d+ = v. dx with the velocity potential.
CHAP. 71 FLUIDS 165
The material derivative drcldt of the circulation may be determined by using (4.60)
which when applied to (7.41) gives
For a barotropic, inviscid fluid with conservative body forces the circulation may be shown
to be a constant. This is known as Kelvin's theorem of constant circulation.
For incompressible potential flow the continuity equation attains the form
and solutions of this Laplace equation provide the velocity components through the defini
tion (7.35). Boundary conditions on velocity must also be satisfied. On a fixed boundary,
for example, d+ldn = O. An important feature of this formulation rests in the fact that
the Laplace equation is linear so that superposition of solutions is possible.
then from (7.48) and (7.49) the stream function and velocity potential are seen to satisfy
the CauchyRiemann conditions
4.1 = q.2 and +,, = +,i (7.50)
By eliminating + and + in turn from (7.50) i t is easily shown that
166 FLUIDS [CHAP. 7
Thus both + and + are harmonic functions when the flow is irrotational. Furthermore, the
complex potential
@(4 = +(x19 x,) + i+(x1, $2) (7.53)
is an analytic function of the complex variable, z = x , + ix, so that its derivative dqldz
defines the c o m p l e x v e l o c i t y
dqldz = vl iv, + (7.54)
Solved Problems
FUNDAMENTALS OF FLUIDS. NEWTONIAN FLUIDS (Sec. 7.17.3)
7.1. Show that the deviator sij for the stress tensor a , of ( 7 . 3 ) is equal to t,, the deviator
of T i j of ( 7 . 3 ) .
From ( 7 4 , aii = 3p + rii and so here
7.2. Determine the mean normal stress o i i / 3 for an incompressible Stokesian (nonlinear)
+
fluid for which T , = a D i j pDikDkj where a and p are constants.
From ( 7 4 , aij = pSij + +
aDij /3DikDkj and so ai< = 3p + aDii + /3DikDki. But Dik =
Dki and Dii = = O for an incompressible fluid so that
4 3 = p + /3DijDij13 = p  2/311D/3
where 11, is the second invariant of the rate of deformation tensor.
7.3. Frictionless adiabatic, or isentropic flow of an ideal gas, is a barotropic flow for which
p = cpk where C and k are constants with k = c(*)/c("), the ratio of specific heat a t
constant pressure to that a t constant volume. Determine the temperaturedensity
and temperaturepressure relationships for such a flow.
Inserting p = Cpk into equation (7.6), the temperaturedensity relationship is #  l / T = RIC,
a constant. Also, since p = (plC)l/k here, (7.6) yields the temperaturepressure relationship a s
p(kl)/k/T = R/Cl/k, a constant.
7.4. Determine the constitutive equation for a Newtonian fluid with zero bulk viscosity,
i.e. with K * = 0.
If K* O , A* =  2 / ~ * / 3 by (7.11) and S O (7.9) becomes aii = pSij  (2/l*/3)SijDkk+ 2/~*Dij
which is expressed in terms of the rate of deformation deviator by
If the deviator stress sij is introduced, this constitutive relation is given by the two equations
S i j = ~ / L * D ;and aii = 3p.
CHAP. 71 FLUIDS 167
7.5. Determine a n expression for the "stress power" cijDijof a Newtonian fluid having
equation (7.9) a s its constitutive relation.
From (7.9) and the stress power definition,
u 23 = pSijDij
o..D.. + h*SijDkkDij+ 2p*DijDij = pDii + h*DiiDjj + 2p*DijDii
In symbolic notation, this expression is written
In symbolic notation,
: D = p(tr D ) + ~ * ( tD)2r + 2p* D' : D'
7.6. Determine the conditions under which the mean normal pressure p,,, = aii/3 is
equal to the thermodynamic pressure p f o r a Newtonian fluid.
With the constitutive equations in the form (7.13) and (7.14), the latter equation gives
P ( , >   P = K *Dii. Thus p(m, = p when K* = O (by (7.11) when A* = Qa*) or when Dii = 0.
o.. 3. =  p j j S i j
i13 + X * S i j ~ k , k+j p*(vi,jj + vj,ij) = P,i (h* + p*)vj,ji + p*vi,jj
and with this expression inserted into (7.16) a direct verification of (7.22) is complete.
Substituting the above equation for qjtogether with (7.20) into the energy equation (7.17), the
result is
P; = [psij h*6ijvk,k + + +
v ~ , ~ ) ] ( v v~j , i ) / 2 kT,ii +
pZ +
which reduces to
Ti = (p~ij + 2p*D;)9dS
for a zero bulk modulus fluid; and upon application
of Gauss' theorem,
/
Fig. 71
168 FLUIDS [CHAP. 7
7.9. In a n axisymmetric flow along the x3 axis the velocity is taken a s a function of x3 and
r where r2 = X : + x2. If the velocity is expressed by v = qê, v3ê3 where 6, is +
the unit radial vector, determine the form of the continuity equation.
Equation (5.4) gives the continuity equation in symbolic notation a saplat V (pv) = 0. +
1 d(rpg) i a(Pv3)
Here the cylindrical form of the operator V may be used to give V (pv) = 
Inserting this into (5.4) and simplifying, the continuity equation becomes r ar 8x3 '
7.10. In a twodimensional flow parallel to the X I X Z plane, v3 and dldxa a r e zero. Determine
the NavierStokes equations for a n incompressible fluid and the form of the con
tinuity equation for this case.
From (7.23) with i = 3, pb, = p,, and when i = 1 , 2 , p'U, = pb,  P,, + P * V , , ~ ~ .The con
tinuity equation (7.15) reduces to v,,, = 0. 4
If body forces were zero and v , = v l ( x l , x,, t), v , = O , p = p ( x l , x,, t ) the necessary equations
would be pGl = dp/axl + +
a * ( d z ~ l / d ~ : d 2 v 1 / d x i ) and dvl/axl = 0.
7.12. A barotropic fluid having the equation of state p = hpk where h and k a r e constants
is a t rest in a gravity field in the x3 direction. Determine the pressure in the fluid
with respect to x3 and po, the pressure a t x3 = 0.
From (7.28), dp/dx3 = pg, d p / d x l = d p l d x , = O. Note that pressure in x , and x z directions
is constant in the absence of body forces b1 and bz. Since here p = (p/h)Uk, pllkdp = gh1fkdx3
and integration gives ( k / ( k  l))pck1)lk = + c. But p = po when x3 = O so that
C = ( k / ( k  1 ) ) ( k  l ) / k . Therefore x3 = ( k p o l ( k l ) g p o ) ( l ( p / p 0 ) ( k  l ) f k ) where po = ( p o / ~ ) ~ ~ k .
po
7.14. If a fluid motion is very slow so that higher order terms in the velocity are negligible,
a limiting case known a s creeping flow results. For this case show that in a steady
incompressible flow with zero body forces the pressure is a harmonic function, i.e.
v 2 p = o.
For incompressible flow the NavierStokes equations (7.23) are
p(avilat + vjvi, +
= pbi  P , ~ p*vi, j j
and for creeping flow these linearize to the form
Hence for steady flow with zero body forces, p,i = Taking the divergence of this equation
yields pJii = P * ~ i , i j jand
; since the continuity equation for incompressible flow i s viai= O, i t follows
that here p., = V 2 p = 0 .
7.15. Express the continuity equation and the NavierStokesDuhem equations in terms of
the velocity potential 4 for an irrotational motion.
By (7.35), v i =  @ , i so that from (7.15) the continuity equation becomes i pV2@ = O. Also
with v i = +,i, (7.22) becomes
Pd,i = Pbi  P,i  (h* p*)+,jji  p*@,iii
or p(a$,i/at + @,k@,ik) = pbi  P,i  (h* + 2p*)$,jji
In symbolic notation this equation is written
pv(a+/at + (v+)2/2) = pb  v p  (A*+ z p * ) v ( v 2 + )
7.16. Deterinine the pressure function P(p) for a barotropic fluid having the equation of
state p = X p k where X and k are constants.
From the definition (7.29),
Also since d p = hkpk1 dp, the same result may be obtained from
Since dxi = d a and P,i dxi = d P the last two terms integrate a t once. Also, along a stream
line, dxi = (viIv)d s where ds is the increment of distance. Thus in the second integral,
v j v i , dxi = v j v i , ( v i / v )ds = vivi, ( v j / v )ds = v i v i , d x j = v i dvi
Therefore S v j v i , dxi =
s v . dvi = +vivi = +v2, and (7.39) is aehieved.
170 FLUIDS [CHAP. 7
7.18. The barotropic fluid of Problem 7.16 flows from a large closed tank through a thin
smooth pipe. If the pressure in the tank is N times the atmospheric pressure, deter
mine the speed of the emerging fluid.
Applying Bernoulli's equation for steady flow between point A , a t rest in fluid of the tank and
+ +
point B, in the emerging free stream, ( 7 3 9 ) assumes the form aA PA & v i = aB PB & v i . + +
But v, = 0, and if gravity is assumed negligible this equation becomes (see Problem 7.16),
7.19. Show that for a barotropic, inviscid fluid with conservative body forces the rate of
change of the circulation is zero (Kelvin's theorem).
From (7.M) :, = § (Gi dxi + vi dvi) and by (7.37), Gi = (a + P),i for the case a t hand. Thus
i.,= $ (qidxi  Paidxi + vi dvJ =  f (& + dP  d(vz12)) =  f P  P / 2 ) = 0 , the
d(Qi inte
grand being a perfect differential.
rh
7.20. Determine the circulation around the square xl = k1,
x2 = 1+1, x3 = O (see Fig. 73) for the twodimensional
+
f l o ~v = ( X I x ~ (x; +
) ~x ~~ ) ~ ~ .
Using the symbolic form of (7.42) with ^n = ê3 and
v x v = (2x1 1 )ê3,
= s:.Sl 1
(2x1  1) d x , dx,
Fig. 73
r, = $vdx
1 1
= (ix2)dx2 + ~  l ( x l + l ) d x l+ (1 x,) dx, + ( x ,  1) dxl = 4
+
7.22. Show that the function + = A(x:  x2 2 x i ) satisfies the Laplace equation and
determine the resulting velocity components.
Substituting +
into (7.46) gives 2A  2 A 4 A = O. From (7.3~9, v , = 2Ax,, v , = 2 A x Z ,
v 3 = 4Ax3. Also, by the analysis of Problem 4.7 the streamlines in the x , plane are represented
by x i x 3 = constant; in the x , plane by xSx3 = constant. Thus the flow is in along the x3 axis
against the x,x, plane (fixed wall).
CHAP. 71 FLUIDS 171
7.23. Show that the stream function +(xI,x2) is constant along any streamline.
From (7.48) and the differential equation of a streamline, dxl/vl = dx21v2 (see Problem 4.7),
=
d~,l+,~ or +,ldxl += d+ = O. Thus + = a constant along any streamline.
7.26. Differentiate the complex potential a(z) = Alx to obtain the velocity components.
Here dWdz = AI22 = A/(xl + ix2)2 which after some algebra becomes d<P/dz =
A(%;  x;)/+ +
i2Ax1x2/+. Thus
v1 = A(%; si)+ and v2. = 2Ax1x2l.P
Note that since @ = A l z = A(%, iz2)/r2, + = Axllr2 and JI = Ax21r2. Also note that
v, = = A ( x ; X:)I+ and v2 = 9.2 = 2 A ~ ~ x ~ 1 7 . 4
MISCELLANEOUS PROBLEMS
7.27. Derive the onedimensional continuity equation for the flow of an inviscid incompres
sible fiuid through a stream tube.
Let V be the volume between arbitrary cross sec
tions A and B of the stream tube shown in Fig. 75.
In integral form, for this volume (5.2) becomes
Sv V * v dV = O since p is constant here. Convert
ing by Gauss' theorem,
LA n o v dS = O where " n s
the outward unit normal to the surface S enclosing V .
ntA
A
n
Since i. v on the lateral surface, the integration re
duces to " " Fig. 75
A
The velocity is assumed unifonn and perpendicular over SA and S,; and since va =  v B n ~ ,
vA iA iB
dS  vB dS = O or v A S A = v B S B = a constant.
172 FLUIDS [CHAP.' 7
7.28. The stress tensor a t a given point for a Newtonian fluid with zero bulk viscosity is
c =
6
(
2 9
2 1
1 4 3
Determine ).
From (7.14), for this fluid p = uii/3 = 6. Then from ( 7 4 ,
.
(4: :)
6 2 1 6 O 2 1
oi; +6 8 or
:: )+ =
7.29. Show that aij and T~~ of (7.3) have the same principal axes.
When written out, (7.8) becomes .oll = p = p
T ~ I ,u 2 ~ +722, u33 = p +
T33, u12 = 7 1 2 , +
u23 = u13 = 7 1 ~ .For principal directions x: of qj, uTz = = uT3 = O and by the last three
equations of (7.39, o; = 7; = O for i # j. Thus Z: are principal axes for T~~ also.
7.30. A dissipation potential Q, is often defined for a Newtonian fluid by the relationship
a, = ( K / ~ ) D ; ~p,*D;D;.
D~~ +
Show that d a D l d D i j = r i j .
Here D~,)+
d+,laDpq = ( ~ / 2 ) [ D ~ ~ ( a D ~ ; l a (aDiilaDpq)Djj] +
2 5 [ D , ' , ( a ~ & 3 ~ , , ) ] . But aDiilaDpq =
~ , 6, , aj4  S i j Spq/3 so that
Sip Si, = S p q and d ~ , ( ~ / a =
+
S ~ ~ 8ijDkk/3)(8ip8jq
d@~laD,, = K D ~ ~ 2p*(Dij  SijSpq/3) = K D + eP*(Dpq
~ ~ SpqDii/3)
~ ~ ~
Finally since K = A* + 2p*/3,
a+,/a~,, = ?,*sPqoii+ 2 p * ~ p q= Tpq
. 7.31. Determine the pressuredensity relationship for the ideal gas discussed in Problem 7.11.
A t s3 = O, p = po and p = po. The ideal gas law (7.6) is here p = pR(To  as3) so that
po = poRT0; and from the pressure elevation relationship p = p,(l  a ~ ~ / T ~ ) g /ofR aProblem 7.11,
p l p o = (T/To)(g/Rai). Thus writing p = po(l  as31To)g/Ra in the form p/po = (T/To)g/Ra, i t is
seen that T / T o = (plpo) Ralo and so P/Po= (p/po)(lRa/g).
7.32. For a barotropic inviscid fluid with conservative body forces show that the material
derivative of the total vorticity,
d
qi dV = viqi dSi. aL
From (4.54) and the results of Problem 4.33, $ i qi dV =.L (ciik ak + qivi) dSj. But here
ak = (R +P),, from (7.37); and by the divergence theorem (1.157),
since the integrand is zero (product of a symmetric and antisymmetric tensor). Hence
7.33. For an incompressible Newtonian fluid moving inside a closed rigid container a t rest,
show that the time rate of change of kinetic energy of the fluid is ,.L*
ing zero body forces. q is the magnitude of the vorticity vector.
CHAP. 'i] FLUIDS 173
From Problem 5.27, the time rate of change of kinetic energy of a continuum is
In this problem the first and third integrals are zero; and for a Newtonian fluid by (7.18),
7.34. Show that for a perfect fluid with negligible body forces the rate of change of cir
culation bc may be given by  ~ ~ ~ j~p , ( a ís i/. ~ ) ,
bc =  6 IP) =
( P . ~ ~d..  i eijk(p,k/p),ini d s
where (7.42) has been used in converting to the surface integral. Differentiating a s indicated,
h = L +
e i j k [ ( l l ~ ) , j p , k P , L ~ P @i
] =  L eijk(lI~).j
~ , dSi
k
Supplementary Problems
7.35. The constitutive equation for an isotropic fluid is given by aij = paij +
K i j p q D p q with Kihq con
stants independent of the coordinates. Show that the principal axes of stress and rate of deformation
coincide.
7.37. Show that the constitutive relations for a Newtonian fluid with zero bulk viscosity may be expressed
by the pair of equations sij = 2r*Dii and oii = 3p.
7.38. Show that in terms of the vorticity vector q the NavierStokes equations may be written
\I = b  V p l p  "*V x q where v* = r*lP is the ,kinematic viscosity. Show that for irrotational
motion this equation reduces to (7.36).
7.39. If a fluid moves radially with the velocity v = v(r, t) where r2 = xixi, show that the equation of
continuity is A
a a + 2 (r%) = 0.
+v 2
at ar r2 ar
7.40. A liquid rotates a s a rigid body with constant angular velocity o about the vertical s3 axis. If
gravity is the only body force, show that plp  o2r212 + gz3 = constant.
174 FLUIDS [CHAP. 7
7.41. For an ideal gas under isothermal conditions (constant temperature = To),
show that p l p , = pIpo =
e(glRTozs) where po and po are the density and pressure a t x3 = 0.
7.42. Show that if body forces are conservative so that bi = nJi, the NavierStokesDuhem equations
for the irrotational motion of a barotropic fluid may be integrated to yield  p ( a @ l a t +
(V@)V2) +
+ +
pQ P +
(i* 2@*)V2+= f ( t ) . (See Problem 7.15.)
7.43. Show that the velocity and vorticity for an inviscid flow having conservative body forces and
constant density satisfy the relation Gi  qjvi,j = O. For steady flow of the same fluid, show that

vjqi,j  qjvi,j
7.44. For a barotropic fluid having p = p(p) and P ( p ) defined by (7..29),show that grad P = grad plp.
7.45. Show that the Bernoulli equation (7.39) for steady motion of an ideal gas takes the form
+ +
(a)Q p ln ( p l p ) v2/2 = constant, for isothermal flow, ( b ) + +
(klk  l ) ( p l p ) v212 = constant,
for isentropic flow.
7.46. Show that the velocity field vl = 2x1x2x3/T1, v 2 = (x:  x;)x3/r4, v3 = x2/r2 where r2 = x: +
x;+ $32is a possible flow for an incompressible fluid. 1s the motion irrotational? Ans. Yes
+ + i$ +
7.47. If the velocity potential +(z) =
ix2 = rei0 show that in polar coordinates 3ar
= '9 and 
r ae
1 a6 =
r ae
*
is an analytic function of the complex variable z = x ,
ar '
7.48. If body forces are zero, show that for irrotational potential flow +,ii = v* = , ~ * l p is the kinematic
viscosity.
Chapter 8
For small strains, these two measures of strain are very nearly equal as seen by (8.2) and it
is often permissible to neglect the difference.
176 PLASTICITY [CHAP. 8
The yield point P, corresponding to the yield stress o, separates the stressstrain curve
of Fig. 81 into an elastic range and a plastic range. Unfortunately, the yield point is not
always welldefined. It is sometimes taken a t the proportional limit, which lies a t the
upper end of the linear portion of the curve. It may also be chosen as the point J, known as
Johnson's apparent elastic limit, and defined as that point where the slope of the curve
attains 50% of its initial value. Various offset methods are also used to define the yield
point, one such being the stress value a t 0.2 per cent permanent strain.
In the initial elastic range, which may be linear or nonlinear, an increase in load causes
the stressstrainstatepoint to move upward along the curve, and a decrease in load, or
unloading causes the point to move downward along the same path. Thus a onetoone
stressstrain relationship exists in the elastic range.
In the plastic range, however, unloading from a point such a s B in Fig. 81 results in the
state point following the path BC which is essentially parallel with the linear elastic portion
of the curve. At C, where the stress reaches zero, the permanent plastic strain cP remains.
The recoverable elastic strain from B is labeled rE in Fig. 81. A reloading from C back to
B would follow very closely the path BC but with a rounding a t B, and with a small
hysteresis loop resulting from the energy loss in the unloadingreloading cycle. Upon a
return to B a load increase is required to cause further deformation, a condition referred
to as work hardening, or strain hardening. It is clear therefore that in the plastic range
the stress depends upon the entire loading, or strain history of the material.
Although it is recognized that temperature will have a definite influence upon the plastic
behavior of a real material, it is customary in much of plasticity to assume isothermal con
ditions and consider temperature as a parameter. Likewise, it is common practice in
traditional plasticity to neglect any effect that rate of loading would have upon the stress
strain curve. Accordingly, plastic deformations are assumed to be timeindependent and
separate from such phenomena as creep and relaxation.
E
PLASTICITY
Rough
L
F
177
I
......................................
Rough
(c) RigidLinear Work Hardening
Rough
(d) Elastic  Linear Work Hardening
Fig. 82
For an isotropic material the yield condition must be independent of direction and may
therefore be expressed as a function of the stress invariants, or alternatively, as a sym
metric function of the principal stresses. Thus (8.3) may appear as
Fig. 83
In stress space, the yield condition (8.5), f2(a,, u,,,olII)= C,, defines a surface, the
socalled yield surface. Since the yield conditions are independent of hydrostatic stress,
such yield surfaces are general cylinders having their generators parallel to 02. Stress
points that lie inside the cylindrical yield surface represent elastic stress states, those
which lie on the yield surface represent incipient plastic stress states. The intersection of
the yield surface with the Rplane is called the yield curve.
In a true view of the nplane, looking along OZ toward the origin 0, the principal stress
axes appear symmetrically placed 120" apart as shown in Fig. 85(a) below. The yield
curves for the Tresca and von Mises yield conditions appear in the nplane a s shown in
Fig. 85!b) and (c) below. In Fig. 85(b), these curves are drawn with reference to (8.7) and
(8.11), using the yield stress in simple tension as the basis. For this situation, the von Mises
circle of radius m 3U, is seen to circumscribe the regular Tresca hexagon. In Fig. 85(c),
the two yield curves are based upon the yield stress k in pure shear. Here the von Mises
circle is inscribed in the Tresca hexagon.
180 PLASTICITY [CHAP. 8
(b)
Fig. 85
The location in the IIplane of the projection of an arbitrary stress point P(u,, a,,, o,,,)
is straightforward since each of the stress space axes makes cosI @ with the IIplane.
Thus the projected deviatoric components are ( m v I , d2/3uII, moIII). The inverse
problem of determining the stress components for an arbitrary point in the nplane is not
unique since the hydrostatic stress component may have any value.
which depende not only upon the stresses, but also upon the plastic strains r; and the work
hardening characteristics represented by the parameter K. Equation (8.15) defines a loading
surface in the sense that f: = O is the yield surface, f: < 0 is a surface in the (elastic)
region inside the yield surface and f r > 0, being outside the yield surface, has no meaning.
Differentiating (8.15) by the chain rule of calculus,
Thus with f: = O and (af :Iaoij) doij < 0, unloading is said to occur; with f: = O and
(af :/aoij) doij = 0, neutra1 loading occurs; and with f i = O and (af:la~,,) doij > 0, loading
occurs. The manner in which the plastic strains c; enter into the function (8.15) when
loading occurs is defined by the hardening rules, two especially simple cases of which are
described in what follows.
The assumption of kotropic hardening under loading conditions postulates that the yield
surface simply increases in size and maintains its original shape. Thus in the IIplane the
yield curves for von Mises and Tresca conditions are the concentric circles and regular
hexagons shown in Fig. 86 below.
CHAP. 81 PLASTICITY 181
In kinematic hardening, the initial yield surface is translated to a new location in stress
space without change in size or shape. Thus (8.4) defining an initial yield surface is
replaced by
f i (oij a,) = O (8.17)
where the aij are coordinates of the center of the new
yield surface. If linear hardening is assumed,
Xi. = c;: (8.18)
and the plastic strain increments related to the stress deviator components by
the resulting equations are known as the PrandtlReuss equations. Equations (8.21) rep
resent the flow rule for an elasticperfectly plastic material. They provide a relationship
between the plastic strain increments and the current stress deviators but do not specify
the strain increment magnitudes.
182 PLASTICITY [CHAP. 8
The name plastic potential function is given to that function of the stress components
g(u,) for which dg dA
de; =  (8.22)
8%
For a socalled stable plastic material such a function exists and is identical to the yield
function. Moreover when the yield function f l ( u i j )= IIr,, (8.22) produces the Prandtl
Reuss equations (8.21).
In terms of the equivalent stress and strain increments defined by (8.24) and (8.25)
respectively, dA of (8.21) becomes
dh =  3 de;,
2 OEQ
Furthermore, if the same material obeys the PrandtlReuss equations (8.21),the plastic work
increment may be expressed as
d W P = a,,dcEQ (8.31)
and (8.21) rewritten in the form
3 dWP
de; =  S..
2 a;,
CHAP. 81 PLASTICITY 183
There are two widely considered hypotheses proposed for computing the current yield
stress under isotropic strain hardening plastic flow. One, known as the workhardening
hypothesis, assumes that the current yield surface depends only upon the total plastic work
done. Thus with the total plastic work given as the integral
for which the precise functional form must be determined experimentally. A second harden
ing hypothesis, known a s the strainhardening hypothesis, assumes that the hardening is a
function of the amount of plastic strain. In terms of the total equivalent strain
( b ) Plastic region
1. Equilibrium equations (2.23), page 49
2. Stressstrain increment relations (8.21)
3. Yield condition (8.8) or (8.11)
4 . Boundary conditions on plastic boundary when such exists
(c) Elasticplastic interf ace
1. Continuity conditions on stress and displacement
and since elastic strains are neglected, the plastic strainrate tensor applicable to the
situation is
. .
'li '12 0
(8.42)
In (8.41) and (8.42) the variables are functions of XI and xz only, and also
With respect to the velocity components, Fig. 810 shows that relative to the a and
/i' lines,
v, = v, cos +  v, sin +
(8.49)
vz = v, sin + + v, cos +
For a n isotropic material, the principal axes of stress and plastic strainrate coincide.
Therefore if x, and x, are slipline directions, ;,, and i,, are zero along the sliplines so t h a t
8.5. Convert the von Mises yield condition (8.10) to its principal stress form a s given in
(8.11).
From (2.72),  1 1 ~=~(slsI1 + SIISIII + sII1s1); and by (2.71), SI = o1  UM, etc., where
UM + +
= (01 oI1 uIII)/3. Hence
8.6. With the rectangular coordinate system OXYZ oriented so that the XY plane coincides
with the nplane and the a,,, axis lies in the YOZ plane (see Fig. 813 and 84), show
that the Mises yield surface intersects the nplane in the Mises circle of Fig. 85(b).
Fig. 813
The table of transformation coefficients between the two sets of axes is readily determined to
be a s shown above. Therefore
,I = x/fi  Y/& + z/fi, o,, = x/fi  Y/& + z / h , QIII = 2 ~ 1+
6 216
which simplifies to the Mises yield circle 3X2 + 3Y2 = 20; of Fig. 85(b).
8.7. Using the transformation equations of Problem 8.6, show that (8.14), a, + a,, + u,,, = 0,
is the equation of the nplane.
Substituting into (8.14) the o's of Problem 8.6, a1 + o,, + aII1 = fiZ = O, or Z = O which
is the XY plane @plane).
with axes a t 45' in the plot. Likewise, from (8.8) and the companion equations a111  a,, = <ry,
uI1  u1 = o,, the Tresca yield condition results in the line segments AB and ED with equations
( o I / o y ) ( o I I I / o y )= 21, DC and F A with equations oI,,/uy = f1, and BC and EF with equations
o I / o y = T I , respectively.
8.9. The von Mises yield condition is referred to in Section 8.3 as the Distortion Energy
Theory. Show that if the distortion energy per unit volume u,dD,is set equal to the
yield constant C, the result is the Mises criterion as given by (8.12).
From Problem 6.26, uTD)is given in terms of the principal stresses by
8.11. For the case of plastic plane strain with r,, = O, de,, = O and u,, = 0, show that the
LevyMises equations (8.19) lead to the conclasion that the Tresca and Mises yield
conditions (when related to pure shear yield stress k) are identical.
Here (8.19) becomes dell = (2õ11 u3,) dh/3, dez2 = (oll + dh/3, O = 203,  011. Thus in the
absence of shear stresses, C , = u l l , ull = 03, = u I 1 / 2 , ( r I I 1 = O = az2. Then from (8.9) the Tresca
yield condition is o,  a,,, = o,, = 2k. Also, from (8.19) for this case, Mises condition becomes
+
(ul1/2)2 (ul1/2)2 +
(ul1)2 = 6k2 or oS1 = 4k2 and oll = 2k.
8.12. Show that the PrandtlReuss equations imply equality of the Lode variable p (see
Problem 8.3)and v = ( 2 d 4  dc;  d r g I ) l ( d r ;  d&).
From equations (8.21),
8.14. Show that when the plastic potential function g(aii) = IIr,, the plastic potential equa
tions (8.22) become the PrandtlReuss equations.
The proof follows directly from the result of Problem 8.13, since aglaa, = sij in this case and
(8.22) reduce to (8.21).
8.15. Expand (8.24) to show that the equivalent stress a,, may be written in the form
of (8.23).
From equation (8.24),
o g Q = 3siisij/2 = 3(aij Sijupp/3)(aij
 Sijaqq/3)/2 = (3aiiuij  uiioii)/2
Expanding this gives
8.16. In plastic potential theory the plastic strain increment vector is normal to the loading
(yield) surface a t a regular point. If [N,,N,, N3] are direction numbers of the normal
to the yield surface f , ( a i j ) , show that d r p l s , = d~;ls,, = d ~ ~ under
~ l the
s Mises
~ ~ yield
~
condition and flow law.
The condition of normality is expressed by N = grad f l which requires N l l ( a f l l a ~ I )=
+ +
N 2 / ( d f l / d o I I )= N31(afllauIII) for the Mises case where f l = (aI  aII)2 ( o I I o I I I ) 2 ( a I I I o J 2 
2a; = O. Here af1/aaI = 2(2u1 a,,  a I I I ) = 6s,, etc., and since the plastic strain increment vector
is along the normal i t follows that deF/sI = d ePI I / s I I= d e g I / s I I I .
8.17. Determine the plastic strain increment ratios for (a) simple tension with a,, = a,,
( b ) biaxial stress with V,, =  ~ , / f i ,uZ2= a Y / f i , u~~= a12 = u~~= u13= O, (C) pure
shear with V,, = ~ , / f l .
(a) Here all = a1 = a,, aI1 = ' a I I I= O and s1 = 2ay/3, s I I = s I I I = oy/3. Thus from Problem
8.16, depl2 =  d e g l l =  d e E I / l .
8.18. Determine the plastic. work increment dWP and the equivalent plastic strain incre
ment de;, for the biaxial stress state a,, =  ~ ~ / fu~~ i =, uy/\/3, = = = (rl2
In principalaxis form, (8.30) becomes d W P = o, de; aI1de: + + a,,, de;,; and for the stress
state given, Problem 8.17 shows that de; = deEI, de: = 0 ; hence
dWP = u,c/\/~ (ay/fi)(c) = 2cay/d
From (8.25),
= {2[(dep deg)2 + (de:  d e g I ) 2+  dep)21}112/3
190 PLASTICITY [CHAP. 8
8.19. Verify (8.32) by showing that for a PrandtlReuss material the plastic work increment
is dWP = o,, d ~ ; , as given in (8.31).
From (8.30), d W P = sijsijdh for a PrandtlReuss material satisfying (8.21). But from (8.27),
P
dh = 3 deEQ/2UEQ for such a material and so d W P = ( 3 ~ ~ ~ 8 ~ ~ / 2 ) ( dwhich
e ; ~ / ubecause
~ ~ ) of the
&EQ.
definition (8.24) gives dWP = U E Q Thus de:, = dWP/uEQ here and ( 8 3 2 ) follows directly
from (8.21).
8.20. For a material obeying the Mises yield condition, the equivalent stress a,, may be
taken as the yield function in the hardening rules (8.36) and (8.36). Show that in this
case uEQFr= H' where Fr and H r are the derivatives of the hardening functions with
respect to their respective arguments.
Here (8.34) becomes U E Q = F ( W P ) and so doEQ= F' dWP. Likewise (8.36) is given here by
OEQ = H(€;,) and so daEQ= H' de;,. Thus F' dWP = H' de;Q; and since from (8.31) (or Problem
8.19) dWP = U E Q i t follows a t once that uEQF1= H'.
8.22. Verify that the Hencky parameter 4 may be expressed as given in (8.39).
Squaring and adding the components in the equation e: = @sij of Problem 8.21 gives e; €5 =
+2sijsij or Q = 4luEQ which when multiplied on each side by 213 becomes
Fig. 815
Here the only nonzero stress is the bending stress ull. In the elastic portion of the beam
(a < x , < a ) , ull = Ecll = E x d R where R is the radius of curvature and E is Young's modulus.
In the plastic portion, ull = uy. Thus
where uy = E a I R , the stress co?dition a t the elasticplastic interface, has been used. From the
result obtained, M = 2bc2uy/3 a t first yield (when a = c ) , and M = bczu, for the fullyplastic
beam (when a = 0).
CHAP. 81 PLASTICITY 191
8.24. Determine the moment for a beam loaded as in Problem 8.23 if the material is a
+
piecewise linear hardening material for which a , , = a , A(€,, u Y / E ) after yield.
The stress distribution for this beam is shown in Fig. 816.
Again c l l = x2/R and so
O 6 r 6 a, and by o,, 
The shear stress u12 is given here by alz = krla for
k for a 5 r 6 e where k is the
yield stress of the material in shear. Thus
1
(k+/a) dr + 2a
S.'kr2 dr = ?qc3
3
a3 / 4 )
8.28. Using the condition that the yield stress in shear k is constant, combine (8.47) with
the equilibrium equations and integrate to prove (8.48).
From the equilibrium equations aall/dxl + ao121axz = O and do12/ax1 + a a 2 2 / a ~ 2= O which
a r e valid here, (8.47) yields
aplaxl  k ( 2 cos 2g)(aglax1)+ k(2 sin 2g)(agldx,) = O
and k ( 2 cos 2g)(a+lax2) = O
k(2 sin 2g)(ag/axl) aplax, I
If x 1 is along a n a line and x , along a ,L3 line, @ = O and these equations become dplax, 
2k(dglaxl) = O along the a line, aplaxz +
2k(agldx2) = O along the ,L3 line. Integrating directly,
p + 2kg = C1 on the a line, p  2kg = C z on the line.
8.29. I n the frictionless extrusion through a square die causing a fifty per cent reduction,
the centered f a n region is composed of straight radial ,8 lines and circular a lines a s
shown in Fig. 819. Determine the velocity components along these slip lines in
terms of the approach velocity U and the polar coordinates r and O.
Fig. 819
Along the straight ,L3 lines, d@= 0; and from (8.53), dv2 = O o r v 2 = constant. From the ,
normal velocity continuity along B C , the constant here must be U cose and so v2 = U cose. Along
the circular a lines, dg = de; and from (8.52),
0
MISCELLANEOUS PROBLEMS
8.30. Show that the vonSMisesyield condition may be expressed in terms of the octahedral
shear stress uoCt (see Problem 2.22) by ao,, = f i ~ , / 3 .
+ +
I n terms of principal stresses 3o,,, = d ( o I  o I I ) 2 ( o I I u I I I ) 2 ( o I I I uI)2 (Problem 2.22)
+ +
and so 9oOCt = (oI  o I I ) 2 ( o I I a I I I ) 2 ( o I I I ar), = 2u$ i n agreement with (8.12).
8.31. Show that equation (8.13) for Mises yield condition may be written as
sS + s:, + s;,, = 2k2
Expanding and rearranging, this may be written s: sSI sSIr  ( s I + + + s I I + s I I I ) 2 / 3= 21~2. But
sI + +
sl, s,,, = IpD O and the required equation follows.
C H A P . 81 PLASTICITY 193
For the state of stress where u and T are constants, determine the
8.34. Show that the PrandtlReuss equations imply incompressible plastic deformation and
write the equations in terms of actual stresses.
P

From (8.21), deii = sii dh = O since sii = IXD O and the incompressibility condition dez = O
+
is attained. In terrns of stresses, d e z = (oij Siiukk/3)d ~ .~ h u sde:, = (2/3)[uli (oz2 q 3 ) / 2 ]d ~ ,
etc., for the normal components and de:, = o,, dA, etc., for the shear components.
8.35. Using the von Mises yield condition, show that in the Uplane the deviator stress com
ponents at yield are
S, = [  2 ~ ,cos (9  ~ / 6 ) ] / 3 , s,
' +
= [2uY cos (9 ~/6)]/3, s,,, = ( 2 u y sin $)I3
where 9 = tanI Y / X in the notation of Problem 8.6.
The radius of the Mises yield circle is &%ry so that by definition X = m o y cose,
Y = m o y sin e a t yield. From the transformation table given in Problem 8.6 together with
o, = s , +
U M , etc., the equations s ,  s,, =  f i=~  ( 2 / f i ) a y cose and S I i  2~111=
6 + +
Y = 20y sin e are obtained. Also, in the nplane, sI s I I s I I 1= O . Solving these three
equations simultaneously yields the desired expressions, a s the student should verify.
qJ
8.37. An elasticperfectly plastic rec
tangular beam is loaded in pure
bending until fully plastic. Deter
mine the residual stress in the
beam upon removal of the bending
6 r l . ) _L
k b  4
m0ment M. Fig. 821
For the fully plastic condition, the moment is (see Problem 8.23) M = bc20y. This moment
would cause a n elastic stress having a = Mel1 = 3oy/2 a t the extreme fibers, since I = 26~313.
x
Thus remova1 of M is equivalent to applying a corresponding negative elastic stress which results
in the residual stress shown in Fig. 822.
30~12 oy/2
fully
plastic
+ negative
elastic
Fig. 822

residual
stress
8.38. A thickwalled cylindrical tube of the dimensions shown in Fig. 823 is subjected to
an interna1 pressure p , . Determine the value of p, a t first yield if the ends of the tube
are closed. Assume (a) von Mises and ( b ) Tresca's yield conditions.
The cylindrical stress components (Fig. 824) are principal stresses and for the elastic analysis
+
) pi(b2/r2 l ) / Q , o(zL)= pilQ where Q = (b2Ia2  1).
may be shown to be o(,,) = pi(b2/r2  1 ) / Q , o ( e o =
(b) For the Tresca yield condition, o(oe) o(,,) = o y since o,, is the intermediate principal stress.
Thus 2pibz/r2 = Qoy and now a t r = a , pi = ( o y / 2 ) ( 1 a2/b2) a t first yield.
CHAP. 81 PLASTICITY
Supplementary Problems
A onedimensional stressstrain law is given by o = Ken where K and n are constants and E is
true strain. Show that the maximum load occurs a t e = n.
Rework Problem 8.4 using the yield stress in shear k in place of u y in the Mises and Tresca yield
conditions. +
Ans. Mises: ( o l f i k)2 (rlk)2 = 1; Tresca: ( ~ / 2 k ) 2 ( ~ / k ) =
2 1 +
Making use of the material presented in Problem 8.6, verify the geometry of Fig. 85(c). .
From the definition of Lode's parameter p (see Problem 8.3) and the Mises yield condition, show
that o,  u,,, = 2 o y / ~ ~ .
In the 11plane where s = tan1 Y I X with X and Y defined in Problem 8.6, show that p = 6tan 0 .
Show that the invariants of the deviator stress IrID = sijsij/2 and IIIED= sijsjkski/3 may be written
= (s; + s;, + s:,,)/2 and III,, = (si + s;, + si,,)/3 respectively.
Show that von Mises yield condition may be written in the form
(011  ~ 2 2 ) ~(o22  ~ 3 3 4) ~
(a33 011)2 + 6(a:2 + ui3 + mil ) = 6k2
Following the procedure of Problem 8.17, determine the plastic strain increment ratios for
(a)biaxial tension with ul1 = o,, = uy, ( b ) tensiontorsion with oll = oy/2, u12 = oY/2.
Ans. (a) deFl = dei, = der3 12 ( b ) deTl /2 = dei2 = der3 = ddeL/3
Verify the following equivalent expressions for the effective plastic strain increment degQ and note
that in each case degQ = deFl for uniaxial tension ull.

The beam of triangular cross section shown in Fig. 825
is subjected to pure bending. Determine the location of
the neutra1 axis (a distance b from top) of the beam when
fully plastic. Ans. b = h l f i
Fig. 825
\ 0 oP/
when referred to the axes rotated about x3 by an angle O
in Fig. 88.
The Maxwell model in viscoelasticity is the combination of a spring and dashpot in series
as shown by Fig. 92(a). The Kelvin or Voigt model is the parallel arrangement shown in
Fig. 92(b). The stressstrain relation (actually involving rates also) for the Maxwell
model is
u+  u = ; '
(9.3)
G 7
and for the Kelvin model is
u = Ge+rl;
The simple Maxwell and Kelvin models are not adequate to completely represent the
behavior of real materials. More complicated models afford a greater flexibility in por
traying the response of actual materials. A threeparameter model constructed from two
springs and one dashpot, and known as the standard linear solid is shown in Fig. 93(a). A
threeparameter viscous model consisting of two dashpots and one spring is shown in
Fig. 93(b). I t should be remarked that from the point of view of the form of their con
stitutive equations a Maxwell unit in parallel with a spring is analogous to the standard
linear solid of Fig. 93(a), and a Maxwell unit in parallel with a dashpot is analogous to the
viscous model of Fig. 93(b).
G9 G,
A fourparameter model consisting of two springs and two dashpots may be regarded
as a Maxwell unit in series with a Kelvin unit as illustrated in Fig. 94 below. Severa1
equivalent forms of this model exist. The fourparameter model is capable of a11 three of
the basic viscoelastic response patterns. Thus it incorporates "instantaneous elastic re
198 LINEAR VISCOELASTICITY [CHAP. 9
sponse" because of the free spring G,, "viscous flow" because of the free dashpot v,, and,
finally, "delayed elastic response" from the Kelvin unit.
Fig. 94
The stressstrain equation for any of the three or four parameter models is of the
general form
+ +
p2a pl& pou = q2Y+ ql; Q o C + (9.7)
where the p,'s and qi's are coefficients made up of combinations of the G's and ?'s, and depend
upon the specific arrangement of the elements in the model. In operator form, (9.7) is
written
{p2a: + plat + p0}u = {q2a: + qlat + qo)€ (98)
Fig. 95
Similarly, a sequence of Maxwell units in parallel as shown in Fig. 96 is called a generalized
Maxwell model. Here the total stress is the resultant of the stresses across each unit; and
so from t9.5),
t 0
Fig. 96
CHAP. 91 LINEAR VISCOELASTICITY 199
For specific models, (9.9) and (9.10) result in equations of the form
p O u + p l ~ + p 2 a +. . . = qOc+ql;+q2;.+ ...
which may be expressed compactly by
where [ U ( t ) ]represents the unit step function applied a t time t, = O. The creep response
of a Kelvin material is determined by solving the differential equation
which results from the introduction of (9.16) into (9.4). Here T = rllG is called the retarda
tion time. For any continuous function of time f ( t ) ,it may be shown that with t' as the
variable of integration,
by means of which (9.17) may be integrated to yield the Kelvin creep response
"o
.(t) = 
G
( 1  et17)[U(t)] (9.19)
The creep loading, together with the creep response for the Kelvin and Maxwell models
(materials) is shown in Fig. 98 below.
200 LINEAR VISCOELASTICITY [CHAP. 9
1,,
uo 
*
t t
(a) Creep Loading ( b ) Creep Response
Fig. 98
The stress relaxation which occurs in a Maxwell material upon application of the strain
with the help of which (9.21) may be integrated to give the Maxwell stress relaxation
The delta function in (9.25) indicates that it would require a n infinite stress to produce a n
instantaneous finite strain in a Kelvin body.
where Ji = l/Gi is called the compliance. If the number of Kelvin units increases indefinitely
so that N + m in such a way that the finite set of constants ( r , Ji) may be replaced by the
continuous compliance function J(T),the Kelvin creep function becomes
where +(t) is called the relaxation function. For the generalized Maxwell model of Fig. 96,
the relaxation function is determined from (9.24) as
Here, as N + .o the function G(T) replaces the set of constants (G, T ~ )and the relaxation
function is defined by
+(t) = G(T)~"'dr Sm (9.31)
The function G(7) is known as the "distribution of relaxation times", or relaxation spectrum.
In linear viscoelasticity, the superposition principle is valid. Thus the total "effect"
of a sum of "causes" is equal to the sum of the "effects" of each of the "causes". Accord
ingly, if the stepped stress history of Fig. 99(a) is applied to a material for which the creep
function is ~k(t),the creep response will be
Therefore the arbitrary stress history u = ~ ( t )of Fig. 99(b) may be analyzed a s an
infinity of step loadings, each of magnitude da and the creep response given by the super
position integral
m
~ ( t ) = Jt $p
9 ( t  t') dt'
Such integrals are often referred to as hereditary integrals since the strain a t any time is
seen to depend upon the entire stress history.
"o '
"1
tl
I
t2
"3

I I
I
t3
 v      
Fig. 99
/e
" + gdt'   
" 
I
I
I
I
t'
(a,
t'
I
I
I
I
+ dt' t 
202 LINEAR VISCOELASTICITY [CHAP. 9
For a material initially "dead", i.e. completely free of stress and strain a t time zero, the
lower limit in (9.33) may be replaced by zero and the creep response expressed as
c(t) = ltd"(t3
dt'
qr(t  t') dt'
Following similar arguments a s above, the stress as a function of time may be rep
resented through a superposition integral involving the strain history e(t) and the relaxation
function +(t). In analogy with (9.33) the stress is given by
.(t) = J
m g +(t  t') dtf
and with regard to a material that is "dead" a t t = 0, the integrals comparable to (9.34)
and (9.35) are respectively
.(t) = Xt%+(tt') dt' (9.37)
and
Since either the creep integral (9.34) or the relaxation integral (9.37) may be used to
specify the viscoelastic characteristics of a given material, it follows that some relationship
must exist between the creep function +(t) and the relaxation function +(t). Such a relation
ship is not easily determined in general, but using the Laplace transform definition
it is possible to show that the transforms G ( s ) and $(t) are related by the equation
o0 sin S
( b ) the loss modulus G, = 
€0
c, cos S
( c ) the storage compliance J, = 
"o
c, sin 6
(d) the loss compliance J, = 
,o
Fig. 910
2. Straindisplacement equations
2fij = (ui,j + uj,i)
or strainratevelocity equations
+
2Gij = (v,,~ V,~)
3. Boundary conditions A
4. Initial conditions
ui (x,, 0) = UO
5. Constitutive equations
(a) Linear differential operator form (9.48)
If the body geometry and loading conditions are sufficiently simple, and if the material
behavior may be represented by one of the simpler models, the field equations above may be
integrated directly (see Problem 9.22). For more general conditions, however, it is con
ventional to seek a solution through the use of the correspondente pr.inciple. Thjs principle
emerges from the analogous form between the governing field equations of elasticity and the
Laplace transforms with respect to time of the basic viscoelastic field equations given above.
A comparison of the pertinent equations for quasistatic isothermal problems is afforded
by the following table in which barred quantities indicate Laplace transforms in accordance
with the definition
f(xk9S ) = xw f ( ~ ,t)est
> dt
From this table it is observed that when G in the elastic equations is replaced by &lP, the
two sets of equations have the same form. Accordingly, if in the solution of the "correspond
ing elastic problem" G is replaced by &lP for the viscoelastic material involved, the result
206 LINEAR VISCOELASTICITY [CHAP. 9
is the Laplace transform of the viscoelastic solution. Inversion of the transformed solution
yields the viscoelastic solution.
The correspondence principle may also be stated f o r problems other than quasistatic
. problems. Furihermore, the form of the constitutive equations need not be the linear dif
ferential operator form but may appear a s in (9.49), (9.50) or (9.51). The particular problem
under study will dictate the appropriate form in which the principle should be used.
Solved Problems
VISCOELASTIC MODELS (Sec. 9.19.3)
9.1. Verify the stressstrain relations f o r the Maxwell and Kelvin models given by (9.3)
and (9.4) respectively.
In the Maxwell model of Fig. 92(a) the total strain is the sum of the strain in the spring plus
+
the strain of the dashpot. Thus e = es ED and also ;= Zs +
GD. Since the stress across each
element is o, ( 9 . 1 ) and (9.2) may be used to give ;= O/G a/?. +
, In the Kelvin model of Fig. 92(b), o = os + a~ and directly from (9.1) and (9.2), o = v ; + Ge.
9.2. Use the operator form of the Kelvin model stressstrain relation to obtain the stress
strain law for the standard linear solid of Fig. 93(a).
Here the total strain is the sum of the strain in the spring plus the strain in the Kelvin unit.
Thus E = es +
e~ or in operator form e = olG1 +
o / { G 2 q2at). From this +
and so GIGze + G l s z ; = ( G 1+ Gg)a + 72;.
9.3. Determine the stressstrain equation for the four parameter model of Fig. 94. Let
T,I~ +
and compare with the result of Problem 9.2.
Here the total strain e = e~ + e~ which in operator form is
e = o/{Gp + vzat) + {at + l / ~ l ) o / G l { a t )
Expanding the operators and collecting terms gives
eet/7 = ( ~ ~ [ U ( t ) ] / ~ ) [ r=
e t '(uO/G)(et/7
/~]~ l)[U(t)]
n. i'
or
et'/7[U(t1)]dt' which by formula
e = (ao/G)(l et/T)[U(t)]
(9.23),
Use of et/' a s the integrating factor in (9.21) gives oet/7 = GeO
9.7. Determine the creep response of the standard linear solid of Fig. 93(a).
Since e = es + EK for this model the creep response from (9.1) and (9.19) is simply
e(t) = [ l / G l + (1/G2)( 1  e  t / 7 ~ ) ] a o [ U ( t ) ]
The same result may be obtained by setting q2 = m in the generalized Kelvin (N = 2) response
2
c = 2
i=l
J i ( l  e  t / r , ) ~ o [ U ( t ) ]or by integrating directly the standard solid stressstrain law. The
student should carry out the details.
A t t = 2r2 the load is removed and o becomes zero a t the same time that the "elasticJJ deformation
uo/G1 is recovered. For t > 272 the response is governed by the equation +
€/r2 = O which is
the stressstrain law for the model with u = O (see Problem 9.2). The solution of this differential
equation is C = C e  T / r ~where C is a constant and !i'
= t  2 ~ A~ t . T = 0, C = C = oO(l e2)/G2
and so
e = aO(l e2)eT1r2/G2 = uo(e2 l ) e  t / r z/G 2 for t > 2 r 2
9.9. The special model shown in Fig. 916 is elongated a t a constant rate ;= ~ , / t , as
indicated in Fig. 917. Determine the stress in the model under this straining.
i r+2
3 ~ ~ t 'dt'
tl
Xt dtf
9.10. Determine by a direct integration of the stressstrain law for the standard linear
solid its stress relaxation under the strain E = e , [ U ( t ) ] .
+ + +
Writing the stressstrain law (see Problem 9.2) a s 6 (G1 G2)uIq2= ~ ~ G ~ ( [ 8 (G1G2[U(t)]lT2)
t)]
for the case a t hand and employing the integrating factor e(Gi+Gz)tlnz i t is seen that
+
Thus by use of (9.18) and (9.23), o = eo(Gl G2et/rz) = eo$(t). Note that this result may also be
obtained by putting q1 + m in (9.30) for the generalized Maxwell model.
CHAP. 91 LINEAR VISCOELA'STICITY 209
9.12. Using the relaxation function +(t)for the model of Problem 9.11, determine the creep
function by means of (9.40).
+
The Laplace transform of $ ( t ) = G1 G2et/Tr is 5 ( s ) = G l / s + G z / ( s+ 1 1 ~ (see
~ ) any stand
ard table of Laplace transforms). Thus from (9.40),
This result may be readily verified by integration of the model's stressstrain equation under creep
loading.
9.14. Using the creep integral (9.34) together with the Kelvin creep function, verify the
result of Problem 9.13.
For the Kelvin body, $ ( t )= (1 etI7)/G and (9.34) becomes
=
X

G o
St
[ [ ~ ( t ) ] ( 1  e  ( t  t l ) l ~dt'
)  [ U ( t tl)]
tl
t
$
(1  e  ( t  t f ) l r ) d t f
I
A straightforward evaluation of these integrals confirms the result presented in Problem 9.13.
The stress may be represented a s a sequence of ramp loadings a s shown in Fig. 921 above.
+
From Problem 9.13, E = ( h / G ) [ t r(etlf  I ) ][ U ( t ) ] for this stress loading. In the présent case
therefore
+
) ( i / G ) [ ( t r(etlT  l ) ) [ U ( t ) 
~(t= ] ( ( t t i ) + r ( e  ( t  t l ) l ~  l ) ) [ U ( t ti)]
((t t 2 t l ) ] + ( ( t 3tl) + ~(e(t3t1)I7 l ) ) [ U ( 
 2tl) + ~ ( e  ( t  z t ~) /l ~) ) [ U (  t 3tl)]l
Note that a s t + m, E + 0.
9.17. Show that the result of Problem 9.16 may also be obtained by simply replacing the
operator d, by .i in equation (9.5)and defining u / e = G*.
After the suggested substitution (9.5) becomes (iolG + l / q ) o = &e from which
U / E = Giol(io + 117) = G ~ w T+ /(~ ~ W T )
a s before.
9.18. Use equation (9.10) for the generalized Maxwell model to illustrate the rule that
"for models in parallel, the complex moduli add".
From Problem 9.17 the complex modulus for the Maxwell model may be written G* = a/€ =
Giwrl(l+ iwr). , Thus writing (9.10) a s
9.19. Verify the relationship Jl = l / G l ( l+ tan28) between the storage modulus and
compliance.
From (9.43) and (9.44), J* = 1/G* and so J1  iJ2 = l / ( G l iG2) = (G1 ~G,)I(G? 62). + +
Thus
+ +
J1 = G ~ / ( G : 02) = l / G 1 ( l (G2/G1)2) = l / G 1 ( l tan2 6) +
9.20. Show that the energy dissipated per cycle is related directly to the loss compliance
J2 by evaluating the integral
S u d over
~ one cycle.
I
For the stress and strain vectors of Fig. 910, the integral
S odr evaluated over one cycle is
ds
dt = l 2 ~ 1 0
(oo sin U cos (ut  6 ) d t
~ ) E ~ W
9.22. A bar made of Kelvin material is pulled in tension so that a,, = a o [ U ( t ) ] , a,, = a33 =

ai,  a,, = a,, = O where ao is constant. Determine the strain ell for this loading.
From (9.48b), 3cii = o o [ U ( t ) ] l Kfor this case; and from ( 9 . 4 8 ~with
) i = j = 1, { P ) ( u l l a11/3)=
+
{ 2 Q ) ( ~ l l eii/Q). But from (9.61, { P ) 1 and { Q ) = {G Tat) for a Kelvin material; so that now
200[U(t)I/3 = 2{G +~ a ~ aO[U(t)]lgK)
l ( ~ ~ ~
The viscoelastic operator for the term ( 1  2 ~ is) { 3 Q ) l { 3 K P + Q ) so that for a Kelvin body
the transformed viscoelastic solution becomes
which may be inverted with help of partia1 fractions and transform tables to give the viscoelastic
stress
9.25. The correspondence principie may be used to obtdin displacements a s well as stresses.
The x displacement of the surface of the elastic half space in Problem 9.24 is given by
W ( z = ~=) P(l v 2 ) l E T r . Determine the viscoelastic displacement of the surface for
the viscoelastic material of that problem.
The viscoelastic operator corresponding to ( 1  v2)lE is { 3 K + 4Q)/4Q(3K + Q ) which for the
Kelvin body causes the transformed displacement to be
The bending stress for a simply supported elastic beam does not depend upon material properties,
so the elastic and viscoelastic bending stress here are the same. The elastic deflection of the beam
is w ( x l ) = pon(xI)/24EZ where a ( x J is a known function. For a Maxwell body, {P) = {a, 117) +
and { Q ) = {Ga,), so that the transformed deflection is
9.27. Show that as t += .o the stress a,, in Problem 9.23 approaches ao (material behaves as
a fluid) if the material is considered incompressible ( V = 112).
From Problem 9.23,
0221t+m = oO(9K  (4G + 3K))/2(4G+ 3 K ) = oo(3K  2G)l(3K + 4G)
which may be written in terms of v as 02,1t,, = voo/(l  v). Thus for v 112, o,,lt+, = o*
CHAP. 91 LINEAR VISCOELASTICITY 213
MISCELLANEOUS PROBLEMS
9.28. Determine the constitutive relation for the 81
KelvinMaxwell type model shown in Fig. .
925 and deduce from the result the Kelvin
and Maxwell stressstrain laws. c<t
Here
a = UM + aK = ;/{dt/G1 + l / v l } + {Gz + q2at}r 82
9.31. The "logarithmic retardation spectrum" L is defined in terms of the retardation spec
trum J by L(ln r ) = r J ( r ) . From this definition determine the creep function +(t)in
terms of L(ln r ) .
Let ln r = h so that eX = T and thus drldh = "e T, or dr = ~ d ( l Tn) . From this, (9.28)
defining g ( t ) becomes q ( t ) =
LW L(ln ~ ) ( let")d(ln
 r). In the same way, if
defines the logarithmic relaxation spectrum, $ ( t ) in (9.81) may be written
H(ln r ) = TG(T)
9.33. Determine the viscoelastic operator form of the elastic constant V (Poisson's ratio)
using the constitutive relations (9.48).
Under a uniaxial tension o,, = o,, (9.48b) gives si/3 = o d 9 K so that ( 9 . 4 8 ~ )for i = j = 1
+
yields cll = ( 3 K P Q)ooI{9KQ). In the same way ( 9 . 4 8 ~ )for i = j = 2 yields cz2 =
(2Q  3PK)oo/{18KQ). Thus in operator form, V = c2,/c11 = (3PK  2Q)/{6KP 2Q). +
The relaxation function +, may be found with the help of (9.40). The result is
t  ~ ) ~ r l ( t  t '
)  h)eVz(ttl)
rs3 = Kbt + i.
(r1  ~
2
2 )
I IU(tl)]dt'
which upon integration gives
033 + +
= (K;Ot 2Z0[(r1 X ) ( I  e~it)/3r, (r,  X ) ( I  erzt)/3r2]/(rl
r , ) ) [ ~ ( t ) ]
9.35. The "creep buckling" of a viscoelastic column may be analyzed within the linear
theory through the correspondence principle. Determine the deflection w(x,, t ) of a
Kelvin pinnedend column by this method.
Fig. 930
9.36. Formulate the steadystate vibration problem for a viscoelastic beam assuming' the
constitutive relations are those given by (9.48).
,
The free vibrations of an elastic beam are governed by the equation E Z ( ~ ~ W / ~ X ; ) +
+
p A ( a 2 ~ l a t 2=) O. From (9.48) the viscoelastic operator for E is {9KQl(3KP Q ) ) ,and if the deflec
tion w ( x l ,t ) = W ( x l ) e ( t ) the resulting viscoelastic differential equation may be split into the space
+
equation d 4 W l d x ~ k4W = O and the time equation (3KP Q)(d2eldt2) (k4ZlpA){9KQ)(~) + = 0.
The solution W i of the space equation represents the ith mode shape, and from the time equation for
N
k = ki the solution ei = 2 Aijehilt where N depends upon the degree of the operator. The total
j=1 m N
solution therefore is w ( x l ,t ) = Wi(xl)AijeAijt in which the Aij are complex.
i=1 j=l
Supplementary Problems
9.37. Determine the constitutive equation for the four parameter mode1 shown in Fig. 931.
Ans. 8 + (Gilv2 + G2/v2+ Gl/7l); + ( G l G z / ~ l ~ z=) u G l f . + (G1G2/72)'
Fig. 931
216 LINEAR VISCOELASTICITY [CHAP. 9
9.38. Determine the creep response of the standard linear solid by direct integration of E +€ 1 =~ ~
+ +
oo[U(t)](Gl G2)/G1v2 uo[8(t)]/Gl. (See Problem 9.7.)
9.39. Deduce the Kelvin and Maxwell stressstrain laws from the results established in Problem 9.5 for
the four parameter model of that problem. (Hint. Let G3 = 0 , etc.)
abmtk1.) A m . $(t) =
+
+(t) = G2 Glet/Tl Fig. 932
9.43. In the model of Problem 9.42 let G1 = G, = G and q, = q3 = q and determine the stress history
of the resulting model when it is subjeeted to the strain sequence shown in Fig. 934.
€0
Ans. u =  (G(2t t,)
tl
+ ~ ( 4( 1 + eti'r)etlT)) for t l < t < 2tl
wo sin (7x11)
Po[U(t)l P,[U(t)l
9.45. A pinnedend viscoelastic column is a Maxwell material for which 8 olr = E;. The initial +
shape of the column is w = w o sin ( ~ z l l when ) the load Po [ U ( t ) ]is applied (see Fig. 936). Deter
mine the subsequent deflection w ( x l , t ) a s a function of P,, the elastic buckling load.
Ans. w ( x l , t ) = w0 sin ( a ~ ~ / l ) e  t ~ ( '  P ~ / P o ) ~
INDEX
Absolute dynamic, Conjugate dyadic, 4
compliance, 202 Conservation of,
modulus, 202 energy, 128
Acceleration, 111 . m a s , 126
Addition and subtraction, Constitutive equations, 132
of matrices, 17 Continuity equation, 126
of (Cartesian) tensors, 1 5 Continuum concept, 44
of vectors, 2 Contraction, 1 6
Adiabatic process, 140 Contravariant tensor, 11
Airy stress function, 147 Convective,
Almansi s t r a i n tensor, 82 derivative, 110
Angle change, 86 r a t e of change, 111
Angular momentum, 128 Conventional stress and strain, 175
Anisotropy, 44, 141 Coordinate transformation, 11
Antisymmetric, Correspondence principle, 205
dyadic, 5 Couplestress vector, 45
matrix, 19 Coupled h e a t equation, 150
tensor, 19 Covariant tensor, 1 2
Axis of elastic symmetry, 142 Creep,
function, 200
Barotropic change, 161 test, 199
Base vectors, 6 Cross product, 3, 5, 1 6
Basis, 6, 9 Cubical dilatation, 90
orthonormal, 6 Curvi1ine:ir coordinates, 7
Bauschinger effect, 176 Cylindrical coordinates, 8
BeltramiMichell equations, 144
Bernoulli equation, 164 Dashpot, 196
Biharmonic equation, 148 Decomposition,
Body forces, 45 polar, 87
Boundary conditions, 144, 146 velocity gradient, 112
Bulk, Deformation, 77
modulus, 143 gradients, 80
viscosity, 161 inelastic, 175
plane, 91
Caloric equation of state, 132 plastic, 175
Cartesian, tensors, 81
coordinates, 6 total (deformation) theory, 183
tensors, 1, 12, 1 3 De1 operator, 22
Cauchy, Density, 44, 126
deformation tensor, 8 1 entropy, 130
strain ellipsoids, 89 strain energy, 141
stress principle, 45 Derivative,
stress quadric, 50 material, 110, 114
CauchyRiemann conditions, 165 of tensors, 22
Circulation, 164 of vectors, 22
Kelvin's theorem of, 165 Deviatoric,
ClausiusDuhem inequality, 131 s t r a i n tensor, 91
Column matrix, 17 stress tensor, 57
Compatibility equations, 92, 114 Diagonal matrix, 17
Complex, Dilatation, 90
niodulus, 202 Direction cosines, 7
potential, 166 Displacement, 78, 83
Compliance, 201 gradient, 80
Coniponent, 1, 7 relative, 83
Compressible fluid, 165 Dissipation function, 132
Configuration, 77 Dissipative stress tensor, 131
Conformable niatrices, 17 Distortion energy theory, 178
INDEX
Work,
Vector, hardening, 176
addition, 2 plastic, 182
displacement, 78
dual, 16 Yield,
field, 22 condition, 177
of a dyadic, 4 curve, 179
position, 12, 77 surface, 179
potential, 126 Young's modulus, 143
products, 3, 4, 16
rotation, 83 Zero,
traction, 46 matrix, 17
transformation law, 14 order tensor, 1
vorticity, 113 vector, 2