You are on page 1of 4

Is Stress at a Point a True Tensor?

G. Chakraborty
Associate Professor
Department of Mechanical Engineering
Indian Institute of Technology Kharagpur
INDIA – 721302
e-mail: goutam@mech.iitkgp.ernet.in

ABSTRACT

Stress in continuum mechanics is traditionally considered to be a true tensor of second rank.


However, certain inconsistencies have been noted with the above tensor character of stress at a
point, which can only be removed if stress is represented as a pseudo tensor. Numerous
evidences both from solid mechanics and fluid mechanics are given to substantiate the claim.

1. INTRODUCTION

Stress at a point within a continuous medium is defined as a tensor of second rank in numerous
treaties and text books [1-3]. Exception is the book written by Brillouin [4] where a different
nature of stress tensor has been suggested. In the last reference stress is defined as a tensor
density, a pseudo tensor which follows a different rule of transformation under change of co-
ordinates. The purpose of the present communication is to investigate Brillouin’s claim further.

2. NATURE OF STRESS TENSOR

Any physical quantity is represented as a tensor whose nature depends on the rule of
transformation its components follow when a different co-ordinate system is used. Second order
tensors , and , defined in obey the following transformation rules, respectively,

, , and (1)

where , and are the components of the same tensor when co-ordinates are changed to
by linear transformation. In equation (1) ’s are defined by the following relation:

(2)

For a non-linear co-ordinate transformation, or , ‘s


and ’s are defined as

and (3)
respectively. A tensor density, on the other hand transforms, as following

, and (4)

where . Thus, a tensor density transforms like a true tensor except for a factor 1/J.
If the stress at a point is taken to be true tensor then the following inconsistencies are
found which have not been taken seriously in the literature.
(1) In a static medium the balance of momentum equation or so called equilibrium equation
can be written as

(5)

where ρ is the material density and is the body force per unit mass. Now, it is
established from tensor analysis that the divergence operation which has been done on
in equation (5) is permissible only if the latter is a tensor density.

(2) During deformation process the mapping between the reference and the current
configuration, defined as , can be considered to be a co-ordinate
transformation with identified as the elements of deformation gradient matrix
F. The stress tensor defined in the current configuration is known as Cauchy stress
tensor while the same defined in the reference configuration is referred as Second Piola-
Kirchoff’s stress tensor and is written as . The relationship between the two tensors is
well established in literature and runs as

(6)

which can be written in the usual matrix notation as

. (7)

Equation (6) or (7) clearly shows that the components of the stress transform like a tensor
density rather than a true tensor.

(3) The well-known thermodynamic relation

(8)

where u is the internal energy density, s is the entropy density, T is the absolute
temperature and is the infinitesimal strain tensor, shows the inconsistency unless the
stress is considered to be a tensor density. In fact all the terms transform like density
except the last term which is a true scalar when the stress tensor is considered to be a true
tensor.

In fact, that stress is a tensor density can be established when the fundamental relationship
between force and stress, namely, force = stress area is written correctly in tensorial notation.
Since force is a covariant vector and area is a skew-symmetric twice contravariant tensor of
second rank the following equation can be written

(9)

where . From equation (9) it appears that the stress tensor is a thrice
covariant tensor which has antisymmetric property with respect to the last two indices. That is

. (10)

The number of indices of the above tensor can be reduced to two with the help of the
permutation tensor (Levi-Civita tensor) as follows:

. (11)

The new quantity with two indices is a pseudo-tensor whose transformation rule is given by the
last one of equation (4). That is the quantity is a tensor density. It is, therefore, seen that the
stress tensor that is represented with two indices is a tensor density.
This particular nature of stress may have profound implication in fluid mechanics. The
pressure at a point within the fluid is related to the stress as

. (12)

If stress is a tensor density then from equation (12) is appears that pressure at a point is also a
scalar density with the following transformation rule

(13)

where the subscript zero refers to the respective quantity in the reference configuration. Equation
(13) implies that although p transforms like a density the ratio behaves like a true scalar. In
fluid mechanics there are several instances to justify the above statement.

(1) The equation of state of an ideal gas, written as,

, (14)

suggests that in order to be tensorially consistent pressure must change like scalar density
since temperature T is a scalar and R is a constant.
(2) Bernoulli’s equation, which represents the energy conservation equation in an ideal fluid,
is written as

. (15)

According to this equation the quantity behaves like a true scalar as discussed above.

3. CONCLUSIONS

It is thus established that stress which is commonly defined as a tensor is, in fact, a pseudo-tensor
which has the characteristics of a tensor density of rank two. With this identification not only
some inconsistencies are removed but also it helps in writing various relations in correct tensorial
form.

REFERENCES

1. A. E. H. Love, A Treatise on the Mathematical Theory of Elasticity, Dover Publications,


ew York, 1944.

2. L. D. Landau and E. M. Lifshitz, Theory of Elasticity, Pergamon Press, 1970.

3. C. Truesdell and R. Toupin, The Classical Field Theories, Springer, 1960.

4. L. Brillouin, Tensors in Mechanics and Elasticity, Academic Press, 1964.

You might also like