Professional Documents
Culture Documents
Applications to Engineering
Petre P. Teodorescu
Treatise on
Classical Elasticity
Theory and Related Problems
Mathematical and Analytical Techniques
with Applications to Engineering
Series Editor
Alan Jeffrey
Newcastle upon Tyne, UK
Deceased 6 June 2010
Treatise on Classical
Elasticity
Theory and Related Problems
123
Petre P. Teodorescu
Faculty of Mathematics
University of Bucharest
Bucharest
Romania
The mechanics of deformable solids brings its contribution both to the knowledge
of the phenomena of the surrounding physical reality, from a theoretical point of
view, and to the determination of the state of strain and stress in various elements
of construction, from practical considerations. The deformable solids have a
particularly complex character; a mathematical modelling of them is not simple
and often leads to inextricable difficulties of computation. One of the most simple
mathematical models and, at the same time, the most used one, is the model of the
elastic body—especially the linear elastic model; despite its simplicity, even this
model of real body may lead to great difficulties of calculation.
In general, the engineering constructions have been based, during the centuries,
on empirical methods; beginning with the seventeenth century, one has obtained a
lot of results, which form what now is called ‘‘Strength of Materials’’, where
simplifying supplementary hypotheses have been introduced. As a matter of fact,
this denomination is not a proper one, because it corresponds only to a mechanical
phenomenon modelled by the so-called ‘‘strength theory’’; we maintained this
denomination, being still currently used. The theory of elasticity, chapter of the
mechanics of deformable solids with a theoretical character, succeeds to express
better the physical phenomenon, giving results closer to the reality, in certain
limits; it became a science only at the middle of ninth century, being in continuous
development even today.
The practical importance of a book on the theory of elasticity, which is—at the
same time—an introduction to the mechanics of deformable solids, consists in
putting in evidence points of view and scientific methods of computation in a
domain in which simplified methods or with a non-accurate limit of validity are
still used. The actual technical progress and the necessity to use a minimum of
materials in various constructions ask for a better determination of the state of
strain and stress which takes place in a civil or mechanical construction; the
engineering design may be thus improved.
The first eight chapters deal with the construction of the mathematical model of
a deformable solid, giving special attention to the linear elastic bodies; the for-
mulation of the fundamental problems is followed by their solution in
v
vi Preface
displacements end stresses. The importance of the concentrated loads is put into
evidence, as well in the case of Cosserat-type bodies. Another group of four
chapters contains static and dynamic spatial problems, treated systematically by
the same method of potential functions.
The following two chapters deal with some special problems: particular cases,
treated in the same systematical manner and the case of anisotropic and non-
homogeneous bodies.
The last two chapters contain introductions to thermoelasticity and linear vis-
coelasticity. Special accent is put on the solving methodology as well as on the
mathematical tool used: vectors, tensors and notions of the field theory. Contin-
uous and discontinuous phenomena and various mechanical quantities are pre-
sented in unitary form by means of the theory of distributions. Some appendices
give the book an autonomy with respect to other works, a special mathematical
knowledge being not necessary.
Concerning the first six chapters, I must mention the kind co-operation of
Professor Vasile Ille, Technical University of Cluj-Napoca, who unfortunately has
passed away. I am grateful to Mariana Gheorghitßă for her valuable help in the
presentation of this book. The excellent cooperation of the team of Springer,
Dordrecht, is gratefully acknowledged.
The book covers a wide number of problems (classical or new ones) as one can
see from its contents. It used the known literature, as well as the original results of
the author and his more than 50 years’ experience as Professor of Mechanics and
Elasticity at the University of Bucharest. It is addressed to a large circle of readers:
mathematicians (especially those involved in applied mathematics), physicists
(particularly those interested in mechanics and its connections), engineers of
various specialities (civil, mechanical engineers, etc., who are scientific
researchers or designers), students in various domains etc.
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Aim of Mechanics of Deformable Solids. . . . . . . . . . . . . . . . 1
1.1.1 General Considerations . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Models in the Mechanics of Deformable Solids . . . . . 4
1.2 Fundamental Computation Hypotheses.
Short Historical Account . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.1 Fundamental Computation Hypotheses . . . . . . . . . . . 9
1.2.2 Short Historical Account. Development Trends . . . . . 15
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3 Mechanics of Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.1 Stress Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.1.1 General Considerations . . . . . . . . . . . . . . . . . . . . . . 73
3.1.2 Variation of Stresses Around a Point . . . . . . . . . . . . 78
3.2 Stress Tensor. Equations of Equilibrium and Motion . . . . . . . 84
3.2.1 Stress Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
vii
viii Contents
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 729
The scope of mechanics of deformable solids is to determine the state of strain and
stress of a solid body subjected to the action of external loads in static or dynamic
equilibrium.
In what follows we make some general considerations and put in evidence the
basic computational hypotheses, which specify the mathematical model of the
considered solids.
In general, the strength of materials deals with the same problems as the
mechanics of deformable solids. But this discipline involves a technical solving
method that for each problem or group of problems formulates separate hypoth-
eses, reflecting more or less the physical reality in isolated cases. On the basis of
these working hypotheses, one obtains the equations which govern the phenome-
non and endeavour thus to simplify the computation as much as possible.
By contrast, the mechanics of deformable solids is a chapter of mechanics,
fundamental science of nature, which is based on a mathematical theory, built on
unitary hypotheses; it is applicable whatever the form of the body, the kind of the
material or the manner of action of the external loads may be. Each problem is
specified by its own limiting conditions (boundary conditions and initial condi-
tions (for the movable bodies)), since the fundamental hypotheses always remain
the same. The computational methods applied are general methods leading to an
accurate solution of the problem (more properly, as accurate as necessary and
possible).
In the strength of materials, the construction elements are simplified, by
reducing them to the medium fiber or to the middle surface, so as to diminish the
number of variables involved in the computation. On the other hand, additional
simplifying hypotheses concerning the distribution of the deformations on a cross
section of the body, e.g., the hypothesis of plane cross section [96], as it has been
given by Jacob Bernoulli [4] for the straight bars (equally applied later in the case
of curved bars). If to this hypothesis is added a linear relation between strain and
stress (Hooke’s law) and the mechanical equilibrium is written for a portion of the
bar, then we shall find the known formulae that give the so-called simple stresses
for a straight bar. Assuming that the strains are small with respect to unity and can
be thus disregarded, it follows that the principle of superposition of effects can be
applied, hence the results obtained can be extended to the case of combined
stresses. Thus, computation methods valid for ordinary beams are obtained.
Let us also consider the hypothesis of the straight line element, due to Kirchhoff
[38] for thin plates and to A.-E.-H. Love [43] for thin shells, which leads to the
usual computational methods for these construction elements.
In case in which the state of strain and stress is to be determined for wall beams,
thick (or moderately thick) plates, blocks or in case of stress concentrations etc.,
more accurate computational methods should be applied. In this case, it is no
longer permitted to consider the equilibrium of a finite element (section), but that
of an infinitesimal element cut out of the body. A local study in the neighbourhood
of every point of the body is completed by a global study, referring to the whole
body. The computational method (exact, approximate and even experimental) is
general and is only restricted by difficulties which are liable to arise in practical
applications.
The methods of the mechanics of deformable solids allow one to verify the
limits of applicability of the results obtained by the methods of the strength of
materials. Chiefly, it permits one to solve problems that cannot be tackled by more
elementary methods, such as all three-dimensional problems.
Hereafter, we shall apply the methods of the mechanics of deformable solids.
4 1 Introduction
From the very start, the models can be divided into two large classes: technical (or
material) models and ideal (or imagined) models; this division is made according
to the manner of construction the models and according to the means by which the
objects under study may be reproduced.
The technical models are created by man, but they exist objectively, indepen-
dently of his awareness, being materialized in metal, wood, electromagnetic fields
etc. Their purpose is to reproduce for a cognitive goal the object under study, so as
to put in evidence its structure or some of its properties. The model can or cannot
retain the physical nature of the object under study or its geometrical likeness. If
the similitude is maintained, but the model differs in its physical nature, we have to
deal with analogic systems. For instance, electrical models can reproduce pro-
cesses analogue to those taking place in the mechanics of deformable solids, which
are qualitatively different, but are described by similar equations. These models, as
well as other of the same kind, are classed as mathematical models.
One can construct such models, e.g., to study the torsion of a cylindrical straight
bar of arbitrary simply or multiply connected cross section. If the bar is homo-
geneous, isotropic and linearly elastic (subjected to infinitesimal deformations),
then the phenomenon is governed by a Poisson type equation in B. de Saint-
Venant’s [12] theory. L. Prandtl showed that the same partial differential equation
is valid in the case of a membrane supported on a given contour and undergoing a
constant internal pressure; if the contour is similar to the frontier of the plane
domain corresponding to the cross section of the straight bar, then we have a
correspondence of the boundary conditions, hence the classical membrane analogy
(or of the soap-bubble film) is obtained. Other analogies for these problems are
used too, i.e.: electrical modelling, optical interference modelling, hydrodynamical
modelling etc.
It is interesting to remark that the above analogies can moreover be extended to
more complex phenomena. Thus, in case of a multiply connected domain (espe-
cially that of a tubular bar with intermediate thin walls), it is possible to use a
1.1 Aim of Mechanics of Deformable Solids 5
observe that this similitude must not be limited only to space relationships but can
be extended also to other aspects of the model and of the object (e.g., the character
of the motion). The intuitiveness of these models is put into evidence on one hand
by the fact that the models themselves, formed by elements sensorial perceptible
(planes, levers, tubes, fluids, vortices etc.) and on the other hand by the fact that
they are intuitive images of the objects themselves. Sometimes, these models are
fixed in the form of schemata.
The models of second order are systems of signs the elements of which are
special signs; the logical relations between them form—at the same time—a
system and are expressed by special signs. In this case, there is no likeness
between the elements of the models and the corresponding elements of the object.
These models have not intuitiveness with respect to geometrical likeness or
physical analogy; they have, by their physical nature, nothing in common with the
nature of the modelled objects. The models of second order reflect the reality on a
gnosiological plane, by virtue of their isomorphism with the reality; a one to one
correspondence is thus assumed between each element and each relation of the
model. These models reproduce the objects under study in a simplified form,
constituting thus (as, in fact, all models do) a certain idealization of the reality.
The types of ideal models mentioned above can be considered as limit cases.
Indeed, there are ideal models combining common features of both model types
which have been described; they contain systems of notions and axioms which
characterize quantitatively and qualitatively the phenomena of nature, for instance
representing mathematical models. Such models are particularly important and
their systematic use has permitted—among other things—the large development of
the mechanics of deformable solids in the last time.
The basic dialectic contradiction of modelling (the model serves to the
knowledge of the object just because it is not identical with the latter one) is
useful, e.g., to put into evidence the properties of continuous deformable solids. In
fact, a model contains the more information concerning the object, when it is more
like this one. Nevertheless, the physical reality is rather intricate; the contradiction
is solved by using a sequence of models which are more and more complete, where
each one brings new contributions to the knowledge of real deformable solids. We
shall endeavour to emphasize the very process of continuously improving methods
of the mechanics of deformable solids, a process constituting—as a matter of
fact—the main tenor of the development of this branch of mechanics.
In general, after a certain ideal model is adopted, it is absolutely necessary to
compare the results obtained by theoretical reasoning with physical reality. If these
results are not satisfactory (sometime this can occur between some limits, which
can be sufficiently narrow), then it is necessary to make corrections or to improve
the chosen model. In fact, this is the manner in which the mechanics of deformable
solids was developed, the word ‘‘model’’ being more and more used by researchers
dealing with this branch of mechanics.
1.1 Aim of Mechanics of Deformable Solids 7
Both the general (classical) mechanics and the mechanics of deformable contin-
uous media (in particular, the mechanics of deformable solids and the mechanics of
fluids) are studying the objective laws of the simplest form of motion, namely the
mechanical motion.
First of all, for the study of the mechanical motion, a representation of space
and time is necessary; thus, in classical mechanics, the physical space is the three-
dimensional Euclidean space E3 , while the time (considered as universal and
characterizing the duration, the succession and the simultaneity of the material
processes) is still assimilable to the one-dimensional Euclidean space E1 . Thus, the
geometric models of space and time, as used in general mechanics, are reflecting
properties of the real space and time as forms of existence of the matter.
A real movable body is generally thought to be rigid and is often reduced to a
particle. In the same manner, the systems of particles studied can be deformable.
For various values of the time t in E1 , we are able to find the position of the body
or of the system of particles in E3 .
Another intervening element is the cause of the mechanical motion (which, for
the sake of simplicity, will be called ‘‘motion’’). The bodies interact mechanically
with one another and in many cases it is difficult to establish the physical nature of
this action; generically, it was called force. This idea, which originates in the
action of a human organism upon the external world, acquires a precise meaning in
classical mechanics. It is the abstract expression of the measure of the transmission
of motion. Without investigating the nature of the respective force, it was math-
ematically modelled with the help of vectors (sliding vectors in case of rigid solids
and bound vectors in case of deformable systems of particles). We must mention
that a mechanical motion could exist even in the absence of any force, namely the
inertial motion (rectilinear and uniform).
The notion of mass of the particle must be introduced; this is a fundamental
property of matter, exists objectively and is independent of the place where it is
measured. I. Newton [56] conceived the mass as the measure of a quantity of
matter. The notions of gravitational mass and inertial mass should be introduced;
numerically, they are equal to each other, as it has been proved by L. Eötvös. In
this manner, we dispose of various possibilities of measuring the mass.
In the classical Newtonian conception, space, time and mass are considered as
independent from one another.
After these elements were introduced, the Newtonian model was born, by
adopting the principles of Newton [56] (the principle of inertia, the principle of
action of forces, the principle of action and reaction, the principle of the paral-
lelogram of forces and the principle of initial conditions, in a modern view, to can
put in evidence its deterministic aspect). The model was verified by direct practice
in the case of bodies moving at relatively low velocities (negligible with respect to
the velocity of light in vacuum).
8 1 Introduction
As any other discipline, the mechanics of deformable solids (and especially the
theory of elasticity) has a few fundamental hypotheses which allow us to simplify
and generalize the phenomena, retaining the essential ones and rending them
accessible to the mathematical computation. It is very important to know these
hypotheses if we want to realize the possibility of practical application of the
results thus obtained.
After reviewing such hypotheses, we shall supply some general data concerning
the development of the mechanics of deformable solids and then insist on the
development of the theory of elasticity.
1.2 Fundamental Computation Hypotheses. Short Historical Account 9
First of all, we present the bodies with which one deals in mechanics of
deformable solids, as well as the fundamental computation hypotheses in the
theory of elasticity. Then we specify the position of the theory of elasticity in the
field of the mechanics of deformable solids.
Hereafter, we shall give a classification of the solid bodies according to the ratio of
their dimensions.
Bodies with a much greater dimension (length) than the other two ones (cor-
responding to their cross section) shall be called bars. To make the notion clearer,
we shall give a constructive definition.
Let there be a curve C of finite length l, namely the bar axis. In the plane normal
to the curve C, at one of its points P, we shall consider a closed curve C bounding
a plane domain D (the bar cross section); the centre of gravity of this domain is
assumed to be on the curve at the point P. If the point P travels through the curve
C, the curve C, which can also be deformable, will generate a surface bounding a
three-dimensional domain which will be called bar (Fig. 1.1). According to the
form of the axis, the bars can be straight or curved. Moreover, the curved bars can
be classified as skew curved bars and plane curved bars.
Both the (average) dimensions a and b of the cross section shall be considered
as being of the same order, provided the condition a; b l is fulfilled. If all three
dimensions are of different order (a b l), we shall have to consider thin wall
bars (Fig. 1.2a). Lastly, if the cross section dimensions are negligible with regard
to the bar length, so that the bar should be perfectly flexible (unable to retain a
bending), then we have a string (Fig. 1.2b).
Bodies having one dimension (the thickness) much smaller than the other two-
dimension (corresponding to the middle surface) shall be called plates.
Let be a surface S of finite area, the middle surface of the plate. On the normal
to the surface S at a point P of the latter one, we shall consider a segment of a line
of length h (the thickness), the middle of which is assumed to be on the surface S,
at the point P. If the point P travels through the surface S, the extremities of the
segment of a line, which can be of variable length, will generate two surfaces
l
C P
D
b
a
10 1 Introduction
a
l
b
Fig. 1.3 Plate
hh
22
P
a
b
a
h
h
b b
(c) (d)
a
h
c
a
b
(i) The solid body we are studying (considered to be at rest with respect to an
inertial (fixed) frame of reference) is subjected to the action of balanced
external loads. If the body is in motion, then we introduce moreover the inertial
forces; therefore, the external loads are in dynamic equilibrium. On the other
hand, each part of the body and each particle detached from it will be subjected
to loads statically or dynamically balanced (load system equivalent to zero),
This hypothesis allows to write the partial differential equations verified by the
stresses within the body and to express the boundary and initial conditions.
(ii) The solid body is considered as a continuous medium (without holes, internal
microscopic cracks etc.). This encourages us to assume that the body defor-
mation will be equally continuous and therefore the strains and the stresses
will be mathematically expressed by continuous functions.
For some points of the body (singular points), where the strains and the stresses
tend to infinity, an additional study must be made; it is important to use the
methods of the theory of distributions in this case. Likewise, the case of the bodies
12 1 Introduction
(vii) When the deformations are propagated, within the solid body, at a very high
velocity (a case recently appeared in a unified theory of elementary parti-
cles), the phenomenon should be studied from the standpoint of the theory of
relativity.
(viii) The influence of the temperature variation is not taken into consideration
neither in the constitutive laws nor in the computation of the deformations.
But it will be introduced in the considerations concerning thermodynamical
principles.
(ix) The body under study does not have initial stresses, which could be due to
initial deformations of the material. They could result from machinery
operations (rolling, drawing etc.), from assembling operations or could be
due to phenomena occurring during the working, before the action of the
external loads, contraction phenomena of concrete etc.
Although these phenomena do exist and cannot be eliminated, to simplify the
computation, one assumes that the initial stresses are missing. Thus, to an
unloaded body (case of vanishing external loads) corresponds a null state of stress.
If the initial stresses cannot be neglected, then one makes a supplementary study to
take into account their influence.
We will assume—in general—that all these hypotheses are taken into account;
if, in particular cases, one of these hypotheses is not respected and if we are placed
in a more general situation (a less restrictive hypothesis), then we make a special
mention about it.
the theory of elasticity has as object of study the determination of the state of strain
and stress of an elastic solid body subjected to the action of certain external loads
in static or dynamic equilibrium. To the fundamental equations of the mechanics
of deformable solids one must add a constitutive law, which characterizes the
respective elastic body.
We mention that, the most times, a solid body may be considered as being
elastic only till a certain intensity of the stresses or till a certain magnitude of the
strains (till the limit of elasticity); after this, it becomes elastic plastic properties.
Thus, our study will be valid for the so-called elastic zone of the respective body.
Certainly, since most ancient times, since men began to build, some problems of
mechanics arose—in general—and of strength of materials—in particular; various
problems about deformable solid bodies (although this property was not quite clear
in their minds) were imposed by daily practice.
We shall supply some general data concerning the development of the
mechanics of deformable solids and then we shall insist on the development of the
theory of elasticity. We shall mention also the researches made in Romania.
As well, we will put in evidence some development trends and the new
problems which are put.
force. According to Galileo, in the built-in cross section should appear normal
stresses uniformly distributed on the whole cross section, which later on proved to
be only a rather vague approximation of the physical reality. Architect Fontana
(1543–1607) applied these results to hoist an obelisk at the Vatican.
Robert Hooke (1635–1703) [32], on the basis of various experiments, stated in
1678 his famous law ‘‘ut tensio sic vis’’ (such is the force, as the extension is).
Edmé Mariotte (1620–1684) verified this law on wood tensile test samples.
Valuable ideas about the notion of elasticity were set forth by Mikhail Vasilievich
Lomonosov (1711–1765).
In the seventeenth and eighteenth centuries, progress was realized, especially in
the field of the mechanics of structures and of the strength of materials. Many of
the acquired results have applications corresponding to the new problems met
within civil engineering and machine construction. We mention moreover some
theoretical results, obtained by Jacob (1654–1705) [4] and Jean (1667–1748)
Bernoulli brothers (to the former is due the hypothesis of the plane cross sections
in straight bars). The latter’s son Daniel Bernoulli (1700–1782), Daniel’s nephew
Jacques Bernoulli (1759–1789) and chiefly Leonhard Euler (1707–1783) and
Joseph-Louis Lagrange (1736–1813) dealt with problems of deformation of thin
elastic bars.
Other theoretical and experimental studies are due to Charles-Augustin de
Coulomb (1736–1806), Jean-Victor Poncelet (1788–1867) and Franz Grashof
(1826–1893). Jean le Rond d’Alembert (1717–1783) and Peter Gustav Lejeune-
Dirichlet (1803–1859) contributed by their studies in the field of mechanics to the
birth of the theory of elasticity.
The first studies about sliding lines were made in the soil mechanics by Ch.-A.
de Coulomb, William-John Macquorn Rankine (1820–1872) and Maurice Lévy
(1839–1910). Their importance was grasped by Henri-Édouard Tresca
(1814–1885) who, in 1864 and 1872, published several notes, concerning a con-
dition of plasticity in which the maximal shear stress is constant at each point of
the elastic-plastic zone, and by Christian Otto Mohr (1835–1918) who, on this
basis, formulated in 1882 a theory of the strength of materials. Adhémar-Jean-
Claude Barré de Saint-Venant (1797–1886) reviewed H.-É. Tresca’s works at the
Academy of Science (Paris) and, on this occasion, developed the fundamental
equations of the plasticity theory, by admitting that the cubical dilatation vanishes
during the plastic deformation (which is thus incompressible), that the principal
directions of the tensor of the rate of strain coincide with the principal directions of
the stress tensor and that the maximal shear stress is constant at each point of the
solid body; and we can assert that it is with this development that the theory of
plasticity was born in 1870.
And, still in the second half of the ninteenth century, the problem of the study of
the solid bodies for which the deformations rate has been taken into account began
to be formulated. Thus, James Clerk Maxwell (1831–1879), William Thomson
(Lord Kelvin) (1842–1907) and Woldemar Voigt (1850–1919) define particular
viscoelastic bodies (bodies of Maxwell or of Kelvin (or Voigt) type; the last two
arrived at analogous results by independent studies), corresponding to certain real
1.2 Fundamental Computation Hypotheses. Short Historical Account 17
bodies. Nevertheless, from a formal standpoint, the rheology was born only at an
international congress in 1929 (it is on this occasion that the name ‘‘rheology’’
(from ‘‘panta rei’’) was introduced); especially, the theory of viscoelasticity has
been developed.
Thus, the bases are set of a new branch of mechanics that deals with the study
of deformable solids.
The bases of the mathematical theory of elasticity were set forth, at the beginning
of nineteenth century, by the works of Louis-Marie-Henri Navier (1785–1836)
[54] and Mikhail Vasilievich Ostrogradskı (1801–1861); but it was Augustin-
Louis Cauchy (1789–1857) who, making use of these results, established the
fundamental equations of elastic bodies (the equations of static and dynamic
equilibrium and the relations between displacements and strains, valid for any
deformable solid bodies) in the form still used today. Siméon-Denis Poisson
(1781–1841) contributed to elucidate these problems, by introducing also the
notion of transverse contraction, while Benoît-Paul-Émile Clapeyron (1799–1869)
establishes a theorem of equivalence between the internal work (of deformation)
and the external work. The formulation in displacements of the fundamental
problems of the theory of elasticity as well as an important treatise on the theory of
elasticity, where is equally formulated the famous problem of the elastic paral-
lelepipedon, which only in the last years obtained solution, are chiefly due to
Gabriel Lamé (1795–1870) [39]; on the other hand, starting from the studies of
Karl Friedrich Gauss (1777–1855), he introduced [40] the systematic use of cur-
vilinear co-ordinates in solving problems of the theory of elasticity.
During the nineteenth century and the beginning of twentieth century, many
researches in the domain of the theory of elasticity have been performed, results
which are very difficult to pass, even summarily, in review.
At the same time, the strength of materials acquires equally a large develop-
ment concerning the study of plates related to the design of railway bridges. As to
the problem of plates, we may recall, besides the name of J.-L. Lagrange, those of
Sophie Germain (1776–1831) and Gustav Robert Kirchhoff (1824–1887) [36],
who continued—theoretically—the experimental research begun by Ernst
Friederich Chladni (1756–1827).
A controvercy (which lasted several decades) about the elastic constants of the
material confronted the supporters of the conception according to which a single
elastic constant is sufficient to characterize the mechanical properties of the material
(L.-M.-H. Navier [54], A.-L. Cauchy and their pupils) against those who held that
two elastic constants are necessary to achieve this end. George Green (1793–1841),
starting from considerations of an energetic nature, showed that, in the general case
of anisotropy, 21 constants are necessary, which in case of isotropic bodies are
reduced to two distinct constants. Adolf Yakovlevich Kupffer (1799–1865), at the
Central Laboratory of Weights and Measures in Sankt Petersburg, Wilhelm
18 1 Introduction
(a third volume was published later only by L. Föppl), these works played an
important part, revealing to very large, especially engineering, circles the realiza-
tions of the theory of elasticity; in their time, the first works were translated into
French and Russian.
The pupils of A.-J.-C. B. de Saint-Venant successfully continued the studies he
had begun. The most important among pupils was Joseph Valentin Boussinesq
(1842–1929); to him is equally due the application of the theory of the potential to
both the static and the dynamic problems of elasticity. As an example,
J. V. Boussinesq [7] establishes the state of stress within the elastic half-space,
undergoing the action of a concentrated normal force or of a distributed load on the
separation plane. Alfred Aimé Flamant (1839–1914) studied the corresponding
two-dimensional problem (the elastic half-plane case). J. H. Michell applied these
results to the study of a whole series of plane problems of the elasticity theory in
polar co-ordinates, chiefly to the influence of the concentrated internal forces.
Another pupil of B. de Saint-Venant, Maurice Lévy (1838–1910), dealt with the
plane problem of the theory of elasticity, especially in the case of an elastic wedge.
Both to him and to J. H. Michell, an important theorem is due, according to which
the state of stress, in the plane case, in a simply connected domain, being given the
loads on the boundary and in the absence of the volume forces, does not depend on
the elastic constants of the material.
Lord Rayleigh brought important contributions to the theory of vibrations
(elastodynamics), his most important results being contained in a treatise on the
theory of sound [70]. The computation method pointed out by him and resumed
later by Walter Ritz (1878–1909) is known today as the Rayleigh-Ritz variational
method. Another variational method, often used in the computation practice, was
later elaborated by Erich Immanuel Trefftz (1888–1937). Starting from the results
supplied by Lord Rayleigh, Horace Lamb (1849–1934) and Augustus-Edward-
Hough Love (1863–1940) entered upon various studies concerning the vibrations
of plates. To the latter is also due an important treatise on the theory of elasticity,
which was published in several ulterior editions and played an important rôle in
the scientific information of several generations and is still successfully used [43].
In the same period, one of the most developed histories of the theory of elas-
ticity until the present time was published by Isaac Todhunter (1820–1884) and
Karl Pearson (1857–1936) [78].
W. Voigt was the first to introduce the notion of tensor in the theory of elas-
ticity. Hermann Ludwig Ferdinand von Helmholtz (1821–1894) was known by his
studies in the domain of kinematics of motion. Heinrich Rudolf Hertz (1857–1894)
researched in the field of hardness of materials and of local stresses (contact
problems).
In the plane problems of the theory of elasticity, a distinction is to be made
between the state of plane stress (short cylinder problem) and the state of plane
strain (long cylinder problem, theoretically of infinite length). To the studies of
G.-B. Airy, J. C. Maxwell, and M. Lévy and A. A. Flamant mentioned earlier,
other studies followed, which raised a particular interest both from theoretical and
practical standpoint. Let us discuss the research of Aksel Wihelmovich Gadolin
1.2 Fundamental Computation Hypotheses. Short Historical Account 21
His treatise [60] on the theory of elasticity was a particularly valuable book for
many generations of Russian elasticians.
In the climate impressed at Göttingen by Felix Klein (1849–1925) and David
Hilbert (1862–1943), many researchers asserted themselves into the field of
mechanics of deformable solids. We mention Ludwig Prandtl (1875–1953) who, in
the field of the theory of elasticity, was especially renowned for his membrane
analogy in the torsion problem of a straight bar. Another remarkable representative
of this school was Theodor von Kárman (1881–1963).
In the theory of elasticity, problems of multiply connected bodies were studied
and established on a mathematical basis—introducing the distorsions—by Vito
Volterra (1860–1940), whose results dating from the beginning of twentieth
century were gathered up in a volume, published subsequently by his son Enrico
Volterra [79].
In the domain of thin wall bars and of thin shells, interesting computational
methods were proposed by Vasilii Zakharovich Vlasov (1906–1958).
The application of the methods of integral equations to the problems of the
theory of elasticity is due to Giuseppe Lauricella (1869–1913). The theories of
strength of materials marked a new upsurge by the work of E. Beltrami, Maksy-
milian Tytus Huber (1872–1950) [34], Heinrich Hencky (1885–1951) and Richard
von Mises (1883–1953). The spatial representation of the principal stresses, due to
Harald Malcolm Westergaard (1888–1950), is much used.
The plane problem of the theory of elasticity acquires a remarkable develop-
ment after the studies of Charles Henri Ribière and Louis-Napoleon George Filon,
who use harmonic functions in the form of Fourier series or Fourier integrals,
obtained by means of certain terms which are product of trigonometric lines and
hyperbolic or exponential lines. Other similar solutions were given at the begin-
ning of twentieth century in the doctoral thesis of Aloys Timpe. These ideas have
acquired a modern form among the operational computation methods. Jacques-
Salomon Hadamard (1865–1963) gave a general method of solution of the two-
dimensional biharmonic problem (the mathematical aspect of the plane problem of
the elasticity theory). G. V. Kolosov proposed another formulation of this problem
with the help of the complex variable functions; this method was subsequently
developed at length by Nikolai Ivanovich Muskhelishvili (1891–1976) and his
pupils, and constitutes one of the finest realizations of the Russian school of
elasticity theory. The above-mentioned results were expounded at length by
N. I. Muskhelishvili [51] in his treatise translated into several languages.
Among the treatises on the theory of elasticity which played an important part
in its development, we mention those of Henri Poincaré (1854–1912) [63] and of
Roberto Marcolongo (1862–1943) [46, 47]; Eugène-Maurice-Pierre Cosserat
(1864–1931) and his brother François Cosserat [13] published a book an the theory
of asymmetric elasticity (a particle of the body has six degrees of freedom), at the
beginning of twentieth century. From 1914, when the first edition appeared in
Russian, up to now, the treatises of Stepan Prokofievich Timoshenko (1878–1972)
had a great influence on the development of many generations; they were trans-
lated into several languages. We particularly mention his book on the theory of
1.2 Fundamental Computation Hypotheses. Short Historical Account 23
elasticity, the last edition of which was published together with J. N. Goodier [77].
His history on the strength of materials, containing also a history of the theory of
elasticity, is equally interesting [75]. In Russia, we point out the treatises of
Mikhail Mitrofanovich Filonenko-Borodich (1885–1962) and of Leonid Samu-
ilovich Leıbenzon (1879–1951) [42]. In Poland, the treatise of M. T. Huber [34] on
the theory of elasticity was wide spread. Among the treatises with a general
character, including also volumes devoted to the theory of elasticity, we mention
the treatise on general mechanics of Paul-Émile Appell (1855–1939) [1], the
cycles of volumes on theoretical physics of Arnold Sommerfeld (1868–1951) [68]
and of Lev Davidovich Landau and Evgenii M. Lifshitz [41].
Today, the theory of elasticity is well developed in multiple directions
including new computational methods, both analytical and numerical; more and
more disciplines are joined with the theory of elasticity in research of a common
character. For instance, to the theory of thermoelasticity, dating from nineteenth
century, are added magnetoelasticity, viscoelasticity, the theory of finite elastic
deformations or magneto-thermoelasticity. Because of this great development, it is
very difficult to mention, even briefly, the most famous men of science who dealt
or deal with these problems. That is why we preferred to mention only the research
which has become classical and to present only the research made until the first
decades of twentieth century.
At its beginnings, the research into the mechanics of deformable solids had a
chiefly technical and practical character; engineers of wide experience both in
design and execution, professors at the Polytechnical School in Bucharest, such as
Elie Radu (1853–1931), Anghel Saligny (1854–1937), Ion Ionescu (1870–1946),
Gheorghe Emanoil Filipescu (1882–1937) or such as Nicolae Profiri (1886–1967),
Aurel A. Belesß (1891–1976), Constantin C. Teodorescu (1892–1972), Cristea
Mateescu (1894–1979) and Mihail Hangan (1897–1964) studied problems of
calculus of structures by method of strength of materials and of statics of structure.
The research on this line continued successfully for the last years, supplying
interesting results in mechanics of structures (statics, stability and dynamics), in
machine construction mechanics etc.
As to the mathematical study of deformable solids, the first research dating
from the beginning of twentieth century, had—chiefly—a characteristic of equa-
tions of mathematical physics due to a few professors of the faculties of science in
the Bucharest and Jassy Universities, but who obtained their doctor titles in France
or in Germany. For instance, the first work in this domain was the doctoral thesis
of Anton Davidoglu (1876–1958) [84, 85], defended at Paris in 1900, concerning
the application of the successive approximations method of Émile Picard
(1856–1941) to the study of some differential equations of fourth order, corre-
sponding to the transverse vibrations of nonhomogeneous elastic bars.
24 1 Introduction
In connection with the same ideas, we bring into relief the doctoral thesis of
Alexandru Myller (1879–1965) [53], defended at Göttingen in 1906; it treats
ordinary differential equations in connection with integral equations and it offers
examples concerning the elastic bar equations, in various supporting cases.
Besides, Simion Sanielevici (1870–1963) [65, 94], defended at Paris, in 1908, a
doctoral thesis on the differential equations of the vibrating string and membranes.
Other mathematical researches into the deformable solids appear beginning with
1929, but are rather sporadic until 1948. In this time interval, the doctoral thesis of
Grigore C. Moisil (1906–1973) [49, 87, 88] appeared, concerning the application of
the non-linear functional analysis to the study of dynamics of strings; we mention
also his work in which some dynamic problems of the theory of elasticity are
studied [50, 89, 90]. On the other hand, we must emphasize the doctoral thesis of
Nicolae Teodorescu (1908–2000), as well as his work concerning the application
of the areal derivative and of the generalized potentials to problems of mechanics of
deformable continuous media [71, 95]. We mention also a paper on the vibrations
of a rectangular membrane published by Mircea Drăganu (1911–1984) [86].
Since this period, the problems of the theory of elasticity were commenced
within the framework of the study of systems of partial differential equations. First
of all, to be able to study the general properties of the systems of partial differential
equations with constant coefficients, the method of the associated matrices to these
systems off equations, elaborated by Gr. C. Moisil [50], was used; by so doing,
Gr. C. Moisil initiated, in 1949, at Bucharest, at the Institute of Mathematics of the
Romanian Academy, the first systematic research into the mathematical study of
the mechanics of deformable solids, thus creating—by the specialists formed
there—a true school in this field. These researches were developed later particu-
larly within the frame of the Mathematics and Mechanics Faculties of Bucharest
and Jassy Universities of the Bucharest and Jassy Mathematics Institutes and of the
Solid Mechanics Institute of the Romanian Academy, as well as—nowadays—in
many Technical Universities and in various research institutes. As in the previous
subsection, we do not go into details as to the present researches in this domain,
but we confine ourselves to the information mentioned above.
Hereafter we shall try to bring into relief, quite succinctly, some of the develop-
ment trends of the theory of elasticity.
We mention, in the first plane, the study of other types of bodies (non-classi-
cal), offering also—in general—an elastic character, just as the study of certain
nonclassical problems of the theory of elasticity.
Thus, most of the studies made on isotropic bodies were extended to aniso-
tropic ones, in general, and to various particular cases of anisotropy. In most of the
cases, these generalizations do not present essential difficulties; obviously, the
form of the results is more intricate, since they depend on the 21 elastic constants
involved. In the static case, the harmonic or the biharmonic equations governing
1.2 Fundamental Computation Hypotheses. Short Historical Account 25
In connection with these bodies, we must equally recall the bodies with
microstructure. It is to be remarked that the structure of certain constructions can
be approximated with the help of such bodies, from the standpoint of static
computation.
Beside the above mentioned physical non-linearity, we deem it equally
important to emphasize the geometric non-linearity, due to nonlinear relations
between strains and displacements. Tensor calculus methods play an increasingly
important part, where the use of curvilinear co-ordinates is essential. Owing to this
non-linearity, yet another non-linearity of a mechanical order appears, concerning
the equations of equilibrium and motion.
Because of the great computational difficulties, various approximations are
used, as a function of the nature and form of the body as well as of the problems
arising. We must distinguish, for instance, between finite and infinitesimal
deformations and finite and infinitesimal rotations; on the other hand, any of the
combinations liable to be made is possible. The classical case is that of infini-
tesimal deformations and of infinitesimal rotations, the last hypothesis being
essential for a correct computation.
Let us remark that the non-linear constitutive laws can also occur by considering
an elastic potential depending not only on the first-order gradient of the displace-
ment (like in the classical case), but on higher-order gradients too; if we confine
ourselves to second-order gradients, then we deal with bodies of second order
(bipolar bodies); more general studies lead to multipolar bodies. We mention also
studies of the elastic potential of these bodies from the dynamical standpoint. The
case of bodies with initial stresses has been brought into relief as well.
Let us mention that the bodies of Cosserat type cannot be bodies of second
order since we have to do with two distinct directions of generalization. The
identity only occurs in the case of bodies with constraint rotations.
In connection with the various generalization trends of the theory of elasticity
and with the introduction of its specific methods and forms in the more embracing
frame of the mechanics of deformable solids, we must mention studies about
dislocations (linear defects in crystals).
The problem possesses a particular importance for the physics of solids; its
theoretical aspects are increasingly of interest in mechanics too. Indeed, in the
frame of the so-called continuous theory of dislocations we endeavour to present,
in a unitary form, the theory of elasticity, the theory of plasticity and the theory of
creep. In Japan, RAAG (Research Association of Applied Geometry) dealt, under
the guidance of Kazuo Kondo, with the unification of various engineering theories,
through the agency of geometry; one of its objectives was mentioned above.
Another current line of research is that of its connection with the thermal,
magnetic, electric etc. fields. Thus, were born thermoelasticity, magnetoelasticity,
electroelasticity etc. On the other hand, there are cases when all these effects have
to be considered simultaneously.
It is to be remarked, for instance, that the thermal effect influences the deformation
of the body, but it is reciprocal; heat propagation can be influenced by the defor-
mation of the body, while Fourier’s equation, e.g., must be completed accordingly.
1.2 Fundamental Computation Hypotheses. Short Historical Account 27
References
A. Books
1. Appell, P.: Traité de mécanique rationnelle, vol. IV. Gauthier-Villars, Paris (1926)
2. Bach, C.: Elasticität und Festigkeit. Die für die Technik wichtigsten Sätze und deren
erfahrungmässige Grundlage, 2nd edn. J. Springer, Berlin (1894)
3. Belluzi, O.: Scienza delle costruzioni, vol. III. N. Zanichelli, Bologna (1956)
4. Bernoulli, J., Œuvres choisies, vol. II. Genève (1744)
5. Bezukhov, N.I.: Teoriya uprugosti (Theory of Elasticity). Gostekhizdat, Moskva (1953)
6. Biezeno, C.B., Grammel R.: Technische Dynamik, vols. I, II. 2nd edn. Berlin (1953)
7. Boussinesq, J.: Applications des potentiels à l’étude de l’équilibre et du mouvement des
solides élastiques, New ed., Al. Blanchard, Paris (1969)
8. Brdička, M.: Mechanika kontinua (Mechanics of Continuum). Nakladat. Českoplov. Akad.
Věd, Praha (1959)
9. Butty, H.: Tratado de Elasticidad TeoricoTehnica, vol. I. Buenos Aires (1946)
10. Castigliano, A.: Théorie de l’équilibre des systèmes élastiques. Fr. Bocca. Ed., Torino (1879)
11. Cesàro, E.: Introdutione alla teoria matematica della Elasticità. Fr. Bocca Ed., Torino (1894)
30 1 Introduction
12. Clebsch, A.: Théorie de l’élasticité des corps solides (traduite par MM. B. de Saint-Venant et
Flamant avec de notes étendues de M. de Saint-Venant). Paris (1883)
13. Cosserat, E., Cosserat, F.: Théorie des corps déformables. E. Hermaun et fils, Paris (1909)
14. Eringen, A.C.: Nonlinear Theory of Continuous Media. McGraw-Hill Book Co., Inc., New
York (1962)
15. Fl}uge, S. (ed.): Handbuch der Physik. VI. Elasticity and Plasticity. Springer, Berlin (1958)
16. F}oppl, A.: Vorlesungen } uber technische Mechanik. III. Festigkeitslehre. B. G. Teubner
Verlag, Leipzig (1898)
17. F}oppl, A.: Vorlesungen } uber technische Mechanik. V. Die wichtigsten Lehren der h} oheren
Elastizitätstheorie. B. G. Teubner Verlag, Leipzig (1907)
18. F}oppl, A. and L., Drang und Zwang. Eine h} ohere Festigkeitslehre f}ur Ingenieure. I, II.
R. Oldenbourg Verlag, München (1920)
19. Franciosi, V.: Scienza delle costruzioni, vol. I. Liguori, Napoli (1965)
20. Fung, V.C.: Foundations of Solid Mechanics. Prentice Hall, Inc., Englewood Cliffs (1965)
21. Galerkin, B.G.: Sobronie sochineniı (Complete Works), vols. I, II. Izd. Akad. Nauk SSSR,
Moskva (1952–1953)
22. Germain, P.: Mécanique des milieux continus. Masson Éd., Paris (1962)
23. Gol’denblat, I.I.: Nelineınye problemy teoriı uprugosti (Non-Linear Problems of the Theory
of Elasticity). Izd. ‘‘Nauka’’, Moskva (1969)
24. Goodier, J.N., Hodge Jr, P.G.: Elasticity and Plasticity. Wiley, New York (1958)
25. Green, A.E., Adkins, J.E.: Large Elastic Deformations and Non Linear Continuum
Mechanics. Clarendon Press, Oxford (1960)
26. Green, A.E., Zerna, W.: Theoretical Elasticity. Clarendon Press, Oxford (1954)
27. Grioli, G.: Mathematical Theory of Elastic Equilibrium (Recent Result). Springer, Berlin
(1962)
28. Gurtin, M.E.: The Linear Theory of Elasticity (in Handbuch der Physik. VIa/2). Springer,
New York (1972)
29. Haimovici, M.: Teoria elasticitătßii (Theory of Elasticity). Ed. did. ped, Bucuresßti (1969)
30. Hermite, R.L’.: Résistance des matériaux, théorique et expérimentale. I. Dunod, Paris (1954)
31. Hertz, H.: Gesammelte Werke, vol. I. Leipzig (1895)
32. Hooke, R.: Lectures de Potentia Restitutive or of Springs Explaining the Power of Springing
Bodies. John Martin, London (1678)
33. Howink, R.: Elastizität, Plastizität und Struktur der Materie. Dresden-Leipzig (1957)
34. Huber, M.T.: Teoria spre_zistości (Theory of Elasticity). I, II. Państ. Wydawn. Naukove,
Warszawa (1954)
35. Kecs, W., Teodorescu, P.P.: Applications of the Theory of Distributions in Mechanics. Ed.
Acad., Bucuresßti, Abacus Press, Tunbridge Wells (1974)
36. Kirchhoff, G.R.: Gesammelte Abhandlungen. Leipzig (1882)
37. Knops, R.J., Payne, L.E.: Uniqueness Theorems in Linear Elasticity. Springer, New York
(1971)
38. Kr}oner, E.: Kontinuumstheorie der Versetzungen und Eigenspannungen. Springer, Berlin
(1958)
39. Lamé, G.: Leçons sur la théorie mathématique de l’élasticité des corps solides. Paris (1852)
40. Lamé, G.: Leçons sur les coordonnées curvilignes. Paris (1859)
41. Landau, L., Lifchitz, E.: Théorie de l’élasticité. Ed. Mir, Moskva (1967)
42. Leıbezon, L.S.: Kurs teoriı uprugosti (Course of Theory of Elasticity). Ogiz, Moskva-
Leningrad (1947)
43. Love, A.E.H.: A Treatise of the Mathematical Theory of Elasticity, 4th edn. Cambridge
University Press, Cambridge (1934)
44. Lurje, A.I.: Räumliche Probleme der Elastizitätstheorie. Akad Verlag, Berlin (1963)
45. Lur’e, A.I.: Teoríya uprugosti (Theory of Elasticity). Izd. ‘‘Nauka’’, Moskva (1970)
46. Marcolongo, R.: Teoria matematica della elasticità. Milano (1902)
47. Marcolongo, R.: Teoria matematica dell’equilibrio dei corpi elastici. Milano (1904)
48. Mohr, O.: Technische Mechanik, 2nd edn. Verlag von W. Ernst & Sohn, Berlin (1914)
References 31
49. Moisil, Gr.C.: La mécanique analytique des systèmes continus. Thèse, Gauthier-Villars, Paris
(1929)
50. Moisil, Gr.C.: Matricele asociate sistemelor de ecuatßii cu derivate partßiale. Introducere în
studiul cercetărilor lui I.N.Lopatinschi (Matrices associated to systems of partial differential
equations. Introduction to the study of Lopatinski’s researches) Ed. Academiei, Bucuresßti
(1950)
51. Muskhelishvili, N.I.: Some Basic Problems of the Mathematical Theory of Elasticity.
P. Noordhoff Ltd., Groningen (1963)
52. M}uller, W.: Theorie der elastichen Verformung. Akad. Verlagsgesellschaft Geest & Portig
K.G, Leipzig (1959)
53. Myller, Al.: Gew}ohnliche Differenzialgleichungen h} ocherer Ordnung in ihrer Beziehung zu
den Integralgleichungen. Inaugural Dissertation zur Erlangung der Doktorw} urde, G}ottingen
(1906)
54. Navier, L.M.H.: Résumé des leçons sur l’application de la mécanique a l’établissement des
constructions et des machines. 3me (édn.) avec de notes et des appendices de B. de Saint-
Venant. Paris (1864)
55. Neumann, F.: Vorlesungen } uber die Theorie der Elasticität der festen Körper und des
Lichtäthers. Leipzig (1885)
56. Newton, I.: Philosophiae naturalis principia mathematica. S. Pepys, Reg. Soc. Praeses,
Londini (1687)
57. Novozhilov, V.V.: Osnovy nelineınoı teoriı uprugosti (Fundamentals of the non-linear theory
of elasticity). Gostekhizdat, Moskva-Leningrad (1948)
58. Novozhilov, V.V.: Teoriya uprugosti (Theory of Elasticity). Sudpromgiz, Moskva (1958)
59. Nowacki, W.: Teoria sprepystości (Theory of Elasticity). Państ. Wydawn. Naukowe,
Warszawa (1970)
60. Papkovich, P.F.: Teoriya uprugosti (Theory of Elasticity). Gostekhizdat, Moskva (1939)
61. Pearson, C.E.: Theoretical Elasticity. Harvard University Press, Cambridge (1959)
62. Planck, M.: Mechanik deformierbarer K} orper, 3rd edn. S. Hirzel, Leipzig (1931)
63. Poincaré, H.: Leçons sur la théorie de l’élasticité. Paris (1892)
64. Prager, W.: Introduction to Mechanics of Continua. Ginn & Co., Boston (1961)
65. Sanielevici, S.: Sur les équations différentielles des cordes et des membranes vibrantes.
Thèse, Gauthier-Villars, Paris (1908)
66. Sokolnikoff, I.S.: Mathematical Theory of Elasticity, 2nd edn. McGraw-Hill Book, Co., Inc.,
New York (1956)
67. Sokolnikoff, I.S.: Tensor Analysis. Theory and Applications to Geometry and Mechanics of
Continua, 2nd edn. Wiley, New York (1964)
68. Sommerfeld, A.: Mechanik der deformierbaren Medien, 6th edn. Akad. Verlags-gesellschaft
Geest & Portig K.-G., Leipzig (1949)
69. Southwell, R.V.: An Introduction to the Theory of Elasticity for Engineers and Physicists,
2nd edn. Oxford University Press, London (1953)
70. Strutt, J.W. (Lord Rayleigh): The Theory of Sound. London (1877)
71. Teodorescu, N.: La dérivée aréolaire et des applications à la physique mathématique. Thèse,
Gauthier-Villars, Paris (1931)
72. Teodorescu, P.P.: Probleme spatßiale în teoria elasticitătßii (Space problems in the theory of
elasticity). Ed. Academiei, Bucuresßti (1970)
73. Teodorescu, P.P.: Dynamics of Linear Elastic. Bodies. Ed. Acad. Bucuresßti, Abacus Press,
Tunbridge Wells, Kent (1975)
74. Thomson, W., Tait, P.G.: Treatise on Natural Phylosophy, vol. I. Cambridge University
Press, Cambridge (1861)
75. Timoshenko, S.P.: History of Strength of Materials. McGraw-Hill Book Co., Inc., New York
(1953)
76. Timoshenko, S.P.: Collected Papers. McGraw-Hill Book Co., Inc., New York (1953)
77. Timoshenko, S.P., Goodier, J.N.: Theory of Elasticity, 2nd edn. McGraw-Hill Book Co., Inc.,
New York (1951)
32 1 Introduction
78. Todhunter, I., Pearson, K.: History of the theory of elasticity, vols. I–III. Cambridge
(1886–1893)
79. Volterra, V., Volterra, R.: Sur les distorsions des corps élastiques (théorie et applications).
Mém. Sci. Math., CXLVII. Gauthier-Villars, Paris (1960)
80. Wang, C.C., Truesdell, C.: Introduction to Rational Elasticity. Noordhoff International
Publishing, Leyden (1973)
81. Wang, Chi.Teh.: Applied Elasticity. McGraw-Hill Publishing Company, New York (1953)
82. Westergaard, H.M.: Theory of Elasticity and Plasticity. Harvard University Press, Cambridge
(1952)
83. Winkler, E.: Die Lehre von der Elastizität und Plastizität. Praha (1867)
B. Papers
84. Davidoglu, A.: Sur l’équation des vibrations transversales des verges élastiques. Thèse. Ann.
Sci. de l’École norm. sup. Paris, 3rd ser., 17, 359 (1900)
85. Davidoglu, A.: Sur une application de la méthode des approximations successives. C. Rend.
hebd. des séauces de l’Acad. Sci. 130, 692; 1341 (1900)
86. Drăganu, M.: Sur une résolution de l’équation différentielle des mouvements vibratoires
d’une membrane rectangulaire par la méthode de la transformation multiple de Laplace.
Mathematica 22, 206 (1946)
87. Moisil, Gr.C.: Sur la dynamique du fil. Bull. Math. de la Soc. Roum. Sci., 31, 2; 170 (1929)
88. Moisil, Gr.C.: Sur le mouvement d’un fil sur une surface. Mathematica 3, 144 (1930)
89. Moisil, Gr.C.: Asupra sistemelor de ecuatßii cu derivate partßiale liniare cu coeficientßi
constantßi (On the systems of linear partial differential equations with constant coefficients).
Bul. Sßt. Acad., ser A, 1, 4; 341 (1949)
90. Moisil, Gr.C.: Teoria preliminară a sistemelor de ecuatßii cu derivate partßiale liniare cu
coeficientßi constantßi (Preliminary theory of systems of linear partial differential equations
with constant coefficients). Bul. Sßt. Acad., S
ßt. Mat.-Fiz., 4, 2; 319 (1952)
91. Myller, Al.: Asupra ecuatßiilor coardelor vibrante (On the equations of vibrating strings). Gaz.
Mat. 14, 6 (1909)
92. Oravas, G.Æ., Lean, L.Mc: Historical development of energetical principles in
elastomechanics. I. From Heraclites to Maxwell. Appl. Mech. Rev. 19, 8; 647 (1966)
93. Oravas, G.Æ., Lean, L.Mc: Historical development of energetical principles in
elastomechanics. II. From Cotterill to Prange. Appl. Mech. Rev. 19, 11; 919 (1966)
94. Sanielevici, S.: Sur l’équation aux dérivées partielles des membranes vibrantes. C. Rend.
hebd. des séances de l’Acad. Sci. 146, 1249; 1387 (1908)
95. Teodorescu, N.: Sur une formulation généralisant l’intégrale de Cauchy et sur les équations
de l’élasticité plane. C. Rend hebd. des séances de l’Acad. Sci. 189, 565 (1930)
96. Teodorescu, P.P.: Asupra ipotezei sectßiunilor plane în rezistentßa materialelor (On the
hypothesis of the plane cross-sections in strength of materials). An. Univ. Bucuresßti, ser. ßst.
nat. 6, 41 (1957)
97. Teodorescu, P.P.: One hundred years of investigations of the plane problem of the theory of
elasticity. Appl. Mech. Rev. 17, 175 (1964) (also in ‘‘Applied Mechanics Surveys’’, Spartan
Books, Washigton, 245 (1966))
98. Teodorescu, P.P.: Entwicklungstendenzen in der Elastizitätstheorie. Mitteilungen der Math.
Gesellschaft der D. D. R. 1, 5 (1970)
}
99. Wieghardt, K.: Uber ein neues Verfahren, verwickelte Spannungsverteilungen in elastischen
K}orpern auf experimentellen Wege zu finden. Mitt. } uber Forsch. Geb. Ing. 49, 15 (1908)
Chapter 2
Geometry and Kinematics of Deformation
To study the deformation of a body, we start from the general case of finite
deformations. After some general considerations, we pass to the geometric aspects
of the deformation, using material co-ordinates, as well as space ones.
!
The displacement MM ¼ u shall be a vector the components of which will be
denoted by u1 ; u2 ; u3 . We can write
xi ¼ xi þ ui ; i ¼ 1; 2; 3: ð2:1Þ
If the transition from the initial state to the actual (final) state does not occur
instantaneously (in theory) or in a very short time (in practice), as it is considered in
general, in the static case, then the phenomenon assumes a dynamic character and
the study of the deformation as a function of time becomes necessary. We will
consider to be in the general case; in the static case, the time variable will disappear.
If the body occupies the domain D before deformation (at the initial moment
t ¼ t0 ) and the domain D after deformation (at a moment t), we shall admit that
the transition from D to D is continuous, so that the equations characterizing the
deformation will take the form (Fig. 2.1b)
xi ¼ xi ðx1 ; x2 ; x3 ; tÞ; i ¼ 1; 2; 3; ð2:2Þ
corresponding to a geometric mapping. We shall also admit that the inverses of
these functions
xj ¼ xj ðx1 ; x2 ; x3 ; tÞ; j ¼ 1; 2; 3; ð2:20 Þ
in this domain. This hypothesis is known under the name of axiom of continuity,
expressing the indestructibility of matter. No domain, to which corresponds a finite
positive volume, can be deformed into a domain of zero or infinite volume. This
implies moreover the impenetrability of matter. The motion (2.2) (or (2.20 ))
transforms any domain into another domain, any surface into another surface and
any curve into another curve.
2.1 Finite Deformations 35
By taking into account the supporting condition of the solid body (at the sup-
porting points, lines or surfaces, the displacements are either zero or have imposed
values), we eliminate the rigid-body motion and the three functions
ui ¼ ui ðx1 ; x2 ; x3 ; tÞ; i ¼ 1; 2; 3; ð2:4Þ
It is necessary to study the variation of the state of strain when we pass from a
point M to a neighbouring one N (Fig. 2.1b). But we must remark from the outset
that this study can be made in two ways.
We can thus consider the variables x1 ; x2 ; x3 , corresponding to the domain D
(the solid in the initial state), as independent variables, named Lagrange
co-ordinates (material co-ordinates); this denomination is due to the fact that,
during the phenomenon of deformation, one follows a material particle of the solid
in its displacement. They are called referential co-ordinates too, because they
specify the position of the particle with respect to the initial state, considered to be
a referential frame. In this computational method, the co-ordinates of the point M ,
which specify the domain D (the solid in the actual state) are the unknown
functions of the problem, that have to be determined; the same thing can be stated
about the displacements (2.4).
As well, we can consider the variables x1 ; x2 and x3 as independent variables,
under the name of Euler co-ordinates (space co-ordinates); this denomination is
36 2 Geometry and Kinematics of Deformation
due to the fact that, at a given point of the space, all the particles which pass
through this point at various moments, because of their displacements, are
examined. The problem is put to determine the position vector r (the unknown
functions (2.20 )) and the displacements in the form
ui ¼ ui ðx1 ; x2 ; x3 ; tÞ; i ¼ 1; 2; 3; ð2:40 Þ
In the case of finite deformations, this will lead to two distinct formulations of
the problems, but these formulations will be identical with each other in the case of
small deformations and rotations.
Let us consider the variation of the state of strain when passing from the point
Mðx1 ; x2 ; x3 Þ to an adjoining point Nðx1 þ dx1 ; x2 þ dx2 ; x3 þ dx3 Þ; the corre-
sponding points after deformation will be M and N , respectively (Fig. 2.2).
We can write the vector relations
! ! ! !
NN MM ¼ M N MN ¼ dr dr; ð2:6Þ
it is easy to see that the displacement components will differ by the components of
! !
the vector NN MM ¼ du, i.e.
oui
dui ¼ dxj ; i ¼ 1; 2; 3; ð2:7Þ
oxj
co-ordinates
n u+du dr*
M*
N
u
dr
M (x1, x2, x3)
r*
r
O
2.1 Finite Deformations 37
and
dr2 ¼ ds2 ¼ dxk dxk ¼ cij dxi dxj ; ð2:80 Þ
where
oxk oxk
cij ¼ ; i; j ¼ 1; 2; 3; ð2:9Þ
oxi oxj
is Green’s deformation tensor TG , represent the distance between the points M and
N (before deformation) and between the points M and N (after deformation),
respectively. We obtained thus
where
1
eij ¼ cij dij ; i; j ¼ 1; 2; 3; ð2:10Þ
2
is the Lagrangian deformation tensor TL .
Taking into account (2.1), it results
oxk ouk
¼ dki þ ; k; i ¼ 1; 2; 3;
oxi oxi
so that, using the expression (2.9), the relation (2.10) becomes
1 oui ouj 1 ouk ouk
eij ¼ þ þ ; i; j ¼ 1; 2; 3: ð2:100 Þ
2 oxj oxi 2 oxi oxj
Taking (2.40 ) into account and referring to the results obtained in the preceding
! !
subsections, the components of the vector NN MM ¼ du will be given by
38 2 Geometry and Kinematics of Deformation
oui
dui ¼ dx ; i ¼ 1; 2; 3; ð2:70 Þ
oxj j
and
where
oxk oxk
cij ¼ ; i; j ¼ 1; 2; 3; ð2:13Þ
oxi oxj
where
1
are the components of a symmetric tensor of second order, which has been
introduced by A.-L. Cauchy for the infinitesimal case and by E. Almansi and G.
Hamel in the case of finite deformations and allows us to define the Cauchy
deformation tensor; it can be expressed also in the matric form
2.1 Finite Deformations 39
2 1 1 3
e e e
6 11 2 12 2 13 7
61 1 7
TE ¼ 6
6 2 e21 e22 e 7: ð2:150 Þ
4 2 23 75
1 1
e e e33
2 31 2 32
2.1.2.3 Strains
We shall define by
ds ds
en ¼ ð2:16Þ
ds
the linear strain (extension) of the element of a line at the point M ðx1 ; x2 ; x3 Þ, in
the direction n. If en [ 0, then we have a stretching, corresponding—in the case of
a one-dimensional state of strain and stress—to a normal tension stress, and, if
en \0, then we shall have a shortening, corresponding—under the same condi-
tions—to a normal compression stress.
Taking into account (2.16), it follows that
ds ¼ ð1 þ en Þds; ð2:160 Þ
therefore e11 , e22 , e33 represent the extensions in the directions of the axes of
co-ordinates, related to the components of the tensor TL by expressions of the form
(2.19) (we use a notation with two indices for reasons which will be seen in the
case of infinitesimal deformations).
Generally, for the direction n we can write
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en ¼ 1 þ 2eij ni nj 1: ð2:190 Þ
! !
Let us now consider two elements of a line MN and MP , parallel to the co-ordinate
axes Ox2 and Ox3 , respectively; let u23 be the angle between these elements after
the deformation (Fig. 2.3). We shall define by
40 2 Geometry and Kinematics of Deformation
P N*
23
dx3 M *
O x2
p
c23 ¼ u23 ð2:20Þ
2
the variation of the right angle formed by two elements of a line, parallel to the
co-ordinate axes Ox2 and Ox3 . So, c23 , c31 and c12 characterize the angular strain
(shearing strain). They are positive when representing a diminution of the right
angle and correspond to positive tangential stresses; otherwise (an increase of the
right angle), the angular strains are negative.
From the triangle M N P we deduce (equally in material co-ordinates), by
using the cosine theorem,
! pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N P 2 ¼ ð1 þ 2e22 Þdx22 þ ð1 þ 2e33 Þdx23 2 ð1 þ 2e22 Þð1 þ 2e33 Þ cos u23 dx2 dx3 ;
and we remark that, by passing from N to P , we obtain dx2 \0. This leads to
2e23
sin c23 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð2:21Þ
1 þ 2e22 1 þ 2e33
In general, for two directions of unit vectors n and m, which make the angle x,
we can write
n m n m
d r d r ¼ d r d r cos x;
n m n m
dr dr ¼ dr dr cosðx cnm Þ;
where we have introduced the angular strain cnm with respect to the angle x; taking
into account (2.9) and (2.10), we have
n m
n
m
n m ox ox n m n m
dr dr ¼ dxk dxk ¼ k k dxi dx j ¼ ð2eij þ dij Þdxi dx j :
oxi oxj
2.1 Finite Deformations 41
observing that
n m
dxi dxj
n ¼ ni ; m ¼ mj ; i; j ¼ 1; 2; 3;
dr dr
it results
ð2eij þ dij Þni mj 2eij ni mj þ cos x
cosðx cnm Þ ¼ ¼ : ð2:210 Þ
ð1 þ en Þð1 þ em Þ ð1 þ en Þð1 þ em Þ
In particular, if the directions n and m are orthogonal one each other, then we
obtain
2eij ni mj
sin cnm ¼ : ð2:2100 Þ
ð1 þ en Þð1 þ em Þ
In space co-ordinates, we get—analogically—relations of the form
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e11 ¼ 1 1 2e11 ; ð2:22Þ
2e23
sin c23 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffipffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; ð2:23Þ
1 2e22 1 2e33
the linear strain em being defined with respect to the distance between two adjacent
points after the deformation (Fig. 2.4a).
ds ds
en ¼ ; ð2:1600 Þ
ds
obviously, the angular strain is defined with respect to the right angle between two
segments of a line after deformation (which before the deformation formed an
angle uyz ) (Fig. 2.4b).
Generally, we can write
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
en ¼ 1 1 2eij ni mj ; ð2:220 Þ
N M* P
N*
-dr *
M
23
M r* N
r
M (x1, x2, x3)
O O x2
(a) (b)
likewise, in space co-ordinates, we can write (with respect to the volume after
deformation)
dx1 dx2 dx3 dx1 ð1 e11 Þdx2 ð1 e22 Þdx3 ð1 e33 Þ
ev ¼ ;
dx1 dx2 dx3
so that
Referring to the mapping relations (2.2) and taking into account (2.1) and the
functional determinant (2.3), we can equally write
2.1 Finite Deformations 43
oxi oui
1 þ ev ¼ det ¼ det dij þ
oxj oxj
ou1 ou1 ou1
1 þ
ox1 ox2 ox3
ou2 ou2 ou2
¼ 1þ ; ð2:26Þ
ox1 ox2 ox3
ou ou3 ou3
3
1þ
ox1 ox2 ox3
analogically, we have
" # " #
oxi oui
1 ev ¼ det ¼ det dij þ
oxj oxj
ou ou1 ou1
1 1
ox1 ox2 ox3
ou ou2 ou2
2
¼ 1 : ð2:260 Þ
ox1 ox2 ox3
ou3 ou3 ou3
1
ox ox2 ox3
1
We can write a relation, analogue to the one in the case of extensions, in the
form
ð1 þ ev Þð1 ev Þ ¼ 1: ð2:27Þ
If we compute these determinants, then we are able to express the cubical
dilatation by means of linear and angular strain components; if the angular strains
vanish, then we shall obtain once again the relations (2.25), (2.250 ), which were
deduced under just these conditions (i.e., by admitting that an elementary paral-
lelepiped, the sides of which are parallel to the axes of co-ordinates, maintains its
form throughout the deformation). This occurs for the three directions, orthogonal
one to each other, for which the linear strains have extreme values; indeed, from
the formulae (2.19), (2.22) it follows that these directions coincide with the
principal direction of the tensors TL and TE , respectively, while the formulae
(2.21) and (2.23) show that the corresponding angular strains vanish.
In the case of infinitesimal deformations, we shall admit that the strain can be
neglected with regard to unity; also, the product of two such strains can be
neglected with respect to a third one.
44 2 Geometry and Kinematics of Deformation
We also consider the hypothesis of the infinitesimal rigid body local rotation
(which we will subsequently define); in this case, the components of the rigid body
local rotation vector are negligible with respect to unity. It is important to mention
that these hypotheses are independent from each other, since the possibility exists
of simultaneous infinitesimal deformations and finite rigid body local rotations or
inversely. On the other hand, the principle of superposition of effects ceases to be
valid, because the superposition order of infinitesimal deformations and finite
deformations, for example, should be taken into account.
In computations, products can also occur between strains and rigid body local
rotations, the omission of which with respect to unity should be made with great
care, observing especially that the ratio between these quantities is to be taken into
account.
The hypotheses of infinitesimal deformations and of infinitesimal rigid body
local rotations are equivalent to the hypotheses of neglecting of the displacement
gradient with respect to unity. The hypothesis corresponds to what we shall call,
shortly, the case of infinitesimal deformations, when the relations (2.100 ) and
(2.140 ) become linear; that is the case of a linear theory from a geometric
standpoint.
Let f ðx1 ; x2 ; x3 ; tÞ be a function of class C1 , expressed by means of material
co-ordinates; using space co-ordinates, we obtain the function in the form
f ðx1 ; x2 ; x3 ; tÞ. The partial derivative of first order with respect to one of the space
variables, e.g., the variable x1 , will be written in the following form
of of ou1 of ou2 of ou3
¼ 1þ þ þ ; ð2:28Þ
ox1 ox1 ox1 ox2 ox1 ox3 ox1
where we took (2.1) into account; neglecting the displacement gradient with
respect to unity, it follows that
of of
¼ : ð2:29Þ
ox1 ox1
Therefore, in the linear theory, the same results, in the same form, will be found
whether the material co-ordinates or the space ones are used.
In what follows we will suppose to be in the case of infinitesimal deformations
and we will use only material co-ordinates.
We will introduce the displacement gradient tensor as a sum of the strain tensor
and the rigid body local rotation tensor. By this occasion, we will put in evidence
the connection with the displacement vector, as well as with the volume strain.
2.2 Infinitesimal Deformations 45
In the case of the linear theory, we can stop at the first terms of an expansion in
series of the expression (2.19) and approximate the sine by means of the corre-
sponding arc in (2.21); we shall have
e11 ¼ e11 ; e22 ¼ e22 ; e33 ¼ e33 ; ð2:30Þ
and shows how the linear strains vary in the neighbourhood of a point; as a matter
of fact, this relation can also be deduced from the fact that the strains are the
components of a tensor, which allows us to write, by analogy,
cnm ¼ 2eij ni mj ; ð2:340 Þ
!
Let us consider the vector MP of modulus
! 1
MP ¼ pffiffiffiffiffiffiffiffi ; ð2:35Þ
en
directed along the external normal n to the considered area element; to simplify the
computation, we shall consider that the origin of the co-ordinate axes is at the
point M, so that
ni
xi ¼ pffiffiffiffiffiffiffiffi ; i ¼ 1; 2; 3: ð2:350 Þ
en
By means of the relation (2.34), we find that the locus of the point P is given by the
equation
eij xi xj ¼ 1; ð2:36Þ
which indicates how en varies; that is why the surface takes the name of Cauchy
linear strain quadric at the point M. We remark that one takes the sign + or -,
according to whether we have to deal with a tension or with a compression, so that
the quadric be real. It results from (2.35) that the extension of any linear element at
M is in inverse proportion with the square of the radius vector of the quadric along
the linear element; hence, the extreme values of these extensions will correspond
to the directions given by the axes of the quadric.
With M1; M2; M3 as principal axes, the equation of the quadric can be written
in the form
e1 x02 02 02
1 þ e2 x2 þ e3 x3 ¼ 1; ð2:360 Þ
where e is an extreme strain; indeed, the relations (2.38) take place only for the
principal directions, and if we choose these directions as co-ordinate axes and
0
consider, e.g., n ¼ x1 ðn1 ¼ 1; n2 ¼ n3 ¼ 0Þ, then we get e ¼ e11 ¼ e1 .
So that this homogeneous system in direction cosines should possess non-trivial
solutions, the following determinant
e11 e e12 e13
det½eij edij e21 e22 e e23 ¼ 0 ð2:39Þ
e31 e32 e33 e
e3 I1 e2 þ I2 e I3 ¼ 0; ð2:40Þ
with
1
I1 ¼ ijk ljk eil ¼ dil eil ¼ eii ¼ e11 þ e22 þ e33 ¼ e1 þ e2 þ e3 ; ð2:400 Þ
2
1 1
I2 ¼ ijk lmk eil ejm ¼ ðeii ejj eij eij Þ ¼ e22 e33 þ e33 e11 þ e11 e22
2 2
ðe223 þ e231 þ e212 Þ ¼ e2 e3 þ e3 e1 þ e1 e2 ; ð2:4000 Þ
1
I3 ¼ ijk lmn eil ejm ekn ¼ det½eij ¼ e11 e22 e33
6
ðe11 e223 þ e22 e231 þ e33 e212 Þ þ 2e23 e31 e12 ¼ e1 e2 e3 : ð2:40000 Þ
The equation (2.40) supplies the values of the principal strains, while the
system (2.38) to which we associate the condition (2.37) supplies then the prin-
cipal direction corresponding to a given extension.
Let e1 ; e2 be the principal extensions, to which correspond the systems of
equations
1 1 2 2
e1 n i ¼ eij n j ; e2 n i ¼ eij n j ; i ¼ 1; 2; 3:
48 2 Geometry and Kinematics of Deformation
2 1
We multiply the first system by n i and the second one by n i ; we then subtract the
equations of the second system from the equations of the first system, take into
account the symmetry of the tensor Te and obtain the relation
1 2
ðe1 e2 Þ n i n i ¼ 0: ð2:41Þ
This condition is fulfilled in the case of a double root
e1 ¼ e2 ; ð2:410 Þ
which shows that the principal directions M1 and M2 are orthogonal to each other.
Therefore, all three principal directions are orthogonal with one another and form a
trirectangular trihedron.
Let us admit that two of the roots of the Eq. (2.40) are complex conjugate
e1 ¼ A þ iB; e2 ¼ A iB;
it results
e1 e2 ¼ 2iB:
Introducing in the system (2.38), we get for the direction cosines complex values
too, i.e.
1 2
n j ¼ aj þ ibj ; n j ¼ aj ibj ; j ¼ 1; 2; 3:
that cannot be fulfilled, except when B ¼ 0; it follows that the roots of the
Eq. (2.40) are always real.
Since the principal directions at a point do not depend on the system of axes
Mx1 x2 x3 , from which we started, the coefficients (2.400 )–(2.40000 ) are three
invariants which play an important rôle in the geometry of the deformation (the
invariants of the strain tensor Te ).
Referring now to the principal axes determined above, the formulae (2.34),
(2.340 ) read
en ¼ e1 n21 þ e2 n22 þ e3 n23 ; ð2:42Þ
It is easy to show that the angular strains at the point M possess extreme values
for the planes bisecting the principal dihedrals 2M3, 3M1, 1M2; these values are
supplied by
c1 ¼ ðe2 e3 Þ; c2 ¼ ðe3 e1 Þ; c3 ¼ ðe1 e2 Þ; ð2:43Þ
while the corresponding linear strains will be
1 1 1
e1 ¼ ðe2 þ e3 Þ; e2 ¼ ðe3 þ e1 Þ; e3 ¼ ðe1 þ e2 Þ: ð2:430 Þ
2 2 2
It is useful to decompose the tensor Te in a sum of two tensors: the spherical tensor
Te0 of components e0 dij , where
1 1
e0 ¼ eii ¼ I1 ; ð2:44Þ
3 3
and the deformation deviator tensor Te0 , defined in the form
e0ij ¼ eij e0 dij ; i; j ¼ 1; 2; 3; ð2:45Þ
0
We notice that for the invariant I2 one obtains
I20 ¼ ðe022 e033 þ e033 e011 þ e011 e022 Þ þ e223 þ e231 þ e212
1
¼ ðe02 þ e02 02 2 2
22 þ e33 Þ þ e23 þ e31 þ e12
2
2 11
1
¼ ½ðe22 e33 Þ2 þ ðe33 e11 Þ2 þ ðe11 e22 Þ2 þ e223 þ e231 þ e212
6
1
¼ ½ðe2 e3 Þ2 þ ðe3 e1 Þ2 þ ðe1 e2 Þ2
6
1 1
¼ ðc21 þ c22 þ c23 Þ ¼ ðe02 þ e02 02 0 0 0 0 0 0
2 þ e3 Þ ¼ ðe2 e3 þ e3 e1 þ e1 e2 Þ; ð2:48Þ
6 2 1
0
the invariant I3 being given by
1
I30 ¼ ð2e1 e2 e3 Þð2e2 e3 e1 Þð2e3 e1 e2 Þ
27
8 1
¼ ðe1 e1 Þðe2 e2 Þðe3 e3 Þ ¼ ðe03 þ e03 03 0 0 0
2 þ e3 Þ ¼ e1 e2 e3 : ð2:480 Þ
27 3 1
In case of the bodies with plastic properties, an important rôle is played by the
equivalent strain (called intensity of deformations too), defined in the form
qffiffiffiffi
2 0
e ¼ pffiffiffi I2 : ð2:49Þ
3
Let us consider an octahedron bounded by eight planes equally inclined on the
three principal axes (four of the planes bounding the octahedron are put in evi-
dence in Fig. 2.5); let n be the unit vector of the external normal to one of these
planes, of direction cosines (with respect to the principal axes) given by
1
n21 ¼ n22 ¼ n23 ¼ :
3
O x2
x1
2.2 Infinitesimal Deformations 51
the last one being the orthogonality condition of the unit vectors n and m.
Searching the direction m corresponding to the maximal singular strain, we obtain
for that one
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
c0 ¼ ðe2 e3 Þ2 þ ðe3 e1 Þ2 þ ðe1 e2 Þ2 ¼ c21 þ c22 þ c23 ; ð2:50Þ
3 3
it is called octahedral angular strain. We get
pffiffiffi
c0 ¼ 2e: ð2:500 Þ
Taking into account (2.49), (2.500 ), we notice that, sometimes, it is useful to
express the invariant I20 in the form
3
I20 ¼ c20 ; ð2:51Þ
8
as well, for the invariant I2 we get
1
I2 ¼ 3 e20 c20 ; ð2:510 Þ
8
u3 + u3,3 dx3
α32
P N*
u3,2 dx2
M* α23
dx3
u3
u3
N
M ( x1, x2, x3) u2
dx2 u2 + u2,2 dx2
O x2
where ‘‘def’’ is the differential operator used above; in an expanded form, we have
e11 ¼ u1; 1 ; e22 ¼ u2; 2 ; e33 ¼ u3; 3 ; ð2:53Þ
c23 ¼ u2; 3 þ u3; 2 ; c31 ¼ u3; 1 þ u1; 3 ; c12 ¼ u1; 2 þ u2; 1 : ð2:530 Þ
These results can be illustrated by studying the deformation of a right angle
NMP, the sides of which are parallel to the axes of co-ordinates, in a plane
x1 ¼ const (Fig. 2.6); we admit that the deformation does not depend on x1 and
occurs in this plane only, which means that we are in a linear case. After defor-
mation, this element becomes N M P :
Thus, we throw into relief the linear strains e22 ; e33 , as well as the angles
! ! ! !
(formed by M N and M P with MN and MP , respectively, and measured from
the segments of a line in their undeformed state)
a23 ¼ u3; 2 ; a32 ¼ u2; 3 ; ð2:5300 Þ
which leads to shearing strains. With remark, for instance, that a23 represents the
sliding we respect to each other of two straight line elements parallel to the
Ox3 -axis; it results, immediately, the angular strain (taking into account the sign
convention adopted in Sect. 2.2.3)
c23 ¼ a23 þ a32 : ð2:53000 Þ
2.2 Infinitesimal Deformations 53
are given by
1
xij ¼ ðuj; i ui; j Þ ¼ u½j; i ; i; j ¼ 1; 2; 3; ð2:56Þ
2
in a developed form, we can write
1 1 1
x23 ¼ ðu3; 2 u2; 3 Þ; x31 ¼ ðu1; 3 u3; 1 Þ; x12 ¼ ðu2; 1 u1; 2 Þ: ð2:560 Þ
2 2 2
To the tensor Tx we attach the axial vector x, the components of which are
given by
1
xi ¼ ijk xjk ; i ¼ 1; 2; 3; ð2:57Þ
2
conversely, we obtain univocally
xij ¼ ijk xk ; i; j ¼ 1; 2; 3; ð2:570 Þ
as it can be easily seen. Taking into account (2.56), we can also write
1 1 1
xk ¼ ijk u½j; i ¼ ijk ðu½j; i þ uðj; iÞ Þ ¼ ijk uj; i ; k ¼ 1; 2; 3;
2 2 2
so that
1
x ¼ curl u; ð2:5700 Þ
2
where x is the rigid body local rotation vector, while u is the displacement vector.
As well, Tx is the corresponding rigid body local rotation tensor (the justification
for these denominations will be seen further).
We notice that we can write the relations (2.7) in the form
dui ¼ aji dxj ð2:58Þ
54 2 Geometry and Kinematics of Deformation
or
aij ¼ eij þ xij ; eij ¼ aði; jÞ ; xij ¼ a½i; j ; i; j ¼ 1; 2; 3; ð2:600 Þ
the symmetric part of this tensor being the strain tensor Te , while its antisymmetric
part is the rigid body local rotation tensor Tx .
The relations (2.800 ) and (2.1200 ) show that, if ds ¼ ds, then the tensors TL and
TE (hence the strain tensor Te too) vanish; the reciprocal is, as well, true. In other
words, the annulment of the tensor Te is the necessary and sufficient condition to
have a rigid body motion of the body in the neighbourhood of the point where this
tensor vanishes. In this case, the displacement gradient remains only with its
antisymmetric part Tx ; we can write
dui ¼ xji dxj ¼ kji xk dxj ; i ¼ 1; 2; 3; ð2:61Þ
The formulae (2.25), (2.250 ) give the volume strain, with respect to the volume
before or after deformation, in material or space co-ordinates, respectively,
assuming that in planes parallel to the co-ordinate axes one has not angular strains.
We shall make some observations concerning these formulae.
The formula (2.19) shows that the directions for which e11 ; e22 ; e33 have extreme
values coincide with the directions for which e11 ; e22 ; e33 have a similar property;
but the latter quantities are the components of a symmetric tensor of second order.
It results that there exist three directions orthogonal one to each other for which the
linear strains have extreme values. The quantities e23 ; e31 ; e12 vanish for these
directions; the formula (2.21) shows that also the angular strains vanish for these
directions. Hence, the directions for which the linear strains have extreme values
specify, two by two, planes in which the angular strains vanish and reciprocally.
Thus, there exists a frame, characterized by three directions orthogonal one to each
other, which has only a rigid solid motion during the phenomenon of deformation;
we can also say that, at each point of the solid, there exists an elementary paral-
lelepiped in the initial state, which remains parallelepiped in the actual state.
The formulae (2.22) and (2.23) show, by an analogous reasoning, that—at each
point of the solid—there exists an elementary parallelepiped in the actual state,
which originates from an elementary parallelepiped in the initial state too. Obvi-
ously, the elementary parallelepiped which has this property is the same, either we
use material or space co-ordinates. It follows that the linear strains e11 ; e22 ; e33 and
e11 ; e22 ; e33 , respectively, have extreme values for the same three directions
orthogonal one to each other.
From the above considerations it results that the formulae (2.52) and (2.520 )
give the volume cubical strain expressed by means of the principal linear strains
(which have extreme values), We can thus write
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ev ¼ 1 þ 2I1 þ 4I2 þ 8I3 1; ð2:63Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ev ¼ 1 1 2I1 þ 4I2 8I3 ; ð2:630 Þ
where I1 ; I2 ; I3 and I1 ; I2 ; I3 are the invariants of the tensor TL and of the tensor TC ,
respectively. We notice that, squaring the relation (2.25), we get
1 þ 2e11 2e12 2e13
ð1 þ ev Þ ¼ 2e21 1 þ 2e22 2e23 ;
2e31 2e32 1 þ 2e33
wherefrom it results the relation (2.63); analogously, we find again the relation
(2.630 ).
56 2 Geometry and Kinematics of Deformation
The relations (2.25), (2.250 ) or the relations (2.63), (2.630 ) show that, in case of
infinitesimal deformations, the volume strain will be given by
h ¼ ev ¼ e1 þ e2 þ e3 ¼ eii ¼ I1 : ð2:64Þ
Taking into account (2.52), we can also write
h ¼ ui;i : ð2:640 Þ
If we have
h¼0 ð2:65Þ
for a certain solid and for certain external loads, then we say that we have to do
with an incompressible state of deformation (called isochore state of deformation
too); if the relation (2.65) takes place for any external loads, we say that we have to
do with an incompressible solid (which is a physical property of the respective
solid).
0
We notice that in case of the deformations deviator, we have I1 ¼ 0. Thus, the
respective deviator leads to an incompressible state of deformation; otherwise, the
spheric tensor leads to a homogeneous deformation (without a change of the form,
but with a change of the volume).
We notice that we can consider the relations (2.52) between strains and
displacements as a system of 6 equations with 3 unknown functions ui ; i ¼ 1; 2; 3.
This system is compatible if the 6 distinct components of the tensor Te verify
certain condition of compatibility.
Starting from the relations (2.52) and (2.56), we can write the components of
the displacement gradient in the form
ui; j ¼ eij xij ; i; j ¼ 1; 2; 3;
assuming that the displacements are functions at least of class C2 , the necessary
and sufficient conditions of compatibility are obtained equating the mixed deriv-
atives of second order of the functions ui . We obtain thus
2.2 Infinitesimal Deformations 57
summing in both members and taking into account the properties of symmetry and
of antisymmetry of the tensors Te and Tx , respectively, it results
xij; k þ xjk; i þ xki; j ¼ 0; i; j; k ¼ 1; 2; 3: ð2:67Þ
Assuming, analogically, that the rigid body local rotations are functions at least of
the class C2 , the necessary and sufficient conditions of compatibility, expressed
only by means of the strains, are obtained equating the mixed derivatives of second
order of the functions xkj . We get thus
or in the form
ilm jkn eij; kl ¼ 0; m; n ¼ 1; 2; 3; ð2:680 Þ
* *
N1 N2
P N R P* R*
M*
Q M S Q* S*
(a) (b)
P*
P* R*
* R*
N2 N 1
*
N
*
Q* S* Q*
S*
M*
M*
(c) (d)
Fig. 2.7 Two neighbouring elements: undeformed state (a), separation (b), interpenetration (c),
continuity (d)
hence the tensor of second order from the first member of the relation (2.680 ) is
symmetric with respect to the indices m and n; this tensor will have 6 distinct
components.
Hence, we can write six distinct equations of compatibility; they are of the form
1
ðe23;1 þ e31;2 þ e12;3 Þ;1 ¼ ðc23;1 þ c31;2 þ c12;3 Þ;1 ¼ e11;23 ;
2
1
ðe31;2 þ e12;3 þ e23;1 Þ;2 ¼ ðc31;2 þ c12;3 þ c23;1 Þ;2 ¼ e22;31 ; ð2:690 Þ
2
1
ðe12;3 þ e23;1 þ e31;2 Þ;3 ¼ ðc12;3 þ c23;1 þ c31;2 Þ;3 ¼ e33;12 :
2
These conditions of compatibility have a particularly important physical
interpretations. Indeed, let be two neighbouring elements of the body, at the
beginning in a undeformed state (Fig. 2.7a). If the conditions of compatibility
would not be fulfilled, then the two elements would be separated after deformation
(Fig. 2.7b) or would be interpenetrated (Fig. 2.7c); if these conditions are fulfilled,
2.2 Infinitesimal Deformations 59
then the elements remain glued one to the other (Fig. 2.7d). Hence, the conditions
(2.68) or (2.680 ) or (2.69), (2.690 ) are, from a mechanical point of view, necessary
and sufficient conditions of continuity of the deformation for a simply connected
domain.
Using the results in the preceding subsection and placing ourselves in the case of a
simply connected domain, we can compute the displacements at an arbitrary point
P, supposing we know them at a point P0 (Fig. 2.8). Thus, starting from the
derivatives of the first order, we get
where we assumed that ui ; eij ; xij depend on the integration variables n1 ; n2 ; n3 ; but
Using the relation (2.6600 ), we get Cesàro’s formulae [2] in the form
ð2:70Þ
The conditions that the above integrals do not depend on the path, in case of a
simply connected domain, are just the continuity equations established in the
previous subsection.
In the particular case in which we assume that all the components of the tensor Te
vanish, the formula (2.70) is reduced to
where we took into account the relation (2.570 ). This corresponds to a rigid body
motion, characterized by the displacement vector u0 and by the rigid body local
rotation vector x0 :
We notice, as well, that the formula (2.71) gives the general integral of the
system of Eq. (2.52), considered to be homogeneous (with vanishing components
of the tensor Te ).
We notice also that two states of displacement, to which corresponds the same
state of deformation, differ by a rigid body motion; this is the theorem of
G. R. Kirchhoff.
If we take the origin of the co-ordinate axes at the point P0 , then the rigid body
motion will be given by
where u0i ; x0i ; i ¼ 1; 2; 3, are the displacements and the rigid local rotations,
respectively, corresponding to a zero tensor Te . In a vector form, we have
!
u ¼ u0 þ x P0 P: ð2:71000 Þ
assuming that the rigid body displacement and local rotation vanish at the origin of
the co-ordinate axes, considered to be the centre of the phenomenon of dilatation.
We also have
xij ¼ ex½j; i ¼ 0; i; j ¼ 1; 2; 3: ð2:75Þ
If the strain tensor has only one non-zero component, which is with different
indices, e.g., e12 ¼ e21 ¼ c=2 (Fig. 2.9a), then we have to do with a simple angular
strain, Cauchy’s quadric being a cylinder with an equilateral hyperbola directrix,
of equation
cx1 x2 ¼ 1; ð2:82Þ
x2 P P*
P* γ N*
Q 2 N
P Q
Q* Q*
N* 2 1
M
N M* M
M*
π
4
x1 O x1
O
(a) (b)
Fig. 2.9 Simple angular strain: in the Ox1 x2 -plane (a), in the O12-plane (b)
or
0 0
e1 x12 þ e2 x22 ¼ 1: ð2:880 Þ
This leads to the state of displacement
ui ¼ ui ðx1 ; x2 Þ; i ¼ 1; 2; u3 ¼ 0: ð2:89Þ
The rigid body local rotation vector remains with only one non-zero component
1
x3 ¼ ðu2;1 u1; 2 Þ; ð2:90Þ
2
while the volume strain is given by
ev ¼ u1; 1 þ u2; 2 : ð2:91Þ
replacing in the relations (2.93), using the formulae (2.920 ) and by the help of the
formula (A.134), we get
1 ousa ousb
es a s b ¼ þ
2 osb osa
1 oxi o2 xi oxi o2 xi
þ þ usc ; a; b ¼ 1; 2; 3: ð2:94Þ
2 osa osb osc osb osa osc
Taking into account the formula (A.1350 ), we may write (without summation
with respect to a:b ¼ 1; 2; 3)
1 ousa ousb
esa sb ¼ þ
2 osb osa
X3
1 1 ohc ohc 1 oha 1 ohb
þ dac þ dbc dab þ usc ð2:940 Þ
2 c¼1 hc osb osa ha osc hb osc
M
r
O x2
x1
66 2 Geometry and Kinematics of Deformation
We can write the relations to pass from the displacements in orthogonal Cartesian
co-ordinates to those in cylindrical co-ordinates in the form
ur ¼ u1 cos h þ u2 sin h;
uh ¼ u1 sin h þ u2 cos h; ð2:97Þ
u z ¼ u3
and inversely
u1 ¼ ur cos h uh sin h;
u1 ¼ ur sin h þ uh cos h; ð2:970 Þ
u3 ¼ uz :
Applying the results in the preceding subsection, there result the relations between
strains and displacements
our
err ¼ ;
or
1 ouh 1 ouh 1
ehh ¼ þ ur ¼ þ ur ; ð2:98Þ
r oh r osh r
ouz
ezz ¼ ;
oz
2.2 Infinitesimal Deformations 67
while those which make the passing from the spherical co-ordinates to the
orthogonal Cartesian ones are
Since the quantities characterizing the state of displacement and the state of strain
are function of time, a study of the kinematics of deformation becomes absolutely
necessary. We will make thus some general considerations and will analyse the
case of infinitesimal deformations.
Taking into account that the velocity is the derivative of the position vector (we
remark that one must consider the position vector in the actual state) with respect
to time (Fig. 2.11)
dr
v¼ ð2:103Þ
dt
and starting from (2.1), where xi are material co-ordinates, independent of time, and
xi ; i ¼ 1; 2; 3, are the space co-ordinates, supplied by (2.5), we can equally write
du ou dxi ou
v¼ ¼ þ : ð2:1030 Þ
dt oxi dt ot
dv d2 r d2 u ov dx ov
a¼ ¼ 2 ¼ 2 ¼ i þ : ð2:104Þ
dt dt dt oxi dt ot
O
2.2 Infinitesimal Deformations 69
In the case of infinitesimal strains and rotations, we shall neglect the non-linear
terms and so we obtain the displacement velocity
ou
v¼ ¼ u;
_ ð2:105Þ
ot
where the partial derivatives with respect to time shall be marked by a point; thus,
the components of the displacement velocity are
vi ¼ u_ i ¼ u_ i ðx1 ; x2 ; x3 ; tÞ; i ¼ 1; 2; 3: ð2:1050 Þ
Neglecting the space derivative, thesubstantial derivative reduces itself to the time
derivative. Besides, we remark that, in the case of infinitesimal deformations, the
relations of the form (2.29) are valid and no distinction is to be made between
material co-ordinates and space ones. Therefore, we shall use only material co-
ordinates and we shall write relations such as (2.1050 ); likewise, in case of strains
and rigid body local rotations, we shall use the relations (2.33) and (2.55).
Concerning the displacement acceleration, we shall also have
ov o2 u
a¼ ¼ v_ ¼ 2 ¼ u
€; ð2:106Þ
ot ot
the components of the displacement acceleration will therefore be
ai ¼ v_ i ¼ €
ui ¼ €
ui ðx1 ; x2 ; x3 ; tÞ; i ¼ 1; 2; 3: ð2:1060 Þ
We can moreover introduce the strain velocities (the Euler tensor of the rate of
strain)
vij ¼ e_ ij ¼ e_ ij ðx1 ; x2 ; x3 ; tÞ; i; j ¼ 1; 2; 3; ð2:107Þ
as well as the rigid body local rotation velocities (theCauchy spin tensor of the
rate of rigid body local rotation)
x_ ij ¼ x_ ij ðx1 ; x2 ; x3 ; tÞ; i; j ¼ 1; 2; 3: ð2:108Þ
Starting from (2.52) and (2.56) and taking into account that we have to deal
with infinitesimal deformations (we use material co-ordinates, independent of
time; we have thus relations of the form o2 f =otox ¼ o2 f =oxot for functions f ¼
f ðx1 ; x2 ; x3 ; tÞ of class C2 ), we can also write the relations
1
e_ ij ¼ ðu_ i; j þ u_ j; i Þ ¼ u_ ði; jÞ ; i; j ¼ 1; 2; 3; ð2:109Þ
2
1
x_ ij ¼ ðu_ j; i u_ i; j Þ ¼ u_ ½j; i ; i; j ¼ 1; 2; 3: ð2:110Þ
2
70 2 Geometry and Kinematics of Deformation
Obviously, for the tensors Te_ and Tx_ we can proceed to a study similar to that
dealt with in Sects. 3.1.1 and 3.1.4; as well, we can introduce the tensor Ta_
supplied by
Ta_ ¼ Gradu ¼ Te_ þ Tx_ ; ð2:111Þ
where the last two tensors represent its symmetrical and its antisymmetrical part,
respectively.
We notice that, in general, the principal axes of the tensor Te_ do not coincide
with the principal axes of the tensor Te , while the principal linear strain velocities
do not represent the derivatives with respect to time of the principal linear strains.
As well, after Euler’s criterion, if
e_ ij ¼ 0; i; j ¼ 1; 2; 3; ð2:112Þ
then we have an irrotational motion; in this case, the field of the displacement
velocities is a field of gradients (the displacement velocities derive from a scalar
potential u ¼ uðx1 ; x2 ; x3 ; tÞ of class C2 ) and we may write
2.2 Infinitesimal Deformations 71
v ¼ u_ ¼ gradu: ð2:117Þ
All the consideration made in Sect. 2.2.2 concerning the continuity of the
deformation and of the rotation and concerning the computation of these quantities
can be repeated, in case of the infinitesimal deformations, for the strain velocities
and for the rotation velocities.
Likewise, in curvilinear co-ordinates, one can use the results supplied in
Sect. 2.2.4.
References
A. Books
B. Papers
11. Beltrami, E.: Sur la théorie de la déformation infiniment petite d’un milieu. C. Rend. hebd. de
séance de l’Acad. Sci. 108, 502 (1889)
}
12. Golitsyn, B.: Uber die Dispersion und Dämpfung der seismischen Oberflächwellen. Bull. de
l’Acad. Imp. des Sci. de Saint-Petersburg, ser. 6, 219 (1912)
13. Galletto, D.: Sull’unicità in presenza di vincoli interni di una condizione cinematica
fondamentale nella teoria delle deformazioni finite. Atti dell’Ist. Veneto di Sci. Lett. e Arti,
Cl. di Sci. Mat. e Nat. 123, 197 (1965)
14. Lamé, G.: Mémoire sur les coordonnées curvilignes. J. de Math. Pures et Appl. 5, 313 (1840)
15. Noll, W.: A mathematical theory of mechanical behaviour of continuous media. Arch. Rat.
Mech. Anal. 2, 195 (1958)
16. Truesdell, C.: The mechanical foundations of elasticity and fluid dynamics. J. Rat. Mech.
Anal. 1, 125 (1952)
Chapter 3
Mechanics of Stresses
We shall continue now the theoretical study of the preceding chapter, by putting in
evidence the stresses which arise because of the deformation of a solid body; both
vector and tensor aspects of the stress are thus emphasized [4].
So as to emphasize the stresses occuring within a solid body, stresses which can be
considered as variations of the internal forces of cohesion, we shall—in general—
carryout an arbitrary section S (which can be curved), open or closed, through this
body, that is thus divided into two parts (Fig. 3.1a). Applying the principle of
equilibrium of parts (static or dynamic equilibrium), we get the main properties of
the stress vector.
(a) A
II
(b) t (c) t
n n n
T n p
P
M M
I I
n n
N n x3 r dA n
A
x1 O x2
Fig. 3.1 Stress vector: section A (a), efforts (b), stress-vectors (c)
n n
by the effort D P, of components D N (along the external normal to the section at
n
the given point) and D T (in the tangent plane to the section at the point M), which
makes connection with the remote side (the right side) (Fig. 3.1b). We admit that
n
the effort D P is of the nature of a force. We obtain thus the mean stress vector in
the neighbourhood of the point M of an area element of external normal of unit
vector n, in the form
n
n DP
pmean ¼ : ð3:1Þ
DA
At the limit, admitting its existence, we get (Fig. 3.1c)
n
n n dP
p ¼ lim pmean ¼ ; ð3:2Þ
DA!0 dA
i.e. the stress vector at the point M, on an area element of external normal n; the
n
value p of this vector represents the stress at the point M.
The notion of stress should be introduced with certain precautions of mathe-
n
matical nature, because one cannot know how D P varies when DA ! 0. In regard
n
to the concentrated loads, for instance, such a definition is nonsense (p ! 1); the
theory of distributions plays a quite important rôle in this case.
We mention that the external (outer) normal n is used usually for solid bodies,
while in case of fluids it is used the internal (inner) normal.
3.1 Stress Vector 75
n
The component of the stress vector p along the direction m is usually denoted
n
by pm . Introducing the components along the three axes of co-ordinates, we may
write
n n n n
n n
p2 ¼ pi pi ¼ p21 þ p22 þ p23 : ð3:3Þ
n n
Ordinarily, we denote the component of p along the external normal n by r and
n
we call it normal stress; likewise, the component of p in the plane tangent at the
n
point M to the section surface S is denoted by s and is called tangential stress
n n
(Fig. 3.1c). These components result from D N and D T by a process of passing to
the limit, analogue to that mentioned above; evidently, we can introduce, in the
n n
same manner, the notions of normal stress vector r and tangential stress vector s.
We have
n n n
p ¼ r þ s; ð3:30 Þ
as well as
n n n
p2 ¼ r 2 þ s 2 : ð3:300 Þ
The normal stress is positive if it corresponds to a phenomenon of tension and
negative in the opposite case (phenomenon of compression). We shall admit
moreover that the tangential stress is positive if it corresponds to a positive angular
strain and negative in the contrary case; this fact should however be related to the
chosen system of co-ordinates.
The components of the stress vector will be functions of the form (we use
material co-ordinates, assuming that we are in the case of infinitesimal
deformations)
n n
pi ¼ pi ðx1 ; x2 ; x3 ; tÞ: ð3:4Þ
n
From a dimensional point of view, the stresses are expressed by ½p ¼ ML1 T2 .
n
By state of stress of the body at the point M we mean the totality of the stresses p
in all the directions n around that point. The state of stress of the whole body will
be given by the states of stress of all the points that form it.
By one-dimensional or linear state of stress (also called antiplane state of
stress) we shall mean that state of stress in which appear stress vectors only on the
area elements of a normal of fixed direction. This is the case of straight bars of
constant cross section, undergoing simple stresses.
76 3 Mechanics of Stresses
(a) x3 (b) x3
~ ~
V F dV F
( M) (dM )
r M r M
O x2 O x2
x1 x1
where
Fi ¼ Fi ðx1 ; x2 ; x3 ; tÞ; i ¼ 1; 2; 3: ð3:8Þ
The component Fi is considered positive if it corresponds to the positive
direction of the co-ordinate axis and negative in the contrary case.
The volume forces can be of various origins; among the most important ones,
we mention the own weight (represented by the weight per unit volume c) and the
forces of inertia.
If DM is the mass corresponding to a volume element DV (Fig. 3.2a), then we
can similarly introduce the mean density
DM
qmean ¼ ; ð3:9Þ
DV
at the limit (Fig. 3.2b), provided it exists, we shall get
dM
q ¼ lim qmean ¼ ; ð3:10Þ
DV!0 dV
which gives the unit mass (density) at the point M, of dimension ½q ¼ ML3 .
The mass of the whole body is given by
ZZZ
M¼ qdV; ð3:11Þ
V
where, generally,
q ¼ qðr; tÞ ¼ qðx1 ; x2 ; x3 ; tÞ; ð3:12Þ
Fi ¼ Fi q€
ui ku_ i ; i ¼ 1; 2; 3; ð3:150 Þ
n
The stress vector p depends on the direction vector n of the area element upon
n
which it acts and of the point at which the computation is made. If p depends only
on the direction n, then we have to deal with a state of uniform stress; the contrary
case is the state of non-uniform stress.
3.1 Stress Vector 79
We shall cut out from the body, around the point M, an element of volume in the
form of a plate, of thickness h, the parallel faces being of area A; one of them has the
outer normal n and the other has the outer normal n0 (Fig. 3.3); obviously, we have
n þ n0 ¼ 0: ð3:16Þ
On the face Aþ , which passes through the point M, the centre of mass of the
respective surface, the stress vectors are reduced to a torsor at M, which has only a
n
force resultant p (we assume that the moment resultant is negligible); taking into
account the hypothesis of continuity of stresses, we can write the stress vector (the
n0
force resultant) at the point M 0 , centre of mass of the face A , in the form p þ g,
n0
where p is the corresponding stress vector, calculated at the point M 0 , while g ! 0
C
together with h ! 0. If F is the resultant of the lost forces of d’Alembert and if p
are the resultant stress vectors on the lateral surface along the contour C, on the
area element hds, then the dynamic equation of equilibrium reads
Z
n n0 C
p A þ ðp þ gÞA þ h p ds þ FAh ¼ 0;
C
M'
-
A
n' n'
p
80 3 Mechanics of Stresses
On the basis of this result, we can state that, by changing the sense (sign) on the
outer normal to the element of area upon which acts a stress vector, the sense
(sign) of the latter one changes too.
i
In particular, we consider the stress vectors p applied at the point O and cor-
responding to the co-ordinate axes Oxi ; i ¼ 1; 2; 3. We denote by
i
rij ¼ p ij ð3:17Þ
i
the component of the vector p along the co-ordinate axis Oxj of unit vector ij
(Fig. 3.4); the components rij ; i; j ¼ 1; 2; 3, will be called stresses. If i ¼ j, then
we get normal stresses, while if i 6¼ j, then we obtain tangential stresses.
We assume that the stresses are positive when the external normal of the face
upon which they act has the same direction as the corresponding co-ordinate axis
(or inversely) and when they are oriented in the positive (or negative) direction of
the axis of co-ordinates to which they are parallel. On the basis of this sign
convention, which corresponds to Cauchy’s theorem, the normal stresses
r11 ; r22 ; r33 will be always positive when they correspond to a tension and neg-
ative when they correspond to a compression. As to the tangential stresses
r23 ; r32 ; r31 ; r13 ; r12 ; r21 , their sign will depend on the chosen system of
co-ordinates; it is to be remarked however, that the above convention agrees with
the fact that the positive tangential stresses correspond to positive angular strains.
M3
x3
1
p
dx 3
11
12 G1
n
21
F3 p3
13
Gn n
G2 p n
22
M (x1,x2, x3) G n
F2 p2 dx2
2 n
p r p1 M2
23 F1 31
O 32 x2
G3
x1 dx1
3
p 33
M1
n
So as to be able to appreciate the manner in which the stress vector p varies when
the direction n varies, in the neighbourhood of the point Mðx1 ; x2 ; x3 Þ, we shall cut
out from the body an infinitesimal three-rectangular tetrahedron MM1 M2 M3 with
the faces MM2 M3 ; MM3 M1 ; MM1 M2 of areas
1
dAi ¼ dAni ¼ ijk dxj dxk ; dxj dxk ¼ dxk dxj ; i ¼ 1; 2; 3; ð3:18Þ
4
parallel to the planes of co-ordinates, the side edges having the dimensions
dxi ; i ¼ 1; 2; 3; the face M1 M2 M3 of area dA will be defined by the outer normal n,
of direction cosines ni ; i ¼ 1; 2; 3 (Fig. 3.4).
In case of a system of any curvilinear axes, we shall consider a tetrahedron the
faces of which are corresponding surfaces of co-ordinates.
On the faces MMj Mk act stress vectors the torsors of which at the gravity
centres Gi ðxi ; xj þ dxj =3; xk þ dxk =3Þ; i 6¼ j 6¼ k 6¼ i; i; j; k ¼ 1; 2; 3, are reduced
only to force resultants (we assume also that the moment resultants, which would
lead to micromoments, are negligible); taking into account the hypothesis of
continuity of stresses and Cauchy’s theorem, we can write the resultant stress
i i i
vector which acts at the point Gi , in the form p þ g, where p is the stress vector
i
corresponding to the Oxi -axis, which acts at the point M, while g ! 0 for h ! 0,
where
h ¼ n1 dx1 ¼ n2 dx2 ¼ n3 dx3 ð3:19Þ
is the height of the tetrahedron with respect to the face M1 M2 M3 . On this face acts
a stress vector the torsor of which, at the gravity centre Gn ðxi þ dxi =3; i ¼ 1; 2; 3Þ,
n n n
leads to a force resultant of the form p þ g, where p is the stress vector which
corresponds to the area element of outer normal n, which acts at the point M,
n
observing that g ! 0 for h ! 0.
At the centre of gravity Gðxi þ dxi =4; i ¼ 1; 2; 3Þ of the tetrahedron act the lost
n
forces of d’Alembert F þ g, where F is the lost force of d’Alembert, corre-
sponding to the point M, while g ! 0 for h ! 0.
The equation of dynamic equilibrium of the tetrahedron element will be of the
form
i i n n h
ð p þ gÞ dAi þ ðp þ gÞ dA þ F þ g dA ¼ 0;
3
taking into account (3.18) and making h ! 0, we get
n i
p ¼ p ni ; ð3:20Þ
82 3 Mechanics of Stresses
relation that shows that the stress vector on any direction around a point can be
expressed by means of the stress vectors corresponding to three directions
orthogonal one to each other. Projecting along the Oxi -axes, we may write
n
pi ¼ rji nj ; i ¼ 1; 2; 3; ð3:200 Þ
where we used the notation (3.17). These relation can be written in the developed
form
n
p1 ¼r11 n1 þ r21 n2 þ r31 n3 ;
n
p2 ¼r12 n1 þ r22 n2 þ r32 n3 ; ð3:2000 Þ
n
p3 ¼r13 n1 þ r23 n2 þ r33 n3 ;
We put the condition that the resultant moment of all the forces which act upon the
infinitesimal tetrahedron considered above be equal to zero; calculating this
moment with regard to the point Gn , we may write
X3
! i i ! h
Gg Gi ð p þ gÞ dAi þ Gn G F þ g dA ¼ 0:
i¼1
3
X
3
dxi i i dxk h
jkl dik ð l þ gl ÞdAi þ jkl
p Fl þ gl dA ¼ 0; j ¼ 1; 2; 3;
i¼1
3 12 3
where dik is Kronecker’a symbol; taking into account (3.18) and (3.19), we get
k k dxk
jkl ð pl þ gl Þ þ jkl Fl þ gl ¼ 0; j ¼ 1; 2; 3:
12
If we make h ! 0 and use the notation (3.17), we get the relation
jkl rkl ¼ 0; j ¼ 1; 2; 3; ð3:21Þ
From a mechanical standpoint, we can assert that, in this case, the existence of a
tangential stress rnm on an area element of external normal n, within a body, entails
the appearance of a tangential stress rmn of a corresponding direction, upon an area
element normal to the first one (of outer normal m). Hence, we may write (Fig. 3.5a)
rmn ¼ rnm : ð3:22Þ
Taking into account the sign convention adopted in Sect. 2.2.1 for the tan-
gential stresses, we observe that one cannot have a situation as that in Fig. 3.5b.
We notice also that the formula (3.22) is valid also in the case in which one of the
area elements lies on the outer surface of the body, being acted upon by a
superficial loading.
This formula has many applications. Let thus be a straight cylinder of Ox3 -axis,
acted upon the lateral surface by a superficial load, which has no components
along the Ox3 -axis. We make a section normal to the considered bar; let now be
two elements of outer normals n and x3 , respectively (Fig. 3.6a). We notice that
n
r3n ¼ p3 ; ð3:23Þ
n
but the tangential stress p3 is equal to zero, so that
r3n ¼ 0 ð3:230 Þ
and we can state that the tangential stress r3t which arises in the cross section is
tangent to its contour (Fig. 3.6b).
n m
Let be now the stress vectors p and p, corresponding to the elements of area of
outer normals n and m, respectively. The components of these vectors along the
directions m and n, respectively, are given by
n n n
pm ¼ p m ¼ pi mi ¼ rji nj mi ;
m m m
pn ¼ p n ¼ pj nj ¼ rij mi nj ;
mn n mn n
n
p
nm nm
84 3 Mechanics of Stresses
n
pm
The fact that the stress vector acts on an oriented element of area shows that, to
deepen this notion, it is necessary to introduce a quantity of tensor character too,
which we shall call stress tensor. We present, in what follows, the main properties
of this tensor and we establish the equations of equilibrium and motion which are
verified by its components.
n
Projecting the stress vector p, corresponding to the element of area of outer normal
n, on the unit vector n (Fig. 3.1c), we obtain
3.2 Stress Tensor. Equations of Equilibrium and Motion 85
n n n
pn ¼ p n ¼ pi ni ¼ rji nj ni ;
thus, the relations (3.25) and (3.28) give the normal and tangential components,
n
respectively, of the stress vector p.
!
Let MQ be the vector of modulus
! 1 ni
MQ ¼ pffiffiffiffiffiffiffinffi ; xi ¼ pffiffiffiffiffiffiffinffi ; i ¼ 1; 2; 3; ð3:29Þ
r r
directed after the outer normal n to the considered element of area. With the help
of the relation (3.25), we find that the locus of the point Q will be supplied by the
equation (for the sake of simplicity, the point M is considered to be the origin of
the axes of co-ordinates)
rij xi xj ¼ 1; ð3:30Þ
n
that shows how r varies; let us remark that we take the sign + or the sign -,
according to whether we have to deal with a tension or with a compression, so that
86 3 Mechanics of Stresses
the surface be real. This quadric is called the quadric of normal stresses (Cauchy’s
quadric) (Fig. 3.8).
Referring the quadric to its principal axes M1; M2; M3, its equation will take
the form
r1 x02 02 02
1 þ r2 x2 þ r3 x3 ¼ 1; ð3:31Þ
where x0i ; i ¼ 1; 2; 3, are the new co-ordinates. It follows that there exist three
principal directions along which the extreme normal stresses r1 r2 r3 develop
and for which the tangential stresses vanish. Generally, these three principal
directions do not coincide with the principal directions of the tensor Te .
If we have r1 r2 r3 [ 0 or if 0 [ r1 r2 r3 , then the quadric is an
ellipsoid. We choose the sign of 1 so that the ellipsoid be real; if two of the
principal normal stresses are equal, then we obtain an ellipsoid of revolution, while
if all the three principal normal stresses are equal, then the ellipsoid is a sphere.
If r1 r2 [ 0 [ r3 , e.g., then we obtain a one or a two-sheet hyperboloid as
we choose the sign of 1 (the locus is formed by both hyperboloids); if two of the
principal normal stresses are equal, then the two hyperboloids are of revolution.
We can have situations too, in which the quadric is degenerate (e.g., in case of a
n
state of plane stress). Imposing the condition r ¼ 0, we get only tangential stresses
along the asymptotic cone.
n
If the direction n is a principal one, to which corresponds the normal stress r,
n
and if we take into account (3.30 ) (where we take s ¼ 0), as well as (3.200 ), then we
are led to the equations
rij rdij nj ¼ 0; i ¼ 1; 2; 3: ð3:32Þ
The homogeneous system has non-trivial solutions if the determinant
2 3
r11 r r12 r13
det rij rdij 4 r21 r22 r r23 5 ¼ 0 ð3:33Þ
r31 r32 r33 r
r3 J1 r2 þ J2 r J3 ¼ 0; ð3:34Þ
n
Q
M pn
2
3.2 Stress Tensor. Equations of Equilibrium and Motion 87
with
1
J1 ¼ ijk ljk ril ¼ dil ril ¼ rii ¼ r11 þ r22 þ r33 ¼ r1 þ r2 þ r3 ; ð3:340 Þ
2
1 1
J2 ¼ ijk lmk ril rjm ¼ rii rjj rij rij ¼ r22 r33 þ r33 r11 þ r11 r22
2 2
ðr223 þ r231 þ r212 Þ ¼ r2 r3 þ r3 r1 þ r1 r2 ; ð3:3400 Þ
1
J3 ¼ ijk lmn ril rjm rkn ¼ det½rij ¼ r11 r22 r33
6
ðr11 r223 þ r22 r231 þ r33 r212 Þ þ 2r23 r31 r12 ¼ r1 r2 r3 : ð3:34000 Þ
The Eq. (3.34) supplies the values of the principal normal stresses, while the
system (3.32), with the additional condition (2.37), supplies the principal direc-
tions corresponding to the chosen principal normal stress.
As it has been specified in Sect. 2.2.1.1 for the tensor Te , one can show, in this
case too, that the three principal directions are three-orthogonal and that the three
roots of the Eq. (3.34) are always real.
The coefficients J1 ; J2 ; J3 are invariant at a change of co-ordinate axes; the first
of these invariants is denoted also by
H ¼ rii ¼ r1 þ r2 þ r3 : ð3:35Þ
Taking into account the above results, we can write the normal stress corre-
sponding to an area element of outer normal n, in the M123 system of axes, in the
form
n
r ¼ r1 n21 þ r2 n22 þ r3 n23 ; ð3:36Þ
the direction cosines being considered with respect to these axes; analogically, the
tangential stress corresponding to the directions of unit vectors n and m is given by
rnm ¼ r1 n1 m1 þ r2 n2 m2 þ r3 n3 m3 : ð3:360 Þ
n
The components of the stress p are
n n n
p1 ¼ r1 n1 ; p2 ¼ r2 n2 ; p3 ¼ r3 n3 ð3:37Þ
and the formula (3.300 ) shows that the tangential stress is given by
n 2
s2 ¼ r21 n21 þ r22 n22 þ r23 n23 r1 n21 þ r2 n22 þ r3 n23 : ð3:38Þ
Observing that the relation
takes place, we can write the expression (3.38) also in the remarkable form
n
s2 ¼ ðr2 r3 Þ2 n22 n23 þ ðr3 r1 Þ2 n23 n21 þ ðr1 r2 Þ2 n21 n22 : ð3:380 Þ
To obtain the principal tangential stresses, we equate to zero the derivatives of
first order of the form
n
U ¼ s2 r2 1 n21 n22 n23 ð3:40Þ
M 4 p3
4 2
4
3
1
90 3 Mechanics of Stresses
By taking into consideration the relation (3.39), the relations (3.37) lead to the
equation
which represents Lamé’s ellipsoid of stresses; the radius vector of every point of
n
the ellipsoid is just the stress vector p. So as to establish the area element on which
such a stress is acting, we shall equally consider the direction surface of stresses
by computing the partial derivatives of first order of this function and taking into
account (3.200 ) and (3.27), we notice that we can write
! ! n
F;i ¼ rij xj ¼ jMQjrij nj ¼ jMQj pi ; i ¼ 1; 2; 3:
Finally, we have
n 1
pi ¼ ! F;i ; i ¼ 1; 2; 3; ð3:47Þ
jMQj
the function Fðx1 ; x2 ; x3 Þ playing thus the rôle of a potential function.
On the other hand, F;i ; i ¼ 1; 2; 3, are the direction parameters of the normal to
n
Cauchy’s quadric at the point Q, while pi ; i ¼ 1; 2; 3, are the components of the
n n
stress vector p. This offers the possibility of a simple construction of the vector p,
as concerns its direction; indeed, if we trace the plane, tangent at the point Q to
n
Cauchy’s quadric, the normal from M to this plane will supply the direction of p
(Fig. 3.8).
3.2 Stress Tensor. Equations of Equilibrium and Motion 91
An element of area of outer normal n being given, the relations (3.480 ) allow to
draw three arcs of circle with the centres at O1 ; O2 ; O3 , of radii given by
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
r1 ¼ ðr2 r3 Þ2 þ n21 ðr1 r2 Þðr1 r3 Þ;
2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
r2 ¼ ðr3 r1 Þ2 þ n22 ðr2 r3 Þðr2 r1 Þ; ð3:50Þ
2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
r3 ¼ ðr1 r2 Þ2 þ n23 ðr3 r1 Þðr3 r2 Þ;
2
n
respectively (Fig. 3.11a); these arcs of circle meet at the point P of co-ordinates r
n
and s, obtaining the components of the corresponding stress vector.
We can use also an entirely graphic method. We draw thus the tangents to the
n
circle C2 at the points at which this one pierces the axis O r and construct semi-
straight lines, inclined to these tangents by the angles a2 and a3 , made by the unit
vector n with the principal axes, respectively (Fig. 3.11b); these semi-right lines
pierce the circles C2 and C3 at the points A2 and A3 and the circles C1 and C2 at the
points B1 and B2 , respectively. One can show easily that the arc of circle of radius
r1 with the centre at O1 passes through the points A2 and A3 , while the arc of circle
or radius r3 with the centre at O3 passes through the points B1 and B2 . The
construction is thus completely specified.
As in case of strains, we decompose the tensor Tr into a sum of two tensors: the
spherical tensor Tr0 , given by r0 dij , where
1 1
r0 ¼ rii ¼ J1 ; ð3:51Þ
3 3
and the stress deviator tensor Tr0 , defined by the relation
n n
(a) (b) A2 B2
C2
P C2
C1 n P r3 B1
r1 r2 C3 3
1
3 2 1 3 2 A3 1
O O1 O2 n O3 n O O1 O2 O3 n
C3
C1
Fig. 3.11 Mohr’s circles: semigraphic method (a), graphic method (b)
The extreme values of the normal components of the latter tensor are given by
the equation
Considering the octahedron in Fig. 2.5, the formula (3.36) shows that the
octahedral normal stress is equal to r0 . Using the same procedure as in
Sect. 2.2.1.2, the octahedral tangential stress reads
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 2
s0 ¼ ðr2 r3 Þ2 þ ðr3 r1 Þ2 þ ðr1 r2 Þ2 ¼ s21 þ s22 þ s23 : ð3:57Þ
3 3
We mention also the relation
pffiffiffi
2
s0 ¼ r: ð3:570 Þ
3
Taking into account (3.56), (3.570 ), we can express the invariant J20 in the form
3
J20 ¼ s20 ; ð3:58Þ
2
while the invariant J2 is given by
2 1 2
J 2 ¼ 3 r 0 s0 ; ð3:580 Þ
2
The study up to this point refers to the behaviour of the stresses in the neigh-
bourhood of a point. In order to complete the local study, we must see what
happens when we pass from an area element through the point M to an area
element through an adjacent point. Thus, we deduce the differential equations of
equilibrium and motion which must be fulfilled by the components of the tensor Tr
and by the components of the vector u (the last ones in the dynamic case).
Let us assume that the considered solid body occupies a domain of volume V,
bounded by the surface S; let be also an arbitrary subdomain of volume V 0 ,
bounded by the surface S0 (Fig. 3.12). Using the principle of equilibrium of parts
for this subdomain (principle which has been used till now too for the dynamic
equilibrium for the elementary domains in form of a plate or of a tetrahedron,
which have been cut from the body), one can write the vector equations of
equilibrium in the form
ZZ ZZZ
n 0
pdS þ FdV 0 ¼ 0; ð3:60Þ
S0 V0
ZZ ZZZ
n
r pdS0 þ r FdV 0 ¼ 0: ð3:61Þ
S0 V0
V
r
O
96 3 Mechanics of Stresses
ZZ ZZZ
n
ijk xj pk dS0 þ ijk xj Fk dV 0 ¼ 0: i ¼ 1; 2; 3: ð3:610 Þ
S0 V0
Starting from the Eq. (3.600 ) and using the formula (3.200 ) and the flux-diver-
gence formula (A.1.960 ), which transforms a surface integral in a volume one, it
follows
ZZZ
rji;j þ Fi dV 0 ¼ 0: i ¼ 1; 2; 3;
V0
assuming that the stresses are functions of class C1 and that the volume forces are
functions of class C0 and observing that the subdomain V 0 is arbitrary, we obtain
the equations of equilibrium in the form
rji;j þ Fi ¼ 0; i ¼ 1; 2; 3: ð3:62Þ
The relation (3.610 ) can be written, analogically, in the form
ZZZ h i
ijk xj rlk ;l þxj Fk dV 0 ¼ 0: i ¼ 1; 2; 3;
V0
x3
33 + 33,3 dx3
32 + dx3
32,3
31 + 31,3 dx3
12 11
dx3
23 + 23,2 dx2
F3
13
21 F2 22 + dx2
22,2
F1
22 13 + 13,1 dx1 G + dx2
21 21,2
23
M (x1,x2,x3)
11 + 11,1 dx1 12 + 12,1 dx1
31
r dx1
32
O 33 x2
x1
dx2
co-ordinate planes (Fig. 3.13). Taking into account the variations of the stresses
from a face of the parallelepiped to a neighbouring parallel face by formulae of
Taylor type and observing that we must pass from the stresses at the point
Mðx1 ; x2 ; x3 Þ to the stresses at the centres of gravity of each adjacent face, as in
Sect. 2.2.2, we obtain the equations of equilibrium (3.62).
If we choose a subdomain bounded by co-ordinate surfaces, in a system of
arbitrary curvilinear co-ordinates, then we obtain equivalent equations of equi-
librium in the considered system of co-ordinates.
The above considerations are valid for an arbitrary deformable continuous
medium, hence also for a fluid one [3, 16–19].
In the dynamical case, we must replace the volume forces by d’Alembert’s lost
forces, defined in Sect. 3.1.1.4. Thus, the Eq. (3.62) allow to write the equations of
motion in the form
rji; j þ Fi ¼ q€
ui ; i ¼ 1; 2; 3; ð3:63Þ
In that follows we present some elementary states of stress which appear fre-
quently in applications, i.e.: the hydrostatic stress, the simple tension or com-
pression and the simple shear. As well, we give some data concerning a state of
plane stress.
We mention that these states of stress correspond to the elementary states of
deformation considered in Sect. 2.2.3.
We call simple state of normal stress that state of stress for which
rij ¼ rdij ; r [ 0; i; j ¼ 1; 2; 3; ð3:66Þ
3.2 Stress Tensor. Equations of Equilibrium and Motion 99
We say that we have to do with a simple tension if the state of stress is reduced to
the normal stress r1 ¼ r [ 0 along the direction of a principal axis (e.g., O1,
1
which is specified by the direction cosines ni ; i ¼ 1; 2; 3, with respect to an ort-
honormed frame Oxi ; i ¼ 1; 2; 3); if r\0, then we have to do with a simple
compression.
The quadric of normal stresses is given by the equation
rx02
1 ¼ 1 ð3:68Þ
If the stress tensor has only one component with distinct indices, e.g., the tan-
gential stress r12 ¼ r21 ¼ s (Fig. 3.14), then we have to do with a simple shear;
the quadric of normal stresses is given by
being an equilateral hyperbola, the state of stress taking place in the Ox1 x2 -plane.
100 3 Mechanics of Stresses
O 2 2
x1
By a rotation of p=4 in the Ox1 x2 -plane we can express the quadric with respect
to the principal axes, obtaining the equation
s x02 02
1 x2 ¼ 1; ð3:700 Þ
hence, in case of a simple shear in the Ox1 x2 -plane, the principal normal stresses
are given by
r1 ¼ r2 ¼ s; r3 ¼ 0; ð3:71Þ
the reciprocal being true.
We can thus state that the simple shear in one direction is equivalent to a simple
tension along a direction which makes an angle p=4 with the first direction and to a
simple compression, of the same magnitude, along a direction normal to that of the
simple tension; this result is put in evidence by considering the equilibrium of the
triangular element in Fig. 3.14.
We call plane state of stress that state of stress for which the principal normal
stress r3 vanishes
r3i ¼ 0; i ¼ 1; 2; 3: ð3:72Þ
The quadric of normal stresses is written in the form
or in the form
r1 x02 02
1 þ r2 x2 ¼ 1; ð3:730 Þ
The density q, introduced in Sect. 2.1.3, must behave so that the mass M of the
solid which we follow in its motion be invariant in time; this is the conservation
principle of mass.
At the actual moment, the mass is expressed in the form
ZZZ
M ¼ q
x
1 ; x
2 ; x
3 ; t dV
; ð3:74Þ
V
M ¼ q
J dV; ð3:7400 Þ
V
where we have introduced the Jacobian (2.3). Equating the two masses, we get
ZZZ
ðq
J qÞdV ¼ 0;
V
assuming that the function under the integral is continuous and observing that the
volume V is arbitrary, it results the relation
Jq
¼ q; ð3:75Þ
102 3 Mechanics of Stresses
V dt V
dt
where the volume V is arbitrary. If the function under the integral is continuous,
then we can write Euler’s continuity condition of mass in the form
dq
þ q
divv ¼ 0: ð3:78Þ
dt
Observing that
dq
oq
dr
oq
¼ þ gradq
¼ þ gradq
v;
dt ot dt ot
divðq
vÞ ¼ q
divv þ gradq
v;
3.2 Stress Tensor. Equations of Equilibrium and Motion 103
þ divðq
vÞ ¼ 0: ð3:780 Þ
ot
If the density q
is constant in time, we obtain
divv ¼ 0; ð3:79Þ
hence the field of displacement velocities is solenoidal.
The relation (3.78) can be expressed also in the form
d
ln q
+ divv ¼ 0: ð3:7800 Þ
dt
From the relation (3.770 ), (3.7800 ), we get
d
lnðJq
Þ ¼ 0;
dt
hence, the product Jq
is constant in time, being equal to q (corresponding to the
initial state). We find thus again the continuity condition of mass given by
d’Alembert.
d P
p ¼
; ð3:81Þ
dA
called Cauchy’s stress vector. As a matter of fact, till now, in the frame of this
chapter, we have used the latter stress vector (being in the case of the infinitesimal
deformations, it was not necessary to use the asterisk for the actual state).
104 3 Mechanics of Stresses
r*
r
O x2
x1
^ ¼ d P
;
dP ð3:82Þ
equivalent with
n n
^ i = d P
i ;
dP i ¼ 1; 2; 3: ð3:820 Þ
Taking into account (3.200 ), (3.80) and (3.81), the relation (3.820 ) leads to
^ji nj dA ¼ r
ji ; n
j dA
;
r i ¼ 1; 2; 3; ð3:8200 Þ
where r^ij are the components of Piola-Kirchhoff stress tensor of first kind, while r
ij
are the components of Cauchy’s stress tensor, which is a tensor of second kind.
Observing that two elements of volume constructed on the elements of area
dA and dA
, respectively, are given by
oxj
ox
i
dV
¼ dA
n
dr
¼ n
i dA
dx
i ¼ n
i dA
dxj ð3:830 Þ
oxj
and that
q
dV
¼ JdV ¼ dV; ð3:84Þ
q
we obtain
q oxj
n
i dA
¼
nj dA
; i ¼ 1; 2; 3; ð3:85Þ
q oxi
q
ox
j
ni dA ¼ n dA ; i ¼ 1; 2; 3: ð3:850 Þ
q j oxi
The relations (3.82’’) and (3.85) allow to express the components of the tensor
Tr^ in the form
q oxi
^ij ¼
r r ; i; j ¼ 1; 2; 3; ð3:86Þ
q
ox
k kj
^ij ¼
rij
rij ; i; j ¼ 1; 2; 3;
r ð3:860 Þ
q oxk
q
oui
rij ¼ ^ij þ
r ^kj ; i; j ¼ 1; 2; 3:
r ð3:870 Þ
q oxk
The Piola-Kirchhoff stress vector of second kind is defined by means of
Cauchy’s stress vector by the relations
n oxi n
d Pi ¼ d Pj ; i ¼ 1; 2; 3; ð3:88Þ
ox
j
dxi ¼ dx ; i ¼ 1; 2; 3:
ox
j j
The relations (3.200 ), (3.800 ) and (3.81) allow to write the relation (3.88) in the
form
oxi
rij nj dA ¼ r n dA ; i ¼ 1; 2; 3; ð3:880 Þ
ox
j jk k
where rij are the components of the Piola-Kirchhoff stress tensor of second kind
(which is called the Piola-Kirchhoff stress tensor too).
106 3 Mechanics of Stresses
Using the relations (3.84), we can express the components of the tensor Tr in
the form
q oxi oxj
rij ¼ r ; i; j ¼ 1; 2; 3; ð3:89Þ
q
ox
k ox
l kl
q
ox
i ox
j
r
ij ¼ rkl ; i; j ¼ 1; 2; 3: ð3:90Þ
q oxk oxl
Unlike the tensor Tr^ , the tensor Tr is a symmetric tensor.
By means of the relations (2.1), we can also write
q ouj oui oui ouj
rij ¼
r
ij
r
il
r
kj þ
r
kl ; i; j ¼ 1; 2; 3; ð3:890 Þ
q oxl oxk oxk oxl
q
ouj oui oui ouj
rij ¼ rij þ ril þ rkj þ rkl ; i; j ¼ 1; 2; 3: ð3:900 Þ
q oxl oxk oxk oxl
Using the relations (3.86), (3.87), (3.89) and (3.90), we can express the Piola-
Kirchhoff tensor one with respect to the other by the relations
oxj
rij ¼ r
^ik ; i; j ¼ 1; 2; 3; ð3:91Þ
ox
k
ox
j
^ij ¼
r rik ; i; j ¼ 1; 2; 3; ð3:92Þ
oxk
which may be written also in the form
ouj
rij ¼ r
^ij r
^ik ; i; j ¼ 1; 2; 3; ð3:910 Þ
ox
k
ouj
^ij ¼ rij þ
r rik ; i; j ¼ 1; 2; 3: ð3:920 Þ
oxk
The Piola-Kirchhoff stress tensors are thus entirely defined.
The formulae (3.900 ) show that, in case of the infinitesimal deformations, the
Piola-Kirchhoff tensor becomes Cauchy’s tensor.
ZZ n
ZZZ
p
i dS0
þ Fi
dV 0
¼ 0; i ¼ 1; 2; 3;
S0
V 0
for an elementary subdomain V 0 in the initial state (Fig. 3.12), where, for the
change of co-ordinates, we took into account the relations (3.80), (3.81) and (3.82)
and the fact that the volume forces (which are given forces), expressed with
respect to the initial state and to the actual state, respectively, verify the relations
1 1
Fi ¼
Fi
; i ¼ 1; 2; 3; ð3:93Þ
q q
corresponding to the equality of the mass forces with regard to the two states.
Using the flux-divergence formula (A.1.960 ), we obtain
ZZZ
o^
rji
þ Fi dV 0 ¼ 0; i ¼ 1; 2; 3;
V 0 oxj
wherefrom, assuming that r ^ji are functions of class C1 , while Fi are functions of
0
class C , we get the equations of equilibrium
o^
rji
þ Fi ¼ 0; i ¼ 1; 2; 3; ð3:94Þ
oxj
where we took into account the fact that, in the initial state, using the material
co-ordinates, the derivatives with respect to time are partial derivatives.
Using the formula (3.52), we can write the equations of equilibrium with the aid
of the tensor Tr in the form
o ox
i
rkj þ Fi ¼ 0; i ¼ 1; 2; 3; ð3:95Þ
oxj oxk
or in the form
Multiplying by oxi =osc , summing with respect to the index i and using the
formula (A.134), we can write the equations of equilibrium in the form
observing that the stress tensor is symmetric ðrsb sc ¼ rsc sb Þ, it results that the
product of this tensor by the skew-symmetric part of the tensor o2 xi =osb osc van-
ishes, so that we can write the equations of equilibrium in the form
orsb sa 1 oxi o2 xi o2 x i oxi o2 xi
þ þ rs b s c þ rs s þ Fsa ¼ 0; a ¼ 1; 2; 3:
osb 2 osa osb osc osc osb osc osc osb b a
Taking into account the formula (A.13600 ). (A.137), we may write the
Eq. (3.96) also in the form (without summation with respect to a)
orsb sa 1 oha 1 ohc
rs s rs s
osb ha osb b a hc osb b a
X3
1 ohb ohb
þ rsb sa þ rsb sb þ Fsa ¼ 0; a ¼ 1; 2; 3: ð3:970 Þ
h osb
b¼1 b
osa
The relations by which one can pass from stresses in orthogonal Cartesian co-
ordinates to stresses in cylindrical co-ordinates read
1 1
rrr ¼ ðr11 þ r22 Þ þ ðr11 r22 Þ cos 2h þ r12 sin 2h;
2 2
1 1 ð3:99Þ
rhh ¼ ðr11 þ r22 Þ ðr11 r22 Þ cos 2h r12 sin 2h;
2 2
rzz ¼ r33 ;
zr
dz
rz
Fz F
r
r Fr
d z z rr
M r
(b)
R
R
FR RR
M
R
R
d
F
d
dR
1
ruh ¼ rhz sin u rrh cos u ¼ ðr11 r22 Þ cos u sin 2h
2
r23 sin u cos h þ r31 sin u sin h þ r12 cos u cos 2h;
1
rhR ¼ rhz cos u rrh sin u ¼ ðr11 r22 Þ sin u sin 2h
2
þ r23 cos u cos h r31 cos u sin h þ r12 sin u cos 2h; ð3:1010 Þ
1 1
rRu ¼ ðrrr rzz Þ sin 2u þ rzr cos 2u ¼ ðr11 þ r22 Þ sin 2u
4 4
1 1
þ ðr11 r22 Þ sin 2u cos 2h r33 sin 2u
4 2
1
þ r23 cos 2u sin h þ r31 cos 2u cos h þ r12 sin 2u sin 2h;
2
where we have put in evidence the linkage with the cylindrical co-ordinates too.
For an elementary subdomain, represented in Fig. 3.17b, the equations of
equilibrium read
orRR 1 orRu 1 orhR
þ þ
oR R ou R sin u oh
1
þ ð2rRR ruu rhh þ rRu cot uÞ þ FR ¼ 0;
R
orRu 1 oruu 1 oruh
þ þ
oR R ou R sin u oh
ð3:102Þ
1
þ ½ðruu rhh Þ cot u þ 3rRu þ Fu ¼ 0;
R
orhR 1 oruh 1 orhh
þ þ
oR R ou R sin u oh
1
þ ð2ruh cot u þ 3rhR Þ þ Fh ¼ 0:
R
References
A. Books
1. Grioli, G.: Mathematical Theory of Elastic Equilibrium (Recent Results). Ergebnisse der
angewandten Mathematik. Springer, Berlin (1962)
2. Mohr, O.: Technische Mechanik, 2nd edn. Verlag von W. Ernst & Sohn, Berlin (1914)
3. Navier, L.M.H.: Résumé des leçons sur l’application de la mécanique à l’établissement des
constructions et des machines, 3rd edn. Avec des notes et des appendices de B. de Saint-
Venant, Paris (1864)
4. Teodorescu, P.P.: Dynamics of Linear Elastic Bodies (Ed. Academiei, Bucuresßti). Abacus
Press, Tunbridge Wells (1975)
References 113
B. Papers
By the ‘‘ut tensio sic vis’’ anagram, published by R. Hooke in 1678, the first model
of a deformable solid, the Hookean model, the ideal (perfect) elastic solid appears;
it is assumed that the solid undergoes a deformation in direct proportion to the
acting force, which disappears at the same time as the latter one (reversible
deformation).
In what follows, we make a general study concerning the construction of a
constitutive law of elastic solids, taking also into account considerations of ther-
modynamical nature, laying stress on the case of linear elasticity.
P
arctan E
O
(a) (b)
4.1 Elastic Models 117
We shall now specify the mathematical model of the elastic body, by adding to the
geometrical and mechanical aspect of the problem, an aspect of physical and
experimental nature; for this end, we shall introduce a constitutive law, a relation
between stresses and strains, a relation that (to avoid computation difficulties),
should be as simple as possible (obviously, within the limits of the admitted
hypotheses concerning the body under consideration).
We shall thus admit as constitutive law Hooke’s law corresponding to the one-
dimensional case. Generally, a stress on a direction yields deformations on all
directions; these deformations can be acquired by superposition of effects (the
linear case), by using the generalized Hooke law (which we shall simply call
Hooke’s law) that shall therefore have a linear form.
Let be a parallelepipedical infinitesimal element within the body, the faces of
which are parallel to the planes of co-ordinates. Let us suppose that only the
normal stress r11 appears (Fig. 4.3a); owing to this stress, a linear strain is born
1
e11 ¼ r11 ; ð4:2Þ
E
where E bears the name of modulus of longitudinal elasticity, as it has been
mentioned in the previous subsection. This elastic constant was introduced and
discussed by Euler; it is known by the name of Young’s modulus, although he did
118 4 Mathematical Models in Mechanics of Deformable Solids
23
11
dx3
dx3
23
23
11
dx 1 32 dx 1
dx2 dx2
(a) (b)
Fig. 4.3 Generalized Hooke’s law: normal stress (a); tangential stress (b)
not define it. Assuming that the stress r11 corresponds to a tension, along the Ox2 -
axis and the Ox3 -axis appear linear strains (shortenings), expressed by
m
e22 ¼ e33 ¼ me11 ¼ r11 ; ð4:20 Þ
E
where m is a coefficient of transverse contraction (Poisson’s ratio). We assume
that, due to the normal stress, there do not appear angular strains.
If on the faces of the parallelepiped appear only the tangential stresses r23 ¼
r32 (Fig. 4.3b), in a plane parallel to the plane Ox2 x3 appears the angular strain
1
c23 ¼ 2e23 ¼ r23 ; ð4:200 Þ
G
where G is the modulus of transverse elasticity. This tangential stress does not give
rise to other strains.
The stresses r22 ; r33 ; r31 ¼ r13 ; r12 ¼ r21 lead to analogue strains. Using the
principle of superposition of effects, the constitutive law (Hooke’s law) reads
1
e11 ¼ ½e11 mðe22 þ e33 Þ;
E
1
e22 ¼ ½e22 mðe33 þ e11 Þ; ð4:3Þ
E
1
e33 ¼ ½e33 mðe11 þ e22 Þ;
E
1
c23 ¼ 2e23 ¼ r23 ;
G
1
c31 ¼ 2e31 ¼ r31 ; ð4:30 Þ
G
1
c12 ¼ 2e12 ¼ r12 ;
G
4.1 Elastic Models 119
where are involved three elastic constants of the material; we shall see that these
constants are linked by a relation, only two of them being independent.
For this, we assume that the principal directions of the stress tensor coincide
with the principal directions of the strain tensor (we will show in Sect. 4.1.3.1 that
this takes place always in case of linearly elastic, isotropic bodies).
In case of a simple shear (see Sect. 3.2.3.3), the state of stress with respect to
the principal axes is expressed in the form
r1 ¼ r2 ¼ s; r3 ¼ 0;
where s is a tangential stress; there corresponds a simple angular strain (see Sect.
2.2.3.3) given by
1
e1 ¼ e2 ¼ c; e3 ¼ 0;
2
where c is an angular strain. Taking into account Hooke’s law (4.3, 4.30 ), we may
write
1 1þm 1
e1 ¼ ðr1 mr2 Þ ¼ s; c¼ s;
E E G
which leads to the relation
E
G¼ ; ð4:4Þ
2ð1 þ mÞ
that takes place between the three elastic constants.
Hooke’s law (4.3), (4.30 ) can be written in the form
1 1 m
eij ¼ ½ð1 þ mÞrij mrkk dij ¼ rij rkk dij ; i; j ¼ 1; 2; 3; ð4:300 Þ
E 2G 1þm
or
1 m
Te ¼ Tr trTr d : ð4:3000 Þ
2G 1þm
l l*
l* l
(a) (b)
The respective diagrams may be conventional or real as, in the computation, the
stress is obtained with respect to the initial cross section (the most used case) or to
the actual cross section. In these diagrams, the strain can be considered in the sense
of Cauchy (e ¼ ðl lÞ=l, where l is the initial length, while l is the actual one) or
in the sense of Hencky (e ¼ lnðl =lÞ); we will consider the first sense (Fig. 4.5a).
The experiments are made in certain standard conditions, that is at the same
temperature, humidity and at the same radioactive radiations; the loading is made
with the same loading velocity, the dimensions of the sample are the same etc.
Only in such conditions the results which are obtained are comparable.
On the characteristic curve, we consider some remarkable points, which mark
the limits of significant parts from the mechanical point of view. We mention thus
the limit stress of elasticity, denoted by re , beyond which the plastic deformations,
which are smaller than the elastic ones, can no more be neglected; theoretically, till
this point the deformation is reversible. The position of this point depends on the
approximation requested by each experiment.
At certain materials, one can put in evidence also a limit stress of propor-
tionality, denoted by rp , till which the curve may be approximated by a straight
line.
Beyond re we find a point called critical stress of plasticity (yield point),
denoted by rY . The deformation process beyond this point, where the plastic
deformations are approximately of the same order of magnitude as the elastic ones,
is called plastic deformation. If the plastic deformations become very large, so that
the elastic ones can be neglected with respect to the first ones, then the corre-
sponding process is called plastic flow. To these two processes correspond two
portions of the characteristic curve, separated by a point of a somewhat conven-
tional position, denoted by rf and called limit stress of flow. For some materials,
the elastic deformations can be neglected after the beginning of the process of
plastic deformation; in this case, the limit of flow coincides practically with the
yield point. This coincidence takes place, e.g., for the materials the characteristic
curve of which has a long horizontal bearing.
Passing from the elastic domain to the plastic one is, customarily, progressive,
gentle. But there exist also some materials for which this passing is sudden; in this
case, the limit of elasticity has no more a conventional character, but corresponds
to a well definite physical phenomenon.
If beyond the yield point the characteristic curve of a material is strictly
increasing, then the respective material has the property of stress hardening;
4.1 Elastic Models 121
A'
O p e
A'
O
(a) (b)
during the plastic deformation, the limit stress of elasticity is increasing for such
materials. Indeed, for an unloading from the point A, e.g., this one is linearly
elastic; beginning again the loading, the diagram which is obtained is perfectly
superposed over that corresponding to the unloading, the new limit stress of
elasticity (corresponding to the point A) being greater than the initial one.
The perfectly elastic unloading represents an idealized schema of the unloading
phenomenon, being—the most times—in a good harmony with the experiment.
Although it does not correspond to the reality in some case, either because an
important phenomenon of hysteresis of the material (by transformation of the
mechanical energy into a caloric one) is put in evidence (Fig. 4.5b) or because the
unloading is an elastic-plastic process.
To represent the property of hardening of a material, the function (4.100 ) must be
continuous, monotone increasing and have a derivative with a finite number of
points of discontinuity.
In some cases, the elastic deformations can be neglected with respect to the
plastic deformations; if the body is a hardening one, then it is called a plastic rigid/
hardening body.
If the stress increases and we reach the point B on the characteristic curve
(Fig. 4.5a), then—the most times—appears a striction (a sudden reduction of the
cross section of the sample), while at the point C takes place the breakage
(corresponding to a conventional diagram); in case of a real diagram, the breakage
is produced at the point C0 . In the considered case, the breakage is a viscous one; if
there is no striction, then the breaking is fragile.
In general, the characteristic curves have their concavity directed towards the
Oe-axis, as in Fig. 4.5; there are also materials for which the concavity is directed
entirely or partially towards the Or-axis, e.g., the natural rubber (Fig. 4.6a).
We have considered above a phenomenon of tension; if the sample is subjected
to compression, then one obtains, the most times, a similar characteristic curve.
But there exist bodies which resist better to compression than to tension, e.g., the
rocks, the concrete etc. (Fig. 4.6b).
122 4 Mathematical Models in Mechanics of Deformable Solids
O
(a) (b)
The study of the characteristic curve of a material is particularly complex and puts
in evidence the most parts of its mechanical properties; in the above considerations
we have made only a short presentation of these properties, being interested to
specify the limit stress of elasticity re , corresponding to the passing from the
elastic zone to the plastic one.
From the Fig. 4.5a, we notice that
e ¼ ee þ ep ; ð4:5Þ
where we put in evidence the elastic and the plastic part of the strain. Taking into
account the relative order of magnitude of the strains, there results various pos-
sibilities of study of the respective bodies in the plastic zone.
The hypothesis of the local dependence of the stresses on the strains leads, both for
loading and unloading, to a relation of the form
rij ¼ rij ðe11 ; e22 ; e33 ; e23 ; e31 ; e12 ; x1 ; x2 ; x3 ; TÞ; i; j ¼ 1; 2; 3; ð4:7Þ
where the functions are considered to be uniform (to a state of strain corresponds a
state of stress), while the absolute temperature T is a parameter.
4.1 Elastic Models 123
The relations (4.7) must be invertible for the elastic bodies, hence
eij ¼ eij ðr11 ; r22 ; r33 ; r23 ; r31 ; r12 ; x1 ; x2 ; x3 ; TÞ; i; j ¼ 1; 2; 3; ð4:70 Þ
this takes place if the Jacobian of the respective relations is non-zero and bounded
oðr11 ; r22 ; . . .; r12 Þ
0 6¼ \1: ð4:700 Þ
oðe11 ; e22 ; . . .; e12 Þ
Thus, the elastic deformation is reversible; if a body is passing for the second
time through a given state of stress, then it passes, at the same time, through the
corresponding state of strain.
If the above functions do not depend on the point, then the body is
homogeneous.
The elastic bodies for which, in certain given conditions (see Sect. 4.1.2.6 too),
there exists an elastic potential are called hyperelastic bodies. The hyperelastic
state depends both on the material and on the conditions of the deformation
process, which can be described only in mechanical terms, obviously the same for
the loading and the unloading of the body. We notice that the theory of elasticity
was developed especially as a theory of hyperelastic bodies, but not any elastic
body is a hyperelastic one. But in the frame of the linear theory, as we will see, any
elastic body is hyperelastic too; such bodies will be called, shortly, elastic bodies.
We mention the existence of some more complex constitutive laws, which
represent more general relations between aspects of geometric-kinematic nature
and aspects of mechanical nature (static or dynamic aspects). Thus, the hypoelastic
bodies are characterized by a one-to-one dependence between the variations of the
stress and the variations of the strain at any moment; for such bodies there does not
exist a natural state, while the components of the stress variation depend on the
component of the strain variation, but not on their velocity variation, because the
hypoelastic bodies do not have viscous properties. The hypoelastic deformation is
reversible and in the neighbourhood of a state without stresses, in case of infini-
tesimal deformations, the respective body has elastic properties. As well, there
exist anelastic bodies for which the deformations are reversible and linearly linked
to stresses, the temporal variable being involved too; the correspondence between
stresses and strains is no more one-to-one.
Another interesting case is that in which the stress tensor is asymmetric (case of
non-Boltzmannian bodies); this leads to the apparition of a supplementary free
rotation, each particle of the body having six degrees of freedom. Thus is born the
theory of asymmetric elasticity; the elastic bodies of Cosserat type are situated in
the frame of this theory.
We admit, in what follows, the existence of a potential W, which depends only
on the components of the tensor Te (and on point, in case on non-homogeneous
bodies), being a function of class C1 with respect to these six variables
W ¼ Wðe11 ; e22 ; . . .; e12 Þ; ð4:8Þ
124 4 Mathematical Models in Mechanics of Deformable Solids
this is Green’s elastic potential (for a hyperelastic body). This form corresponds to
the reversible character of the deformation and to the hypothesis of local depen-
dence. We assume, as well, that the state of stress is given by Green’s formulae [3]
1 oW
rij ¼ 1 þ dij ð!Þ; i; j ¼ 1; 2; 3; ð4:9Þ
2 oeij
the condition imposed to the indices is put because the elastic potential depends
only on six independent variables, the tensor Te being symmetric. Taking into
account Green’s formulae (4.9), we may write
dW ¼ rij deij ð4:100 Þ
or
W ¼ WðI1 ; I2 ; I3 ; w; h; uÞ; ð4:800 Þ
where w; h; u are Euler’s angles which specify the principal axes; if W does not
depend explicitly on these angles, then the body is isotropic.
The formulae (4.8) (or (4.80 ) or (4.800 )) and (4.9) constitute a constitutive law for
the elastic bodies (law of hyperelastic nature).
The elementary work of a force P, which acts upon the elastic body at a point of
position vector r (in the actual state), is given by
4.1 Elastic Models 125
dW ¼ P dr ; ð4:11Þ
taking into account (2.1) and observing that this work is expressed by means of
space co-ordinates, we may also write
dW ¼ P du ¼ Pi dui : ð4:12Þ
Let be a state of strain and stress characterized by the quantities ui ; eij ; rij ,
i; j ¼ 1; 2; 3, due to arbitrary external loads; in case of a differential variation, the
new state of strain and stress is characterized by the quantities
ui þ dui ; eij þ deij ; rij þ drij ; i; j ¼ 1; 2; 3:
Let us consider a parallelepipedic infinitesimal element, with the faces parallel
to the co-ordinate planes, acted upon only by the stress r11 (Fig. 4.3a); in this case,
the elementary work reads
ðr11 dx2 dx3 Þdðe11 dx1 Þ ¼ ðr11 dx2 dx3 Þde11 dx1 ¼ r11 de11 dV;
where we took into account that the distance between the two faces of the par-
allelepiped which are parallel to the plane Ox2 x3 vary by e11 dx1 . Analogically, if
on the faces of the elementary parallelepiped appear only the tangential stresses
r23 ¼ r32 (Fig. 4.3b), then the elementary work is given by
ðr23 dx1 dx3 Þdða23 dx2 Þ þ ðr32 dx1 dx2 Þdða32 dx3 Þ
¼ r23 dc23 dV ¼ 2r23 de23 dV ¼ ðr23 de23 þ r32 de32 ÞdV;
where we took into account the considerations made in Sect. 2.2.1.3 (see Fig. 2.6
too). By analogous procedure for the stresses r22 ; r33 ; r31 ¼ r13 ; r12 ¼ r21 , we get
the elementary work, corresponding to an element of volume equal to unity, in the
form
dW ¼ rij deij ; ð4:13Þ
this is called the elementary work of deformation (the volume density of strain
energy).
Taking into account (4.100 ), we notice that (4.13) has also another mechanical
significance (corresponding to the elastic potential W); for this reason, one uses the
same notation.
The dual expression
dWc ¼ eij drij ð4:14Þ
is called the elementary complementary work of deformation; if eij and rij are seen
as independent parameters, we notice that
dW þ dWc ¼ dðeij rij Þ: ð4:15Þ
126 4 Mathematical Models in Mechanics of Deformable Solids
In what follows, we will consider only quantities with respect to the unit
volume, without specifying this; the values corresponding to the whole domain
occupied by the body are obtained by integration on V. The elementary variations
from a state to another one are denoted by d, being differentials (exact or not ones);
the non-elementary variations are obtained by integrating elementary variations on
paths C in the thermodynamic parameters space (figurative space).
dU þ dT þ dV ¼ dW þ dQ þ dQ0 ; ð4:18Þ
128 4 Mathematical Models in Mechanics of Deformable Solids
In particular, in the case of an isolated system, there results, for any elementary
process,
ZZZ
dUdV ¼ 0; ð4:20Þ
V
while the internal energy does not vary; the result remains valid for a non-ele-
mentary process (by integration on an arbitrary path in the figurative space)
Z ZZZ
dUdV ¼ 0: ð4:200 Þ
C V
If two systems are separated by a frontier which allows the heat to pass freely
from one to the other, then we say that the two systems are in thermal contact.
Otherwise, if the frontier is preventing any change of heat, then one says that it is
thermic insulating; if a system is bounded by a thermic insulating frontier, then it
is thermic insulated.
A thermodynamic process which takes place in a thermic insulated system is
called adiabatic; we may write
dQ ¼ 0; dU ¼ dW; ð4:21Þ
as well as
ZZZ ZZZ ZZZ
dQdV ¼ 0; dUdV ¼ dWdV: ð4:210 Þ
V V V
In this case, there will not exist a flux of heat, the system being in thermic
equilibrium.
We notice also that the first principle of thermodynamics allows to establish the
mechanical equivalent of heat; thus from a dimensional point of view, we have
½calorie ¼ ML2 T2 .
4.1 Elastic Models 129
for the variation di S, the equality taking place only for the reversible processes; in
case of the inequality the process is irreversible.
From (4.22)–(4.230 ) it result that
dQ
dS ¼ ; ð4:24Þ
T
for a reversible system, as well as
130 4 Mathematical Models in Mechanics of Deformable Solids
dQ
dS [ ; ð4:240 Þ
T
for an irreversible system; these relations represent the second principle of
thermodynamics.
For the whole system, one can write
ZZZ ZZZ
dQ
dSdV dV: ð4:2400 Þ
V V T
In case of a reversible system, the relation (4.24) for the function of state S
shows, from the mathematical point of view, that the elementary flux of heat dQ
admits as integral factor the inverse of the absolute temperature T.
This principle can be obtained starting from various hypotheses of physical
nature. Thus, Clausius obtains this result assuming that it is not possible without
the intervention of another process. But Kelvin obtains it starting from another
idea: it is impossible that the only consequence of a sequence of changes of a
system does not consist in the whole transformation of a quantity of heat in work.
Carathéodory uses a more general hypothesis; he assumes that it is impossible to
reach all the thermodynamic states in the neighbourhood of a given arbitrary initial
state by an adiabatic process. The second principle of thermodynamics is equiv-
alent to the impossibility to construct a perpetum mobile.
The two principles stated above form the basis of classical thermodynamics;
they are experimentally justified for a real system at a macroscopic scale.
The thermodynamic processes for which T ¼ const, hence
dT ¼ 0; ð4:25Þ
We get
dH ¼ dU dW dWc ¼ dQ dWc ¼ dWc þ TdS; ð4:36Þ
where we took into account the formula (4.15) and the relations (4.19), (4.24);
moreover, this is a new form of the first principle of thermodynamics. We notice
that the elementary work is, in our case, just the elementary work of deformation
introduced in Sect. 4.1.2.2.
We obtained thus
dG ¼ dWc S dT; ð4:37Þ
for a reversible process, and
dG\ dWc S dT; ð4:370 Þ
We notice that, in case of a quasistatic process of elastic deformation one can write
dU ¼ rij deij þ TdS; ð4:40Þ
4.1 Elastic Models 133
as well as
dF ¼ rij deij SdT: ð4:400 Þ
We observe that, in each of these two cases, we have to deal only with seven
independent variables e11 ; e22 ; e33 ; e23 ¼ e32 ; e31 ¼ e13 ; e12 ¼ e21 and S or T,
respectively. The process is reversible, so that in the above formulae we have total
differentials; it results that
1 oU oU
rij ¼ ð1 þ dij Þ ð!Þ; i; j ¼ 1; 2; 3; T¼ ; ð4:41Þ
2 oeij oS
or
1 oF oF
rij ¼ ð1 þ dij Þ ð!Þ; i; j ¼ 1; 2; 3; S¼ ; ð4:410 Þ
2 oeij oT
or a function of state
F ¼ Fðe11 ; e22 ; . . .; e12 ; T0 Þ; ð4:420 Þ
where S0 and T0 are the constant values of the entropy and of the absolute tem-
perature, respectively, corresponding to the conditions in which the process is
developed.
The first case is of interest for the dynamic problems, where rise elastic
vibrations of small amplitude, the process being sufficiently quick, so that the
change of heat be practically not possible; the second case, important for static
problems, corresponds to slow loadings and unloadings of the body (quasistatic
processes), when it exists time for absorption or release of heat, so that the tem-
perature does remain constant.
In both cases (for S ¼ S0 or for T ¼ T0 ) there exists a function of state called
elastic potential or volume density of strain energy; as state of reference one
chooses that state in which both the stresses and the strains vanish at a temperature
T ¼ T0 . The elastic bodies which have this property are called hyperelastic.
If the process of unloading of an elastic solid is effected on the same way as the
process of loading, then the variation of the work on the whole cycle vanishes; but,
in general, the process of unloading takes place on another way, dW is not an exact
differential, while the first principle of thermodynamics allows to state that
134 4 Mathematical Models in Mechanics of Deformable Solids
Z Z
dW ¼ dQ 6¼ 0; ð4:43Þ
C C
where the integral is taken along the way travelled through. This leads to the
increase of the entropy and to decrease of the free energy, as well as to the
transformation of a part of the work of loading in heat (dissipation of the energy).
In case of the hyperelastic deformation, the existence of a function of state
which plays the rôle of elastic potential shows that for any cycle one has
Z Z
dW ¼ dQ ¼ 0; ð4:430 Þ
C C
as well as
dG ¼ eij drij SdT; ð4:440 Þ
in each of these cases we have to deal with seven independent variables, i.e.:
r11 ; r22 ; r33 ; r23 ¼ r32 ; r31 ¼ r13 ; r12 ¼ r21 and S or T, respectively. The process
is reversible, so that in the above formulae we have exact differentials and we may
write
1 oH oH
eij ¼ ð1 þ dij Þ ð!Þ; T¼ ð4:45Þ
2 orij oS
4.1 Elastic Models 135
or
1 oG oF
eij ¼ ð1 þ dij Þ ð!Þ; S¼ : ð4:450 Þ
2 orij oT
In case of an adiabatic process (hence, isentropic), it results, from (4.39), that
one can build a function of state
H ¼ Hðr11 ; r22 ; . . .; r12 ; S0 Þ; ð4:46Þ
where the condition imposed to the indices is due to the fact that the comple-
mentary elastic potential depends only on six independent variables, the tensor Tr
being symmetric; taking into account Castigliano’s formulae (4.48), one may write
136 4 Mathematical Models in Mechanics of Deformable Solids
summing now for all the value i; j ¼ 1; 2; 3. The introduction of the dual expres-
sion (4.14) is thus justified.
In the case of a one-dimensional experiment, the elastic potentials W and Wc
have an interesting significance, in connection with the characteristic curve of the
material. Thus, if we stop at a point Pðe; rÞ on this curve (Fig. 4.7), then the elastic
potential W represents the area of the domain between the characteristic curve, the
Oe-axis and the parallel at P to the Or-axis; as well, the complementary elastic
potential Wc represents the area of the domain between the characteristic curve, the
Or-axis and the parallel at P to the Oe-axis. These results are valid excepting an
additive constant; we assumed that the elastic potentials are equal to zero for a
vanishing state of strain and stress. We notice that, in this case,
W þ Wc ¼ re; ð4:50Þ
then we have
eij drij ¼ rij deij ;
wherefrom
1
rij deij ¼ d rij eij :
2
this is equivalent to
dS 0; U ¼ const, ð4:520 Þ
for any neighbourhood, in which we will have an entropy S þ dS smaller than that
corresponding to the stable state of equilibrium.
One can further state that for the stable thermodynamic equilibrium of an
isolated system it is necessary and sufficient that the internal energy be a minimum
with respect to all the neighbouring states which have the same entropy; one can
write
U ¼ min; S ¼ const; ð4:53Þ
and the minimum of which can be thus considered equal to zero, must be a positive
definite function.
If we take into account (4.52) or (4.53), then the relation (4.30) shows that
Helmholtz’s function F reaches also a minimum for the stable equilibrium of the
thermodynamic system. In this case too, the corresponding elastic potential will be
a positive definite function.
As a consequence of these results, we assume, in what follows, that the elastic
potential W, corresponding to the elastic body with hyperelastic properties, is a
positive definite function. This corresponds to the fact that the deformation process
needs an expense of work.
where we have observed that the elastic potential depends on six independent
variables. Taking into account Green’s formulae (4.9), we may write this potential
in the form (4.510 ), the sum being made for all the values i; j ¼ 1; 2; 3; using the
considerations in Sect. 4.1.2.6, it results the relation (4.51).
Comparing the relations (4.510 ) and (4.54), one obtains the constitutive law
rij ¼ Hijkl ekl ; i; j ¼ 1; 2; 3; ð4:56Þ
of the form (4.6); one can obtain this result also by using Green’s formulae (4.9).
Observing that the tensors Tr and Te are symmetric, it results that
4.1 Elastic Models 139
hence, from the 34 ¼ 81, components of the tensor TH remain only 36 independent
coefficients. As well, from the expression (4.54), one sees that
Hklij ¼ Hijkl ; i; j; k; l ¼ 1; 2; 3; ð4:570 Þ
taking into account the symmetry of the tensor Te , the matrix of the tensor TH
reads
2 3
H1111 H1122 H1133 2H1123 2H1131 2H1112
6 H2211 H2222 H2233 2H2223 2H2231 2H2212 7
6 7
6 H3311 H3322 H3333 2H3323 2H3331 2H3312 7
TH 66 H2311 H2322 H2333
7: ð4:5600 Þ
6 2H2323 2H2331 2H2312 7
7
4 H3111 H3122 H3133 2H3123 2H3131 2H3112 5
H1211 H1222 H1233 2H1223 2H1231 2H1212
Assuming that
det½Hijkl 6¼ 0; ð4:58Þ
where Cijkl are 81 coefficients of elastic deformation (of compliance), from which
21 are distinct; in this case, the elastic potential can be written in the form
1
W ¼ Cijkl rij rkl ð4:540 Þ
2
too, as a positive definite quadratic form in stresses, corresponding to the com-
plementary elastic potential.
Castigliano’s formulae
1 oW
eij ¼ ð1 þ dij Þ ð!Þ; i; j ¼ 1; 2; 3; ð4:59Þ
2 orij
take place.
140 4 Mathematical Models in Mechanics of Deformable Solids
The necessary and sufficient condition for the quadratic form (4.54) be positive
definite is given be Sylvester’s criterion; the diagonal minors of the matrix of the
coefficients Hijkl must be positive, i.e.
1 1 H1111 H1122 1 2
H1111 [ 0; ¼ ðH1111 H2222 H1122 Þ [ 0; . . .; ð4:60Þ
2 4 H2211 H2222 4
order), one cannot write interesting relations for the coefficients in the latter system
of axes; assuming that the system of axes Mx1 x2 x3 is just the system of principal
axes, there result the conditions
These results are valid only for the point M and may not be valid at another point.
Among the types of the most important anisotropic elastic bodies which have
this property we mention the orthotropic bodies, for which the planes of elastic
symmetry are principal planes, at the point of intersection of them, as well as the
bodies with transverse isotropy for which the monotropy axis is a principal one at
all the points of the axis; as well, for all the points of the isotropic bodies the
principal axes of the tensors Te and Tr coincide. For the bodies with a plane of
elastic symmetry, as well as for the bodies with an axis of elastic symmetry of the
third or of the fourth order such a property does no more hold.
If the mechanical properties of an elastic body maintain their form after symmetric
directions with respect to a plane, we say that this body is with a plane of elastic
symmetry. Choosing the plane x3 ¼ 0 as a plane of elastic symmetry, it results that,
passing to the system of axes Ox01 x02 x03 , specified by the relations
142 4 Mathematical Models in Mechanics of Deformable Solids
Taking into account (4.67, 4.670 ), we notice that the relations (4.68, 4.680 ) hold
simultaneously if and only if H1123 ¼ H1131 ¼ 0; repeating this reasoning for the
other relations which form Hooke’s law too, we state that the elastic coefficients
must verify the conditions
Hii23 ¼ Hii31 ¼ H23ii ¼ H31ii ¼ 0ð!Þ; i ¼ 1; 2; 3; ð4:69Þ
We assume now that the plane x1 ¼ 0 is also a plane of elastic symmetry; in this
case, the elastic coefficients must verify the conditions
4.1 Elastic Models 143
In case of a body with central symmetry, the mechanical properties must be the
same in any direction around a point of the body. In what follows we start from the
case of the orthotropic body, considered above.
Effecting a rotation of p=2 with respect to the Ox1 -axis, specified by the
relations
as well as a rotation of p=2 with respect to the Ox2 -axis, specified by the relations
hold, so that the body does remain with the same mechanical properties with
respect to the new obtained directions.
Now, effecting a rotation of p=4 with respect to the Ox3 -axis, specified by the
relations
pffiffiffi pffiffiffi
2 2
x01 ¼ ðx1 þ x2 Þ; x02 ¼ ðx1 x2 Þ; x03 ¼ x3 ;
2 2
one states, analogously, that the conditions
144 4 Mathematical Models in Mechanics of Deformable Solids
1
H2323 ¼ ðH1111 H1122 Þ ð4:730 Þ
2
must be fulfilled, so that the mechanical properties do remain the same with
respect to the new directions. Denoting
H1111 ¼ k; H2323 ¼ l; ð4:74Þ
we observe that one can write the constitutive law of a linearly elastic body with
central symmetry (Hooke’s law) in the form
rij ¼ kell dij þ 2leij ; i; j ¼ 1; 2; 3; ð4:75Þ
r11 ¼ kh þ 2le11 ;
r22 ¼ kh þ 2le22 ; ð4:7500 Þ
r33 ¼ kh þ 2le33 ;
Summing in both members, we find that the first invariants of the two tensors are
linked by the relation
4.1 Elastic Models 145
we remember that the principal axes of the stress tensor coincide with the principal
axes of the strain tensor.
Solving with respect to the strains, one can write Hooke’s law also in the form
k 1
eij ¼ Hdij þ rij ; i; j ¼ 1; 2; 3: ð4:77Þ
2lð3k þ 2lÞ 2l
We assume that a simple axial tension, as it has been defined in Sect. 3.2.3.2, takes
place; in this case, the state of stress is expressed in the form
r11 ¼ r ¼ const; r22 ¼ r33 ¼ r23 ¼ r31 ¼ r12 ¼ 0: ð4:78Þ
We notice that, in the absence of the volume forces, the equations of equilibrium
(3.610 ) are identically satisfied.
Hooke’s law (4.77) leads to the state of strain
kþl k
e11 ¼ r; e22 ¼ e33 ¼ r; e23 ¼ e31 ¼ e12 ¼ 0;
lð3k þ 2lÞ 2lð3k þ 2lÞ
ð4:79Þ
which satisfies Saint-Venant’s conditions of continuity (2.68).
Let be a finite right cylinder of axis Ox1 , subjected to the simple axial load
considered above (Fig. 4.8). We notice that this takes place if the cylinder is acted
upon only on the end cross sections by external loads of intensity equal to r. If
r [ 0, (e11 [ 0; e22 ¼ e33 \0), then we have to do with a simple tension, for which
the cylinder is extended, while the cross section is diminished (the case in
Fig. 4.8); if r\0, (e11 \0; e22 ¼ e33 [ 0), then there corresponds a simple com-
pression, case in which the cylinder is shortened, while its cross section increases.
Taking into account the form (4.2), (4.20 ) of Hooke’s law, it results that
lð3k þ 2lÞ k
E¼ ; m¼ ; ð4:80Þ
kþl 2ðk þ lÞ
these are the technical elastic constants of the material.
σ
σ O x1
146 4 Mathematical Models in Mechanics of Deformable Solids
Table 4.1 Numerical data concerning the elastic constants of some important materials
Material 105 E 105 G 105 K 105 k m
ðMPa) ðMPa) ðMPa) ðMPaÞ
Grey cast iron 1.20 0.480 0.800 0.480 0.25
Carbon steel 2.10 0.840 1.400 0.840 0.25
Tough copper 1.10 0.416 1.018 0.740 0.32
Bronze 1.15 0.432 1.127 0.838 0.33
Brass 0.95 0.349 1.130 0.900 0.36
Laminated aluminium 0.69 0.259 0.676 0.502 0.33
Lead 0.17 0.059 0.354 0.314 0.42
Glass 0.56 0.224 0.373 0.224 0.25
Concrete 200 0.21 0.089 0.106 0.089 0.17
Rubber 0.00008 0.00003 0.00040 0.00040 0.47
Celluloid 0.00020 0.00007 0.00022 0.00017 0.35
148 4 Mathematical Models in Mechanics of Deformable Solids
2 3
k þ 2l k k 0 0 0
6 k k þ 2l k 0 0 0 7
6 7
6 k k k þ 2l 0 0 0 7
TH ¼ 6
6 0
7: ð4:890 Þ
6 0 0 2l 0 0 7
7
4 0 0 0 0 2l 0 5
0 0 0 0 0 2l
Using the spheric tensors of strain and stress, as well as the deviators of strain and
stress, which have been introduced in Sects. 2.2.1.2 and 3.2.1.3, we notice that
Hooke’s law can be decomposed in the form
r0 ¼ 3Ke0 ; ð4:92Þ
thus, some important properties of the elastic constants of the material, as well as
an interesting decomposition of its deformations are put into evidence. One can
state that the constitutive law (4.92) corresponds to a change of volume of the
elastic solid, without a change of form, while the constitutive law (4.920 ) corre-
sponds to a change of form of the elastic solid, the volume remaining constant
(because I10 ¼ 0).
We notice that one can write the elastic potential in the form
1 2l 2 1 h
W¼ kþ h þ l ðe22 e33 Þ2 þ ðe33 e11 Þ2
2 3 3
i
2
þðe11 e22 Þ þ 6ðe223 þ e231 þ e212 Þ
too; taking into account the results given in Sect. 2.2.1.2 concerning the deviator
of strains, it results
W ¼ Wt þ Wf ; ð4:93Þ
where
1
Wt ¼ KI12 ð4:94Þ
2
is the unit work of deformation in case of a change of volume without a change of
form, while
Wf ¼ 2lI20 ð4:95Þ
is the unit work of deformation in case of a change of form, the volume remaining
constant.
Analogically, we can express the two works, in stresses, in the form
1 2
Wt ¼ J ; ð4:940 Þ
18K 1
1 0
Wf ¼ J : ð4:950 Þ
2l 2
We have the relations
150 4 Mathematical Models in Mechanics of Deformable Solids
J1 ¼ 3KI1 ; ð4:96Þ
If, for a given state of stress of the deformable body, we have H ¼ 0, then it
follows, from the formula (4.76), that h ¼ 0 whatever Poisson’s ratio m would be;
we have to do with a state of incompressible deformation (state of isochore
deformation). But this is no more a property depending on the nature of the
material, as in the preceding case; such a state of deformation may occur in any
body, depending on the respective state of strain and stress.
Let us admit that the hypothesis of homogeneity of the body is no longer valid.
Generally, we can have to deal with two types of non-homogeneous bodies: bodies
with a discontinuous non-homogeneity (e.g., stratified bodies) and bodies with a
continuous non-homogeneity [22, 23].
In the first of these cases, distinct elastic constants are introduced for every
subdomain occupied by a homogeneous part of the given elastic body (which
occupies the whole considered domain). The coefficients referring to other prop-
erties of the material should be similarly specified.
In the case of bodies with a continuous non-homogeneity, the elastic constants
become functions of point; thus, we have E ¼ Eðx1 ; x2 ; x3 Þ, m ¼ mðx1 ; x2 ; x3 Þ and,
consequently, k ¼ kðx1 ; x2 ; x3 Þ, l ¼ lðx1 ; x2 ; x3 Þ. Besides, the density can be
q ¼ qðx1 ; x2 ; x3 Þ.
Because the coefficient of transverse contraction varies, in general, between
close limits, one considers often that m ¼ const, remaining with only one elastic
coefficient function of point.
It is useful to introduce the reduced stresses
1
rij ¼ rij ; i; j ¼ 1; 2; 3; ð4:100Þ
E
which are non-dimensional quantities. Hooke’s law may be written in the form
eij ¼ ð1 þ mÞrij mrll dij ; i; j ¼ 1; 2; 3; ð4:101Þ
E ¼ E0 e f ; ð4:104Þ
k ¼ k0 eh ; ð4:108Þ
where h ¼ hðx1 ; x2 ; x3 Þ is a continuous function, differentiable as many time as
necessary; as in the previous case, if h ¼ f , then we shall have k ¼ const.
1
err ¼ ½rrr mðrhh þ rzz Þ;
E
1
ehh ¼ ½rhh mðrzz þ rrr Þ; ð4:110Þ
E
1
ezz ¼ ½rzz mðrrr þ rhh Þ;
E
1 1 1
chz ¼ rhz ; czr ¼ rzr ; crh ¼ rrh ; ð4:1100 Þ
G G G
while in the case of spherical co-ordinates we can write
1
eRR ¼ ½rRR mðruu þ rhh Þ;
E
1
euu ¼ ½ruu mðrhh þ rRR Þ; ð4:111Þ
E
1
ehh ¼ ½rhh mðrRR þ ruu Þ;
E
1 1 1
cuh ¼ ruh ; chR ¼ rhR ; cRu ¼ rRu : ð4:1110 Þ
G G G
with
E
b¼ a: ð4:116Þ
1 2m
One can use the results obtained above for anisotropic bodies too; we take different
coefficients of linear dilatation for the directions which are considered.
More general, if a body is subjected to initial strains e0ij , then the linear constitutive
law becomes
1
eij e0ij ¼ ½ð1 þ mÞrij mrll dij ; i; j ¼ 1; 2; 3: ð4:117Þ
E
We can express these relations also in the form
If the initial strains verify the equations of continuity, then do not appear
stresses due to them. Otherwise, they represent incompatibilities and there appear
stresses, called initial stresses.
The results above are used for the distribution of stresses in the multiply
connected bodies, with the aid of the cuts. As well, they can be used also for the
bodies with initial stresses due to the processing of the material of which is made
the body, due to the contraction during the solidification, due to errors by
assembling etc. They can occur also in case of defects in crystals (e.g., disloca-
tions), which introduce incompatibilities.
In the following we do not take into consideration the initial stresses, assuming
that to a unloaded body corresponds a state of zero stress. In general, to can speak
about the state of stress of a body, one must admit that there exists a state of that
body—the natural state of stress—for which all the stresses vanish; as well, to can
speak about the state of strain of a body, one must choose a certain form of the
body—the initial state of strain—for which all the quantities which characterize
the deformation are, by definition, equal to zero.
It is necessary to suppose that neither the external loads vanish for the natural
state of stress, nor that this state corresponds to the initial state of strain. However,
this is the hypothesis assumed by Lamé and which we admit also in that follows.
4.1 Elastic Models 155
while the Eulerian deformation tensor TE and Cauchy’s stress tensor Tr have
been introduced.
The deformable solids which, by unloading, do not return to the initial form are
inelastic solids (the thermodynamic process is irreversible); their properties of
plasticity and viscosity are thus put in evidence. In what follows we make a concise
presentation of some inelastic deformable solids, returning then later to these
problems; the scope of these considerations is only that to have a general view on
the constitutive laws of deformable solids, laws which complete their mathemat-
ical model.
In the model considered in the preceding paragraph, we have taken out from the
complex phenomenon the principal part, neglecting other aspects; thus, the influ-
ence of peak loads, hence the consideration of plastic properties of the material
can play an essential rôle.
156 4 Mathematical Models in Mechanics of Deformable Solids
We introduce some models in which appear plastic properties, in the frame of a one-
dimensional loading, presenting characteristic curves of the idealized material.
Thus, the perfect plastic (rigid plastic) body (Fig. 4.9a) will be represented by a
rigid solid, who slides with friction on a horizontal plane (Fig. 4.9b); when the force
P attains the maximal intensity of the sliding friction, the body begins to move,
arising a displacement D (which corresponds to the limit stress of elasticity re ).
One obtains thus Saint-Venant’s model; the respective constitutive law reads
r ¼ re ; e 0: ð4:120Þ
Linking in parallel a Hooke model with a Saint-Venant model, one obtains the
model which corresponds to the plastic rigid/linearly hardening body (the hard-
ening modulus being just the elasticity modulus E1 of the spring of the model)
(Fig. 4.10a and b); one may write the constitutive law
r ¼ re þ E1 e; e 0: ð4:121Þ
As well, linking in series the above elementary models, one obtains the model
of the elastic-perfect plastic body (Fig. 4.11a and b); it results the constitutive law
Ee for 0 e rEe ;
r¼ ð4:122Þ
re for e rEe :
Using the functions introduced by the formulae (A.165) and (A.167), we can
write the above constitutive laws, in the following remarkable forms
r ¼ re hðeÞ; ð4:1200 Þ
r ¼ EehðeÞ E e h e
h Er
i E
re
e
¼ Ee hðeÞ h e þ re h e
E E
re
¼ E eþ e ; ð4:1220 Þ
E þ
4.2 Inelastic Bodies 157
O
(a) (b)
O
(a) (b)
arctan E 1 P
O
(a) (b)
arctan E
O
(a) (b)
158 4 Mathematical Models in Mechanics of Deformable Solids
re
re
r ¼ EehðeÞ ðE E1 Þ e h e
E E
re
¼ Eeþ ðE E1 Þ e ; ð4:1230 Þ
E þ
which can be useful in various practical computations.
Although their intuitivity, these models could not be developed for bodies with
more complex plastic properties, because between the phenomenon of plastic flow
and the sliding phenomenon there exist important differences.
More complex constitutive laws are of the form (4.100 ) (or (4.1000 )) and can
correspond to bodies with elastic and plastic properties.
In the considerations made in the preceding subsection we notice that the defor-
mation of a solid body is composed by an elastic part and by a plastic one, being of
the form (4.5). A special rôle is played by the moment at which appears the plastic
deformation ep , hence the position of the point of ordinate re . The construction of a
theory of plasticity in the one-dimensional case contains thus a constitutive law, as
well as the specification of the point at which one passes from the elastic zone to
the plastic one, by means of the limit stress of elasticity re .
Sometimes, the plastic deformation can be neglected with respect to the elastic ones;
other times, they are of the same order of magnitude, but at times the plastic defor-
mations are greater then the elastic ones. The study of these states of plastic deformation
can be effected using various theorems of plasticity. In the actual state, each theory
can be applied to a small number of materials or phenomena and only between certain
limits of variation of the parameters on which depend the deformations.
A basic concept in a three-dimensional theory of plasticity is that of loading
surface, which represents the frontier between the elastic and the plastic domains
in the stress space; this surface is correspondent of the limit stress of elasticity of
the one-dimensional case. We notice that the stress space rij is a space with 9
dimensions; one can use also a space with only 6 dimensions, because the tensor
Tr is symmetric. Frequently, one uses the deviators of stresses and strains; one is
thus led to a space of deviators which has 5 dimensions. The most times, one uses
the three-dimensional space of the principal normal stresses r1 ; r2 ; r3 . In this
space, the loading surface, which is an open surface, puts in evidence the domain
of the elastic state, containing the origin, and an external domain, which has not
mechanical interpretation in case of a perfect plastic material or which represents
possible plastic states in the future, in case of a hardening material; the points of
the loading surface correspond to the plastic states of stress.
The components of the strain tensor will be written in the form
where the elastic part and the plastic one of the strain are put in evidence.
The equation of the loading surface is given by
then a unloading takes place, the increasing of stresses leading to the passing from
a plastic state to an elastic one, the plastic deformation remaining the same. If
of
f ¼ 0; r_ ij ¼ 0; ð4:1260 Þ
orij
then one has a neutral loading, which corresponds to the passing from a plastic
state to another plastic one, the plastic deformation remaining the same. If
of
f ¼ 0; r_ ij [ 0; ð4:12600 Þ
orij
then a loading takes place, in which one passes from a plastic state of stress to
another state of plastic stress, the plastic deformation varying too; the last con-
dition is without sense in case of perfect plastic materials.
A variation drij of the stress is represented by a vector with a point of appli-
cation on the loading surface. In case of a unloading, this vector is directed
towards the interior of the surface, in case of a neutral loading it is tangent to the
loading surface, while in case of a loading it is directed towards its exterior; the
vector of =orij is always directed towards the exterior of the loading surface
(Fig. 4.13).
Initially, for the plastic non-deformed material, the loading surface depends
only on the components of the stress tensor. Because for the most materials the
hydrostatic pressure has a negligible influence on the plastic deformation, the
loading depends, especially, on the deviator tensor of stresses r0ij . For an isotropic
body, the equation of this surface depends only on the invariants of the deviator,
being of the form
this equation represents, in the space Or1 r2 r3 , a cylinder with the generatrices
parallel to the straight line r1 ¼ r2 ¼ r3 , normal to the octahedral plane
r1 þ r2 þ r3 ¼ 0. A relation of the form (4.127), corresponding to the initial
surface of loading, represents a condition of plasticity.
In particular, one can use Mises’s condition, valid for many polycrystalline
materials, for which—during the plastic flow—the relation
160 4 Mathematical Models in Mechanics of Deformable Solids
0 unloading ij
ij
neutral loading
1 2
O 0 O 0
(a) (b)
O 0 O 0
(a) (b)
Z
ðrij r0ij Þ deij 0: ð4:130Þ
_
ABC
Because the work corresponding to the strains eeij vanishes along the closed
cycle A ! B ! C ! A, it results (stability in great)
I
ðrij r0ij Þ depij 0: ð4:1300 Þ
In case of a hardening material, the equality takes place only in the absence of
plastic deformations. Because the plastic deformation\takes place only on the path
B ! C, the last inequality becomes (stability in small)
We consider now another cycle with the initial point B. For the loading process
B ! C we may write
drij deij 0; ð4:131Þ
r_ ij e_ pij 0: ð4:1320 Þ
The equality takes place only in case of neutral loadings; as well, from the
above relations, it results that the loading surface can never be concave.
We notice also that, beside regular points, the loading surface can have sin-
gular points (vertices etc.); concerning these points must be made a special study.
To complete the constitutive law of a body with plastic properties, we intro-
duce, beside a condition of plasticity, a law of hardening too, that is relations
which describe the mode of deformation and displacement of the loading surface
during the plastic deformation. The most used hardening law is that of isotropic
hardening, on the basis of which, during the plastic deformation, the loading
surface is uniformly deformed in all directions; analytically, one can write such a
law in the form
f ðJ_ 2 ; J_ 3 Þ ¼ k2 ; ð4:133Þ
where k ¼ kðqÞ is an increasing variable parameter, while q is a positive scalar
(Fig. 4.18a). For instance, if the loading surface is that corresponding to Mises’s
4.2 Inelastic Bodies 163
(a) (b)
where aij are the co-ordinates of the centre of the loading surface while k ¼ const
(Fig. 4.18b); e.g., we can have aij ¼ cepij , where c is a characteristic constant of the
material. One can use also a hardening law which corresponds to an expansion and
a translation of the form
which can often correspond better to the reality; however, no one of the mentioned
hardening laws covers all the forms taken by the physical reality.
As well, the third element which characterizes the constitutive law is a law of
flow, i.e. a relation between stresses and strains; in such a relation often appear the
deviators of the respective tensors.
A law of flow is, in general, a differential (non-integrable) relation between
stresses and strains; thus, it constitutes a relation between the infinitesimal increase
of the stresses and strains, stresses and some parameters corresponding to the
plastic state. We make following hypotheses:
(i) The body is isotropic.
(ii) The volume strain is elastic and small, being proportional to the mean nor-
malstress; we have
r0 ¼ 3Ke0 ð4:134Þ
or
dr0 ¼ 3Kde0 : ð4:1340 Þ
164 4 Mathematical Models in Mechanics of Deformable Solids
and
(iv) The stress deviator Dr is in direct proportion with the deviator of the plastic
increasing strains
where dk is an infinitesimal scalar coefficient. This hypothesis shows that the state
of stress determines the instantaneous increase of the plastic strain.
Taking into account (4.137), we notice that e0p p
ij ¼ eij , so that one can write the
relation (4.138) in the form
introducing (4.139) and using the relation (2.55), (2.57), (2.570 ), we get
taking into account (4.136) and introducing the stress and strain deviators, we can
write the law of flow (4.141) in the form
1 dW p
de0ij ¼ dr0ij þ 0 r0ij ; i; j ¼ 1; 2; 3; ð4:142Þ
2l 2J2
4.2 Inelastic Bodies 165
1
de0 ¼ dr0 : ð4:1420 Þ
3K
In case of infinitesimal deformations, it results the law
W_p 0
e_ ij ¼ e_ 0ij þ r ; i; j ¼ 1; 2; 3; ð4:143Þ
2J20 ij
1
e_ 0 ¼ r_ 0 : ð4:1440 Þ
3K
These relations constitute the Prandtl-Reuss law of flow. In this case, it is cus-
tomary to use Mises’s condition of plasticity; the law of flow (4.144) becomes
1 0 W _p
e_ 0ij ¼ r_ ij þ 2 r0ij ; i; j ¼ 1; 2; 3; ð4:14400 Þ
2l 2se
_ p is given by
where W
_ p ¼ rij e_ pij ¼ r0ij e_ 0p
W ij : ð4:145Þ
These relations are useful in the case in which the elastic deformations cannot be
neglected.
Assuming that the modulus of transverse elasticity tends to infinite ðl ! 1Þ,
that is if the material becomes rigid in the elastic domain, one obtains the Saint-
Venant-Lévy-Mises law of flow in the form
W_p 0
e_ 0ij ¼ r ; i; j ¼ 1; 2; 3; ð4:146Þ
2s2e ij
e_ 0 ¼ 0: ð4:1460 Þ
From (4.146) it results that the principal axes of the stress and of the strain velocity
tensors coincide. One obtains
W_p 0 0
r0ij e_ 0ij ¼ r r
2s2e ij ij
too; using Mises’s condition of plasticity, it results
_ p ¼ r0ij e_ 0ij ;
W ð4:1450 Þ
This law is applied when the plastic deformations are much more greater than
the elastic ones. Thus, it can be applied to describe the processes of metal working:
rolling, drawing, cutting etc.
Assuming that the plastic work of deformation is given by a relation of the form
rÞ
dUð
dW p ¼ r;
d ð4:148Þ
r
d
where we made to intervene the intensity of stresses, and denoting
1 rÞ
dUð
0 ¼ rÞ;
Fð
J2 r
d
we obtain
dk ¼ Fð
rÞd
r;
the law of flow (4.141) becomes
rÞr0ij d
deij ¼ deeij þ Fð r; i; j ¼ 1; 2; 3: ð4:1410 Þ
The hypotheses at the basis of the physical law corresponding to the theory of
plastic deformation are:
(i) The body is isotropic.
(ii) The mean normal stress is in direct proportion with the mean strain, the
coefficient of proportionality being the same both in the elastic and in the
plastic state
(iii) The deviators of the stress and strain tensors are in direct proportion.
The Nadai-Hencky-Ilyushin law can be written in the form
1
e0 ¼ r0 ; ð4:1490 Þ
3K
where w ¼ wð r; eÞ is a function which must be specified on an experimental way
for each case. The law may be written in the form
1
eij ¼ r0 dij þ wr0ij ; i; j ¼ 1; 2; 3; ð4:14900 Þ
3K
too; solving with respect to the stresses, it results
1 0
rij ¼ 3Ke0 dij þ e ; i; j ¼ 1; 2; 3: ð4:149000 Þ
w ij
One has
Taking
1
w¼ ¼ const; ð4:152Þ
2l
one obtains Hooke’s linearly elastic law, for which the unit work of deformation is
given by
3
W ¼ ð3Ke20 þ le2 Þ ð4:153Þ
2
and for which Green’s formulae (4.9) are valid.
Using Mises’s condition of plasticity, one gets
pffiffiffi
3 e
w¼ : ð4:154Þ
2 se
The unit work of deformation corresponding to a deformation potential reads
9 pffiffiffi
W ¼ Ke20 þ 3see: ð4:155Þ
2
The relation between stresses and strain takes the form
2 se
rij ¼ 3Ke0 dij ¼ pffiffiffi e0ij ; i; j ¼ 1; 2; 3; ð4:156Þ
3 e
in case of an incompressible body, we have
2 se
rij r0 dij ¼ pffiffiffi eij ; i; j ¼ 1; 2; 3: ð4:1560 Þ
3 e
The formulae of (4.9) type remain, further, valid.
In case of a hardening body, we assume that
1
w¼ gð
rÞ; ð4:157Þ
2
rÞ is a function which must be specified; from (4.1500 ) it results
gð
where
1
e ¼ r gð
rÞ: ð4:158Þ
3
Between stresses and strains takes place the relation
1 1
eij ¼ r0 dij þ rÞr0ij ;
gð i; j ¼ 1; 2; 3; ð4:159Þ
3K 2
introducing a new function gðeÞ, defined by the relation
1
gðeÞ ¼ ; ð4:160Þ
gð
rÞ
4.2 Inelastic Bodies 169
we have
Another factor which must be taken into consideration is the time factor. Direct
observations have shown, e.g., that the earth is settled in time under the action of
the constructions that it supports (which are deformed too). Various bodies con-
sidered to be solid (as steel, concrete, mountain glacier etc.) flow. From intuition
one passes to abstraction, to considerations of new models [1, 7, 24].
To can describe the phenomena in which appears the time variable we will use
a relation between stresses and strains of the form
f ðr; e; tÞ ¼ 0; ð4:162Þ
From the linear theory of elasticity and the dynamics of viscous fluids one
obtains thus the linear theory of viscoelasticity, which implies two hypotheses:
(i) The bodies undergo infinitesimal deformations.
(ii) The mechanical properties of the bodies are subjected to Boltzmann’s prin-
ciple of superposition of effects.
Thus, the relation (4.162) represents a differential, integral or integro-differ-
ential equation.
O t1 t
(a)
(t) = 0 (t; 0)
1
0
(t )
E
0
O t1 t
(b)
hence if after a long time the sample has no more a residual deformation, then the
body is called with delayed elasticity.
Let us consider the same loading of the bar at the moment t ¼ t0 , where we
take, analogically, t0 ¼ 0; assuming now that the strain e0 remains constant
(Fig. 4.21a), then the stress r decreases with time. This phenomenon is called
relaxation; to study it, we consider the function
1
wðt; e0 Þ ¼ eðtÞ; ð4:167Þ
e0
which we call relaxation function and where we considered that the moment
t0 ¼ 0 is origin for the time variable.
The stress decreases with time after a relation of the form (Fig. 4.21b)
rðtÞ ¼ rð0 þ 0Þ þ rr ðtÞ ¼ Ee0 þ e0 #ðtÞ; ð4:168Þ
O t1 t
(a)
(t) = 0 (t; 0)
(t) 0
E 0
O t1 t
(b)
if the stress rðsÞ is a continuous function in the interval ð1; t and if we replace
the sum by an integral, then we can express the strain at the moment t in the form
Z t
drðsÞ
eðtÞ ¼ uðt sÞds: ð4:170Þ
1 ds
Z t Z t
drðsÞ drðt sÞ
eðtÞ ¼ hðtÞ uðt sÞds ¼ hðtÞ uðtÞds; ð4:1700 Þ
0 ds 0 ds
where hðtÞ is Heaviside’s distribution, given by (A.165); this relation represents
the correspondent of Hooke’s law in the viscoelastic case.
Analogically, using the relaxation function, we get
Z t
deðsÞ
rðtÞ ¼ wðt sÞds ð4:171Þ
1 ds
or
Z t Z t
deðsÞ deðt sÞ
rðtÞ ¼ hðtÞ wðt sÞds ¼ hðtÞ wðsÞds: ð4:1710 Þ
0 ds 0 ds
These relations, which correspond to Boltzmann’s principle, are known as
Boltzmann’s integral relations [1] and can be considered as integral equations if
one of the quantities eðtÞ, rðtÞ is unknown, while the other one appears as given. In
this case, the functions u and w are the nuclei of Volterra’s type equations.
We can express the loading at the moment t0 ¼ 0 in the form
rðtÞ ¼ r0 hðtÞ; ð4:172Þ
replacing in (4.170) and taking into account (A.166), we find again the relation
(4.164). An analogous result is obtained with a strain of the form
eðtÞ ¼ e0 hðtÞ; ð4:173Þ
In case of the loading-unloading process previously described (creep-inverse
creep), we may write (Fig. 4.20a)
rðtÞ ¼ r0 ½hðtÞ hðt t1 Þ; ð4:174Þ
obtaining
eðtÞ ¼ r0 ½uðtÞhðtÞ uðt t1 Þhðt t1 Þ: ð4:175Þ
where the parameter t ¼ g=E represents the retardation time (or the creep time).
Assuming that the body is in the natural state for t 0, hence putting the initial
conditions eð0Þ ¼ rð0Þ ¼ 0, we can integrate the differential equation of first order
(4.176); we get thus
Z Z
hðtÞ t hðtÞ t
eðtÞ ¼ rðsÞeðtsÞ=t ds ¼ rðt sÞes=t ds: ð4:177Þ
g 0 g 0
Putting rðtÞ given by (4.172) in (4.177), we find the creep function in the form
eðtÞ 1
uðtÞ ¼ ¼ ð1 et=t ÞhðtÞ: ð4:178Þ
r0 E
The strain eðtÞ will vary after the law
r0 r0
eðtÞ ¼ ð1 et=t ÞhðtÞ ¼ ð1 et=t Þþ ; ð4:179Þ
E E
its graphic being given in Fig. 4.23a. For t ! 1 we obtain e ¼ r0 =E.
If we admit a loading-unloading process to which corresponds a creep-inverse
creep phenomenon, then we take rðtÞ of the form (4.174) in (4.177) and obtain
(r
0
ð1 et=t Þ for t t1 ;
eðtÞ ¼ E
ð4:180Þ
e1 et=t for t t1 ;
where the strain
r0 t1 =t
e1 ¼ ðe 1Þ ð4:1800 Þ
E
is constant; the corresponding graphic is given in Fig. 4.23b. The strain tends to
zero for t ! 1; the rapidity by which the strain tends to zero depends on the
retardation time t , justifying thus its name.
P
4.2 Inelastic Bodies 175
O t O t1 t
(a) (b)
Taking eðtÞ in the form (4.173), we find the relaxation function corresponding
to Maxwell’s model
wðtÞ ¼ EhðtÞet=t ; ð4:184Þ
t plays here the rôle of a relaxation time. The graphic of this function is given in
Fig. 4.25a.
Taking rðtÞ of the form (4.172) in the constitutive law (4.182), one obtains the
creep function corresponding to Maxwell’s model; observing that
d d
tþ ¼ ½thðtÞ ¼ hðtÞ þ tdðtÞ ¼ hðtÞ; ð4:185Þ
dt dt
on the basis of the relations (A.167) and (A.169), we get
1 t
1h tþ i
uðtÞ ¼ 1 þ hðtÞ ¼ hðtÞ þ ; ð4:186Þ
E t E t
the creep function being defined by the relation (4.164). Its graphic is a straight
line of slope g, which, at the moment t ¼ 0, leads to uð0 þ 0Þ ¼ 1=E (Fig. 4.25b).
If we take rðtÞ of the form (4.174), we obtain
1 h t
t t1
i
uðtÞ ¼ 1 þ hðtÞ 1 þ hðt t1 Þ
E t t
1 n t
t1 o
¼ 1 þ ½hðtÞ hðt t1 Þ þ hðt t1 Þ ; ð4:187Þ
E t t
there corresponds the graphic in Fig. 4.25b.
Although these models have been obtained in an intuitive way, they are par-
ticularly useful, facilitating the scientific imagination, to may build up more
complex models, reflecting also other aspects of the reality. It is important that
these models can explain the phenomena of creep and relaxation too, which occur
with time.
(t)
(t)
E
1
E
1
E E
e t1
O t1 t O t1 t
(a) (b)
1 2 n
with
1 Xn
1 X n
¼ ; Ki ¼ 1; ð4:1910 Þ
E0K EK i¼1
i¼1 i
178 4 Mathematical Models in Mechanics of Deformable Solids
with
Z 1 Z 1
1 dk
¼ ; KðkÞdk ¼ 1: ð4:1920 Þ
EK 0 EðkÞ 0
Thus, the creep function corresponding to a generalized Kelvin model with con-
tinuous retardation spectrum is given by
Z
hðtÞ 1
uðtÞ ¼ KðkÞð1 et=k Þdk: ð4:193Þ
EK 0
In an analogous way, one can group in parallel n simple Maxwell models of
relaxation times ti ¼ gi =Ei ; i ¼ 1; 2; . . .; n (Fig. 4.27), obtaining thus the gen-
eralized Maxwell model; the total stress is given by
X
n
rðtÞ ¼ ri ðtÞ; ð4:194Þ
i¼1
1 2 n-1 n
P
4.2 Inelastic Bodies 179
X
n Zt
deðsÞ ðtsÞ=ti
rðtÞ ¼ hðtÞ Ei e ds: ð4:19400 Þ
i¼1
ds
0
If one takes eðtÞ of the form (4.173), then one obtains the relaxation function
X
n
wðtÞ ¼ hðtÞ Ei et=ti : ð4:196Þ
i¼1
with
Z 1 Z 1
EM ¼ EðkÞdk; FðkÞdk ¼ 1: ð4:1980 Þ
0 0
Starting from the simple Kelvin and Maxwell models, to which we associate
Hooke and Newton models, we can build up other models of viscoelastic bodies,
useful for various particular bodies too; these models will be models with more
than two parameters.
We can thus set up a model with three parameters, obtained by grouping into
series a Hooke model with a Kelvin one (Fig. 4.28a); the corresponding consti-
tutive law reads
E1 þ E2 d 1 d
þ rðtÞ ¼ E1 þ eðtÞ; ð4:200Þ
g2 dt t1 dt
180 4 Mathematical Models in Mechanics of Deformable Solids
P P
(a) (b)
the corresponding graphic being given in Fig. 4.29; we notice that we have a jump
equal to r0 =E1 at the moment t ¼ t1 , the strain eðtÞ tending exponentially to zero.
Analogically, one obtains Zener’s model, by grouping in parallel a Hooke
model and a Maxwell one (Fig. 4.28b); the corresponding constitutive law reads
1 d 1 E2 d
þ rðtÞ ¼ E 1 þ 1 þ eðtÞ; ð4:203Þ
t1 dt t2 E1 dt
0
E1
1
1
0
E1
O t1 t
(a)
(t)
O t1 t
(b)
P P
(a) (b)
E2 2
A model often used is the Burgers model obtained by linking into series a
Kelvin model and a Maxwell one (Fig. 4.31); in an operational way, the consti-
tutive law is of the form
1 d d=dt
e_ ðtÞ ¼ 1 þ t1 þ rðtÞ: ð4:208Þ
g1 dt E2 ð1 þ t2 d=dtÞ
The creep function reads
1 1 t=t2
uðtÞ ¼ ðt þ tÞ þ ð1 e Þ hðtÞ: ð4:209Þ
g1 1 E2
If rðtÞ is given by (4.174), then one obtains eðtÞ in the form (4.207) (Fig. 4.32).
On the same line, one can link into series a generalized Kelvin model and a
Newton one or one can link in parallel a generalized Maxwell model with a Hooke
model and a Newton one; the corresponding curves of creep and relaxation may be
thus closer to those obtained experimentally for some particular bodies.
We can successfully effect a study of the models presented above by means of
the Laplace transform; using the results in Sect. A.3.4.2, we can introduce the
transforms of the stress and of the strain in the form
Z 1
L½rðtÞ ¼ r~ðpÞ ¼ rðtÞept dt;
0
Z 1 ð4:210Þ
L½eðtÞ ¼ ~eðpÞ ¼ eðtÞept dt;
0
where p is the new complex variable, corresponding to the time variable. Thus,
e.g., the constitutive law (4.176), corresponding to Kelvin’s model, becomes
~ðpÞ
r
~eðpÞ ¼ ; ð4:211Þ
gð1=t þ pÞ
effecting the inverse Laplace transform, we get the constitutive law (4.177).
Analogically, the Laplace transforms of the creep and relaxation functions
will be
4.2 Inelastic Bodies 183
E1
1
1
E1
1t
1
1
O t1 t
(a)
(t)
O t1 t
(b)
1
L½uðtÞ ¼ u
~ ðpÞ ¼ ; ð4:212Þ
Epð1 þ t pÞ
~ E
L½wðtÞ ¼ wðpÞ ¼ ð1 þ t pÞ: ð4:2120 Þ
p
We notice that the relation
p2 u ~
~ ðtÞwðpÞ ¼ 1; ð4:213Þ
by means of which one can pass from the creep function to the relaxation one and
conversely; this property is valid for all the bodies which admit Boltzmann’s
principle of superposition and which verify the relations (4.1700 ), (4.1710 ).
which take also the form of Boltzmann’s integral relations (4.1700 ), (4.1710 ).
Applying the Laplace transform, it results
~ðpÞ QðPÞ
r
¼ ; ð4:216Þ
~eðpÞ PðpÞ
starting from Boltzmann’s relations, one obtains the Laplace transforms of the
creep and relaxation functions
1 ~eðpÞ 1 PðpÞ
~ ðpÞ ¼
u ¼ ; ð4:217Þ
pr~ðpÞ p QðpÞ
~ 1r~ðpÞ 1 QðpÞ
wðpÞ ¼ ¼ ; ð4:2170 Þ
p ~eðpÞ p PðpÞ
which verify the relation (4.213). Using the decomposition in simple fractions, we
obtain, easily, the inverse Laplace transforms too.
We consider now the case of a three-dimensional loading, where appear the
tensors Tr and Te with all their components; we notice, from the very beginning,
that it is important to use the decomposition of these tensors in their spheric and
deviator parts. Thus, in case of Hooke’s model, we use the constitutive law in the
form (4.92), (4.920 ).
Using the mode of construction of a Kelvin model in the one-dimensional case,
we obtain a viscoelastic model of generalized Kelvin type, specified by the con-
stitutive law
0 d
rij ðtÞ ¼ 2l 1 þ t e0 ðtÞ; i; j ¼ 1; 2; 3; ð4:218Þ
dt ij
d 0 d 0
1 þ t1 rij ðtÞ ¼ 2l 1 þ t2 e ðtÞ; i; j ¼ 1; 2; 3; ð4:220Þ
dt dt ij
where l; tk ; k ¼ 1; 2, are the constants of the material, to which is associated the
elastic relation (4.92); this model corresponds to the Zener model of the one-
dimensional case.
Using the operators
n0
X n00
X d
P1 ðDÞ ¼ a0i Di ; P2 ðDÞ ¼ a00i Di ; D¼ ; ð4:221Þ
i¼1 i¼1
dt
m0
X m00
X d
Q1 ðDÞ ¼ b0j D j ; Q2 ðDÞ ¼ b00j D j ; D¼ ; ð4:2210 Þ
j¼1 j¼1
dt
where a0i ; a00i ; b0j ; b00j are the constant coefficients of the material, we may write the
general differential constitutive law of a linear viscoelastic body in the form
P1 ðDÞ ¼ P2 ðDÞ ¼ 1;
ð4:223Þ
Q1 ðDÞ ¼ 2l; Q2 ðDÞ ¼ 3K;
P1 ðDÞ ¼ P2 ðDÞ ¼ 1;
d ð4:224Þ
Q1 ðDÞ ¼ 2l 1 þ t ; Q2 ðDÞ ¼ 3K;
dt
~0ij ðpÞ ¼ 2~
r lðpÞ~e0ij ðpÞ; i; j ¼ 1; 2; 3; ð4:226Þ
~ e0 ðpÞ;
~0 ðpÞ ¼ 3KðpÞ~
r ð4:2260 Þ
where we used the notations
186 4 Mathematical Models in Mechanics of Deformable Solids
Q1 ðpÞ ~ Q2 ðpÞ
~ðpÞ ¼
l ; KðpÞ ¼ ; ð4:22600 Þ
2P1 ðpÞ 3P2 ðpÞ
this constitutive law has the same form as that in case of linearly elastic bodies,
being written in the space of Laplace transforms. Obviously, we may write the
constitutive law in this space also in the form
~ij ðpÞ ¼ ~
r rll ðpÞdij þ 2~
kðpÞ~ lðpÞ~eij ðpÞ; i; j ¼ 1; 2; 3; ð4:227Þ
where
~ 1 Q2 ðpÞ Q1 ðpÞ Q1 ðpÞ
kðpÞ ¼ ~ðpÞ ¼
; l ; ð4:2270 Þ
3 P2 ðpÞ P1 ðpÞ 2P1 ðpÞ
~e0 ðpÞ ¼ p~
u2 ðpÞ~
r0 ðpÞ; ð4:2280 Þ
where
P1 ðpÞ P2 ðpÞ
~ 1 ðpÞ ¼
u ~ ðpÞ ¼
;u ; ð4:22800 Þ
pQ1 ðpÞ 2 pQ2 ðpÞ
r ~ ðpÞ~e0 ðpÞ; i; j ¼ 1; 2; 3;
~0ij ðpÞ ¼ pw ð4:230Þ
1 ij
~ ðpÞ~e0 ðpÞ;
~0 ðpÞ ¼ pw
r ð4:2300 Þ
2
where
~ ðpÞ ¼ Q1 ðpÞ ; w
w ~ ðpÞ ¼ Q2 ðpÞ ; ð4:23000 Þ
1
pP1 ðpÞ 2 pP2 ðpÞ
4.2 Inelastic Bodies 187
p2 u ~ ðpÞ ¼ 1;
~ k ðpÞw k ¼ 1; 2; ð4:232Þ
k
take place.
One can use much more complex linear models too, where the constitutive law
is an integro-differential one.
We notice that one can combine the property of viscosity with that of plasticity,
obtaining thus models of viscoplastic bodies. We mention thus Bingham’s model
for which the relation between stresses and strains is of the form
References
A. Books
B. Papers
}
22. Teodorescu P.P.: Uber das kinetische Problem nichthomogener elastischer K} orper. Bull.
Acad. Pol. Sci., sér. Sci. Technol. 12, 867 (1964)
23. Teodorescu, P.P., Predeleanu, M.: Quelques considérations sur le problème des corps
élastiques hétérogènes. In: Proceedings of IUTAM-Symposium. Non-Homogeneity in
Easticity and Plasticity, Warszawa, 1958, Perganon Press, 31 (1959) [Bull. Acad. Pol. Sci.,
sér. Sci. Technol. 7, 81 (1959)]
24. Voigt, W.: Theoretische Studien } uber die Elastizitätsverhältnisse der Krystalle. I, II. Abh. der
K}onigl. Ges. Wiss., G}ottingen, 34 (1887)
Chapter 5
General Equations of the Theory of
Elasticity. Formulation of Problems
The local study made up to the present time allows us to define the state of strain
and stress at an arbitrary point of any solid body (the set of all the possible states of
strain and stress), without taking into account the specific conditions of the given
problem (external loads acting upon the given body, as well as the conditions in
which the latter happens to be at a given time). The aim of this chapter is just to
complete this study by a global one.
If we refer to a given body and a given state of strain and stress, then we must take
into account the limit conditions of the problem, that, in the static case, are
boundary conditions, while, in the dynamic case, are boundary conditions and
initial conditions. These conditions state precisely the solution of the problem,
Therefore, the local study must be finished off by a global study, liable to make the
connection between the considered body and the other bodies at any time, also
taking into account the motion of the respective body at a given time.
Moreover, we shall state the formulations for both static and dynamic problems
of the theory of elasticity in the classical case: homogeneous, isotropic, linearly
elastic bodies, subjected to an infinitesimal state of deformation.
A preliminary study of the system of partial differential equations of elasticity
was made by Gr. C. Moisil [28], who used the method of matrices associated with
systems of partial differential equations, elaborated in [7]; thus, various differential
equations to be verified by each of the unknown functions by means of certain
potential functions are emphasized.
We shall deal firstly with the case of a static state of strain and stress, specifying
the basic system of equations; the three fundamental problems are thus put in
evidence.
or in the form
1
eij ¼ ½ð1 þ mÞrij mrll dij ; i; j ¼ 1; 2; 3: ð5:30 Þ
E
These equations, which must be verified on the domain occupied by the elastic
body, form the fundamental system of equations of the linear elasticity; if we add
the boundary conditions, then they form the complete system of equations of the
classical theory of elasticity in the static case.
In displacements (first basic problem of the theory of elasticity) the conditions are
put on the boundary (the frontier C of the domain D) in the form (Fig. 5.1a)
n
ui ¼ ui ; i ¼ 1; 2; 3; ð5:4Þ
n n
where the displacement u; of components ui ; i ¼ 1; 2; 3; of the point of external
normal n is given by the linkage with other bodies. The cases in which the
conditions in displacements can be put on all the contour are uncommon; even-
tually, one puts conditions for certain derivatives of the displacements (e.g., for
strains).
5.1 General Equations of the Theory of Elasticity 193
(a) x3 (b) x3
n
u n p
n n
M M
r dA r dA
x1 x2 x1 x2
One must mention that these conditions are distinct from the conditions of fixity
(conditions of support) of the body, which are put at various points to determine its
rigid body motion.
The most times we know the external loads which act on the contour. At each
point of the contour must take place a mechanical equilibrium between the internal
n n
stresses rij and the external loads p (of components pi ; i ¼ 1; 2; 3; on the element
of area of external normal n); hence, we must put the condition that, on the contour
C, the stresses, obtained by computation, be equated by the given external loads
(Fig. 5.1b), using the relations
n
p i¼ rji nj ; i ¼ 1; 2; 3; ð5:5Þ
or the relations
n
p i ¼ ri ni ð!Þ; i ¼ 1; 2; 3; ð5:50 Þ
corresponding to the principal directions. If the normal and the tangential com-
ponents of the external load are given, then we can use relations of the form (3.25),
(3.28); we notice also that, if on an element of area of external normal n we have
not tangential components of the external load, then the direction n is a principal
one, the other principal directions being contained in the tangent plane.
If we succeed to integrate the system of equations considered above, verifying
the boundary conditions too, then the state of strain and stress is completely
determined, the problem of elastostatics being thus solved. The above problem, in
which the conditions on the boundary are put in stresses, is called the second basic
problem of the theory of elasticity.
As well (and this happens frequently) one can put conditions in displacements
on some portions of the contour (frontier Cu ) and conditions in stresses on other
portions of it (frontier Cr ; Cu [ Cr ¼ C; Cu \ Cr ¼ ;) (the mixed problem of the
theory of elasticity). One can conceive also more elaborate mixed problems, in
which, e.g., at the same point both conditions in displacements and in stresses are
put.
The problems of the theory of elasticity can be easier solved by choosing as
unknowns only the displacements or only the stresses, as the boundary conditions
194 5 General Equations. Formulation of Problems
are put, and eliminating the other unknowns between the 15 equations mentioned
above. One is thus led to a solution in displacements or to a solution in stresses of
these problems.
On can imagine a solution in strains of the problem (introducing, eventually, the
local rotations of rigid body) too, using also the equations of continuity (2.68); but
this solution is not interesting, from a practical point of view, because the con-
ditions on the contour are not put, customarily, in this form.
In a practical solution of the problems, we use that representation which is
better fitted to the boundary conditions; thus, in case of the first basic problem as
well as in case of the mixed one, a solution in displacements is convenient, while
in case of the second basic problem one can use successfully a solution in stresses.
In case of the mixed problem, it is useful to can express the conditions in
stresses by means of the displacements; starting from Hooke’s law, written in the
form
m 1 m
rij ¼ 2l eij þ ell dij ¼ 2l ðui;j þ uj;i Þ þ ul;l dij ;
1 2m 2 1 2m
it results
n n 1 m
p ¼ pk ik ¼ rjk nj ik ¼ 2l ðuj;k uk;j Þ nj þ uk;j nj þ ul;l djk nj ik ;
2 1 2m
so that the stress vector along the direction of unit vector n is given by
n ou 1 m
p ¼ 2l þ n x curl u þ n div u : ð5:500 Þ
on 2 1 2m
The use of potential functions is particularly important to formulate the prob-
lems of the theory of elasticity; in case of a formulation in displacements one has
to do with displacement functions, while in case of a formulation in stresses with
stress functions. We will pay a special attention to such formulations; most of the
problems will be studied on this way.
In a solution in displacements of the static problem, we mention, first of all, the
formulation given by B. G. Galerkin [16], who uses three biharmonic displace-
ment functions; these functions may be considered as the components of a vector
(Galerkin’s vector). Using Almansi’s formula (A.100), one can state that in this
representation intervene six harmonic functions which must be determinate with
the aid of six conditions at a point of the contour. A triple functional indetermi-
nation occurs; the state of displacement is univocally determined, but the dis-
placement functions have not the same property.
Another formulation in displacements has been obtained by P. F. Papkovich
[33] in 1932, using four harmonic functions; it has been previously found by G.
D. Grodski, but published only in 1935, after being mentioned in 1927 by the
occasion of a conference. The conference has been forgotten, although Papkovich
recognized later the priority. This result has been independently formulated by
5.1 General Equations of the Theory of Elasticity 195
The basic problems of the theory of elasticity are the same in the dynamic case as
in the static one; but we must notice that both the displacements and strains, as
well as the stresses are functions not only of point but also of the time variable,
which involves the initial conditions too.
6 relations between strains and displacements (5.2) and 6 finite relations between
stresses and strains (5.3) or (5.30 ).
These equations form the basic system of equations of linear elastodynamics; if
we add also the limit conditions, they form the complete system of equations of the
classical theory of elasticity in the dynamic case.
The three basic problems of the theory of elasticity are defined as in the static case;
the conditions on the contour are put in the form (5.4), in case of the first basic
problem, or in one of the forms (5.5), (5.50 ), in case of the second basic problem,
5.1 General Equations of the Theory of Elasticity 197
for t t0 ; where t0 is the initial moment. In case of the mixed problem, one
proceeds analogically.
To the boundary conditions one adds the initial conditions
ui ¼ u0i ; u_ i ¼ u_ 0i ; i ¼ 1; 2; 3; ð5:7Þ
which must be fulfilled in the domain D occupied by the body and on the contour
C, where they must connect to the boundary conditions mentioned above; one can
put initial conditions also for the stress tensor Tr as well as for the velocity stress
tensor Tr_ , in the form
rij ¼ r0ij ; r_ ij ¼ r_ 0ij ; i; j ¼ 1; 2; 3: ð5:8Þ
We have denoted by u0i ; u_ 0i the displacements and the displacement velocities at the
initial moment t ¼ t0 ; respectively; analogically, r0ij and r_ 0ij represent the stresses
and the stress velocities, at the same initial moment, respectively. These functions
depend on the space variables x1 ; x2 ; x3 : To these conditions one adds also the
conditions of support (conditions of fixity) of the body.
The conditions on the contour, the initial conditions and the fixity conditions
form the limit conditions of the problem.
If we succeed to integrate the above considered system of equations, putting the
limit conditions too, then the state of strain and stress is completely determined;
the problem is thus solved.
As in the static case, we are led to a solution in displacements or to a solution in
stresses of the problem, as the unknown functions are mostly involved in the limit
conditions.
The formulation of the problems of elastodynamics constituted, since the
beginning of nineteenth century, the object of many researches; especially, general
formulations were sought for, in the case of homogeneous, isotropic, linearly
elastic bodies, of G. Lamé’s equations [6] in the dynamic case. These equations are
to be found by C. L. M.-H. Navier in the particular case when k ¼ l; as well as
later by G. Lamé and B. Clapeyron and by G. G. Stokes. In the mechanics of fluids
they are known as the Navier-Stokes equations.
In this direction we mention, firstly, the representations with the help of a vector
and of a scalar displacement function, based upon the decomposition of a vector
into the sum of a gradient and a curl. The first of such a representation was given by
S. D. Poisson in 1829 in the particular case in which m ¼ 1=4; without emphasizing
the completeness of the representation or the fact that the solenoidal part of the
representation is a curl; it was found later on by G. Lamé [6] in 1852, for an
arbitrary Poisson’s ratio. A. Clebsch resumed the problem in 1863 and gave a
demonstration, yet insufficient to ensure the completeness of the representation; the
subsequent demonstration of W. Thomson did not conclude either. Rigorous
demonstrations of this result were given later on by C. Somigliana [43] in 1892 and
by O. Tedone in 1897; we mention also the corresponding demonstration of
P. Duhem made in 1898 and resumed in 1960 in a more concise form by
E. Sternberg [45]. We shall call this formulations the Lamé-Clebsch representation.
198 5 General Equations. Formulation of Problems
which we brought also into relief the completeness, in case of a simply connected
domain, we started, as Schaefer did, from the equations of motion, imposing the
condition that the other equations should be verified too.
Another method of tackling these problems, by solving them in stresses, consist
of performing in the opposite direction the method indicated above: i.e. starting
from the Beltrami type equations and imposing the condition that the resulted
representation should verify both the motion equations and the constitutive law.
We [11] gave thus a new representation in stresses either with the help of three
functions verifying the double waves equation or with the help of four functions,
each of them verifying a simple waves equation. The passage from a representa-
tion to another one can be made by means of formulae of Boggio type (see A.2.7).
These representations generalise those previously given by us [47], in the static
case. General results on this subject can be found in the synthesis work [49].
The use of the potential functions is particular important for the formulation of the
problems of the theory of elasticity; in case of a formulation in displacements, we
have to deal with displacement functions, while in case of a formulation in stresses,
we introduce stress functions. We mention that one cannot represent the general
solution of a problem of static or dynamic elasticity by means of only one potential
function.
We shall attach a special importance to the formulations in displacements or in
stresses, the most of the limit problems which will be considered being studied in
such a manner.
or of the form
1 1
Du þ grad div u þ F ¼ 0: ð5:130 Þ
1 2m G
In the absence of the volume forces, it results
1
Du þ grad div u ¼ 0: ð5:120 Þ
1 2m
If we apply the operator div to the Eq. (5.13) or to the Eq. (5.130 ) and if we take
into consideration (5.11), then we find that the volume strain must verify the
equation
1 1 2m
Dh ¼ div F ¼ div F; ð5:14Þ
k þ 2l 2ð1 mÞG
5.2 Statical Problem. Potential Functions 201
and
1
D uþ r div u ¼ 0; ð5:1200 Þ
2ð1 2mÞ
respectively.
Applying Laplace’s operator to the Eq. (5.13) and taking into account (5.14),
we find that the displacement vector must verify the equation
kþl 1
DDu ¼ grad div F DF; ð5:16Þ
lðk þ 2lÞ l
taking into account the relation (A.920 ) between the differential operators of sec-
ond order, we can write the Eq. (5.16) in the form
1 1
DDu ¼ grad divF þ curl curl F ð5:160 Þ
k þ 2l l
too, which can be useful in some cases. In case of a solenoidal field of volume
forces we may write
1
DDu ¼ curl curl F; ð5:17Þ
l
while, in case of an irrotational field of volume forces, we get
1
DDu ¼ grad div F: ð5:170 Þ
k þ 2l
In the absence of the volume forces, the displacement vector is a biharmonic
function
DDu ¼ 0: ð5:18Þ
We must mention that the set of integrals of the Eq. (5.16) includes the set of
integrals of the Eq. (5.13), the first equation being not sufficient to solve the
problem. We must integrate the Eq. (5.13) or the Eq. (5.130 ) to get a solution in
displacements.
202 5 General Equations. Formulation of Problems
Starting from the homogeneous Eq. (5.100 ), one can introduce a function by means
of the relation
2G div u ¼ ð1 2mÞDu; ð5:19Þ
with
Dv ¼ 0: ð5:21Þ
Taking into account the relation (5.19), we get
div v div grad u ¼ ð1 2mÞDu
or
2ð1 mÞDu ¼ div v: ð5:22Þ
We introduce now a vector C ¼ Cðx1 ; x2 ; x3 Þ; with the aid or the relation
v ¼ 2ð1 mÞDC; ð5:23Þ
the Eq. (5.21) shows that this vector must verify the biharmonic equation
DDC ¼ 0: ð5:24Þ
We are led to
Dðu div CÞ ¼ 0;
wherefrom
u ¼ div C þ C0 ;
5.2 Statical Problem. Potential Functions 203
with
DC0 ¼ 0:
By means of Almansi’s formula (A.100), which allows the representation of a
biharmonic function with the aid of two harmonic functions, we can introduce the
function C0 in the harmonic part of the biharmonic components of the biharmonic
vector C.
The representation (5.20) leads now to Galerkin’s representation
2Gu ¼ 2ð1 mÞD C grad div C; ð5:25Þ
The above given representation is complete for a simply connected domain (any
displacement vector which verifies Lamé’s Eq. (5.120 ) can be represented in the
form (5.25)). One must make a supplementary study of the relations (5.19 and
5.23) for a multiply connected domain, in other words one must specify the
conditions in which the functions u and C are univocally determined; certain
complementary terms, corresponding to some singularities, are thus introduced.
Using Cauchy’s relations (5.2) and Hooke’s law (5.3), we obtain the state of
stress in the form
rij ¼ 2ð1 mÞDCði;jÞ þ ðmDdij oi oj ÞCk;k ; i; j ¼ 1; 2; 3: ð5:26Þ
If the volume forces are non-zero, then we can use the same representation
(5.25); Galerkin’s vector must verify the equation
1
DDC þ F ¼ 0; ð5:27Þ
1m
as it can be easily seen, by replacing the representation (5.25) in the Eq. (5.130 ).
We notice that a particular integral is sufficient to introduce the influence of the
volume forces.
204 5 General Equations. Formulation of Problems
Taking into account Almansi’s formulae (A.100), (A.1000 ), which allow the
representation of a biharmonic function by means of two harmonic ones, it results
that in Galerkin’s representation intervene six harmonic functions, which must be
determined with the aid of three conditions at one point of the contour. One has
thus a triple functional indetermination; the state of displacement is univocally
determined, as it will be seen in the next chapter, but the displacement function has
not this property.
Taking into account the formula (A.1030 ) and observing that the vector v verifies the
harmonic Eq. (5.21), it results that the relation (5.22) allows to write u in the form
1
u¼ ðv þ r vÞ; ð5:28Þ
4ð1 mÞ 0
with
Dv0 ¼ 0; ð5:29Þ
Dvj ¼ 0; j ¼ 1; 2; 3: ð5:210 Þ
If we eliminate the displacements between the relations (5.2), then we obtain the
equations of continuity in the form
mij nkl eik; jl ¼ 0; m; n ¼ 1; 2; 3; ð5:36Þ
or
ð1 þ mÞmij nkl rik;jl mDrll dmn þ mrll;mn ¼ 0; m; n ¼ 1; 2; 3; ð5:37Þ
or
ð1 mÞDrll ð1 þ mÞrij;ij ¼ 0; ð5:38Þ
taking into account the equation of equilibrium (5.1) too, we may write
rij;ij ¼ Fk;k ; ð5:39Þ
so that we obtain the equation verified by the sum of normal stresses in the form
1þm 1þm
Drii ¼ DH ¼ Fi;i ¼ div F: ð5:40Þ
1m 1m
We notice that one could obtain this result starting from (5.14) and using
Hooke’s law.
In the absence of volume forces or in the case of a solenoidal field of volume
forces, the sum of the normal stresses becomes a harmonic function
DH ¼ 0: ð5:41Þ
Taking into account (A.37), we can write the Eq. (5.37) in the form
or in the form
which results from the equations of equilibrium (5.1), as well as the relations
(5.39), (5.40), the Eq. (5.370 ) read
1 m
Drij þ H;ij ¼ 2Fði;jÞ Fk;k dij ; i; j ¼ 1; 2; 3; ð5:42Þ
1þm 1m
one obtains thus the Beltrami-Michell equations of continuity in stresses (obtained
by Beltrami in the absence of the volume forces and by Michell in case of arbitrary
volume forces).
In a developed form, we have
1 m
Dr11 þ ðr11 þ r22 þ r33 Þ;11 ¼ 2F1;1 ðF1;1 þ F2;2 þ F3;3 Þ;
1þm 1m
1 m
Dr22 þ ðr11 þ r22 þ r33 Þ;22 ¼ 2F2;2 ðF1;1 þ F2;2 þ F3;3 Þ; ð5:43Þ
1þm 1m
1 m
Dr33 þ ðr11 þ r22 þ r33 Þ;33 ¼ 2F3;3 ðF1;1 þ F2;2 þ F3;3 Þ;
1þm 1m
1
Dr23 þ ðr11 þ r22 þ r33 Þ;23 ¼ ðF2;3 þ F3;2 Þ;
1þm
1
Dr31 þ ðr11 þ r22 þ r33 Þ;31 ¼ ðF3;1 þ F1;3 Þ; ð5:430 Þ
1þm
1
Dr12 þ ðr11 þ r22 þ r33 Þ;12 ¼ ðF1;2 þ F2;1 Þ:
1þm
Applying Laplace’s operator to the Eq. (5.42), we obtain the equations
1
DDrij ¼ 2DFði;jÞ þ ðoi oj mDdij ÞFk;k ; i; j ¼ 1; 2; 3; ð5:44Þ
1þm
which are verified by the components of the stress tensor; in the absence of the
volume forces, the stresses are biharmonic functions
DDrij ¼ 0; i; j ¼ 1; 2; 3; ð5:45Þ
We notice that the stresses must be functions of class C2 ; to can verify the
Eq. (5.42); if we wish the Eq. (5.44) be verified too, then these functions must be
of class C4 .
We also must notice that the set of integrals of the Eq. (5.44) contains the set of
integrals of the Eq. (5.42), the first equations being not sufficient to solve the
problem. For a solution in stresses one must integrate the system formed by the
equations of equilibrium (5.1) and by the Eq. (5.42). One cannot use only the latter
equations because they do not contain the equations of equilibrium (these differ-
ential equations are of the first order and represent conditions much more
restrictive that the ones imposed by the Beltrami-Michell equations, which are
differential equations of second order).
In the absence of the volume forces, the equations of equilibrium have the form
rji;j ¼ 0; i ¼ 1; 2; 3: ð5:10 Þ
We notice that these equations correspond to the conditions of equating to zero the
divergence of the tensor Tr ; hence, this tensor must be the curl of another tensor
and we may write
rij ¼ jkl fli;k ; ð5:46Þ
so that
ijm jkl fli;k ¼ 0; ð5:470 Þ
analogically, we get
ijm jkl fli ¼ knp gnm;p ;
where gnm is an asymmetric tensor too. Taking into account (A.38), we may write
this relation in the form
fmk ¼ fii dkm þ kpn gnm;p
where Fij is Finzi’s symmetric tensor, the components of which must be functions
of class C3 : This representation has been given by Beltrami, but its tensor form is
due to Finzi. The given demonstration ensures its completeness for the equations
of equilibrium (5.10 ), in case of a simply connected domain.
In case of non–zero volume forces, the Eq. (5.1) allow particular integrals of the
form
Z Z Z
r11 ¼ F1 dx1 ; r22 ¼ F2 dx2 ; r33 ¼ F3 dx3 ; ð5:52Þ
it results, necessarily,
1
Fij0 ¼ kði;jÞ ¼ ðki;j þ kj;i Þ; ð5:550 Þ
2
where ki ; i ¼ 1; 2; 3; are the components of an arbitrary vector of class C3
We notice thus that if the potential functions Fij lead to a certain state of stress,
then the potential functions Fij þ Fij0 ; i; j ¼ 1; 2; 3; give the same state of stress.
We can choose always the functions ki so that Fij þ Fij0 ¼ 0 for i 6¼ j or for i ¼ j;
obtaining Maxwell’s representation or Morera’s one, respectively; indeed, this
0 0 0
corresponds to the fact that if the components F23 ; F31 ; F12 of the symmetric
0 0 0 0
tensor Fij are given, then the other components F11 ; F22 ; F33 are determined by the
Eq. (5.55) and conversely. Hence, Maxwell’s and Morera’s representations are, as
well, complete for a simply connected domain.
We also mention that between Maxwell’s and Morera’s potential functions one
has the relations (the other relations are obtained by circular permutations)
F22;33 þ F33;22 ¼ 2F23;23 ; ð5:56Þ
One observes that the representations (5.57) and (5.570 ) verify the Eqs. (5.560 ) and
(5.56), respectively. Thus, one can also determine the functions of two variables
which appear by the indefinite integration (which may be effected, the potential
functions being of class C3 ) and which have not been put in evidence yet.
Hence, if we take into account only the equations of equilibrium (5.10 ), then we
can represent the state of stress by means of only three potential functions of class
C3 :
where the potential functions Hij ¼ Hij ðx1 ; x2 ; x3 Þ; i; j ¼ 1; 2; 3; are the compo-
nents of Schaefer’s tensor; both these functions and the function X ¼ Xðx1 ; x2 ; x3 Þ
will be of class C3 : From
1
Hij ¼ Fij þ ðFll þ XÞdij ð5:580 Þ
2
one observes that the components of Schaefer’s tensor TH are univocally deter-
mined, excepting the arbitrary function X:
Introducing the expressions (5.58) in the representation (5.51), we get the
representation
rij ¼ ikl jmn Hkm;ln þ DðHll XÞdij ðHll XÞ;ij ; i; j ¼ 1; 2; 3; ð5:59Þ
where we took into account the formula (A.38). If the stresses given by (5.59)
verify the Beltrami Eq. (5.420 ) too, then we may write
DDX ¼ 0: ð5:620 Þ
We notice that the condition (5.62) is satisfied if the function X verifies the
equation
1
DX ¼ Hkl;kl ; ð5:63Þ
1m
taking into account (5.61), the condition (5.620 ) is verified too.
Returning to the relations (5.58) between Finzi’s and Schaefer’s tensor, we
notice that for i 6¼ j we have
Fij ¼ Hij ; i 6¼ j; ð5:64Þ
while the second case corresponds to Maxwell’s stress functions, which must be
biharmonic
DDFij ¼ 0; i ¼ j: ð5:650 Þ
Taking into account (5.61), (5.63) and (A.37), the representation (5.59) reads
rij ¼ mDXdij þ X;ij 2Xkði;jÞk ; ð5:66Þ
this being Schaefer’s representation; the seven stress functions must be of class
C4 .
We notice that the sum of normal stresses is given by
rll ¼ ð1 þ mÞDX; ð5:67Þ
taking into account (5.30 ), we can write the strains in the form
1
eij ¼ ðX;ij 2Hkði;jÞk Þ; i; j ¼ 1; 2; 3: ð5:68Þ
2G
5.2 Statical Problem. Potential Functions 213
In this case, Cauchy’s relations (5.2) show that the state of displacement is given
by
2Gui ¼ X;i 2Hij;j : ð5:69Þ
Although supplementary conditions have been put in the above demonstration,
the representation (5.66), (5.69) is complete for a simply connected domain; to put
this in evidence, we will show the connection which can be made between this
representation and the representations in displacements given above.
Introducing the notation
Ui ¼ Hij;j ; i ¼ 1; 2; 3; ð5:70Þ
we notice that
X ¼ divC; ð5:76Þ
where we have neglected a function which satisfies the harmonic equation,
because it can be included in the biharmonic vector C.
If in the representation
2Gu ¼ gradX 2U ð5:720 Þ
we take into account (5.75), (5.76), then we find again Galerkin’s representation
(5.25); because this latter representation is complete for a simply connected
domain, it results that Schaefer’s representation is complete too, in the same
conditions.
214 5 General Equations. Formulation of Problems
If the volume forces are non-zero, then we may use the same representations
(5.66), (5.69) or (5.71), (5.72). In the latter case, we notice that the equations of
equilibrium and Beltrami-Michell equations are verified if the vector U is given by
the equation
DU ¼ F; ð5:730 Þ
while the function X is given by the Eq. (5.74); applying the operator D to this
equation, we notice that the function X must satisfy the equation
1
DDX ¼ divF; ð5:740 Þ
1m
From the point of view of the particular solution which corresponds to given
volume forces, we observe that one can replace the Eq. (5.74) by the Eq. (5.740 ).
Indeed, the Eq. (5.740 ) is equivalent to the equation
D½ð1 mÞDX divU ¼ 0;
which leads to
ð1 mÞDX divU ¼ u;
U ¼ U þ U0 ; ð5:7300 Þ
Duij ¼ 0; i; j ¼ 1; 2; 3: ð5:87Þ
In the case of non-zero volume forces, one can use the same representation,
adding to the biharmonic functions Fij particular integrals of the equations
1 h m i
DDFij ¼ 2Fði;jÞ þ Fk;k dij ; i; j ¼ 1; 2; 3; ð5:910 Þ
2þm 1m
thus, the Beltrami-Michell Eq. (5.42) are identically verified.
On the basis of Almansi’s formula (A.100), we may write the biharmonic
functions Fij in the form
1 1
Fij ¼ Wij þ ðxi Uj þ xj Ui Þ ; i; j ¼ 1; 2; 3; ð5:93Þ
2ð2 þ mÞ 2
in this case, we get a new representation for the state of stress, in the form
1
rij ¼ Uði;jÞ ðxk Uk þ U0 Þ;ij ; i; j ¼ 1; 2; 3; ð5:94Þ
2ð2 þ mÞ
where we have introduced the function U0 ¼ U0 ðx1 ; x2 ; x3 Þ given by
U0 ¼ Wll : ð5:95Þ
We notice that the functions Ui ¼ Ui ðx1 ; x2 ; x3 Þ; i ¼ 1; 2; 3, are the components
of the vector U ¼ Uðx1 ; x2 ; x3 Þ; we must have
DU ¼ 0; DU0 ¼ 0 ð5:96Þ
both the vector potential and the scalar one being functions of class C4 :
The formula (5.30 ) allows to write the state of strain in the form
1 1 m
eij ¼ Uði;jÞ ðxk Uk þ U0 Þ;ij Uk;k dij ; i; j ¼ 1; 2; 3; ð5:97Þ
2G 2ð2 þ mÞ 2þm
218 5 General Equations. Formulation of Problems
where we have not put in evidence functions of two variables which appear by
integration.
To find the conditions in which the state of stress given by (5.93) corresponds to
the basic system of equations of the theory of elasticity, this representation must
verify the equations of equilibrium (5.10 ) too; we get thus the equations
D½ð2 þ mÞFij;j Fkk;i ¼ 0; i ¼ 1; 2; 3; ð5:99Þ
which lead to
ð2 þ mÞFij;j Fkk;i ¼ ui ; i ¼ 1; 2; 3; ð5:990 Þ
where ui ; i ¼ 1; 2; 3; are harmonic functions. We may write this system also in the
form
1
Fij;j þ Fik;k ¼ ½F11 þ F22 þ F33 ð2 þ mÞFii ;i ð!Þ;
2þm ð5:9900 Þ
i 6¼ j 6¼ k 6¼ i; i ¼ 1; 2; 3;
the function ui being contained in the harmonic part of the biharmonic functions
Fij ; i 6¼ j; i; j ¼ 1; 2; 3:
One can show that the general integral of the system of homogeneous equations
Fij;j þ Fik;k ¼ 0 ð!Þ; i 6¼ j 6¼ k 6¼ i; i ¼ 1; 2; 3; ð5:100Þ
is of the form
Fij ¼ ðfik fjk Þ;k ð!Þ; i 6¼ j 6¼ k 6¼ i, i; j ¼ 1; 2; 3; ð5:1000 Þ
where the functions fji ¼ fij ¼ fij ðxi ; xj Þ; i 6¼ j; i; j ¼ 1; 2; 3, of class C5 depend each
one on only two variables and are biharmonic too
DDfij ¼ 0; i; j ¼ 1; 2; 3: ð5:10000 Þ
One adds a particular integral of the system of Eq. (5.10000 ), which can be easily
obtained, observing that this system is equivalent to the system
1
Fij;ij ¼ Fkk;kk ðFii;ii þ Fjj;jj þ Fkk;kk Þ
2
1 1
þ DðF11 þ F22 þ F33 Þ ðF11 þ F22 þ F33 Þ;kk ð!Þ;
2þm 2
i 6¼ j 6¼ k 6¼ i; i ¼ 1; 2; 3; ð5:10000 Þ
As in the static case, the use of potential functions is very important in the dynamic
case too. In the following, we present formulations in displacements, by means of
displacement functions, and formulations in stresses, with the aid of stress func-
tions. In this study, we restrain us to the case of small nondamped motions.
Eliminating the stresses and the strains, one obtains a formulation in displacements
of the problems of elastodynamics, by means of Lamé’s equations. Starting from
these equations, we deduce the Somigliana-Iacovache, Sternberg-Eubanks and
Lamé-Clebsch representations.
Eliminating the stresses between the equations of motion (5.6) and Hooke’s law
(5.3), we get the equations
kell;i þ2leij;j þFi ¼q€
ui ; i ¼ 1; 2; 3; ð5:101Þ
1 o2
hi ¼ D ; i ¼ 1; 2; ð5:103Þ
c2i ot2
takes place; thus, the propagation velocity of the longitudinal waves c1 is strictly
greater than the propagation velocity of the transverse waves c2 (c1 [ c2 ). In the
pffiffiffiffiffiffiffi
limit cases, one obtains c1 ¼ 2c2 for m ¼ 0 or c1 ¼ 1 for m ¼ 1=2 but we remark
that, in these limit cases, we must start from Hooke’s law written in the corre-
sponding form.
We notice that between the operators (5.104) introduced above there exists a
relation of the form
2ð1 mÞh1 ð1 2mÞh2 ¼ D: ð5:106Þ
The vector form of the Eq. (5.102) reads
lh2 u þ ðk þ lÞ grad div u þ F ¼ 0: ð5:1020 Þ
Using the technical elastic constants, we have
1 1
h 2 ui þ h;i þ Fi ¼ 0; i ¼ 1; 2; 3; ð5:107Þ
1 2m G
or
1 1
h2 u þ grad div u þ F ¼ 0: ð5:1070 Þ
1 2m G
In a developed form, Lamé’s equations become
1 1
h 2 u1 þ ðu1;1 þ u2;2 þ u3;3 Þ;1 þ F1 ¼ 0;
1 2m G
1 1
h 2 u2 þ ðu1;1 þ u2;2 þ u3;3 Þ;2 þ F2 ¼ 0; ð5:10700 Þ
1 2m G
1 1
h 2 u3 þ ðu1;1 þ u2;2 þ u3;3 Þ;3 þ F3 ¼ 0:
1 2m G
In the absence of volume forces, Lamé’s equations depend only on an elastic
constant of the material, as in the static case; they read
1
h2 ui þ h;i ¼ 0; i ¼ 1; 2; 3; ð5:108Þ
1 2m
or
1
h2 u þ grad div u ¼ 0: ð5:1080 Þ
1 2m
If we apply the operator div to the Eq. (5.1020 ) or to the Eq. (5.1070 ) and take
into account the relation (5.106), then it results the equation
1 1 2m
h1 h ¼ div F ¼ div F; ð5:109Þ
k þ 2l 2ð1 mÞ
5.3 Dynamical Problem. Potential Functions 221
which must be verified by the volume strain; in the absence of the volume forces or
in case of a solenoidal field of volume forces, the cubic strain will verify the
equation of longitudinal waves
h1 h ¼ 0: ð5:110Þ
Applying the operator h1 to the Eq. (5.1020 ) and taking into account (5.109),
one sees that the displacement vector must verify the equation
kþl 1
h1 h2 u ¼ grad div F h1 F; ð5:111Þ
lðk þ 2lÞ l
taking into account the relation (A.920 ) between the differential operators of sec-
ond order; one can make observations analogue to those in Sect. 5.2.1.1.
In the absence of volume forces, the displacement vector verifies the double
waves equation
h1 h2 u ¼ 0: ð5:112Þ
We notice that the index 1 corresponds to the primary waves (P waves); as it
will be later seen, these waves are irrotational, longitudinal waves of dilatation.
The index 2 corresponds to the secondary waves (S waves); these waves are shear,
transverse waves which correspond to a deformation under constant volume, hence
to an incompressible (isochore) deformation.
We mention that the set of integrals of the Eq. (5.111) includes the set of
integrals of the Eq. (5.1020 ); but the Eq. (5.111) is not sufficient to solve the
problem. In a solution in displacements one must integrate the equation (5.1020 ).
To verify the Eq. (5.102), it is sufficient that the displacements be functions of
class C2 : The equations of continuity (2.68) are identically verified, after replacing
the strains by Cauchy’s relations (5.2), if the displacements are functions of class
C3 ; finally, to can verify the double wave Eq. (5.112) the displacements must be
functions of class C4 : The components of the volume force can be of class C0 ;
provided we do not raise the problem of Eq. (5.109) or of Eq. (5.111); otherwise,
they must be of class C1 or of class C2 ; respectively.
We remark that the Eq. (5.1020 ) can be equally written in the form
1
c22 h2 u þ ðc21 c22 Þ grad div u þ F ¼ 0; ð5:113Þ
q
where we introduced the velocities of wave propagation and the mass force
ð1=qÞF.
Taking into account the relation (A.920 ), we can write Lamé’s equation in the form
ðk þ 2lÞh1 u þ ðk þ lÞ curl curl u þ F ¼ 0 ð5:114Þ
or in the form
1 1
h1 u þ curl curl u þ F¼0 ð5:1140 Þ
2ð1 mÞ k þ 2l
222 5 General Equations. Formulation of Problems
As in the static case, starting from the homogeneous Eq. (5.1080 ), one can intro-
duce a function u by means of the relation
2G div u ¼ ð1 2mÞh2 u; ð5:116Þ
with
h1 C0 ¼ 0:
Taking into account Boggio’s formula (A.107), which allows to write the vector
C as a sum of two vectors, verifying each one a simple wave equation, one can
introduce the function C0 in one of the components of the vector which verifies the
waves equation of index 1.
The representation (5.118) leads now to the Somigliana-Iacovache representation
2G u ¼ 2ð1 mÞh1 C grad div C; ð5:122Þ
where the functions Ci ¼ Ci ðx1 ; x2 ; x3 ; tÞ of class C4 must verify the double waves
equations
h1 h2 Ci ¼ 0; i ¼ 1; 2; 3: ð5:1210 Þ
Using Cauchy’s relations (5.2) and Hooke’s law (5.3), one obtains the state of
stress in the form
rij ¼ 2ð1 mÞh1 Cði;jÞ þ ðmh2 dij oi oj ÞCk;k ; i; j ¼ 1; 2; 3: ð5:123Þ
In the case in which the volume forces are non-zero, one can use the same
representation (5.122), where the Somigliana-Iacovache vector must verify the
equation
1
h1 h2 C þ F ¼ 0; ð5:124Þ
1m
as it can be easily seen by replacing the representation (5.122) in the Eq. (5.1070 ); a
particular integral is sufficient to introduce the influence of the volume forces.
We notice that, taking into account (5.106) and (A.920 ), we may write the
representation (5.122) also in the form
2G u ¼ ð1 2mÞh2 C curl curl C: ð5:12200 Þ
224 5 General Equations. Formulation of Problems
As in the static case, if one uses Boggio’s formula (A.107), then we notice that
in the Somigliana-Iacovache representation intervene six functions which verify
simple wave equations and must be determined by means of three conditions at a
point of the contour. There appears thus a triple functional indetermination; the
state of displacement is univocally determined, but the displacement functions do
not have the same property.
As in the Sect. 5.2.1.3, we notice that the function in the representation (5.117) is
of the form (5.120) and get the first Sternberg-Eubanks representation in the form
(5.122). In this representation one uses a vector potential v ¼ vðx1 ; x2 ; x3 ; tÞ, which
verifies the transverse wave Eq. (5.118), as well as a scalar potential
v0 ¼ v0 ðx1 ; x2 ; x3 ; tÞ, which, taking into account the relation (A.108), must verify
the equation
h1 v0 þ r h1 v ¼ 0; ð5:125Þ
thus, the two potentials cannot be uncoupled, as in the static case. The first
Sternberg-Eubanks representation is, as well, complete for a simply connected
domain.
The state of stress is given by the formulae (5.125).
If the volume forces are non-zero, then one can use the same representation
(5.125), the vector potential verifying the Eq. (5.126), while the scalar potential is
given by the Eq. (5.125) too.
Taking into account the expressions (5.103) of d’Alembert’s operators and the
Eq. (5.126), one may write the Eq. (5.125) also in the form
1þmq
h 1 v0 þ r€
v ¼ 2r F: ð5:1250 Þ
1 mE
Using the formula (A.107), one can write the Somigliana-Iacovache vector in
the form
C ¼ , þ ,0 ð5:126Þ
too, where
h2 , ¼ 0; h1 ,0 ¼ 0: ð5:1260 Þ
Denoting by
,0 ¼ div,0 ð5:127Þ
a scalar potential which verifies the longitudinal wave equation
h1 ,0 ¼ 0; ð5:1270 Þ
5.3 Dynamical Problem. Potential Functions 225
introducing the functions (5.126) in the representation (5.122) and taking into
account (5.1250 ), we obtain a new representation in displacements of the form
2Gu ¼ 2ð1 mÞh1 , gradðdiv, þ ,0 Þ; ð5:128Þ
where , is a vector potential, which verifies the longitudinal wave equation, while
,0 is a scalar potential, which verifies the transverse wave equation.
Taking into account the relation (5.106), the expression (5.103) of d’Alembert’s
operator and the Eq. (5.1270 ), one gets the second Sternberg–Eubanks represen-
tation in the form
q
2Gu ¼ ,€ gradðdiv , þ ,0 Þ: ð5:129Þ
l
One obtains Sternberg’s representation too, equivalent with the above one, in
the form
2Gu ¼ D, gradðdiv , þ ,0 Þ: ð5:130Þ
All these representations are complete for a simply connected domain. The
potential functions , ¼ ,ðx1 ; x2 ; x3 ; tÞ and ,0 ¼ ,0 ðx1 ; x2 ; x3 ; tÞ must be of class C4 :
In components, we have
2Gui ¼ D,i ð,j;j þ ,0 Þ;i ; i ¼ 1; 2; 3; ð5:1300 Þ
One must notice that, in what concerns these last results, one cannot obtain the
static case by a simple particularization of the dynamic problem. It is true that
some results can be obtained by using such a particularization, but nor all results
can be obtained in this way. This happens because a representation of Almansi
type is not a representation of Boggio type, the two representations being essen-
tially different.
Observing that the cubic strain is given by
2Gh ¼ D,0 ; ð5:131Þ
we get the state of stress in the form
m
rij ¼ D,ði;jÞ ð,k;k þ,0 Þ;ij D,0 dij ; i; j ¼ 1; 2; 3: ð5:132Þ
1 2m
If the volume forces are not zero, then we can express them univocally in the form
F ¼ DF0 ; ð5:133Þ
226 5 General Equations. Formulation of Problems
Taking into account the notation (5.127), the representation (5.130) reads
u ¼ div ,0 ð5:136Þ
and the vector potential W ¼ Wðx1 ; x2 ; x3 ; tÞ with the aid of the relation
W ¼ curl ,; ð5:137Þ
We also notice that the vector potential must verify the condition
div W ¼ 0: ð5:14000 Þ
5.3 Dynamical Problem. Potential Functions 227
Hence, the corresponding field is a solenoidal one. But this condition is not nec-
essary to verify identically Lamé’s equations; taking into account this notice and
the fact that the representation in the previous subsection is complete, it results that
the representation (5.138), with or without the condition (5.14000 ) is complete.
In components, we may write
2Gui ¼ u;i þ ijk wk;j ; i ¼ 1; 2; 3; ð5:1380 Þ
1
h2 r23 þ H;23 ¼ ðF2;3 þ F3;2 Þ;
1þm
1
h2 r31 þ H;31 ¼ ðF3;1 þ F1;3 Þ; ð5:1500 Þ
1þm
1
h2 r12 þ H;12 ¼ ðF1;2 þ F2;1 Þ:
1þm
Applying the operator h1 to the Eq. (5.149) and taking into account d’Alem-
bert’s operators (5.103), we get the equations
1
h1 h2 rij ¼ 2h1 Fði;jÞ þ ðoi oj mh2 dij ÞFk;k ; i; j ¼ 1; 2; 3; ð5:151Þ
1m
which are verified by the components of the stress tensor; in the absence of volume
forces, the stresses verify the double wave equation
h1 h2 rij ¼ 0; i; j ¼ 1; 2; 3; ð5:152Þ
We notice that the stresses must be functions of class C2 to can verify the
Eq. (5.149); if we wish to verify the Eq. (5.151) too, then these functions must be
of class C4 :
We also notice that the set of integrals of the Eq. (5.151) contains the set of
integrals of the Eq. (5.149), the first equations being not sufficient to solve the
problem. For a solution in stresses one must integrate the system formed by the
equations of Beltrami-Michell type (5.149) and by the equations of motion (5.6)
(one must take into account the equations of motion too, because they represent
more restrictive conditions than those imposed by the Beltrami-Michell type
equations). To this system one must associate Hooke’s law (5.30 ), where we take
into account Cauchy’s relations (5.2), because in the equations of motion appear
the displacements too.
We start from the equations of motion (5.6), in the absence of the volume forces,
and express the displacement vector in the form
ui ¼ uji;jkk ¼ DUji;j ; i ¼ 1; 2; 3; ð5:153Þ
ðrij q€
uki;kj Þ;j ¼ 0; i ¼ 1; 2; 3;
using the considerations made in Sect. 5.2.2.2 and taking into account (A.38), we
obtain
1
rij ¼ ikl jmn Fkm;ln þ q D€
uij þ 2€
u½kj;ki þ ijk mnp u
€ mn;pk ; ð5:15400 Þ
2
where we have put in evidence the antisymmetric part of the tensor uij .
Introducing the notations
DuðijÞ ¼ F ij ; ð5:156Þ
which correspond to the symmetric and to the antisymmetric part of uij ; respec-
tively, we may write
1 0 0 0
Duji þ 2u½kj;kl þ ijk mnp umn;pk ¼ F ij ijk Fkl;l þ 2jkl Fki;l þ ijk Fli;k :
2
We also introduce the notation
5.3 Dynamical Problem. Potential Functions 231
0 0 0
ikl F kj;l ¼ ijk Fkl;l þ 2jkl Fki;l þ ijk Fll;k ;
€ €
Þ; i; j ¼ 1; 2; 3;
ij þ ikl F
rij ¼ ikl jmn Fkm;ln þ qðF kj;l ð5:157Þ
€ €
þ F
ij þ ikl F €
rij ¼ ikl jmn Fkm;ln þ qðF kj;l jkl ki;l Þ; i; j ¼ 1; 2; 3; ð5:1570 Þ
too, because the new term introduced leads to rij;j ¼ 0 and does not influence the
equations of motion; but the tensor
þ F
ikl F
kj;l jkl ki;l
is symmetric with respect to the indices i and j; so that it can be contained in the
ij ; without losing the generality. Finally we obtain the Beltrami-Finzi type
tensor F
representation, given by Teodorescu,
€
ij ; i; j ¼ 1; 2; 3;
rij ¼ ikl jmn Fkm;ln þ qF ð5:159Þ
ij;j ; i ¼ 1; 2; 3:
ui ¼ F ð5:1590 Þ
where F1 ; F
2; F
3 are functions of class C1 with respect to space variables and of
2
class C with respect to time, while F1 ; F2 ; F3 are integrable functions.
r ¼F þF €
þ qF ; ð5:161Þ
22 33;11 11;33 22
€ ;
r33 ¼F11;22 þ F22;11 þ qF 33
€ Þ;
r23 ¼ ðF23;1 þ F31;2 þ F12;3 Þ;1 þ qF 23
r ¼ ðF
31 þF 31;2 þ F Þ þ qF
12;3
€ Þ;23;1 ;2 31 ð5:1620 Þ
€ Þ;
r12 ¼ ðF12;3 þ F23;1 þ F31;2 Þ;3 þ qF 12
12;2 þ F
u1 ¼ F 31;3 ; u2 ¼ F
23;3 þ F
12;1 ; u3 ¼ F
31;2 þ F
23;2 : ð5:16200 Þ
On the way indicated in Sect. 5.2.2.4, one can obtain a representation analogue to
that given by Schaefer in the static case, starting from the representation (5.159),
5.3 Dynamical Problem. Potential Functions 233
(5.590 ) of Beltrami-Finzi type. We choose thus the components of the tensor stress
functions Fij ; F ij ; in the form (5.150) and in the form
ij þ ðH
ij ¼ H
F ij ;
ll XÞd ð5:163Þ
ij ¼ H
where the potential functions Hij ¼ Hij ðx1 ; x2 ; x3 ; tÞ, H ij ðx1 ; x2 ; x3 ; tÞ are
0
given by (5.150 ) and by
H ij þ 1 ðF
ij ¼ F ij ;
ll þ XÞd ð5:1630 Þ
2
respectively; these functions are univocally determined, excepting the arbitrary
¼ Xðx
functions X ¼ Xðx1 ; x2 ; x3 ; tÞ and X 1 ; x2 ; x3 ; tÞ.
Introducing in the representations (5.160), (5.1590 ), we get the representation
ij;j þ ðH
ui ¼ H ;
ll XÞ ð5:1640 Þ
;i
ij;ij þ DðH
H ¼ ui;i ¼ H
ll XÞ; ð5:16400 Þ
if the stresses (5.164) satisfy the Eq. (5.1490 ) too, then one obtains
Dh2 X þ qh2 X€
m q € € þH €€
€ kl;kl þ 2qH €€
þ 2
ðDHll 2DX ll 3qXÞ ¼ 0: ð5:1670 Þ
2ð1 m Þ l
Supposing that following relations
Hll þ 2lH ¼0
ll ¼ 0; X þ 2lX ð5:168Þ
1
h1 X ¼ Hkl;kl ; ð5:170Þ
ð1 mÞ
taking into account (5.166), we notice that the function X must satisfy the double
waves equation
h1 h2 X ¼ 0; ð5:1700 Þ
1 € € d : ð5:171Þ
ll DXÞ ð1 2mÞqX
þ ½Hkl;kl mDHll þ ð1 mÞðqH
1þm ij
Taking into account the conditions (5.166), (5.168), (5.170), one may write the
conditions (5.171) in the form
5.3 Dynamical Problem. Potential Functions 235
1 q € €
kði;jÞk ;
Hll;ij þ DHij 2Hkði;jÞk Hll dij qHij ¼ H ð5:1710 Þ
2l 2l
while the second case corresponds to Maxwell’s stress functions. Which must
satisfy the double waves equation
h1 h2 Fij ¼ 0; i ¼ j: ð5:1730 Þ
Taking into account (5.166), (5.168), (5.170), (5.172) and (A.37), the repre-
sentation (5.164), (5.16400 ) becomes
rij ¼ mh2 Xdij þ X;ij 2Hkði;jÞk ; i; j ¼ 1; 2; 3; ð5:174Þ
one takes into account (5.76) and (5.179), then one finds again the Somigliana-
Iacovache representation (5.122); the latter representation being complete for a
simply connected domain, it results that Teodorescu’s representation is complete
in the same conditions too.
In a developed form, one can write the state of displacement in the form (5.72),
the state of stress being given by
As in the static case, from the point of view of the particular solution, corre-
sponding to given volume forces, we notice that one can replace the Eq. (5.178) by
the Eq. (5.182); indeed, the last equation is equivalent to the equation
h2 ½ð1 mÞh1 X Ui;i ¼ 0;
which leads to
ð1 mÞh1 X Ui;i ¼ u;
where the function u satisfies the transverse waves equation. Taking into account
(5.181), one can make a decomposition of the form (5.7300 ), where the function U
satisfies the transverse waves equation, while U0 is a particular integral. We notice
that the function u may be included, as in the static case, in the components of the
vector U; hence, we may use the Eq. (5.181) to obtain the particular integral
corresponding to the given volume forces.
An important particular case is that of conservative forces, of the form (5.77),
where v ¼ vðx1 ; x2 ; x3 ; tÞ is a function of class C2 with respect to the space vari-
ables; one can choose a particular integral of the form (5.78), where U ¼
1 ; x2 ; x3 ; tÞ is a function of class C with respect to the space variables and of
Uðx 4
ð1 mÞh1 x ¼ DU
2ð1 mÞh1 U; ð5:1840 Þ
it results, in this case, that the function x must be a particular integral of the
equation
1 2m
h1 x þ v ¼ 0: ð5:185Þ
1m
Thus, the particular solution can be obtained by means of only one stress function
x, which must be of class C3 with respect to space variables and of class C2 with
238 5 General Equations. Formulation of Problems
respect to time. If we wish to satisfy the Beltrami-Michell type equations too, then
the function x must be of class C4 with regard to space variables.
The considerations made in Sect. 5.2.2.4 concerning the boundary conditions
remain valid in this case too.
As in the static case, one can obtain a representation in stresses of the dynamical
problem of the theory of elasticity starting from the Beltrami type Eq. (5.1490 ); the
condition to verify the equations of motion and Hooke’s law too is then put.
One introduces a function F by the relation
H ¼ ð1 þ mÞh2 F; ð5:186Þ
in this case, the Eq. (5.1490 ) show that the stresses must be of the form
m q o2
rij ¼ oi oj þ d ij F þ uij ; i; j ¼ 1; 2; 3; ð5:187Þ
2ð1 mÞ l ot2
where uij ¼ uji are functions which verify the transverse waves equation
h2 uij ¼ 0; i; j ¼ 1; 2; 3: ð5:188Þ
which is complete for the system of equations of Beltrami type, in case of a simply
connected domain; the six stress functions Fij must be of class C4 .
In case of non-zero volume forces, one can use the same representation, the
functions Fij being particular integrals of the equations
1 h m i
h1 h2 Fij ¼ Fij þ Fk;k dij ; i; j ¼ 1; 2; 3; ð5:1920 Þ
2þm 1m
thus, the Eq. (5.149) of the Beltrami-Michell type are identically verified.
Using Boggio’s formula (A.107), we may write the functions Fij in the form
where
h1 Wij ¼ 0; h2 Uij ¼ 0; i; j ¼ 1; 2; 3; ð5:1940 Þ
thus, we obtain a new representation for the state of stress in the form
m q o2
rij ¼ ð2 þ mÞh1 Uij oi oj þ dij ðUll þ U0 Þ; i; j ¼ 1; 2; 3; ð5:195Þ
2ð1 mÞ l ot2
both the tensor potential and the scalar one being functions of class C4 with respect
to all variables.
In a developed form, one can write the state of stress in the form
2
o m q o2
r11 ¼ð2 þ mÞh1 U11 þ ðU11 þ U22 þ U33 þ U0 Þ;
ox21 2ð1 mÞ l ot2
2
o m q o2
r22 ¼ð2 þ mÞh1 U22 þ ðU11 þ U22 þ U33 þ U0 Þ; ð5:1950 Þ
ox22 2ð1 mÞ l ot2
2
o m q o2
r33 ¼ð2 þ mÞh1 U33 þ ðU11 þ U22 þ U33 þ U0 Þ;
ox23 2ð1 mÞ l ot2
240 5 General Equations. Formulation of Problems
To show the conditions in which the state of stress given by (5.183) corresponds
to the basic system of equations of elastodynamics, one can make a study analogue
to that in the static case (see Sect. 5.2.2.5).
References
A. Books
1. Finzi, B., Pastori, M.: Calcolo tensoriale e applicazioni, 2nd edn. Zanichelli, Bologna (1961)
2. Gurtin, M.E.: The Linear Theory of Elasticity. Encyclopedia of Physics, vol. VI a/2, Springer,
Berlin-Heidelberg-New York (1972)
3. Haimovici, M.: Teoria elasticitătßii (The Theory of Elasticity). Ed. did. ped, Bucuresßti (1969)
4. Jaunzemis, W.: Continuum Mechanics. The Macmillan Comp, New York-London (1967)
5. Kecs, W., Teodorescu, P.P.: Applications of the Theory of Distributions in Mechanics. Ed.
Academiei, Bucuresßti, Abacus Press, Tunbridge Wells, Kent (1974)
6. Lamé,G.: Leçons sur la théorie mathématique de l’élasticité des corps solides. Paris (1852)
7. Moisil, Gr. C.: Matricele asociate sistemelor de ecuatßii cu derivate partßiale. Introducere în
studiul cercetărilor lui I. N. Lopatinschi (Matrices Associated to Systems of Partial
Derivative Equations. Introduction to the Study of I. N. Lopatinski’s Researches).
Ed. Academiei, Bucuresßti (1950)
8. Nowacki, W.: Dynamics of Elastic Systems. Chapman and Hall Ltd., London (1963)
9. Nowacki, W.: Teoria spre_zistošci (Theory of Elasticity). Państ. Wyd. Nauk, Warsawa (1970)
10. Sedov, L.I.: Mehanika splošnoi sredy (Mechanics of Deformable Media). I. II., Izd. Nauka,
Moskva (1970)
11. Teodorescu, P.P.: Dynamics of Linear Elastic Bodies. Ed. Academiei, Bucuresßti, Abacus
Press, Tunbridge Wells, Kent (1975)
B. Papers
12. Beltrami, E.: Osservazioni sulla nota precedente. Atii R. Accad. dei Lincei. Rend. Cl. Sci. fis.
mat. e nat., ser. 5(3), 141 (1892)
13. Blokh, V.I.: Funktsii napryzhenii v teorii uprugosti (Stress functions in the theory of
elasticity). Prikl. mat. i mekh 14, 415 (1950)
14. Bondarenko, B.A.: Ob odnom klass reshenii dinamicheskikh uravnenii teorii uprugosti (On a
class of solutions of dynamical equations in the theory of elasticity). Akad. Nauk. Uzbek.
SSR, Trudy Inst. Mat. Mekh. 21, 41 (1957)
15. Finzi, B.: Integrazione delle equazioni indefinite della meccanica dei sistemi continui. Atti.
R. Acad. dei Lincei, Rend., Cl. Sci. fis. mat e nat., ser. 6, 19, 578, 620 (1934)
References 241
}
38. Rieder, G.: Uber die Spezialisierung des Schaeferschen Spannungsfunktionenansatzes in der
räumlichen Elastizitätstheorie. Z. A. M. M. 44, 329 (1964)
39. Schaefer, H.: Die Spannungsfunktionen des dreidimensionalen Kontinuums und des
elastischen Körpers. Z. A. M. M. 33, 356 (1953)
40. Schaefer, H.: Die Spannungsfunktionen einer Dyname. Abh. Braunschweig. Wiss. Ges 7, 107
(1955)
41. Schaefer, H.: Die Spannungsfunktionen des dreidimensionalen Kontinuums: statische
Deutung und Randwerte. Ing.–Aechiv, 28, 291 (1959)
42. Schuler, K.W., Fosdick, R.L.: Generalized Beltrami Stress Functions. Dept. Mech. Rept.,
Illinois Inst. Techn. (1967)
43. Somigliana, C.: Sulle espressioni analitiche generali dei movimenti oscillatori.
Atti. R. Accad. dei Lincei, Rend., Cl. Sci. fis., mat., ser. 5, 1, 111 (1892)
44. Soós, E.: The Galerkin vector for the dynamic problems of an elastic isotropic and non-
homegeneous body. Rev. Roum. Math. Pures et Appl. 10, 855 (1965)
45. Sternberg, E.: On the integration of the equations of motion in the classical theory of
elasticity. Arch. Rat. Mech. Analysis 6, 34 (1960)
46. Sternberg, E., Eubanks, R.A.: On stress functions for elastokinetics and the integration of the
repeated wave equation. Quart. Appl. Math 15, 149 (1957)
47. Teodorescu, P.P.: Sur une solution générale du problème en espace de la théorie de
l’élasticité. IXme Congrès IUTAM, Bruxelles, 1956, Actes, 5, 155 (1957)
48. Teodorescu, P.P.: Sur une représentation par potentiels dans le problème tridimensionnel de
l’élastodynamique. C. Rend. hebd. des séances de l’Acad. Sci., 250, 1972 (1960)
}
49. Teodorescu, P.P.: Uber das kinetische Problem nichthomogener elastischer K} orper. Bull.
Acad. Pol., sér. Sci. Techn., 12, 595 [867] (1964)
50. Teodorescu, P.P.: Schwingungen der elastischen Kontinua. III Konferenz f} ur nichtlineare
Schwingungen, Berlin, 1964, Abh. der deutschen Akad. Wiss., Kl. Math., Phys. u. Techn. 2,
29 (1965)
51. Teodorescu, P.P.: Sur le tenseur de Finzi et sur quelques de ses applications et
généralisations. Ann. Mat. Pura ed Appl., ser. IV, 84, 225 (1970)
52. Teodorescu, P.P.: Sur l’introduction des fonctions-potentiel en élasticité linéaire. An. Univ.
Bucuresßti, mat.–mec. 20, 131 (1971)
}
53. Teodorescu, P.P.: Uber ein Analogon der Schaeferschen Darstellung in der Elastokinetik und
einige Anwendungen. Z. A. M. M., 52, Sonderheft, T 154 (1972)
54. Teodorescu, P.P.: Stress functions in three-dimensional elastodynamics. Acta Mech. 14, 103
(1972)
55. Truesdell, C.: invariant and complete stress functions for general continua. Arch. Rat. Mech.
Anal. 4, 1 (1959/1960)
56. Weber, C.: Spannungsfunktionen des dreidimensionalen Kontinuums. Z. A. M. M. 28, 193
(1948)
57. Wo_zniak, Cz.: Introduction to dynamics of deformable bodies. Arch. Mech. Stos. 19, 647
(1967)
Chapter 6
Principles and General Theorems
of the Theory of Elasticity. Computation
Methods
We shall state in what follows some principles and general theorems of the theory
of elasticity, laying stress chiefly on the ideas connected with the notions of work
of deformation and external work.
Hereafter, we shall make a survey of some of the most important methods used
in the theory of elasticity, insisting on the method of Fourier representations, as
well as on the methods based on the theory of distributions.
6.1.1 Work
In what follows we introduce the notion of work, putting in evidence the external
work and the work of deformation.
The bodies submitted to the action of external loads undergo certain deformations,
while the point of application of these loads move about, by certain paths, and
yield a work, called work of external loads or external work.
We have seen in Sect. 4.1.2.2 that the elementary work, corresponding to a
force P; which imparts to the application point a displacement u; is expressed in
space co-ordinates (we consider to be in the case of infinitesimal deformations) in
the form
dW ¼ P du ¼ P udt
_ ¼ Pi dui ¼ Pi u_ i dt; ð6:1Þ
thus, the external work corresponding to the volume forces F and to the superficial
n
forces p is given by
ZZZ ZZ
dWe n
¼ F udV_ þ p udS
_
dt
ZZZ V ZZ S
n
¼ Fi u_ i dV þ pi u_ i dS: ð6:2Þ
V S
The formula is valid in the dynamical case. In the statical one, we may assume
n
that, in a relatively small interval of time, the intensity of the external forces Fi ; pi
increases from zero till the final intensity; the displacements ui increase, as well,
from zero to their final magnitude. Such an action is called statical and the work
does not depend on time; in this case, the work of the force of component Pi ;
which produces the displacement ui ; which corresponds to an element of volume
or of area, is given by
Z ui
i d
P ui ð!Þ; ð6:3Þ
0
where P i is an intermediary value of the external force, while ui is the corre-
sponding displacement. Because the phenomenon is linear, we can assume that
i ¼ ki P0i ; Pi ¼ ki P0i ð!Þ;
P ð6:4Þ
If the force is suddenly applied, having from the very beginning its entire
intensity, while the displacement instantaneously appears, the factor 1/2 disappears
in the above relation; such an action is called dynamical, although it is a particular
case of such an action. In this case, one can write
which leads to
Z 1
Pi u_ i dt ¼ P0i u0i ð!Þ; ð6:60 Þ
1
The work affected by the external loads is stored in the elastic body and given
back, by the unloading and the returning to its initial form, in the form of work of
internal forces or internal work; the latter is called the work of deformation too,
corresponding to the strain energy.
We shall denote by Ws the specific work of deformation, i.e. the work per unit
volume in the neighbourhood of a point of the body, produced by the state of stress
due to the state of strain of the body at the respective point. We shall give the name
of elementary work of deformation to the corresponding work of a volume element
dV ¼ dx1 dx2 dx3 of the body and will have
dWi ¼ Ws dV; ð6:9Þ
246 6 Principles and General Theorems. Computation Methods
we may express the this work in the form (4.510 ), which represents the formula of
Clapeyron [19], or under one of the formulae (4.90)–(4.90000 ), (4.93)–(4.95), (4.98).
If we refer to the principal directions, then we can write
1
W ¼ kh2 þ lðe21 þ e22 þ e23 Þ
2
1
¼ ½ð1 þ mÞðr21 þ r22 þ r23 Þ mH2
2E
1
¼ ðr1 e1 þ r2 e2 þ r3 e3 Þ: ð6:12Þ
2
These results have the same form both in the case of statical and dynamical
actions.
The formulae of Green [23] are given by the relations (4.9)–(4.900 ), which express
the fact that the partial derivative of the specific work of deformation with respect
to a strain is equal to the corresponding stress. As well, the formulae of Castigliano
[2, 18] (4.59)–(4.5900 ) show that the partial derivative of the specific work of
deformation with respect to a stress is equal to the corresponding strain.
One observes, from (4.9) to (4.100 ), that if Green’s formulae take place, then the
elementary work of deformation is an exact differential, wherefrom the tensor Tr
is conservative. As well Castigliano’s formulae (4.59) correspond to the fact that
the tensor Te is conservative.
We will assume now that the external work, given by (6.7), is an exact dif-
n
ferential; hence, Fi and pi must be the partial derivatives with respect to the
corresponding displacements of a potential function in these displacements. In
case of the action of a concentrated volume force F ¼ PdðrÞ or of a concentrated
n
superficial force p ¼ PdðrÞ it results
dWe ¼ P du ¼ Pi dui ; ð6:13Þ
6.1 Principles and General Theorems of the Theory of Elasticity 247
if the force P is conservative in the sense considered above, then it results that
oWe
Pi ¼ ; i ¼ 1; 2; 3: ð6:130 Þ
oui
In a linear statical theory, we have
O
248 6 Principles and General Theorems. Computation Methods
equal to the external concentrated force which acts at this point on the direction of
the displacement.
The formula (6.18) allows to determine the constraint forces at the fixity points
of the elastic solids, where one has d ¼ 0:
In the case of another type of concentrated load (e.g., a rotational concentrated
moment) the theorems hold too, appearing a corresponding type of displacement
(e.g., a rotation).
In the dynamical case, the relations (6.17), (6.170 ) remain still valid at a given
moment t.
If we start from the formula (6.2) and take into account the formulae (3.200 ), (3.63)
and the Gauss-Ostrogradskiı formula (A.960 ), which allows to pass from a surface
integral to a volume one, one may write
ZZ ZZ ZZZ ZZZ
n
pi u_ i dS ¼ rji u_ i nj dS rji u_ i ;j dV ¼ rji;j u_ i dV
S S V V
ZZZ ZZZ ZZZ
þ rji ð_eji þ x_ ji ÞdV ¼ ðq€ui Fi )u_ i dV þ rij e_ ij dV;
V V V
if the natural state of stresses and the initial state of deformations correspond to the
initial moment t ¼ 0; then one obtains the generalized principle of Clapeyron in
the form
We ¼ T þ Wi : ð6:20Þ
This result constitutes, in fact, a consequence of the general theorem of energy,
representing a conservation relation of it. We have obtained this principle as a
theorem.
In the statical case, if the body is in equilibrium at any moment t; we have
T ¼ 0 and obtain the relation
We ¼ Wi ; ð6:200 Þ
which represents the principle of Clapeyron [19].
we remark that the factor 1/2 does not appear here, because the external loads and
the stresses act, in this case, with their full intensity from the very start. By using
the relations (6.23), (6.230 ), one may obtain the relation (6.22) on a way analogous
to that in case of the proof of Clapeyron’s relation.
If in this principle we substitute d’Alembert’s lost forces to the volume forces,
we shall obtain the d’Alembert-Lagrange principle in the form
ZZZ ZZZ
dWe ¼ dWi þ q€ui dui dV ¼ dWi ðq€ui Þdui dV; ð6:24Þ
V V
therefore, the external work, yielded by the external loads through the agency of a
system of virtual displacements (that drive the body out of its position of elastic
equilibrium) is equal to the difference between the internal work produced by the
state of stress of the body through the agency of the virtual state of strain corre-
sponding to the assumed system of virtual displacement (relations (6.21)) and the
work of the forces of inertia corresponding to the same system of virtual
displacements.
We have to mention that the principle of virtual displacements and the d’Al-
embert-Lagrange principle express the condition of static equilibrium and the
condition of dynamic equilibrium and the limit conditions (the equations of con-
tinuity being verified by the manner of applying the principles), respectively.
In the static case, it has been shown that this principle may be applied to all
systems which verify Betti’s reciprocity principle (see Sect. 6.1.2.7).
We remark, on the other hand, that the internal work yielded by the unloading
of the body is equal in absolute value but of an opposite sign to the work yielded
when loading it. In the statement presented above, we took into consideration the
work yielded by the unloading; if we make use of the internal work produced by
the external loads dWi0 ¼ dWi we can express the principle of virtual displace-
ments in the form
which shows that the total work corresponding to an elastic solid, in case of virtual
displacements, vanishes. In the dynamic case, the d’Alembert-Lagrange principle
reads
ZZZ
0
dWe þ dWi þ ðq€ui Þdui dV ¼ 0; ð6:240 Þ
V
If, in state of the virtual displacements, we take into consideration the virtual
variations of the state of stress (defined analogically), for which the equations of
equilibrium ði ¼ 1; 2; 3Þ
drij ;j þdFi ¼ 0; ð6:25Þ
Thus, the external work, yielded by some system of virtual variations of the
external loads through the agency of the displacements of the body, is equal to the
internal work, yielded by the corresponding virtual variation (see the relations
(6.25), (6.250 )) of the state of stress through the agency of the state of strain of the
body.
The principle of the virtual variations of the state of stress expresses the con-
ditions of continuity of the body (the equations of equilibrium and the conditions
on the boundary being included in the manner of application of the principle). This
principle too may be written in the form (6.220 ).
Let us consider an elastic body the states of which vary continuously between the
times t0 and t1 : Integrating the relation (6.24) on the assumed time interval, we
obtain
Z t1 Z t1 Z t1 ZZZ
dWe dt ¼ dWi dt þ dt q€ui dui dV;
t0 t0 t0 V
252 6 Principles and General Theorems. Computation Methods
the second integral in the right member of this relation can be related to the
variation of the kinetic energy. Indeed, we may introduce the kinetic energy in the
form (6.19), wherefrom
ZZZ ZZZ ZZZ
o
dT ¼ qu_ i du_ i dV ¼ q ð€ui dui ÞdV q€ui dui dV;
V V ot V
if we admit that the virtual displacements dui are such that they vanish at times t0
and t1 (they are synchronous virtual displacements) it follows that
Z t1 Z t1 ZZZ
dTdt ¼ dt q€ui dui dV; ð6:27Þ
t0 t0 V
Since in (6.23) we consider that the external loads do not vary, we can remove the
operator d from under the integral (volume integral); we introduce the potential
energy
P ¼ Wi 2We ; ð6:29Þ
where We is the external work, corresponding to the static case, supplied by (6.7).
By so doing, the relation (6.28) leads to Hamilton’s principle in the form
Z t1
d ðP T Þdt ¼ 0: ð6:30Þ
t0
The above integral takes the name of action; Hamilton’s principle affirms that
the action is steady, while the motion of the elastic body corresponds to the
extremals of this functional, in case of synchronous variations of the
displacements.
Because in (6.23) we suppose that the external loads do not vary, while in (6.26)
we consider that the displacements have the same property, we can remove the
operator d from under the integral sign, so that we may write, in the static case,
dP ¼ 0; ð6:31Þ
6.1 Principles and General Theorems of the Theory of Elasticity 253
where the potential energy is given by (6.29). This relation represents the principle
of the minimum potential energy; indeed, one obtains for P an extreme value and
one can prove that it corresponds to a minimum.
According to this principle, the real displacements ui ; corresponding to the
deformation of the elastic body, are those for which the potential energy is minimum
(Lagrange’s principle); Wi is expressed here by means of the components of the
displacement vector.
As well, from all the states of stress statically possible (which equilibrate the
external loads), takes place that state of stress for which the potential energy is
minimum (Castigliano’s principle); Wi is expressed here by means of the com-
ponents of the stress tensor.
We mention that, in case of a non-linear relation between stresses and strains,
the expressions (6.230 ) and (6.260 ) of the first variation of Wi will be different. In
this case, the energy obtained by using a virtual variation of the state of stress is
called complementary energy; thus, Castigliano’s principle will be the principle of
the minimum complementary energy. The potential energy and the complementary
energy have the same value only in the case of a linear constitutive law (see also
Sect. 4.1.2.6).
It is convenient to apply the principle of the minimum complementary energy in
case of a solution in stresses of the problems of the theory of elasticity; as well, the
principle of the minimum potential energy is useful in case of a solution in
displacements.
We consider the states of strain and stress the variations of which do not affect the
external loads or their corresponding displacements; we have, in this case,
n
dFi ¼ dpi ¼ 0; i ¼ 1; 2; 3; ð6:32Þ
or
dui ¼ 0; i ¼ 1; 2; 3: ð6:320 Þ
Taking into account (6.23) or (6.26), it results
dWi ¼ 0; ð6:33Þ
relation which expresses the principle of the minimum internal work. In this case,
we can assert that among all the states of stress (or of strain) statically corre-
sponding to the given external loads, takes place only that state of stress (or of
strain) which minimizes the internal work (the deformation be produced with a
minimum expense of energy).
As a consequence of the results stated above, we notice as well that the
hyperstatic efforts, to which correspond null displacements, are determined by the
condition that the internal work be minimum for these efforts.
254 6 Principles and General Theorems. Computation Methods
Considering two states of strain and stress of an elastic body, marked by (0 ) and (00 ),
corresponding to two distinct systems of external loads, we shall be able to express
the work yielded by one the loads through the agency of the displacements cor-
responding to the other load in the form (one observes that the factor 1/2 does not
appear, because the external loading acts with the whole intensity along the
displacements)
ZZZ ZZ n ZZZ ZZ n
0 00 0 00 00 0
Fi ui dV þ pi ui dS; Fi ui dV þ p00i u0i dS;
V S V S
where we have used the relations (3.200 ), (3.210 ), (3.61) and the Gauss-Ostro-
gradskiı formula (A.950 ). Thus, there result the relations
ZZZ ZZ n ZZZ
Fi0 u00i dV þ p0i u00i dS ¼ r0ij e00ij dV; ð6:34Þ
V S V
ZZZ ZZ n
ZZZ
Fi00 u0i dV þ p00i u0i dS ¼ r00ij e0ij dV: ð6:340 Þ
V S V
Taking into account Hooke’s law (4.56), valid for a linearly elastic body with
linear anisotropy, one has the relation
so that the right members of the relations (6.34), (6.340 ) are equal. Thus, it results
ZZZ ZZ n ZZZ ZZ n
Fi0 u00i dV þ p0i u00i dS ¼ Fi00 u0i dV þ p00i u0i dS; ð6:36Þ
V S V S
we thus obtain the principle of reciprocity of the work (Betti’s principle [17]) in
the static case. This principle asserts the fact that the work effected by a system of
external loads by corresponding displacements of another system of external loads
is equal to the work effected by the second system of external loads by the
displacements corresponding to the first system.
6.1 Principles and General Theorems of the Theory of Elasticity 255
Replacing the volume forces by the lost forces of d’Alembert, we may write
ZZZ ZZ n
0
0 00
Fi q€
ui ui dV þ p0i u00i dS
V S
ZZZ ZZ n
00
¼ Fi q€ u00i u0i dV þ p00i u0i dS; ð6:37Þ
V S
we get thus the principle of reciprocity of the work in the dynamic case, where the
system of external loads includes the inertia forces too.
D. Graffi [21, 22] showed that, in the case of homogeneous initial conditions, one
can state a theorem of reciprocity in the form (the time variable was brought into
relief, by assuming that t0 ¼ 0)
ZZZ Z t ZZ Z t
0 00 n
dV Fi ðt sÞui ðsÞds þ dS p0i ðt sÞu00i ðsÞds
V 0 S 0
ZZZ Z t ZZ Z t
00 0 n
¼ dV Fi ðt sÞui ðsÞds þ dS p00i ðt sÞu0i ðsÞds: ð6:38Þ
V 0 S 0
the first index of the displacement shows the point where it is produced, while the
second index shows the point where the external load that yields this displacement
is acting.
If dij is the component of dij along the direction of Pi , while dji is the component
of dji along the direction of Pj , one may write the relation (6.39) also in the form
hence, dij are coefficients of influence for the lines of influence of the displacements
in case of movable concentrated forces.
In case of n points at which we calculate the displacements, we may solve the
system (6.41); we obtain
X
n
Pi ¼ cij dj ; ð6:410 Þ
j¼1
obviously, in the case of a non bounded domain these conditions are no more
necessary.
We mention that the impossibility to prove a theorem of existence may arise
because, for the given body and in the given conditions of loading, the considered
equations do not describe accurately the physical phenomenon.
Data about this problem are to be found in the monograph of W. Nowacki [12],
in the monograph of V. D. Kupradzhe, T. G. Gegelya, M. C. Basheleıshvili and T.
V. Burchuladzhe [8] and in the monograph of M. Gurtin [5].
Another question arises about the number of solutions that this problem admits.
In any case, if a solution is found, indifferent on what way, it means that it exists;
but it is important to know if this solution is unique or not. To this, a theorem of
uniqueness due to G. R. Kirchhoff [27], in the static case, which affirms that given
a perfectly elastic body, occupying a finite domain, isotropic, homogeneous, avoid
of given initial stresses, to which can be applied the superposition of effects, on
which acts a system of external loads, which increase continuously from zero to the
maximum value, with corresponding strains and rigid body local rotations neg-
ligible with respect to unity and with no influence whatsoever on the equations of
equilibrium, a single state of strain and stress possible. We mention that the
domain occupied by the body must be simply connected (otherwise, initial stresses
would appear).
We remark that, in the absence of the external loads (superficial loads and
volume forces), the state of strain and stress vanish if these are not initial stresses.
Indeed, from the above hypotheses it results that We ¼ 0; while Clapeyron’s
principle (6.200 ) shows that one must have Wi ¼ 0: But Wi is given by (6.10); from
the formulae (6.12) we see that the elastic potential W is a definite positive
quadratic form, both in stresses and in strains. Consequently, when Wi vanishes,
258 6 Principles and General Theorems. Computation Methods
W vanishes too, but this only occurs simultaneously with the vanishing of all the
components of the strain tensor or concomitantly with the vanishing of all the
components of the stress tensor, which had to be proved; one may have at most a
displacement and a rotation of rigid body. This result is valid also in case of null
displacements on the boundary.
We shall now suppose that to a system of given external loads correspond,
under the same boundary conditions, two states of strain and stress. By subtracting
the two loadings (volume forces and superficial loads, including eventually con-
ditions in displacements on the boundary), as well as the two states of strain and
stress (application of the principle of superposition of effect), we obtain a state of
strain and stress which corresponds to external loads equal to zero (including
boundary conditions equal to zero), hence, a null state of strain and stress. By
virtue of the above conclusions, the two states of strain and stress supposed to take
place will coincide, making abstraction of a global motion of rigid body. Hence, if
the problem of elastostatics has a solution, it will be a single one; the theorem of
uniqueness is thus proved.
We remark that the above result is valid both for the two fundamental problems
as well as for the mixed fundamental problem. The nondetermined motion of rigid
body appears only in case of the second fundamental problem (conditions in
stresses on the boundary); it can be specified by conditions of fixity of the elastic
solid. In the case of the other two problems, this motion is determined by the
conditions on the boundary where displacements does appear.
The theorem remains valid in more general conditions, e.g., in the case of
anisotropic bodies; as well, it is valid for infinite domains, if one puts certain
supplementary conditions of regularity of the solution at infinity. In case of
multiply connected domains one puts, as well, certain supplementary conditions.
The above results are valid as long as the assumptions made in establishing
them are satisfied, especially the hypothesis which states that W is a positive
definite quadratic form. But it is possible that, owing to the instability of the
material (by passing into the plastic domain, by the appearance of yield phe-
nomena etc.), W no longer assumes a positive definite quadratic form. On the other
hand, fundamental changes are liable to occur in the equations of the problem; we
can thus have finite deformations. The theorem of uniqueness is valid as long as
the elastic displacements do not affect the action of the external loads. There are
however cases when these deformations, not negligible any longer, must be taken
into account when writing the equations of equilibrium. In these cases, we cannot
prove that the solution is unique; there can be several possible forms of static
equilibrium for the same system of external loads. This leads us to problems of
elastic instability (eigenvalue problems or branching problems). We can be
equally led to this kind of phenomena in case of the action of non-conservative
external loads, as well as in the case when these loads are functionals of the body
deformation or of the story of this deformation.
We must mention that, in all the cases when the principle of superposition of
effects is applicable, the state of strain and stress, corresponding to the action of
the external loads, is not affected by the eventual existing of initial stresses and can
6.1 Principles and General Theorems of the Theory of Elasticity 259
be computed as in the absence of such stresses. The total stress is obtained, in this
case, by superposition of effects. But, if this principle is not applicable, then the
state of stress due to the external loads can no longer be determined if the initial
conditions are not known from the very beginning. These stresses can be produced
by temperature variations or by many other causes.
In the dynamic case, a theorem of uniqueness due to Fr. Neumann [10], who
states that, given a perfectly elastic body, isotropic, homogeneous, devoid of initial
stresses, to which can be applied the principle of superpositions of effects, on
which a system of external loads acts, admitting given initial conditions, with
corresponding strains and rigid body local rotations negligible with respect to
unity and with no influence whatsoever on the equations of motion, a single state of
strain and stress is possible.
We remark that, in the absence of the external loads (superficial loads and
volume forces) and under homogeneous (vanishing) initial condition, the state of
strain and stress vanish if there are no initial stresses. Indeed, from the above
hypotheses, it follows that We ¼ 0; while the generalized Clapeyron’s principle
(6.20) shows that we must have T þ Wi ¼ 0; since both these quantities are
positive, it follows that we must simultaneously have T ¼ Wi ¼ 0: As in the static
case, from Wi ¼ 0 it result that all the components of both strain and stress tensors
must vanish. Likewise, from the vanishing of the kinetic energy T; it results that
the displacement velocities must vanish. Therefore, the state of strain and stress is
reduced to a rigid body motion that, owing to the homogeneous initial conditions,
vanishes to.
We shall now suppose that to a system of given external loads correspond,
under assumed initial conditions, two states of strain and stress under the same
boundary conditions. By subtracting the two loadings (volume forces, superficial
loads, hence boundary conditions in displacements too), we shall find a state of
strain and stress, corresponding to external loads equal to zero (including boundary
conditions equal to zero) and to initial conditions equal to zero too, therefore—by
virtue of the above conclusions—to a state of strain and stress equal to zero; the
above two states of strain and stress will coincide, while if the problem of elas-
todynamics has a solution, then it will be a single one and the theorem of
uniqueness is proved.
In the dynamic case, one can make considerations analogous to those made in
the static case.
L. T. Wheeler and E. Sternberg [48] extend the theorem of uniqueness, dem-
onstrated in the case of some finite domains, to infinite domains. The demon-
stration of this result is based on the generalization of a theorem of energetical
nature, given by S. Zaremba [49] for the scalar waves equation, independently
rediscovered by A. Rabinovich [31], discussed later on by Fritz John and resumed
by K. O. Friedrichs and H. Lewy [20] and by R. Courant [3, 4]. Similar results
were supplied, in the anisotropic case, by L. T. Wheeler [47].
With the results given by M. E. Gurtin and E. Sternberg [24] in the case of finite
domains, it can be shown that the theorem of uniqueness for the second funda-
mental problem remains equally valid in the case of infinite domains, even when
260 6 Principles and General Theorems. Computation Methods
the two wave propagation velocities, which have real values, are no more related
by (5.5). We mention moreover that the results given in [24] were extended by
M. E. Gurtin and R. A. Toupin [25] to anisotropic bodies. R. J. Knops and
L. E. Payne [28] presented moreover, in the case of finite domains, a theorem of
uniqueness for the weak solutions in elastodynamics.
To solve the chief problems raised by the practice, a particularly important prin-
ciple is usually applied: the principle of Barré de Saint-Venant [32], stated by him
in 1855.
According to Saint-Venant, this principle is stated as follows: If upon an elastic
body acts a system of external loads in static equilibrium, the state of strain and
stress within in is practically equal to zero, except in a zone of the order of
magnitude of the boundary on which these loads are acting (Fig. 6.3)
We shall bring into relief the following corollary, practically important in
practical applications: The state of strain and stress within an elastic body, that
admits the principle of superposition of the effects, are practically independent
from the manner of applying the external loads, excepting in the zone neigh-
bouring the boundary on which these loads are applied (Fig. 6.4a, b, c).
We must mention that this principle can only be applied when the strain and the
rigid body local rotations can be neglected with respect to unity; for instance, it is
inapplicable, in the above mentioned form, to bodies with thin walls.
Saint-Venant’s principle is widely applied in practice. It allows, for instance, to
schematize the actual loading mode of the elements of construction, giving a
mathematical formulation of the problems raised; this principle allows us to obtain
most of the state of strain and stress of the body. So as to find the state of strain and
stress in the zones where the external loads are applied, the study must be com-
pleted by a local one, that helps to determine the local states of strain and stress
born there. Such a problem takes the name of contact problem and involves special
methods of approximating the physical phenomenon.
The above formulation of the principle of Saint-Venant has the disadvantage of
an insufficient accuracy and of a rather empirical character. Later on, justifications
of a theoretical nature of this principle were searched for and endeavours were
ZZ ZZ
n n
pi dA ¼ 0; xj pi dA ¼ 0; i; j ¼ 1; 2; 3: ð6:44Þ
S S
6.1.4.1 Properties
We shall call a simply connected domain that one for which each closed surface
R in its interior (which has not common points with the boundary) can be reduced
to a point, by a continuous deformation. In the three-dimensional case, e.g., one
may define another type of connection, replacing the closed surface R by a closed
curve C. Thus, e.g., the domain contained between two concentric spheres is
simply connected according to the second definition but is doubly connected
according to the first one; in the following discussion we shall confine ourselves to
the first of the above given definitions. In the two-dimensional case, the simply
connection is obviously defined by means of a closed curve C.
Let now be a domain D, bounded by a surface S and having n 1 closed
surfaces S1 ; S2 ; . . .; Sn1 ; within the domain and outside of one another; we
assume, obviously, that these surfaces do not posses multiple points. Let equally
be the closed surface R1 within the domain D and enclosing within it the surface
S1 ; we remark that the surface R1 can be reduced to the sole surface S1 ; by
continuous deformation, contrary to the closed surface R; that can be reduced to a
point (Fig. 6.5a; for sake of simplicity, it will correspond to the plane case). This
property of the multiply connected domains is a fundamental one and helps to
define them; in our case, the domain is n-connected.
6.1 Principles and General Theorems of the Theory of Elasticity 263
(a) (b)
S1 1
S1 D
S2 S n-1
S2 S n-1
D
S S
(c) (d)
S1
S1 S n-1 c1 S n-1
D c2 D
S2 S2 S
S
Fig. 6.5 Multiply connected domain (a). Cuts: reduction to a simply connected domain (b, c),
equivalent cuts (d)
(k)
u+
The strains will be continuous and single-valued functions, since the contri-
bution of the difference of displacements between one part and the other of the cut
vanishes. Taking into account the formulae (2.71)–(2.71000 ) which emphasize the
rigid body local motion, we can write
ðkÞ ðkÞ
uþ u ¼ uðkÞ þ xðkÞ rðkÞ ; k ¼ 1; 2; . . .; n 1; ð6:45Þ
ðkÞ ðkÞ
xþ x ¼ xðkÞ ; k ¼ 1; 2; . . .; n 1; ð6:46Þ
where rðkÞ is the position vector of a point of the cut; in components, one may write
ðkÞ ðkÞ ðkÞ ðkÞ
uþj uj ¼ uj jlm xl xðkÞ
m ; j ¼ 1; 2; 3; k ¼ 1; 2; . . .; n 1; ð6:450 Þ
ðkÞ ðkÞ ðkÞ
xþj xj ¼ xj ; j ¼ 1; 2; 3; k ¼ 1; 2; . . .; n 1: ð6:460 Þ
ðkÞ ðkÞ
The constant vectors uðkÞ and xðkÞ and the constant scalars uj and xj ; j ¼
1; 2; 3; respectively, are characteristic for every cut ck ; k ¼ 1; 2; . . .; n 1; being
the same for each point of a cut, as well as for two equivalent cuts.
Inversely, if, in the absence of the external loads, in a multiply connected body,
in which the necessary number of cuts is performed so as to transform it into a
simply connected body, we displace the faces of each cut by a translation uðkÞ and
a rotation xðkÞ and then suppress the cuts by sticking their faces in their new
positions after the displacement, then a state of stress will be born due exclusively
to these displacements. These displacements are called distorsions, while the
stresses they yield are called initial stresses.
In this case, the theorems of uniqueness can no more be applied. A multiply
connected elastic body, under no external loads, is not necessarily in a natural
state. Besides, we can affirm that the state of strain and stress of a multiply
connected elastic body is determined by the external loads and by the six com-
ponents of the distorsion with regard to each of the cuts transforming the multiply
connected body into a simply connected one. Since the components of the dis-
torsion can be arbitrarily chosen, this leads to a multiple-valued solution.
6.1 Principles and General Theorems of the Theory of Elasticity 265
A thorough study of these problems was performed, in the static case, at the
very beginning of twentieth century century by Vito Volterra [45, 46]. He showed
that, to an arbitrary distribution of distorsions, corresponds a unique state of strain
in a multiply connected body undergoing the action of external loads. This
theorem of uniqueness corresponds to the theorems of Kirchhoff and Neumann for
simply connected bodies in the static and in the dynamic case, respectively.
Results in this direction are to be found equally in the volume [16] published by
E. Volterra in 1960.
To compute the internal work in case of a multiply connected body, to the
expression (6.10) of the internal work of a simply connected domain we must add
an additional term in the form
ZZ
1Xn1
ðkÞ
Wi ¼ ðuþ uðkÞ ðkÞ
Þ p dS
ðkÞ
2 k¼1 SðkÞ
ZZ
1Xn1
ðkÞ ðkÞ ðkÞ
¼ ðuþj uj Þ pj dSðkÞ ; ð6:47Þ
2 k¼1 SðkÞ
where SðkÞ is the area of the corresponding cut. Introducing the torsor of the stress
vectors which appear on the cut cðkÞ
ZZ
PðkÞ ¼ pðkÞ dSðkÞ ; k ¼ 1; 2; . . .; n 1; ð6:48Þ
SðkÞ
ZZ
MðkÞ ¼ rðkÞ pðkÞ dSðkÞ ; k ¼ 1; 2; . . .; n 1; ð6:49Þ
SðkÞ
and, taking into account that in a scalar triple product one can interchange the
scalar and the vector products, we may write the additional term in the form
1Xn1
Wi ¼ uðkÞ PðkÞ þ xðkÞ MðkÞ ; ð6:470 Þ
2 k¼1
if
ZZ
ðkÞ ðkÞ
Pj ¼ pj dSðkÞ ; j ¼ 1; 2; 3; k ¼ 1; 2; . . .; n 1; ð6:480 Þ
SðkÞ
ZZ
ðkÞ ðkÞ
Mj ¼ jlm xl pðkÞ ðkÞ
m dS ; j ¼ 1; 2; 3; k ¼ 1; 2; . . .; n 1; ð6:490 Þ
SðkÞ
6.2.1.2 Particular Integrals for the Harmonic and for the Biharmonic
Equations
we are led to
d dUðuÞ
DUðuÞ ¼ 2 UðuÞ þ 2u ¼ 0;
du du
pffiffiffi
we get the integrals 1= u and const, hence the harmonic functions
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
; const; R ¼ x21 þ x22 þ x23 6¼ 0; ð6:52Þ
R
268 6 Principles and General Theorems. Computation Methods
where R is the polar radius. These integrals are useful in case of a spherical
symmetry of pole O. By using Almansi’s formulae, we get also the biharmonic
functions
xi
R; R2 ; ; i ¼ 1; 2; 3; R 6¼ 0: ð6:53Þ
R
The condition R 6¼ 0 is essential; indeed, if we have to do with distributions, the
results in Sect. A.3.4.3 allow to write
1
D ¼ 4pdðRÞ; DDR ¼ 8pdðRÞ: ð6:54Þ
R
Assuming other relations between the three variables, one may obtain, ana-
logically, other particular integrals.
A particularly fruitful method is that of separation of variables; we may thus
choose a function of the form
Uðx1 ; x2 ; x3 Þ ¼ uðx1 ; x2 Þwðx3 Þ; ð6:55Þ
obtaining
d2 wðx3 Þ
DUðx1 ; x2 ; x3 Þ ¼ wðx3 ÞDuðx1 ; x2 Þ þ uðx1 ; x2 Þ ¼ 0;
dx23
d2 wðx3 Þ
k2 wðx3 Þ ¼ 0;
dx23
ðD k2 Þuðx1 ; x2 Þ ¼ 0;
Other harmonic functions are obtained by replacing some lines ‘‘sin’’ by ‘‘cos’’
or ‘‘sinh’’ by ‘‘cosh’’ or interchanging the rôle of the variables.
By means of the formula (A.1.100) of Almansi, one may then obtain particu-
larly useful biharmonic functions. Starting from these results, one can construct
Fourier representations (Fourier series or Fourier integrals), important for many
boundary value problems; the properties of evenness with respect to certain
variables are easily put in evidence.
Another method to obtain particular harmonic or biharmonic integrals consists
in the differentiation of the already obtained integrals with respect to one of the
variables; thus, differentiating 1=R; we get the harmonic functions
xi
; i ¼ 1; 2; 3; R 6¼ 0: ð6:58Þ
R2
Taking into account (6.54), in case of a distribution, one has
x o
i
D 2 ¼ 4p dðRÞ; i ¼ 1; 2; 3: ð6:580 Þ
R oxi
Because the harmonic and the biharmonic equations are linear partial differ-
ential equations, one may apply the principle of superposition of the effects; any
linear combination of particular integrals will be an integral of the respective
equation. Thus, one obtains, e.g. the Fourier representations mentioned above.
We shall pay a particular attention, in what follows, to the harmonic and bihar-
monic polynomials, due to their importance in various applications.
One can easily verify that following monomials are harmonic functions
const; x1 ; x2 ; x3 ; x2 x3 ; x3 x1 ; x1 x2 ; x1 x2 x3 ð6:59Þ
being the only monomials which have this property.
In general, let be a homogeneous polynomial of nth degree
XXX
Pn ðx1 ; x2 ; x3 Þ ¼ Ai1 i2 i3 xi11 xi22 xi33 ; i1 þ i2 þ i3 ¼ n; ð6:60Þ
i1 i2 i3
where all the indices are nonnegative. We notice that for i1 fix one has i2 þ i3 ¼
n i1 corresponding n i1 þ 1 different coefficients; making i1 to take all pos-
sible values, we find the total number of the coefficients to be determined, i.e.
270 6 Principles and General Theorems. Computation Methods
1
ðn þ 1Þ þ n þ ðn 1Þ þ þ 2 þ 1 ¼ ðn þ 2Þðn þ 1Þ: ð6:61Þ
2
The condition of harmonicity leads to
XXX
Ai1 i2 i3 i1 ði1 1Þx1i1 2 xi22 xi33 þ i2 ði2 1Þxi11 xi22 2 xi33
i1 i2 i3
þi3 ði3 1Þxi11 xi22 x3i3 2 ¼ 0;
making a change of indices for the second and the third term in the square
brackets, one may write
XXX
½i1 ði1 1ÞAi1 i2 i3 þ ði2 þ 2Þði2 þ 1ÞAi1 2;i2 þ2;i3
i1 i2 i3
This polynomial is identically null if all its coefficients are identically null. By a
change of indices, we find that between the ðn þ 2Þðn þ 1Þ=2 coefficients must
take place the relations
ði1 þ 2Þði1 þ 1ÞAi1 þ2;i2 ;i3 þ ði2 þ 2Þði2 þ 1ÞAi1 ;i2 þ2;i3
þði3 þ 2Þði3 þ 1ÞAi1 ;i2 ;i3 þ2 ¼ 0; ð6:62Þ
there are n þ 1 coefficients on the first line and n coefficients on the second line.
The other coefficients Ai1 i2 i3 (for the other values of the index i1 ) may be obtained
as functions of the above ones by means of the recurrence relation (6.62).
We find thus that, for an even index i1 , the coefficients are given by
X
q
Clq 2ðq1Þ
A2q;j;k ¼ ð1Þq Clþ2ðq1Þ C2l
kþ2l A0;jþ2ðqlÞ;kþ2l ;
l¼0 C2l
2q
2q þ j þ k ¼ n; ð6:64Þ
which may be verified by complete induction; here Crs is the symbol of the
combination of s things, r at a time. Analogically, for an odd index i1 one has
6.2 Computation Methods 271
ð1Þq X Cq 2ðqlÞ 2l
q l
A2qþ1;j;k ¼ Cjþ2ðqlÞ Ckþ2l A1;jþ2ðqlÞ;kþ2l ;
2q þ 1 l¼0 C2l
2q
2q þ j þ k ¼ n 1; ð6:640 Þ
If n is an even number, then the polynomial Pn can be even with respect to all
three variables, where one may take the arbitrary constants A0;2r;n2r ; r ¼
0; 1; . . .; n=2; which are n=2 þ 1 ones, or may be even with respect to one variable
and odd with respect to the other variables, choosing thus the arbitrary constants
A0;2rþ1;n2r1 ; A1;2r;n2r1 ; A1;2rþ1;n2r2 ; r ¼ 0; 1; . . .; n=2 1; being n=2 of
each one. Analogically, if n is an odd number, then the polynomial Pn can be odd
with respect to all three variables and one can choose the arbitrary constants
A1;2rþ1;n2r2 ; r ¼ 0; 1; . . .; ðn 3Þ=2; which are ðn 1Þ=2 ones, or may be odd
only with respect to one variable and even with respect to the other variables,
choosing the arbitrary constants A1;2r;n2r1 ; A0;2rþ1;n2r1 ; A0;2r;n2r ; r ¼
0; 1; . . .; ðn 1Þ=2; being ðn þ 1Þ=2 of each one.
One can construct the 2n þ 1 linearly independent harmonic polynomials by
means of the formulae (6.64), (6.640 ). Thus, an even polynomial with respect to the
three variables, corresponding to the coefficient A0;2r;n2r ; reads
n=2 X
X n=2
Csþqr
q 2ðrsÞ 2ðsþqrÞ n2ðqþsÞ
Pn ðx1 ; x2 ; x3 Þ ¼ ð1Þq 2ðsþqrÞ
C2r Cn2ðqþsÞ x2q 2s
1 x2 x3 ; ð6:65Þ
q s C2q
where q þ s
n=2; one may have r ¼ 0; 1; . . .; n=2. The other polynomials may be
expressed analogically.
Starting from the above results, one may easily obtain homogeneous bihar-
monic polynomials.
Thus, starting from the harmonic polynomials (6.59) and using Almansi’s
representation, we obtain the biharmonic monomials
x21 ; x22 ; x23 ; x31 ; x32 ; x33 ; x1 x22 ; x1 x23 ; x2 x23 ; x2 x21 ; x3 x21 ; x3 x22 ;
x1 x32 ; x1 x33 ; x2 x33 ; x2 x31 ; x3 x31 ; x3 x32 ; x21 x2 x3 ; x22 x3 x1 ; x23 x1 x2 ;
x31 x2 x3 ; x32 x3 x1 x33 x1 x2 : ð6:66Þ
Using the formula (A.1.100), one may express a homogeneous biharmonic
polynomial of nth degree in the form
where Pn ; Pn2 are homogeneous harmonic polynomials; thus, one can state that a
homogeneous biharmonic polynomial of degree n [ 1; in three variables, depends
on ð2n þ 1Þ þ 2ðn 2Þ þ 1 ¼ 2ð2n 1Þ arbitrary constants, which one may
choose on the same considerations as in case of harmonic polynomials.
If the homogeneous biharmonic polynomial Qn has a certain evenness with
respect to the variables x1 ; x2 ; x3 ; then the constituent harmonic polynomials have
272 6 Principles and General Theorems. Computation Methods
the same evenness. Based on this observation and proceeding as above, one may
state following results: If n is an even number, then the polynomial Qn may be
decomposed in an even polynomial with respect to all three variables, which
depends on n þ 1 arbitrary constants, and three polynomials even only with respect
of one variable and odd with respect to the other variables, which depend each one
on n 1 arbitrary constants. As well, if n is an odd number, then the polynomial Q
is decomposed in an odd polynomial with respect to all three variables, which
depends on n 2 arbitrary constants, and three polynomials odd with respect to
one variable and even with respect to the other variables, which depend each one
on n arbitrary constants.
The properties of evenness or oddness with respect to the variables which occur
are particularly important in the case in which the body admits planes with
properties of geometrical and mechanical symmetry.
where
2pl 2pm
a1l ¼ ; a2m ¼ ; l:m ¼ 1; 2; 3; . . .; ð6:71Þ
L1 L2
L1 and L2 being the lengths of the periods on the two directions. The unknown
parameters are given by
Z Z
1
d00 ¼ pðn1 ; n2 Þdn1 dn2 ; ð6:72Þ
L1 L2 L1 L2
Z Z
2
bl0 ¼ pðn1 ; n2 Þ sin a1l n1 dn1 dn2 ;
L1 L2 L1 L2
Z Z
2
dl0 ¼ pðn1 ; n2 Þ cos a1l n1 dn1 dn2 ;
L1 L2 L1 L2
Z Z ð6:720 Þ
2
c0m ¼ pðn1 ; n2 Þ sin a2m n2 dn1 dn2 ;
L1 L2 L1 L2
Z Z
2
d0m ¼ pðn1 ; n2 Þ cos a2m n2 dn1 dn2 ;
L1 L2 L1 L2
274 6 Principles and General Theorems. Computation Methods
Z Z
4
alm ¼ pðn1 ; n2 Þ sin a1l n1 sin a2m n2 dn1 dn2 ;
L1 L2 L1 L2
Z Z
4
blm ¼ pðn1 ; n2 Þ sin a1l n1 cos a2m n2 dn1 dn2 ;
L1 L2 L1 L2
Z Z ð6:720 Þ
4
clm ¼ pðn1 ; n2 Þ cos a1l n1 sin a2m n2 dn1 dn2 ;
L1 L2 L1 L2
Z Z
4
dlm ¼ pðn1 ; n2 Þ cos a1l n1 cos a2m n2 dn1 dn2 :
L1 L2 L1 L2
The free term d00 represents the mean loading on a rectangular surface of
dimensions equal to the periods L1 and L2 .
At a point of discontinuity, the sum Sðx1 ; x2 Þof the series represents the
arithmetic mean of the two limits at the right and at the left, corresponding to the
two variables,
1
Sðx1 ; x2 Þ ¼ ½pðx1 0; x2 0Þ þ pðx1 0; x2 þ 0Þ
4
þ pðx1 þ 0; x2 0Þ þ pðx1 þ 0; x2 þ 0Þ: ð6:73Þ
with
Z L1 =2 Z L2 =2
16
alm ¼ pðn1 ; n2 Þ sin a1l n1 sin a2m n2 dn1 dn2 ; ð6:740 Þ
L1 L2 0 0
Analogically, one may use triple Fourier series, useful for periodic volume
loads (of periods L1 ; L2 ; L3 along the three directions). A volume loading F ¼
Fðx1 ; x2 ; x3 Þ; with properties of antisymmetry with the three planes of co-ordinates,
e.g., may be represented in the form
XXX
Fðx1 ; x2 ; x3 Þ ¼ almn sin a1l x1 sin a2m x2 sin a3m x3 ; ð6:75Þ
l m n
where
6.2 Computation Methods 275
Z L1 =2 Z L2 =2 Z L3 =2
64
alm ¼ Fðn1 ; n2 ; n3 Þ
L1 L2 L3 0 0 0
sin a1l n1 sin a2m n2 sin a3n n3 dn1 dn2 dn3 ; ð6:750 Þ
too, where
Z Z
1
klm ¼ pðn1 ; n2 Þeiða1l n1 þa2m n2 Þ dn1 dn2 ; ð6:760 Þ
L1 L2 L1 L2
the indices l and m may take any positive or negative entire value.
Analogically, for local loadings one uses double Fourier integrals
Z 1Z 1Z 1Z 1
1
pðx1 ; x2 Þ ¼ 2 pðn1 ; n2 Þ
4p 1 1 1 1 ð6:77Þ
ei½ða1 ðn1 x1 Þþa2 ðn2 x2 Þ da1 da2 dn1 dn2 :
Passing to trigonometric lines and noting that an integral between symmetric limits
of an odd function vanishes, one may represent the loading in the form
Z Z
1 1 1
pðx1 ; x2 Þ ¼ 2 pðn1 ; n2 Þ
p 0 0
cos a1 ðn1 x1 Þ cos a2 ðn2 x2 Þda1 da2 dn1 dn2 ð6:770 Þ
too; these results may be applied in the case in which the function pðx1 ; x2 Þ
accomplishes the conditions of Lejeune-Dirichlet and is absolutely integrable in
the whole plane, i.e. the integral
Z 1Z 1
pðn1 ; n2 Þdn1 dn2 ð6:78Þ
1 1
has sense. At the points of discontinuity, the above integral represents the arith-
metic mean of the four limits at the right and at the left too.
In the case of a symmetric loading with respect to both axes of co-ordinates, we
use an even function with respect to both variables
276 6 Principles and General Theorems. Computation Methods
Z 1 Z 1
pðx1 ; x2 Þ ¼ dða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2 ; ð6:79Þ
0 0
where
Z 1 Z 1
1
dða1 ; a2 Þ ¼ pðn1 ; n2 Þ cos a1 n1 cos a2 n2 dn1 dn2 ; ð6:790 Þ
4p2 0 0
with
Z Z Z
8 1 1 1
dða1 ; a2 ; a3 Þ ¼ Fðn1 ; n2 ; n3 Þ
p2 0 0 0
cos a1 n1 cos a2 n2 cos a3 n3 dn1 dn2 dn3 : ð6:800 Þ
ð6:81Þ
corresponding to a loading local along the direction x1 and periodic along the
direction x2 ; the variable coefficients of this representation are
Z L1 =2 Z 1
4
a0 ða1 Þ ¼ pðn1 ; n2 Þ cos a1 n1 dn1 dn2 ; ð6:810 Þ
pL1 0 0
Z L1 =2 Z 1
8
am ða1 Þ ¼ pðn1 ; n2 Þ cos a1 n1 cos a2 n2 dn1 dn2 : ð6:8100 Þ
pL1 0 0
6.2.1.5 Integrals for the Simple and for the Double Wave Equations
certain given limit conditions. To this end, it is useful to give some results con-
cerning the integration of these equations of hyperbolic type. In Sect. A.1.2.7 have
been presented Boggio’s relations which allow to construct the solution of the
double wave equation by means of the solutions of two simple wave equations. We
have thus to deal with the equation
h1 h2 Uðx1 ; x2 ; x3 ; tÞ ¼ 0: ð6:82Þ
If we choose a function U of the form (the method of separation of variables)
Uðx1 ; x2 ; x3 ; tÞ ¼ uðx1 ; x2 ; x3 Þ#ðtÞ; ð6:83Þ
then the equation (6.82) shows that the temporal function must satisfy an equation
of the form
€ x2 #ðtÞ ¼ 0:
#ðtÞ ð6:84Þ
Taking the sign -, it results an aperiodic motion; if we have x ¼ 0; then one
obtains a critical aperiodic motion. Both cases are less interesting in practice.
Taking the sign +, one obtains periodic vibrations. The temporal function will be
thus of the form
#ðtÞ ¼ A cos xt þ B sin xt: ð6:840 Þ
In general, the potential function will be of the form
X
1
Uðx1 ; x2 ; x3 ; tÞ ¼ Ui cosðxi t ui Þ; ð6:85Þ
i¼1
Z 1
Uðx1 ; x2 ; x3 ; tÞ ¼ Us0 ðx1 ; x2 ; x3 Þ cos stds
1
Z 1
þ Us00 ðx1 ; x2 ; x3 Þ sin stds: ð6:89Þ
1
To simplify the putting of the boundary conditions, the external loads will be
similarly expressed. The limit
lim PDt; ð6:890 Þ
P!1
Dt!0
sin a1i x1 sin a2i x2 ea3i x3 ; sin a1i x1 sin a2i x2 sinh a3i x3 ; ð6:91Þ
where
ea1i x1 ea2i x2 sin a3i x3 ; sinh a1i x1 sinh a2i x2 sin a3i x3 ;
ea1i x1 sinh a2i x2 sin a3i x3 ; ð6:92Þ
where
So as to find the potential functions and to solve the problems of the theory of
elasticity, two general computation methods are used: the inverse method and the
direct method.
280 6 Principles and General Theorems. Computation Methods
The variational methods are approximate ones, based on the extremal of a func-
tional. To do this, we choose certain expressions which depend on a number
(theoretically infinite, practically finite) of arbitrary parameters for the functions
282 6 Principles and General Theorems. Computation Methods
in the form
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ZZZ
s 1
As ¼ As dV ; ð6:96Þ
V V
0
where Fðx; y; y Þ represents a known real function of arguments x; y and
y0 dy=dx; the value of this integral depends on the choice of the function y ¼
yðxÞ; wherefrom it results the notation used and the denomination of functional.
We assume that the admissible arguments yðxÞ are of class C2 and that at the ends
of the interval ½x0 ; x1 they take the values y0 ; y1 ; in this case, the set fyðxÞg of
admissible arguments yðxÞ may be seen as a family of smooth curves, passing
through the points ðx0 ; y0 Þ and ðx1 ; y1 Þ; from which we must choose one which
minimizes the functional IðyÞ. A necessary condition which may determine this
curve is the Euler-Poisson equation
dFy0
Fy ¼ 0; ð6:970 Þ
dx
associated to the variational problem IðyÞ ¼ min, where Fy ; Fy0 represent the
partial derivatives with regard to the respective arguments; one obtains thus
d2 y dy
F y0 y0 2
þ Fy0 y þ Fy0 x Fy ¼ 0: ð6:970 Þ
dx dx
But this condition is only necessary; one must then verify if the solution of the
differential equation effective minimizes the functional IðyÞ.
Let be now a functional of the form
ZZ
IðuÞ Fðx1 ; x2 ; u; u1 ; u2 Þ dx1 dx2 ð6:98Þ
D
284 6 Principles and General Theorems. Computation Methods
in the domain D.
Analogically, the functional
ZZ
IðuÞ Fðx1 ; x2 ; u; u1 ; u2 ; u11 ; u12 ; u22 Þ dx1 dx2 ð6:100Þ
D
o o o2 o2 o2
Fu Fu 1 Fu2 þ 2 Fu11 þ Fu12 þ 2 Fu22 ¼ 0: ð6:1000 Þ
ox1 ox2 ox1 ox1 ox2 ox2
As well, for the functional
ZZ
IðuÞ Fðx1 ; x2 ; x3 ; u; u1 ; u2 ; u3 Þ dx1 dx2 dx3 ; ð6:101Þ
D
ZZZ
1
A02
2 ¼ ½DDðf0 þ f Þ2 dV; ð6:102Þ
V V
we must put the condition that these functions be approximate the best possible in
the interior of the given domain.
n n
As well, let be given the normal component p and the tangential component q of
the external load on the element of area dS, of external normal n; the mean
quadratic error may be calculated in the form
ZZ ZZ
002 1 n n 2 1 n n
A2 ¼ ðp rÞ dS þ ðq sÞ2 dS; ð6:1020 Þ
S S S S
n n
where r and s are the normal and the tangential stress, respectively, corresponding
to the same element of area, given by potential functions of the form (6.94).
In the first method mentioned at the previous Subsection (the most general one),
we calculate the mean quadratic error in the form (we observe that one can have
several components of the form (6.102) of the mean quadratic error, taking into
account all the potential functions used)
oA22
¼ 0; i ¼ 1; 2; . . .; n; ð6:104Þ
oai
which form a linear system
X
n
cij aj ¼ di ; i ¼ 1; 2; . . .; n; ð6:1040 Þ
j¼1
One may obtain the correction function by using the principle of the minimum
potential energy or the principle of the minimum complementary energy; the
respective method is called Ritz’s method.
If the boundary conditions are put in displacements and one solves the problem
in displacements, then it is convenient to use the principle of minimum potential
energy (Lagrange); we express, in this case, Wi by means of the components ui of
the displacement, using the formula (4.90). We choose (in case of linear
elastostatics)
nj
X
uj ðx1 ; x2 ; x3 Þ ¼ u0j ðx1 ; x2 ; x3 Þ þ akj ukj ðx1 ; x2 ; x3 Þ; ð6:105Þ
k¼1
Taking into account the principle of virtual displacements (from which results
Lagrange’s principle), the variations dakj of the displacement are arbitrary. We
obtain thus the equations
ZZZ ZZ
oWi k n
¼ u F
j j dV ukj pj dS ¼ 0 ð!Þ; j ¼ 1; 2; 3; k ¼ 1; 2; . . .; nj ;
oakj V S
ð6:108Þ
which represents the canonical form of the Lagrange-Ritz equations for linear
elastostatics. We obtain thus a system of n1 þ n2 þ n3 linear algebraic equations,
which determine the unknown parameters akj .
If we put the boundary conditions in stresses and solve the problem in stresses,
then we use the principle of minimum of complementary energy; the internal work
will be expressed by means of stresses in the form (4.9000 ). We write (in the frame
of linear elastostatics too)
njk
X
rjk ðx1 ; x2 ; x3 Þ ¼ r0jk ðx1 ; x2 ; x3 Þ þ aljk rljk ðx1 ; x2 ; x3 Þ; ð6:109Þ
l¼1
6.2 Computation Methods 287
Taking into account the principle of virtual variations of the state of stress (from
which Castigliano’s principle is obtained), one can consider the variations daljk as
arbitrary; if the relations (6.25), (6.250 ) are verified, then we get the equations
ZZZ
oWi
ð1 þ djk Þ l þ ðuk rljk;j þ uj rlkj;k Þ dV
oajk V
ZZ
ðuk nj þuj nk Þ dS ¼ 0 ð!Þ; j; k ¼ 1; 2; 3; l ¼ 1; 2; . . .; njk ; ð6:112Þ
S
which represent the canonical form of the Castigliano-Ritz equations for the linear
elastostatics. The parameters aljk are thus given by a system of n11 þ n22 þ n33 þ
n23 þ n31 þ n12 linear algebraic equations.
If the functions r0jk verify the equilibrium equations, while the functions rljk are
particular integrals of these equations in the absence of the volume forces, the
latter ones do not intervene any more in the statement of the principle of minimum
complementary energy; hence, the Eq. (6.112) take a simpler form, without
volume integrals. If the functions r0jk verify the boundary conditions too and the
functions rljk verify homogeneous (zero) conditions on the boundary, then the
Eq. (6.112) take a simpler form, neither these conditions intervening. We obtain
thus the equations
oWi
¼ 0; j; k ¼ 1; 2; 3; l ¼ 1; 2; . . .; njk ; ð6:113Þ
oaljk
is searched and the approximation one; by putting conditions of extremum for this
functional (least squares method), one can determine the unknown parameters.
One can also apply the Bubnov-Galerkin method, putting the condition that the
gradient of the difference mentioned above be orthogonal with certain functions,
e.g. with the functions generating the expression of approximation. In practice we
shall use some methods or other ones, as the functions f0 and fi do satisfy or not all
the limit conditions or are or not particular integrals of the potential differential
equation of the problem. As well, if it is necessary for simplifying the computa-
tion, then we may use for the same problem ideas from different methods.
It is interesting to mention that Ritz’s method approximates the value of the
function in excess, while the Trefftz one leads just to a contrary approximation;
hence, the value of the searched function may be bordered between two limits.
Besides the computation methods mentioned above as the most important ones,
many other methods have been developed, which may be interesting in various
particular cases. Thus, the method of the least squares has an interesting extension
in the method of the least products; thus, the integral
ZZZ
½DDðf0 þ f ÞðF f0 f Þ dV ð6:114Þ
V
must have a minimum, for which the constructed function represents a better
approximation.
hence, the field of vectors equivalent to the bound vector V may be expressed in
the form
QðrÞ ¼ VdðrÞ: ð6:116Þ
This field of vectors may be considered as volume density of the bound vector V.
If the bound vector is applied at a point A of position vector r0 ; then the
equivalent vector field is given by
QðrÞ ¼ Vdðr r0 Þ: ð6:1160 Þ
To represent the bound vector V, we can use an arbitrary representative
sequence d; because that one does not intervene in the final result by its concrete
form, but only by its properties.
Let us consider now the field of parallel vectors
Qe ðx1 ; x2 ; x3 Þ ¼ Vfe ðx1 ; x2 ; x3 Þ;
defined on the sphere Xe with the centre at the origin and of radius e; passing to the
limit, in the sense of the theory of distributions, we obtain
ZZZ
r dX ¼ 0; ð6:119Þ
Xe
the representation (6.116) being thus justified from the mechanical point of view.
Let be V1 ; V2 ; . . .; Vn a system of n bound vectors, applied at the point Aðr0 Þ;
the equivalent vector field will be given by
Qi ðrÞ ¼ Vi dðr r0 Þ; i ¼ 1; 2; . . .; n: ð6:120Þ
applied at the same point Ai ; one may write the equivalent vector field in the form
X
n
QðrÞ ¼ Qi ðrÞ ¼ Rdðr r0 Þ: ð6:122Þ
i¼1
Hence, the properties of the bound vectors, applied at the very same point, are
preserved in their representation by means of distributions.
The concentrated loads, applied to a deformable solid, may be mathematically
modelled by means of bound vectors. Thus, taking into account the above con-
siderations, one may represent the equivalent load of a concentrated force F,
applied at a point A of position vector r0 , in the form
QðrÞ ¼ Fdðr r0 Þ; ð6:123Þ
the support of the equivalent load QðrÞ is the point A.
is the vector radius, with the notation used in case of the spherical co-ordinates,
then the components of the displacement vector must tend to zero for R ! 1.
6.2 Computation Methods 291
Taking into account the formulae (A.237) and (A.2370 ), we may write
1
D ¼ 4pdðx1 ; x2 ; x3 Þ; ð6:128Þ
R
analogically, using a formula of the form (A.3.41) for the differentiation of the
function under the Fourier transform operator with respect to the variables xi , we
get
292 6 Principles and General Theorems. Computation Methods
hx x i
1 i j ai aj
dij F F ¼ 8p ; i; j ¼ 1; 2; 3: ð6:1290 Þ
R R ðak ak Þ2
We notice that between the elastic constants of the material take place the
relations
kþl 1 k þ 3l 3 4m
¼ ; ¼ :
k þ 2l 2ð1 mÞ k þ 2l 3ð1 mÞ
Starting from the formulae (6.127) and using the above results, we get
1 1h xi xj i
uij ðx1 ; x2 ; x3 Þ ¼ ð3 4mÞdij þ 2 ; i; j ¼ 1; 2; 3; ð6:130Þ
16pð1 mÞG R R
where we have introduced the convolution product with respect to the space
variables.
Using the matrices
2 3 2 3 2 3
u11 u12 u13 u1 F1
U 4 u21 u22 u23 5; u 4 u2 5; Q 4 F2 5; ð6:132Þ
u31 u32 u33 u3 F3
i.e. the fundamental solution matrix U, the displacement state matrix u and the
volume forces field matrix Q, we also may write
u ¼ U Q: ð6:1310 Þ
What concerns the numerical computation, the values corresponding to regular
distributions are calculated as usual, while the singular distributions are approxi-
mated by means of representative sequences. Obviously, for the numerical cal-
culation of various convolution products, the electronic computer may be very
useful.
6.2 Computation Methods 293
In the case of the elastic space subjected to the action of a concentrated force P
at the point Aðx01 ; x02 ; x03 Þ; we introduce the equivalent field of components
R being the position vector applied at the point at which acts the force P. We can
get this result on another way too, using a proof due to N. S ßandru [36]; thus, on
behalf of the formula (A.3.13), we can write the Eqs. (5.34), (5.35), which give the
vector potential and the scalar potential, respectively, in the Papkovich-Neuber
representation (4.4.22), in the form
Dv þ 2Pdðx1 ; x2 ; x3 Þ ¼ 0; Dv0 ¼ 0:
Assuming that the function v is regular at infinite and taking into account (6.128),
we get
1 P
v¼ ; v ¼ 0; ð6:136Þ
2p R 0
so that the representation (5.30) leads to the same result.
If Fj ; j ¼ 1; 2; 3; are locally integrable functions, we can write
Z 1Z 1Z 1
ui ðx1 ; x2 ; x3 Þ ¼ uij ðx1 n1 ; x2 n2 ; x3 n3 Þ
1 1 1
Fj ðn1 ; n2 ; n3 Þdn1 dn2 dn3 ; i ¼ 1; 2; 3; ð6:137Þ
2m
sijk ¼ G uik;j þ ujk;i þ ulk;l dij ; ð6:138Þ
1 2m
1 xi
¼ 2; ð6:139Þ
R ;i R
x x 1 xi xj xk
i j
¼ x d
i jk þ x d
j ik 3 ; ð6:1390 Þ
R3 ;k R3 R2
we obtain, finally
1
sijk ðx1 ; x2 ; x3 Þ ¼
2pð1 mÞ
1 h xi xj xk i
3 ð1 2mÞðxk dij xi djk xj dij Þ 3 2 ; i; j; k ¼ 1; 2; 3:
R R
ð6:1380 Þ
The state of stress corresponding to the arbitrary volume forces
Fk ðx1 ; x2 ; x3 Þ; k ¼ 1; 2; 3; will be given by
rij ðx1 ; x2 ; x3 Þ ¼ sijk ðx1 ; x2 ; x3 Þ Fk ðx1 ; x2 ; x3 Þ; i; j ¼ 1; 2; 3; ð6:140Þ
where we have introduced, as well, the convolution product with respect to the
space variables.
We notice that
sijk ¼ sjik ; i; j; k ¼ 1; 2; 3; ð6:1400 Þ
i.e. the fundamental solution matrix S and the stress tensor matrix r; respectively,
and taking into account the notation (6.132), we may write
r¼SQ ð6:141Þ
If the case of the elastic space subjected to the action of a concentrated force P,
expressed by the equivalent load (6.133), we obtain the state of stress
6.2 Computation Methods 295
We start, for the dynamic problem, from the considerations made in Sect. 5.2 and
in Sect. 4.3. We will use Lamé’s equations to study the problem of the elastic
space 1\x1 ; x2 ; x3 \1; acted upon by arbitrary volume forces, in a solution
in displacements, or the Eqs. (5.182), (5.183) in a solution in stresses, for
arbitrary volume forces too; in the case of conservative volume forces, one may
use the Eq. (5.94).
Using the Laplace transform in distributions with respect to the time t, one may
express a particular solution of the Eq. (5.182) in the form
1 1 1 pR=c2
U¼ L e L½F ; ð6:144Þ
4p R
296 6 Principles and General Theorems. Computation Methods
In both cases, the product of convolution operates with regard to the space
variables.
Analogically, a particular solution of the equation (5.94) is of the form
1 2m 1 1 pR=c1
x¼ L e L½v ; ð6:146Þ
4ð1 mÞp R
o ~
o
ui ðx1 ; x2 ; x3 ; tÞ ¼ ui ðx1 ; x2 ; x3 ; tÞ þ u0i ðx1 ; x2 ; x3 ; tÞdðtÞ;
ot ot
o2 ~
o2
u i ðx 1 ; x 2 ; x 3 ; tÞ ¼ ui ðx1 ; x2 ; x3 ; tÞ þ u_ 0i ðx1 ; x2 ; x3 ; tÞdðtÞ
ot2 ot2
_
þ u0i ðx1 ; x2 ; x3 ; tÞdðtÞ;
o2 ~
o2
ui ðx1 ; x2 ; x3 ; tÞ ¼ ui ðx1 ; x2 ; x3 ; tÞ;
oxj oxk oxj oxk
Observing that
kþl 1 1 1
F1 ¼ F
½ðk þ 2lÞak ak þ qp2 ðlak ak þ qp2 Þ qp2 ak ak þ p2 =c21
1 1 pR=c1 pR=c2
¼ e e ;
ak ak þ p2 =c22 4pqp2 R
where we have introduced the wave propagation velocities, given by the formulae
(5.105), (5.1050 ), and using the properties of the convolution product concerning
the differentiation, we may express the Laplace transforms of the generalized
displacements in the form
1
L½ui ðx1 ; x2 ; x3 ; tÞ ¼ ½L½F i þ q pu0i þ u_ 0i
4pG
1 h i
epR=c2 þ c22 ðL½F j Þ;j þ qðpu0j þ u_ 0j Þ;j
R
1 1 pR=c1
pR=c2
2 e e ; ð6:152Þ
p R ;i
where the convolution product concerns the space variables.
The inverse Laplace transformation leads to the generalized state of displace-
ment in the form
298 6 Principles and General Theorems. Computation Methods
1 1
ui ðx1 ; x2 ; x3 ; tÞ ¼ L1 L½F i epR=c2
4pG R
0 1 _ R 0 1 R
þ q ui d t þ u_ j d t
R c2 R c2
" #
1 1
þ c22 L1 L½F j 2 epR=c1 epR=c2
p R ;ij
1 R R
þ q u0j;j h t h t
R c1 c2 ;i
1 R R
þu_ 0j;j t t ; i ¼ 1; 2; 3:
R c1 þ c2 þ ;i
ð6:153Þ
In the case of homogeneous (null) initial conditions, we get
(
1 1 1 pR=c
ui ðx1 ; x2 ; x3 ; tÞ ¼ L L½F i e 2
4pG R
" #)
1 1 pR=c1
2 1 pR=c2
þ c2 L L½F j 2 e e ; i ¼ 1; 2; 3;
p R ;ij
ð6:154Þ
while, in the absence of the volume forces, we have
(
1 1 0 1_ R 0 1 R
ui ðx1 ; x2 ; x3 ; tÞ ¼ u d t þ _
u d t
4p c22 i R c2 i
R c2
1 R R
þ u0j;j h t h t
R c1 c2 ;i
)
1 R R
þu_ 0j;j t t ; i ¼ 1; 2; 3: ð6:155Þ
R c1 þ c2 þ ;i
1 cos bi R
Ui ¼ P t ; i ¼ 1; 2; 3; ð6:157Þ
4p R c2
6.2 Computation Methods 299
1 2 1 R R
X¼ c2 cos bj PðtÞ t t ; ð6:158Þ
2p ðtÞ R c1 þ c2 þ ; j
where the convolution product is performed with respect to the time variable. We
obtain thus the generalized state of displacement
(
cos bj 1 R
ui ðx1 ; x2 ; x3 ; tÞ ¼ P t dij
4pG R c2
)
2 1 R R
þc2 PðtÞ t t ; i ¼ 1; 2; 3;
ðtÞ R c1 þ c2 þ ; ij
ð6:159Þ
where
P0 cos bj 1 R
ui ðx1 ; x2 ; x3 ; tÞ ¼ d t d
4pG R c2 ij
)
2 1 R R
þc2 t t ; i ¼ 1; 2; 3: ð6:162Þ
R c1 þ c2 þ ; ij
For P0 ¼ 1; we get the fundamental solution tensor in the sense of the theory
distributions, of components
(
1 1 R
uij ðx1 ; x2 ; x3 ; tÞ ¼ d t d
4pG R c2 ij
)
2 1 R R
þc2 t t ; i; j ¼ 1; 2; 3; ð6:163Þ
R c1 þ c2 þ ; ij
where the convolution product is performed with respect both to the space vari-
ables and the temporal one.
300 6 Principles and General Theorems. Computation Methods
In the absence of the volume forces, e.g. we find again the formulae (6.155).
It is interesting to notice that, for a usual function PðtÞ; corresponding to a force
which acts at the origin along the Ox1 -axis, G. C. Stokes [35] established, in 1849,
the formula (6.159), in the form
"
1 1 x21 R
u1 ðx1 ; x2 ; x3 ; tÞ ¼ P t
4pqR c21 R2 c1
Z 1=c2 #
1 x21 R x21
þ 2 1 2 P t 13 2 kPðt kRÞdk ;
c2 R c2 R 1=c1
"
x1 x2 1 R 1 R
u2 ðx1 ; x2 ; x3 ; tÞ ¼ P t P t
4pqR c21 c1 c22 c2
Z #
1=c2
þ3 kPðt kRÞdk ; ð6:165Þ
1=c1
"
x1 x3 1 R 1 R
u3 ðx1 ; x2 ; x3 ; tÞ ¼ P t P t
4pqR c21 c1 c22 c2
Z 1=c2 #
þ3 kPðt kRÞdk :
1=c1
P0 cos bj 2 R
ui ðx1 ; x2 ; x3 ; tÞ ¼ d t d
8pG R c2 ij
( "
2
2 #) )
1 R R
þ c22 t t ; i ¼ 1; 2; 3: ð6:167Þ
R c1 þ c2 þ
;ij
If the domain occupied by the elastic body is infinite, but has at least a part of
the frontier at a finite distance or it is finite, then one can no more use the
computation method indicated above. In this case, we must use a procedure
6.2 Computation Methods 301
analogous to that given by the formulae (6.147), (6.148) in what concerns the
space variables too; the derivatives in the sense of the theory of distributions, with
respect to the space variables, will no more be equal to the corresponding ones, in
the usual sense; one must introduce distributions concentrated on surfaces.
The iterative computation methods can be very useful in some cases. Starting from
a first approximate form of the solution, various degrees of approximation are then
obtained. The two important mathematical problems about it are: the demon-
stration of the convergence of the method and the estimation of the error (the
remainder computation).
Hereafter, we shall indicate such a method of successive approximations in case
of the non-homogeneous bodies, by using the results given by P. P. Teodorescu
[38, 39] in the dynamic case, that generalize those supplied by P. P. Teodorescu
and M. Predeleanu [43] in the static case.
Using the considerations made in Sect. 4.1.3.8, one can write the equilibrium
equations (in the static case) in the form
rij;j þ f;j rij þ Fi ¼ 0; i ¼ 1; 2; 3; ð6:168Þ
If we neglect the terms with variable coefficients from the equilibrium equa-
tions (6.168), then the latter ones will take a form similar to that of the equilibrium
equations in the case of the homogeneous bodies (3.610 ); if, in the frame of a
302 6 Principles and General Theorems. Computation Methods
solution in stresses of the problem, we add Hooke’s law (4.101) and Saint-
Venant’s equations (2.68), then we shall find the approximation of zeroth order,
which solves the problem in the case of a homogeneous body, undergoing given
volume forces and boundary conditions (6.169).
So as to find the approximation of kth order if the approximation of ðk 1Þth
order, that supplies the reduced conventional volume forces
ðkÞ ðk1Þ
F i ¼ f;j rij ; i ¼ 1; 2; 3; ð6:172Þ
is supposed known, the problem is solved now in the case of a homogeneous body
ðkÞ
acted upon by the volume forces F i ; i ¼ 1; 2; 3; with homogeneous boundary
conditions (vanishing surface loads).
Thus, the solution in stresses of the elastostatic problem of a non–homogeneous
body can be approximated by a succession of solutions in stresses of elastostatic
problems of the very same body, considered as homogeneous and acted upon by
certain conventional loads, being submitted to certain conventional boundary
conditions (zero ones, excepting the zeroth approximation).
The convergence of the above mentioned iterative process can be practically
ðnÞ
estimated by its physical interpretation—the conventional volume forces F i ; i ¼
1; 2; 3; must yield a state of stress, negligible with respect to the looked real state
of stress.
An analogous solution in displacements of the problem can be supplied by
starting from the Lamé equations (5.13).
In the dynamic case one may use an analogous method of successive approx-
imations if one admits that the reduced density and the reduced damping coefficient
are constant coefficients (q ¼ const and k ¼ const), as defined in Sect. 4.1.3.8;
to this end, the non-homogeneity corresponding to q and k must be of the same
type as the non-homogeneity of the modulus of longitudinal elasticity.
Some of the computation methods presented above (e.g., the variational methods)
allow to obtain, with an as well as we wish approximation the potential functions
or the state of strain and stress at any point in the interior of the body. Other
methods (e.g., the method of finite differences) allow to obtain approximate values
of those functions at a finite number of points in the interior of the body, satisfying
the boundary conditions at a finite number of points too.
One can imagine a computation method which has common parts with each of
the methods considered above. The method consists in searching a function, of the
most simple form, which does verify the boundary conditions at a finite number of
points; this function, which satisfies a certain partial differential equation, will give
the state of strain and stress at any point in the interior of the body.
6.2 Computation Methods 303
References
A. Books
1. Bers, L., John, F., Schechter, M.: Partial Differential Equations. Interscience Publisher, New
York (1964)
2. Castigliano, A.: Théorie de l’équilibre des systèmes élastiques. Turin (1879)
3. Courant, R.: Partial Differential Equations. Interscience Publisher, New York (1962)
4. Courant, R., Friedrichs, K.O.: Supersonic Flow and Shock Waves. Interscience Publisher,
New York (1948)
5. Gurtin, M.E.: The Linear Theory of Elasticity. In: Encyclopedia of Physics, VIa/2. Springer,
New York (1972)
6. Kecs, W., Teodorescu, P.P.: Application of the Theory of Distributions in Mechanics. Ed.
Academiei, Bucuresßti, Abacus Press, Tunbridge Wells, Kent (1974)
7. Kirchhoff, G.R.: Gesammelte Abhandlungen. Leipzig (1882)
8. Kupradzhe, V.D., Gegelya, T.G., Basheishvili, M.O., Burchuladzhe, T.V.: Trehmernye
zadachi matematicheskoı teorii uprugosti (Three Dimensional Problems of the Mathematical
Theory of Elasticity). Izd. Tbilisskogo Univ, Tbilissi (1968)
9. Love, A.E.H.: A Teatrise on the Mathematical Theory of Elasticity, 2nd edn. University
Press, Cambridge (1934)
10. Neumann, F.: Vorlesungen } uber die Theorie der Elasticität der festen Körper und der
Lichtäthers. B. G. Teubner, Leipzig (1885)
11. Nowacki, W.: Dynamics of Elastic Systems. Chapman and Hall Ltd., London (1963)
12. Nowacki, W.: Teoria spre_zystości (Theory of Elasticity). Państ. Wydawn. Naukowe,
Warszawa (1970)
13. Sneddon, I.N.: Fourier Transforms. McGraw Hill Book Co., Inc., New York (1951)
14. Teodorescu, P.P.: Probleme spatßiale în teoria elasticitătßii (Space Problems in the Theory of
Elasticity). Ed. Academiei, Bucuresßti (1970)
15. Teodorescu, P.P.: Dynamics of Linear Elastic Bodies. Ed. Academiei, Bucuresßti, Abacus
Press, Tunbridge Wells, Kent (1975)
16. Volterra, V., Volterra, E.: Sur les distorsions des corps élastiques (théorie et applications).
Mém. Sci. Math. CXLVII. Gauthier Villars, Paris (1960)
B. Papers
17. Betti. E.: Teoria dell’elasticità. Nuovo Cimento, 7–8, 5, 69, 158; 9, 34 (1872); 10, 58 (1873)
18. Castigliano, A.: Atti R. Accad. Sci. Torino (1875)
19. Clapeyron, B.P.E.: Mémoire sur le travail des forces élastiques dans un corps solide déformé
par l’action des forces extérieures. C. R. des séances de l’Acad. des Sci. 46, 208 (1858)
20. Friedrichs, K.O., Lewy, H.: Uber } die Eindentigkeit und das Abhängigkeitsgebiet der
Lösungen beim Anfangswertproblem linearer hyperbolischer Differentialgleichungen. Math.
Annalen 98, 192 (1928)
References 305
21. Graffi, D.: Sul teorema di reciprocità nella dinamica dei corpi elastici. Mem. della Accad.
delle Sci., Bologna, ser. 10, 4, 103 (1946/47)
22. Graffi, D.: Sui teoremi di reciprocità nei fenomeni non stazionari. Atti della Accad. delle Sci.,
Bologna, ser. 11, 10, 33 (1963)
23. Green, G.: On the propagation of light in crystallized media. Trans. Cambridge Phil. Soc. 7,
121 (1841)
24. Gurtin, M.E., Sternberg, E.: A note on uniqueness in classical elastodynamics. Quart. Appl.
Math. 19, 169 (1961)
25. Gurtin, M.E., Toupin, R.A.: A uniqueness theorem for the displacement boundary—value
problem of linear elastodynamics. Quart. Appl. Math. 23, 29 (1965)
26. Iesßan, D.: On the Reciprocity Theorem in the Linear Elastodynamics. An. ßst. Univ. ‘‘Al.
I. Cuza’’, Iasßi (ser. nouă), sectß. I, Mat., 18, 193 (1972)
27. Kirchhoff, G.R.: Zur Theorie der Lichtstrahlen. Sitzungsberichte der k. Akad. der Wiss.,
Berlin, 2, 641 (1882)
28. Knops, R.J., Payne, L.E.: Uniqueness in classical elastodynamics. Arch. Rat. Mech. Anal. 27,
349 (1968)
29. Love, A.E.H.: The propagations of wave motion in an isotropic elastic solid medium. In:
Proceedings of the London Mathematical Society, ser. 2, 1, 291 (1904)
30. Maxwell, J.C.: On the reciprocal figures, frames and diagrams of forces. Trans. Roy. Soc.
Edinburgh 26, 1 (1870)
31. Rubinovicz, A.L.: Herstellung von Lösungen gemischter Randwertprobleme bei
Hyperbolischen Differenzialgleichungen zweiter Ordnung durch Zusammenst} uckelung aus
Lösungen einfacher gemischter Randwertaufgaben. Monatshefte f} ur Math. u. Phys. 30, 65
(1920)
32. Saint Venant, B. de: Mémoire sur la torsion des prismes. Mém. Acad. savants étrangers
(1885)
33. Somigliana, C.: Sulle equazioni della elasticità. Ann. di Matem., ser. 2, 17, 37 (1889)
34. Somigliana, C.: Sopra alcune formule fondamentali della dinamica dei mezzi isotropi. Atti.
R. Accad. delle Sci., Torino, 41, 869, 1071 (1905/6); 42, 387 (1906/7)
35. Stokes, G.G.: On the dynamical theory of diffraction. Trans. Cambridge Phil. Soc. 9, 1 (1849)
ßandru, N.: O deıstvii peremenyh sil v neogranichennom prostransve (On the Action of
36. S
Variable Forces in the Infinite Space). Bull. Acad. Pol. Sci., ser. Sci. Techn. 12, 45 (1964)
37. Tedone, O.: Sulle vibrazioni dei corpi solidi, omogenei ed isotropi. Mem. R. Accad. delle
Sci., Torino, 47, 181 (1896/97)
38. Teodorescu, P.P.: Uber} das kinetische Problem nichthomogener elastischer Körper. Bull. de
l’Acad. Pol. des Sci., sér. Sci. Techn. 12, 867 (1964)
39. Teodorescu, P.P.: Schwingungen der elastischen Kontinua. III. Konferenz } uber nichtlineare
Schwingungen, Berlin, 1964, Abh. der deutschen Akad. der Wiss. zu Berlin, Kl. f} ur Math.,
Phys. u. Techn., Akad. Verlag, Berlin, 29 (1965)
40. Teodorescu, P.P.: Uber} das dreidimensionale Problem der Elastokinetik. Z.A.M.M. 45, 513
(1965)
41. Teodorescu, P.P.: Sur quelques problèmes dynamiques de la théorie de l’élasticité. Rev.
Roum. Math. Pures Appl. 11, 773 (1966)
42. Teodorescu, P.P.: Stress functions in three dimensional elastodynamics. Acta Mech. 14, 103
(1972)
43. Teodorescu, P.P., Predeleanu, M.: Quelques considérations sur le problèmw des corps
élastiques hétérogènes. IUTAM Symp., Non Homogenity in Elasticity and Plasticity, 1958,
Bull Acad. Pol. Sci., sér. Sci. Techn., 7, 81 (1959)
44. Volterra, V.: Sur les vibrations des corps élastiques isotropes. Acta Mathematica 16, 161
(1894)
45. Volterra, V.: Sull’equilibrio dei corpi elastici più volte connessi. Atti R. Accad. dei Lincei,
Rendiconti, Cl. Sci. fis., mat. e nat., ser. 5, 14, 193 (1905)
306 6 Principles and General Theorems. Computation Methods
46. Volterra, V.: Sur l’quilibre des corps élastiques multiplement connexes. Ann. École Norm.
Sup., 2 me sér., 24, 401 (1907)
47. Wheeler, L.T.: Some results in the linear dynamical theory of anisotropic elastic solids.
Quart. Appl. Math. 28, 91 (1970)
48. Wheeler, L.T., Sternberg, E.: Some theorems in classical elastodynamics. Arch. Rat. Mech.
Anal. 31, 51 (1968)
49. Zaremba, S.: Sopra un teorema d’unicità relativo alla equazione delle onde sferiche. Atti R.
Accad. dei Linceu, ser. 5, Rendicanti, Cl. Sci. fis., mat. e nat., 24, 904 (1915)
Chapter 7
Introduction to the Theory of Cosserat
Type Bodies
We deal hereafter with a synthesis study on the Cosserat type bodies, insisting on
the linear elastic bodies in the case of small deformations and rotations. After some
hystorical considerations, the fundamental equations are presented and formula-
tions in displacements and stresses of the problems are given. Some important
results are then given, e.g., theorems of reciprocity, fundamental solutions corre-
sponding to concentrated loads etc.
After a short review of the problems and of the development of the Cosserat type
bodies, the state of deformation and the states of stress and couple-stress are
considered; then, the constitutive laws are put in evidence.
Mounting the scale of generalization, the Cosserat type bodiea with dislocations
and the bodies of second degree with initial stresses and deformations [196–198]
are obtained. Another generalization of the bipolar bodies is constituted by the
multipolar ones (of Green-Rivlin type [64, 67, 148]).
Such problems as the asymmetry of the stress tensor have been mentioned in the
19th century by A. L. Cauchy [33, 34], in 1850 and 1851, B. de Saint-Venant
[150], in 1869, S.-D. Poisson [144], in 1842, Lord Kelvin [8] and W. Voigt [206],
in 1887. But the first systematic works in this direction have been published in
1909 by the brothers E. Cosserat and F. Cosserat [2], the name of which has been
later put in connection with these bodies; as a matter of fact, the Cosserat brothers
began their studies earlier, in 1896 [41, 42]. But their studies remain unnoticed for
many years, when—after 1950—the problem became a new development.
In the meantime, one can mention the studies of E. Hellinger [5], et K. Heun
[6], in 1914, as well as those of F. Klein [94] and E. Noether [131], in 1918, which
bring not hing new. In his thesis, in 1929, T. J. Jaramillo [7] puts in evidence the
fact that the action density depends on the gradients of second order too.
The possibility of asymmetry of the stress tensor has also been mentioned by E.
Reissner [146], in 1944, but on a wrong way. One must also remember that in his
lectures, in 1899, at Clark University, L. Boltzmann [1] considers the property of
symmetry of this tensor as an axion; G. Hamel [4] gives the name of Boltzmann
axiom to this property, the bodies which have properties of mechanical asymmetry
being non-Boltzmannian ones. The bodies of Cosserat type are just such bodies.
After 1950, many various researches on the energetic state of the crystalline net
have been made, having—as a consequence—the asymmetry of the stress tensor
rij ; indeed, if the work of deformation reads
1
W ¼ Aijkl ui; j uk;l ; ð7:1Þ
2
then it results
rij ¼ Aijkl uk;l : ð7:2Þ
One can mention the studies of Y. Le Corre [37–40] and of J. Laval [104] in this
direction. R. Tiffen and A. C. Stevenson [200] make somewhat different consid-
erations. N. Joel and W. A. Wooster [83, 84], E. S. Rajagopal [145] and
E. S. Krishnan and E. S. Rajagopal [97] make the criticism of these results. One
may state that the majority of these authors are not aware of the work of the
Cosserat brothers.
S. Bodaszewski [26], based on the studies of W. Burzynski [29] of 1949,
introduced, in 1953, the inertia of rotation. Other initial studies on the asymmetry
of the stress tensor are due to N. Oshima [137, 138].
7.1 General Equations 309
If one attaches to any point of the deformable body a rigid orthogonal trihedron,
then one obtains a polar body (with orientation) and one can speak about the
rotation of a point. A body having such properties is called a Cosserat type body,
where any point of which is an infinitesimal rigid; the apparition of couple-stresses
is thus explained.
In 1958, J. L. Ericksen and C. Truesdell [49] have introduced the deformable
director trihedrons, being thus led to the generalized bodies of Cosserat type. As a
continuation of a paper of C. Truesdell [204], in 1952, and of another one of
W. Noll [132], in 1958, these general theories are presented in two volumes of the
Encyclopedia of Physics by C. Truesdell and R. A. Toupin [18], in 1960, and by C.
Truesdell and W. Noll [17], in 1965. Some following studies of F. A. McClintock,
P. A. André, K. R. Schwedt and R. E. Stoeckly [36] or of F. A. McClintock [35]
do not bring anything new.
In 1958 appears an important study of W. Günther [71] on the statics and on the
kinematics of a Cosserat type body; one makes the connection with the theory of
dislocations too. Other interesting researches in this direction, after 1958, are due
to E. Kröner [9, 98–100], F. Hehl and E. Kröner [76] and R. Stojanović [172].
C. Teodosiu [194, 195] deals with the determination of stresses and couple-stresses
within bodies with dislocations. E. Reissner and F. Y. M. Wan [147] deal
with Günther’s theories in a variational form, generalizing thus some results of
P. M. Naghdi [121].
We mention also the studies of fluids of Cosserat type made by J. L. Ericksen
[45–48] and by E. L. Aero, A. N. Bulugin and E. V. Kuvshinskiı̆ [21]. Studies
concerning the plastic, elastic-plastic or viscoelastic bodies with asymmetric
mechanical properties have been made by M. Misßicu [114–119], V. A. Lomakin
and L. N. Savov [105], Savov [156] and V. D. Kubenko and N. A. Shul’ga [101].
A study on a geometrical basis has been made by Y. Yamamoto [212].
The problem of the constitutive laws, especially linear ones, has been the object
of many researches. After the paper of W. Günther [71], C. Truesdell and
R. A. Toupin [18], G. Grioli [3, 68, 69] and E. L. Aero and E. V. Kuvshinskiı̆ [22]
dealt with bodies with constraint rotations. H. Schaefer [160] considers the bodies
with free rotations in the plane case; E. V. Kuvshinskiı̆ and E. L. Aero [102], in
1963, and V. A. Palmov [139] and A. C. Eringen and E. S. Suhubi [52] (dealing
with bodies with microstructure), in 1964, consider bodies with free rotations in
the three-dimensional case. Interesting studies on the constitutive laws are due
to R. Stojanović and L. Vujośević [176] and to R. Stojanović, S. Djurić and
L. Vujośević [174].
In 1966, J. Schijve [165] makes experimental researches to determine an elastic
constant l, having the dimension of a length; he obtains, for certain metals, l ¼ 0:1
mm. G. Adomeit [20] and E. Sooś and P. P. Teodorescu [167] deal with the
determination of elastic constants in bodies of generalized Cosserat type.
S. Kaliski [85, 86] deals with a model where may appear asymmetric stresses
rij (a spatial frame). W. H. Hoppman and F. O. F. Shahman [79] give a model of
isotropic elastic body with three constants.
310 7 Introduction to the Theory of Cosserat Type Bodies
Other important problems studied are those of coupled fields. We mention thus
the work of S. Kaliski, Z. Plochocki and D. Rogula [87], of P. D. Kelly [88] and of
R. C. Dixon and A. C. Eringen [43], who deal with the coupling of the electro-
magnetic field and the field of deformations; works in this direction are due also to
J. B. Alblas [23], who considers Cosserat type bodies with electronic spin. With
the coupling of the thermic field and the field of deformations dealt chiefly
W. Nowacki [133–136].
In 1963, R. D. Mindlin [197] generalizes the classic conception of elastic body,
introducing the model of an linearly elastic body with microstructure. An year
later, R. A. Toupin [192] considers, in the same manner, the theories where appear
the couple-stresses; similar ideas can be found in the papers of A. C. Eringen and
E. S. Suhubi [52].
Detailed studies, where appear dislocations too, are due to C. Teodosiu [186–
189]. We mention, in this direction, the researches of I. A. Kunin [103] and of R.
A. Toupin [203]; R. D. Mindlin [109] comes back later to this problem.
More sophisticated models are conceived by A. E. Green and R. S. Rivlin [64,
67, 148]; there are multipolar bodies, where one takes into account gradients of
higher order of the displacements. Cz. Woźniak [209] develops a theory of fibre
bodies and A. C. Eringen [50] deals with microfluids.
In Italy, G. Grioli [3, 68–70] considers especially the problem of thermody-
namic potentials for bodies of Cosserat type. Various other studies are due to
A. Bressan [28] and to D. Galletto [54–57], the latter one dealing also with
incompressible systems with reversible transformations.
Books on linearly elastic Cosserat type bodies have been written by H. Schaefer
[13] (a cycle of lectures presented at the University of Trieste, dealing especially
with the contribution of the German school to the problem) and by G. N. Savin
[12] (lectures given at the University of Kiev on the plane problem of asymmetric
elasticity, treated by means of functions of complex variables). In Romania we
mention the syntheses of D. Iesßan [16] and N. S ßandru [16], as well the books of
P. P. Teodorescu [14, 15].
We remark also an excellent synthesis due to H. Schaefer [163], in which the
hystorical process of the development of the problem is put into evidence. Another
synthesis paper, dealing with the general problems of polar bodies has been written
by W. Barański, K. Wilmański and Cz. Woźniak [25].
Various studies on problems of a general character are due to W. T. Koiter [95]
(asymmetric elasticity with constraint rotations), Cz. Woźniak and M. Jukowski
[210] (model of an elastic subsoil with couple-stresses), Z. Weselowski [208], A.
E. Green, P. M. Naghdi and W. L. Wainright [66], P. M. Naghdi [122] (a static-
geometric analogy for bodies with couple-stresses), V. A. Gordon and L. A. Tol-
okonnikov [61], B. V. Gorskiı̆ [62] and R. D. Mindlin [111] (relation between the
second gradient of displacements and couple-stresses).
In 1967, an IUTAM-Symposium on the theory of dislocations and on the
Cosserat type bodies took place in Freudenstadt and Stuttgart in Germany; we
7.1 General Equations 311
We have presented shortly only some problems in this direction, but their number
is greater. We will remember other studies concerning bodies with asymmetric
elasticity.
L. P. Vinokurov and N. I. Derevianko [205] dealt with straight bars which are
not subjected to torsion; M. Sokolovski [166] considers the torsion of these bars.
The thin plates have been studied by O. Hoffman [78]. The first studies on thin
shells have been made by W. G} unther [72]; a thesis in this direction has been
written by K. D. Schade [157].
Kirichenco [93] deals with spherical bodies with axial symmetry; analogous
studies have been made by Ju. N. Podil’chuk and A. M. Kirichenko [143].
Anisotropic bodies have been considered by E. V. Kuvchinskiı̆ and E. L. Aero
[102] and by Kessel [89].
Another group of researches deal with the concentration of stresses and couple-
stresses. Let us mention a first study of R. D. Mindlin [106]. H. Neuber [128–130]
continues his previous studies [10] for bodies with symmetric elasticity.
R. Muki and E. Sternberg [120] deal with analogous problems; as well,
E. Sternberg and R. Muki [170] consider concentrations of stresses around cracks.
Concentrations of stresses around cylindrical inclusions have been studied by
Y. Weitsmann [207].
The chief results in asymmetric thermoelasticity are due to W. Nowacki [133–
136], who generalizes the reciprocity theorems of V. Ionescu-Cazimir [81, 82]; the
Green’s functions, corresponding to Nowacki’s formulation have been found by
J. Wyrwiński [211]. F. Pietras and J. Wyrwiński [142] considered the plane
thermoplastic problem in the anisotropic case.
The interest of the researches has been observed in the direction of the dynamical
problems too. D. C. Gazis and R. F. Wallis [59, 60] deal with the propagation of
weaves in crystals. G. Rymarz [149], G. Adomeit [19] and K. F. Graff and Yih-Hsing
Pao [63] consider the surface waves of Rayleigh type. Other results have been given
by R. Stojanović, S. Djurić and L. Vujośević [175] and by S. Djurić [44]. G. N. Savin
and N. A. Shul’ga [155] study the plane dynamic problem.
312 7 Introduction to the Theory of Cosserat Type Bodies
7.1.1.3 Conclusion
As a conclusions, we observe that the theory of Cosserat type bodies has been
much developed in the second part of the twentieth century; one can put in evi-
dence, firstly, the researches in the United States then the studies in Germany
(chiefly in Braunschweig, Karlsruhe and Clausthal-Zellerfeld), the Russian results
(especially in Kiev and Moskow), the Italian research (in Padova), the Romanian
studies (in Bucharest), the Polish studies (in Warsaw) etc.
An important direction of research is the theory of the most general possible
bodies (multipolar bodies, bodies of generalized Cosserat type, with or without
internal defects, with or without initial deformations); another direction of study
consists in the theory of linearly elastic body (in the case of great or small
deformations or rotations).
Nevertheless, this last direction of study generalizes the classical theory of
elasticity; thus, the methods which are generally used may by applied in this case too.
One can easily study the two-dimensional problems, using the results obtained by us
[182, 183] in the symmetric case, various plane problems (using the methods given
by us [179], previously), as well as ideas given by us [178] in a synthesis paper.
But a problem which remains to be considered consists in the experimental
determination of the elastic constants; one must obtain numerical values for the
mechanical properties of the material. It has not been made much in this direction.
It is true that one can imagine plane or spatial structures (e.g., frames of various
constructions), which may be modelled as a Cosserat type body. But it is not
sufficient. To make arise the confidence in such a theory, one must show that it can
be applied at least to a small number of bodies, in certain circumstances, being
absolutely necessary at least for a small number of cases. Thus, experimental
researches very thoroughly made, very difficult too, are to be expected for the
consolidation of a theory the beauty of which influenced its actual development.
In the case of Cosserat type bodies, besides the displacements, the strains and the
local rotations of rigid body, appear the free rotations and the characteristics of the
deformation; the state of deformation is thus characterized by the displacement
vector u and by the free rotation vector U.
Let us consider, in a more general case, that—beside the local rotation of rigid
body, represented by the vector x—we assume the existence of a rotation of vector
U, of components Ui ¼ Ui ðx1 ; x2 ; x3 Þ; i ¼ 1; 2; 3: This rotation is called free
7.1 General Equations 313
where Ui ¼ Ui ðx1 ; x2 ; x3 Þ are the components of the free rotation vector U; con-
versely, we have
1
Ui ¼ ijk Ujk : ð7:40 Þ
2
Let us introduce the local rotation vector
X¼xU ð7:5Þ
which represents the local rotation, without the contribution of the free rotation U;
its components are Xi ¼ Xi ðx1 ; x2 ; x3 Þ; i ¼ 1; 2; 3. The antisymmetric local rota-
tion tensor of second order
2 3
0 X12 X31
TX ½Xij 4 X12 0 X23 5; ð7:6Þ
X31 X23 0
of components
Xij ¼ Xij ðx1 ; x2 ; x3 Þ ¼ ijk Xk ; ð7:7Þ
32
1
N*
P
1
N **
23
23
M* N
M ( x2, x3)
O x2
Let us take again the deformation of the angular element NMP, considered in
Sect. 2.2.1.3, the sides of which are parallel to the co-ordinate axes in the plane
x1 ¼ const (Fig. 7.1); analogically, we admit that the deformation does not depend
on x1 and takes place only in this plane, being in a linear case.
After deformation, without the contribution of the free rotation U, this element
becomes N M P , the angles b23 and b32 , made with the co-ordinate axes and
measured from these axes, being put into evidence; if we take into consideration
the influence of the free rotation U1 (corresponding to a positive rotation in the
x2 Ox3 -plane) too, then the angular element becomes N M P , the sides of which
make the angles a23 and a32 with the co-ordinate axes. The angle a23 represents,
e.g., the unit sliding of two elements of line parallel to the axis Ox3 , one with
regard to the other.
One may easily write
a23 ¼ b23 þ U1 ¼ u3;2 ; a32 ¼ b32 U1 ¼ u2;3 ;
where we used also the formulae (1.60). We get thus the relations
bij ¼ uj;i Uk ; bji ¼ ui;j þ Uk ; i 6¼ j 6¼ k 6¼ i;
we have, finally,
bij ¼ uj;i ijk Uk ; i; j; k ¼ 1; 2; 3: ð7:9Þ
7.1 General Equations 315
1 1
Xij ¼ bij bji ¼ b½ij ¼ uj;i ui;j Uij : ð7:110 Þ
2 2
It results that the symmetric part of the tensor Tb is the tensor Te , while its
antisymmetric part is the tensor TX ; hence
Tb ¼ Te þ TX ; ð7:12Þ
We discussed the above problems in the static case; in the dynamic case, all the
quantities which are involved depend also on the temporal variable t.
In the general case, the relations between characteristics of the deformation, dis-
placements and rotations can be assumed as a system of 18 equations (7.9) and
(7.130 ) with six unknowns ui ; Ui ; i ¼ 1; 2; 3. So as this system be compatible, the
characteristics of the deformation must verify certain conditions of compatibility.
Supposing that Ui are functions of class C2 , one may write (Schwarz’s theorem)
Uj;ik ¼ Uj;ki ;
wherefrom
uij;k ukj;i ¼ 0; i; j; k ¼ 1; 2; 3; ð7:15Þ
Analogically, assuming that ui are functions of class C2 , the relations (7.9) lead
to
bij;l blj;i ¼ jkl uik ijk ulk ; i; j; l ¼ 1; 2; 3; ð7:16Þ
where the relations (7.130 ) have been taken into consideration; one may write these
relations in one of the forms
ilm ðbij;l jkl uik Þ ¼ 0; j; m ¼ 1; 2; 3; ð7:160 Þ
Taking into account the decomposition (7.12) and that the product of a
symmetric tensor and an antisymmetric one, with respect to the same indices,
vanishes, one also obtains
ijl Xij;l þ 2ukk ¼ 0 ð7:180 Þ
* *
N1 N2
P N R P *
R*
** **
N1 N2
P ** Q* R **
* S*
M
Q M S
**
(a) Q (b) S **
P*
P* R* **
P
P** R ** R **
*
Q* ** R*
* N2* ** **N1 N2
Q N2 N1 S* ** *
**
N1 N S **
Q S ** Q**
S*
M* M *
(c) (d)
Fig. 7.2 Deformation of two elements: before deformation (a), separation (b), interpenetration
(c), continuity (d)
We observe that the relation (7.16) can be written also in the form
eij;l elj;i þ Xij;l þ Xjl;i ¼ jkl uik ijk ulk ; ð7:16000 Þ
We notice that this relation is equivalent to the relation (7.180 ). From (7.16000 ) and
(7.18000 ), one obtains
eij;l elj;i þ Xil;j ¼ ikl ujk ;
multiplying by iml and taking into account the relation (A.39), we have, finally,
1
uij ¼ jkl 2o½k eli Xkl;i ; ð7:1900 Þ
2
a relation equivalent to the relation (7.190 ), as it can be easily proved. This relation
may be written also in the form
1
uij ¼ jkl eli;k Xkl;i : ð7:19000 Þ
2
Observing that (see Fig. 2.8)
Z Z Z
Ui ¼ U0i þ _ dUi ¼ U0i þ _ Ui;j dxj ¼ _ uji dxj ; ð7:20Þ
P0 P P0 P P0 P
7.1 General Equations 319
one can calculate the free rotations at an arbitrary point P, if one knows these
quantities at a point P0 , in case of a simply connected domain. Taking into account
(7.190 ), one becomes, finally.
Z Z
1
Ui ¼ U0i jkl _ Xjk;l dxi þ ikl _ blj;k dxj : ð7:200 Þ
2 P0 P P0 P
where Uk ¼ Uk ðn1 ; n2 ; n3 Þ. Integrating by parts and taking into account the relation
(7.130 ), we obtain the Cesàro type formulae
Z h
i
ui ¼ u0i ijk xj x0j U0k þ _ bli ijk xj x0j ulk dxl ; ð7:21Þ
P0 P
0
by the aid of (7.19 ), the relation (7.21) takes the form
ui ¼ u0i ijk xj x0j U0k
Z
0
1
þ _ bki xj xj 2o½i bjk ijk lmn Xlm;n dxk : ð7:210 Þ
P0 P 2
Both the rotations and the displacements are thus expressed by means of the
components of the tensor Tb and of the displacements and rotations of a given
point P0 . The integrals do not depend on the path if the equations of continuity of
the deformations are verified.
Using the relations (7.19000 ), we may write the Cesàro type formulae in the form
Z Z
0 1
Ui ¼ Ui þ ikl _ elj;k dxj ikl _ Xkl;j dxj ; ð7:2000 Þ
P0 P 2 P0 P
ui ¼ u0i ijk xj x0j U0k
Z
þ _ bki xj x0j 2o½i ejk Xij;k dxk ; ð7:2100 Þ
P0 P
equivalent with those obtained above.
Observing that
Z
_ Xkl;j dxj ¼ Xkl X0kl ;
P0 P
Z
0
0
0
_ Xki þ xj xj Xij;k dxk ¼ xj xj Xij ;
P0 P
320 7 Introduction to the Theory of Cosserat Type Bodies
The antisymmetric part of the tensor Tb could be integrated; hence, the sym-
metric part leads to the classical formulae, as it has been shown by P. P. Teodo-
rescu [189].
If
Tu ¼ 0; Tb ¼ 0; ð7:22Þ
then one obtains the motion of rigid body. The formula (2.71000 ) remains valid; the
rotation is given by
U ¼ x0 : ð7:23Þ
TX ¼ 0; ð7:24Þ
then we say that we have to do with constrained rotations; in this case, the vector
U represents the constrained rotation vector while the tensor TU is the constrained
rotation tensor.
There does not exist a relation between the vectors u and U in the case of free
rotations; but, in case of constrained rotations, one has
1
U¼ curl u; ð7:25Þ
2
which is equivalent to
1
ijk Uk ¼ ðuj;i ui;j Þ ð7:250 Þ
2
or to
b½ij ¼ 0; ð7:2500 Þ
so that
U ¼ x: ð7:25000 Þ
7.1 General Equations 321
The classical Cesàro formulae (2.70), (2.700 ) are thus valid; in case of a simply
connected domain, the integrals do not depend on the path if the Saint-Venant
conditions are fulfilled. These conditions which—in general—are only necessary
conditions of compatibility, become now also sufficient ones.
d2 U o2 U €
¼ 2 ¼ U; ð7:28Þ
dt2 ot
of components U €i ¼ U€ i ðx1 ; x2 ; x3 ; tÞ; i ¼ 1; 2; 3, in the case of infinitesimal
deformations, in material co-ordinates too.
The vector relation (7.5) allows to introduce the rate of local rotations vector
_ ¼ x_ U;
X _ ð7:29Þ
We have thus seen how the symmetrical part and the antisymmetrical part of the
tensor Tb_ were thrown into relief; we can equally emphasize the symmetrical part
and the antisymmetrical one of the tensor Tu_ , although offering a remarkable
physical interpretation.
In the case of constrained rotations ðTX_ ¼ 0Þ one has
_ ¼ 1 curl u_ ¼ 1 curl v:
U ð7:31Þ
2 2
In general, after effecting a section S through a body (Fig. 3.1a), on one of the parts
(on the left one), on the area element DA in the tangent plane at point M, beside the
n n n
stress D P, there appears D M, of components DMn (along the external normal to
n
the section at the respective point) and DMt (in the plane tangent to the section at
the point M), that makes the relation with the remote part (the right one)
n
(Fig. 7.3a). We admit that the effort D M is of the nature of a moment. We obtain
thus the mean couple-stress vector in the neighbourghood of the point M for the
area element of external normal n, in the form
x3 t
t n n
n mt m
n
Mt M
M
I
M n
I n
r dA mn n
A Mn n
x1 O x2
(a) (b)
n
n DM
mmean ¼ ; ð7:32Þ
DA
at the limit, admitting that this exists, we find (Fig. 7.3b)
n
n n dM
m ¼ lim mmean ¼ ; ð7:33Þ
D A!0 dA
i.e. the couple-stress vector at the point M of an area element of external normal n.
n
The magnitude m of this vector represents the couple-stress at the point M. This
notion should be introduced with the same precautions, from a mathematical
standpoint, as the stress notion.
n
The component along the direction m of the couple-stress vector m shall be
n
noted with mm . By introducing the components along the three axes of co-ordi-
n n
nates mi ¼ mi ðx1 ; x2 ; x3 Þ; i ¼ 1; 2; 3, we shall be able to write
n n n n
n n
m2 ¼ mi mi ¼ m21 þ m22 þ m23 : ð7:34Þ
n n
The component of m along the external normal n shall be denoted by mn and
n
called normal (torsion) couple-stress; on the other hand, the component of m in
n
the plane tangent at M to the section S shall be noted by mt and shall be called
n n
tangential (bending) couple-stress. These components result from DMn and DMt
by a process of tending to the limit, analogous to the above indicated one
(Fig. 7.3b).
n
We shall equally introduce the notions of normal couple-stress vector mn and
n
tangential couple-stress vector mt . We shall have
n n n
m ¼ mn þ mt ; ð7:340 Þ
as well as
n n n
m2 ¼ m2n þ m2t : ð7:3400 Þ
The normal and tangential couple-stresses lead to the torsion and bending,
respectively. The normal couple-stress shall be considered positive if it corre-
n
sponds to a positive rotation (the vector mt in the same direction as n; an observer
along n will see the rotation from right to left) and negative in the opposite case.
As to the tangential couple-stress, its sign shall be specified in connection with a
system of axes of co-ordinates.
n
Now, by state of stress of the body, beside the totality of the stresses p in all
n
directions n, we shall equally mean the totality of the couple-stresses m in the
same direction. So as to understand the manner in which the couple-stresses
324 7 Introduction to the Theory of Cosserat Type Bodies
appear, we shall admit that, on an area element DA, of external normal n, the
internal forces are not uniformly distributed; that is why their resultant is not
n n
reduced to the sole stress D P but consists also in the couple-stress D M, which
leads to the couple-streses. We can even admit the existence of moments of higher
order (evidently with weaker and weaker effects), which leads to higher order
couple-stresses (hyperstresses); we shall confine ourselves to couple-stresses of the
first order (stresses of second order).
n
From a dimensional standpoint, the couple-stresses are ½m ¼ MT2 .
n n
As in the case of the stress vector p, the couple-stress vector m depends on the
n n
direction n. When p and m depend only on the direction n, we say that we have to
deal with a uniform state of stress, that will be determined by simple algebraic
equations of equilibrium; if in this case we do not take into account the volume
forces and the volume moments, we may confine ourselves to a global study of the
body. In the contrary case, we have a non-uniform state of stress and couple-stress
for which certain differential equations must be established; a local study of the
considered body should be carried out so as to complete the study made in the
previous section.
As in the classical case, the lost moments can be substituted to the volume
moments (introduced in the same manner as d’Alembert’s lost forces). Thus, we
introduce the volume inertia couple dðI dU=dtÞ=dt (analogous to the volume
inertia force); as a matter of fact, the substantial derivative of the angular
momentum is thus used. Here I is a rotation inertia moment per unit volume that
characterizes the body mass in a rotation motion.
In the case of homogeneous body, we have I ¼ const or I ¼ I ðtÞ; in the first
of that cases, the volume inertia couple is I d2 U=dt2 . But, in general, we have
I ¼ I ðx1 ; x2 ; x3 ; tÞ.
In the particular case of the infinitesimal deformations, for the substantial
derivatives, time derivatives are substituted, that supply the volume inertia couple
_
dðI dUÞ=dt; finally, when I ¼ I ðx1 ; x2 ; x3 Þ, therefore independent of time
(eventually a non-homogeneous body), the volume inertia couple assumes the
simple form I U. €
We can equally introduce the volume damping couple proportional to the
angular velocity, in the form k0 dU=dt, where the sign—appears owing to the fact
that this couple opposes the motion; here k0 is a damping coefficient per unit
volume, that—in general—can take the form k0 ¼ k0 ðx1 ; x2 ; x3 ; tÞ. In the case of
infinitesimal deformations the volume damping couple will be k0 U. _
In the case of infinitesimal, deformations that we shall consider later, the lost
moment will be supplied by
M €
¼ M I U; ð7:35Þ
€ k0 U;
M ¼ M IU _ ð7:350 Þ
in the case of small damped motions. In general, we shall consider the first of these
case.
In the case of Cosserat type bodies, the components of the stress tensor are
introduced in the same way as in the classic case, by the formula (2.1.17). The
theorem of Cauchy, expressed by the relation (3.160 ). as well as the relation (3.20),
which specifies the variation of the stresses around a point, remain valid too.
The stress tensor Tr is expressed by (3.26) and has the components (3.27) in the
dynamic case. The normal and the tangential components of the stress vector are
still given by the relation (3.25) and (3.28), respectively.
The quadric of the normal stresses is given by the Eq. (3.30), as in the classic
case. It depends solely on the symmetric part of the stress tensor; its antisymmetric
part brings no contribution to the determination of the principal normal stresses or
to the principal directions either. Therefore, in the results from Sect. 3.2.1.1, we
must substitute the symmetrical parts ðrij þ rji Þ=2 to the tangential stresses
rij ; i 6¼ j; obviously, here also we shall obtain three triorthogonal directions, since
the equation of the third degree supplying the values of the principal normal
stresses has always real roots. About this quadric, a discussion can be made,
similar to that in the above-mentioned subsection. As to the tangential stresses
corresponding to the principal directions, they are reduced to the antisymmetric
part of the tensor Tr (they do not vanish any more). For this reason, the corre-
sponding stress vectors are no more normal to the quadric of the normal stresses at
the points under consideration and we cannot any longer build Lamé’s stress
ellipsoid.
We shall consider again the equilibrium of a small tetrahedron MM1 M2 M3 cut
out of a body (as in Sect. 3.1.2.2); by taking also into account the lost moments
M i ; i ¼ 1; 2; 3, acting at the centre of gravity G of the tetrahedron, we remark that
MG M
i ¼ Mi þ gi ; i ¼ 1; 2; 3;
with gi ! 0 for h ! 0; we can write the moment equation with regard to an axis
parallel to the axis Oxi , passing through G0 . At the limit, with h ! 0, the volume
forces vanish from the computation and one obtains
Hence, the antisymmetric part of the stress tensor does not vanish any more,
so that the tensor Tr is an asymmetric one. This result is only valid when the
couple-stresses are not taken into account.
326 7 Introduction to the Theory of Cosserat Type Bodies
M3
x3
1
m
dx3
11
12 G1
21
M3 mn 3
13
Gn n
G2 m n
22
n
M (x1,x2, x3)G M2 m2 dx2
2 n
m r m1 M2
23 M1 31
O 32 x2
G3
x1 dx1
3
m 33
M1
i
Let us consider now the couple-stress vectors m, corresponding to the co-ordinate
axis Oxi ; we denote by
i
lij ¼ m ij ð7:37Þ
its component along the axis Oxj of unit vector ij . We obtain thus the components
lij ¼ lij ðx1 ; x2 ; x3 ; tÞ; i; j ¼ 1; 2; 3 of the couple-stress tensor
2 3
l11 l12 l13
Tl ½lij 4 l21 l22 l23 5; ð7:38Þ
l31 l32 l33
Let now be the tetrahedron MM1 M2 M3 on the faces of which we admit that
couple-stresses are acting (Fig. 7.4). Proceeding as in the classical case for
stresses, we get the relations
n
mi ¼ lij nj ; i ¼ 1; 2; 3; ð7:39Þ
As in the case of the tensor Tr , we can draw a quadric of the normal couple-
stresses, supplied by
the sign will be ? or - according to whether the rotation due to the couple-stress
is positive or negative, so that the resulting surface be real. We remark that in this
case only the symmetric part of the tensor Tl occurs, its antisymmetric part having
no influence on the results.
We find thus three orthogonal principal directions, since the equation of the
third degree that supplies the values of the principal couple-stresses always has
real roots l1 l2 l3 .
As concerns the appearing invariants, we shall only emphasize the first
invariant, which is the most important,
W ¼ lii ¼ l1 þ l2 þ l3 : ð7:42Þ
As in the classical case, we must see what occurs when we pass from an area
element going through the point M to an area element going through an adjacent
point. We shall thus start from a parallelepipedical element cut out of the body,
acted upon not only by stresses but by couple-stresses too.
Taking into account the variation of stresses and couple-stresses when passing
from a parallelipiped face to a neighbouring one, by formulae of Taylor type, and
considering also the volume forces and the volume moments, the inertia forces
and the inertia moments (i.e. the lost forces and the lost moments) (Fig. 7.5), and
remarking moreover that we must introduce relations permitting to pass from the
328 7 Introduction to the Theory of Cosserat Type Bodies
x3
33 + 33,3 dx3
32 + 32,3 dx3
31 + 31,3 dx3
12 11
dx3
23 + 23,2 dx2
M3
13
21 M1 M2 22 + 22,2 dx2
13 + 13,1 dx1 G
22 +
21 21,2 dx2
23
M (x1,x2, x3)
11 + 11,1 dx1 12 + 12,1 dx1
31
r dx1
32
O 33 x2
x1
dx2
centres of gravity of every adjacent face to the point Mðx1 ; x2 ; x3 Þ, we obtain the
equations of motion for stresses and couple-stresses in the form
rji;j þ Fi ¼ q€
ui ; ð7:43Þ
for i ¼ 1; 2; 3. These equations of motion are valid assuming that the stress tensor
is an asymmetric one.
These equations can also be obtained by using the flux-divergence formula
(A.95).
In the case of damped motions, the above equations become
rji;j þ Fi ¼ q€
ui þ ku_ i ; ð7:44Þ
for i ¼ 1; 2; 3.
In the static case, one obtains the equilibrium equations
rji;j þ Fi ¼ 0; ð7:45Þ
with i ¼ 1; 2; 3.
7.1 General Equations 329
€ i;
lji;j þ ikl rkl ¼ I U ð7:460 Þ
which have the same form as the relations (7.43). If the contrary occurs, then the
stress tensor is asymmetrical.
where the components of the tensor Tr are only related to the components of the
tensor Tb , while the components of the tensor Tl depend only on the components
of the tensor Tu .
330 7 Introduction to the Theory of Cosserat Type Bodies
Here h and w are given by (2.64) and (7.14); k and l are the elastic constants of
Lamé, while a is a third constant with the same dimensions ½a ¼ ML1 T2 . The
elastic constants k0 ; l0 and a0 correspond to the elastic constant k; l and a and play
a similar rôle; from the dimensional standpoint, all these constants have the same
dimensions as the above ones, but multiplied by a length squared ½k0 ¼ ½l0 ¼ ½a0
¼ MLT2 . Generally, a linearly elastic body of the Cosserat type with free rota-
tions depends on six independent elastic constants.
If k0 ¼ l0 ¼ a0 ¼ 0, while a 6¼ 0, we are in the case of asymmetrical elasticity
without couple-stresses, with free rotations, in presence of a volume moment M.
We remark thus that the presence of the volume moments involves the appearance
of free rotations in the case of bodies with three elastic constants ðk; l; aÞ. In the
dynamic case, the relations (7.36) are verified.
Summing up the symmetric parts of the constitutive Eqs. (7.47), (7.470 ), we get
the relations
H ¼ ð3k þ 2lÞh; ð7:48Þ
l0 ð3k0 þ 2l0 Þ 0 k0
E0 ¼ ; m ¼ ð7:50Þ
k0 þ l0 2ðk0 þ l0 Þ
are introduced, the first one of them having the dimension ½E0 ¼ LMT2 , while
the second one is dimensionless.
These relations can be equally expressed inversely in the form
m0 E 0 E0
k0 ¼ ; l0
¼ ¼ G0 : ð7:500 Þ
ð1 þ m0 Þð1 2m0 Þ 2ð1 þ m0 Þ
It is often useful to introduce an elastic constant l, with the dimension of a
length ð½l ¼ LÞ, expressed by
ðl þ aÞðl0 þ a0 Þ
l2 ¼ ; ð7:51Þ
4la
7.1 General Equations 331
k0 þ 2l0
h2 ¼ : ð7:52Þ
4a
In the case of asymmetrical elasticity without couple-stresses ðk0 ¼ l0 ¼ a0 ¼ 0;
a 6¼ 0Þ, we have l ¼ h ¼ 0; and the same occurs in the case of classical elasticity.
The Eqs. (7.45), (7.450 ) and the Eqs. (7.44), (7.440 ) form a complete system of 18
equations for the linearly elastic bodies of the Cosserat type, in the static and in the
dynamic case, respectively; to these equations, we must add the limit conditions
(boundary conditions and initial conditions).
332 7 Introduction to the Theory of Cosserat Type Bodies
x3 x3
n
p n
n u n
n n
m
r M r M
dA dA
x1 x2 x1 x2
O O
(a) (b)
Fig. 7.6 Fundamental problems. Boundary conditions: first (a), second (b)
In the case of the first fundamental problems, the boundary conditions are put in
displacements and rotations, in the form (Fig. 7.6a)
ui ¼ i ; i ¼ 1; 2; 3;
ui ; U i ¼ U ð7:54Þ
for a solution in displacements and rotations. The second fundamental problem
requires a solution in stresses and couple-stresses, so that the boundary conditions
are put consequently in the form (Fig. 7.6b)
n n
pi ¼ rji nj ; mi ¼ lji nj ; i ¼ 1; 2; 3: ð7:55Þ
In the case of the mixed problem, one puts conditions of the form (7.54) on one
part of the contour and conditions of the form (7.55) an another part of it.
W. T. Koiter [95] has shown that, in case of the Cosserat type bodies with
constrained rotations, the number of boundary conditions at a point of the contour
is reduced from 6 to 5.
In case of a dynamic problem, initial conditions of the form
where u0i ; U0i ; u_ 0i ; U_ 0i are the displacements and the rotations and their velocities,
respectively, at an arbitrary point of the body.
In 1966, N. S
ßandru [153] gives a representation of Galerkin’s type for the vector
unknowns of the problem
in this case, the state of displacement and rotation is specified by only one scalar
potential
u ¼ gradK0 ; U ¼ 0; ð7:610 Þ
where
ðk þ 2lÞDK0 þ P0 ¼ 0: ð7:6100 Þ
In the case of a solenoidal field of volume forces, hence if
F ¼ curlP; ð7:62Þ
334 7 Introduction to the Theory of Cosserat Type Bodies
M ¼ gradP00 ; ð7:63Þ
u ¼ 0; U ¼ gradK00 ; ð7:630 Þ
M ¼ curlP0 ð7:64Þ
one introduces a vector potential, which leads to the state of displacement and
rotation
u ¼ ½ðl0 þ a0 ÞD 4av
ðk þ l aÞðl0 þ a0 ÞD 4aðk þ lÞ
gradðr v þ v0 Þ 2a curlv0 ; ð7:65Þ
2ðk þ 2lÞ
7.2 Formulations of the Static and Dynamic Problems. General Theorems 335
In the case of constrained rotations ða ¼ k0 ¼ 0Þ, the Lamé type Eqs. (7.58),
(7.580 ) become
lDu þ ðk þ lÞ grad divu þ F ¼ 0; ð7:66Þ
Hereafter we shall deal with certain formulations of the problems of the Cosserat
type bodies with the help of potential functions; we shall thus consider both the
cases of free and constrained rotations.
1 o2 0 1 o2 1a
h2 ¼ D ; h 2 ¼ D 02 ; ð7:70Þ
c22 ot2 c02
2 ot 2 l l
1 o2 1 o2 1
h1 ¼ D 2 2
; h01 ¼ D 02 2 2 ; ð7:71Þ
c1 ot c1 ot h
k þ 2l 02 k0 þ 2l0
c21 ¼ ; c1 ¼ : ð7:710 Þ
q I
We remark that these operators are related to one another by the relations
ðk þ 2lÞh1 ðl þ aÞh2 ¼ ðk þ l aÞD; ð7:72Þ
i.e. the equation that should be verified by first invariants of the tensors Te and Tu ;
the first equation coincides with the corresponding one in the classical case.
In the absence of the volume forces and moments, the Lamé type equations
become
ðl þ aÞh2 u þ ðk þ l aÞ grad div u þ 2a curlU ¼ 0; ð7:74Þ
h1 h ¼ 0; h01 w ¼ 0: ð7:75Þ
0 0
Taking into account the relation (A.91 ), the Eqs. (7.74), (7.74 ) can be equally
expressed in the form
ðk þ 2lÞh1 u þ ðk þ l aÞ curl curlu þ 2a curlU ¼ 0; ð7:76Þ
7.2.2.2 S
ßandru’s Representation
N. S
ßandru [153] gave a representation of Iacovache [80] type of the solutions of
the Eqs. (7.69), (7.690 ).
Eliminating the rotation vector and the displacement vector, respectively,
between the Eqs. (7.74), (7.740 ), we obtain
or in the form
with
h1 h01 DC ¼ 0: ð7:85Þ
In this case, it follows that
u ¼ divC; ð7:86Þ
where we neglected a function C0 which verifies the equation
7.2 Formulations of the Static and Dynamic Problems. General Theorems 339
h1 h01 C0 ¼ 0 ð7:87Þ
that is complete for the Eq. (7.78) in the case of a simply connected domain (this
result has been obtained step by step, starting from the Eq. (7.78)); the vector
potential C ¼ Cðx1 ; x2 ; x3 ; tÞ should verify the Eq. (7.85).
Similarly, for the function U, which verifies an equation analogous to the Eq.
(7.78), we get the complete representation
analogically, we obtain
U ¼ ðk0 þ 2l0 Þðl þ aÞh01 h2 C0 D2 grad divC0 D01 curl curlC0 : ð7:890 Þ
We remark, however, that the representation (7.88), (7.880 ) should verify the
Eqs. (7.74), (7.740 ), from which we started (the Eqs. (7.77), (7.770 ) are only their
consequences and represent necessary conditions).
Taking into account the relations
the Eqs. (7.75) show that the potential functions C and C0 must verify the equations
which represent more restricted conditions than (7.85) and (7.850 ); as a matter of
fact, these conditions will nevertheless be fulfilled. We remark moreover that we
have neglected a component in the form of a curl of the vectors C and C0 ,
respectively, that vanishes in the relations (7.90).
In this case, we remark that, in the representations (7.89), (7.890 ), we can neglect
the terms in curl, verifying the Eqs. (7.77), (7.770 ) by means of the conditions
(7.900 ). Moreover, we remark that we can introduce additional terms such as
340 7 Introduction to the Theory of Cosserat Type Bodies
u ¼ curlW; ð7:91Þ
U ¼ curlW0 ; ð7:910 Þ
DW ¼ 0; DW0 ¼ 0; ð7:92Þ
so that the Eqs. (7.77), (7.770 ) should be verified. In point of fact, these additional
terms correspond to the passage from the Eqs. (7.74), (7.740 ) to the Eqs. (7.77),
(7.770 ).
Now, laying the condition that the Eqs. (7.74), (7.740 ) should be verified by the
representation (7.89), (7.890 ) with the addition of (7.91), (7.910 ), respectively, we
obtain
Taking into account (7.900 ), we see that these conditions are fulfilled when
the conditions (7.92) will also be fulfilled, since they agree with the conditions
(7.900 ).
In this case, we obtain the representation of N. ßSandru
where the vector potentials C and C0 are functions of class C6 that verify the
Eq. (7.900 ) in the absence of the volume forces and moments; this representation is
complete for a simply connected domain.
If the volume forces and moments are not vanishing, i.e. in the case of the
system of Eqs. (7.69), (7.690 ), we shall use the same representation, and the
potential functions must verify the equations
From (7.900 ), (7.95), (7.950 ) we see that, in the absence of the volume forces and
moments, the displacement and the rotation vectors verify the equations
h1 Du ¼ 0; h01 DU ¼ 0: ð7:97Þ
Since the operators h1 ; h01 and D are prime with one another, we can use
Boggio’s theorem; it follows that
u ¼ u 1 þ u2 ; U ¼ U 1 þ U 2 ; ð7:98Þ
with
h1 u1 ¼ 0; h01 U1 ¼ 0: ð7:99Þ
which corresponds to a decomposition into two types of waves. We remark that the
displacements verifying the first Eq. (7.99) correspond to the classical case (lon-
gitudinal waves).
where
ðk þ 2lÞh1 K0 þ P0 ¼ 0: ð7:1020 Þ
In the case of a solenoidal field of volume forces, therefore of the form
F ¼ curlP; ð7:103Þ
M ¼ gradP00 ; ð7:106Þ
u ¼ 0; U ¼ gradK00 ; ð7:107Þ
the scalar potential K00 ¼ K00 ðx1 ; x2 ; x3 ; tÞ being supplied by the equation
M ¼ curlP0 ; ð7:108Þ
DK0 þ P0 ¼ 0: ð7:1090 Þ
By the method of superposition of effects, we can obtain the results corre-
sponding to a general loading case with volume forces and moments.
C ¼ , þ ,1 ; C0 ¼ ,0 þ ,01 ; ð7:110Þ
where
7.2 Formulations of the Static and Dynamic Problems. General Theorems 343
D, ¼ 0; h1 ,1 ¼ 0; ð7:111Þ
corresponding to irrotational waves that verify the Eqs. (7.99); on the other hand,
the state of displacement and rotation expressed solely by the vector potentials ,
and ,0 is solenoidal (h ¼ 0; w ¼ 0, therefore incompressible) and verifies the Eqs.
(7.990 ). This can constitute an extension of the decomposition into two wave types,
from the classical case, a fact realized in the previous subsection too.
Taking into account the relations (7.72), (7.720 ), (7.79), (7.7900 ), the represen-
tation (7.112), (7.1120 ) can be written as follows
and making use of the differential relations (A.910 ), we can moreover express the
state of displacement and rotation in the form
which may constitute a generalization, for the Cosserat type bodies, of the classical
Lamé-Clebsch representation. We introduce thus two vector potentials , ¼
,ðx1 ; x2 ; x3 ; tÞ and ,0 ¼ ,0 ðx1 ; x2 ; x3 ; tÞ of class C6 and two scalar potentials ,0 ¼
,0 ðx1 ; x2 ; x3 ; tÞ and ,00 ¼ ,00 ðx1 ; x2 ; x3 ; tÞ of class C3 .
With the help of the results of previous subsection, we can write a represen-
tation of the same type in the form
where the scalar potentials should be functions of class C3 , verifying the equations
h1 K0 ¼ 0; h01 K0 0 ¼ 0 ð7:119Þ
and the vector potentials should be of class C5 and satisfy the equations
DK ¼ 0; DK0 ¼ 0: ð7:1190 Þ
Obviously, the representations (7.117), (7.1170 ) and (7.118), (7.1180 ) are
equivalent.
v0 ¼ 2ðk þ 2lÞdivC r v;
ð7:1200 Þ
v00 ¼ 2ðk0 þ 2l0 ÞdivC0 r v0 :
Taking into account the Eqs. (7.900 ), we see that the vector potentials should verify
the equations
7.2 Formulations of the Static and Dynamic Problems. General Theorems 345
Dv ¼ 0; Dv0 ¼ 0; ð7:121Þ
likewise, with the help of the notations (7.120), the relations (7.1200 ) lead to the
conditions
h1 v0 ¼ 2divv h1 ðr vÞ;
ð7:122Þ
h01 v00 ¼ 2divv0 h01 ðr v0 Þ;
h1 v0 þ r h1 v ¼ 0;
ð7:1220 Þ
h01 v00 þ r h01 v0 ¼ 0:
The state of displacement and rotation becomes
u ¼ ðl0 þ a0 Þh02 v
1
D0 gradðr v þ v0 Þ 2a curlv0 ;
2ðk þ 2lÞ 2
ð7:123Þ
U ¼ ðl þ aÞh2 v0
1
D2 gradðr v0 þ v00 Þ 2a curlv;
2ðk0 þ 2l0 Þ
where both the vector and scalar potentials must be functions of class C5 . One
obtains thus the Sternberg-Eubanks type representation.
Hereafter, we shall give some theorems concerning the uniqueness of the solution of
the equations of linearly elastic bodies of Cosserat type, using the monograph of W.
Nowacki [11]; as well, we shall enounce, without demonstration, some theorems
of reciprocity.
On the basis of this theorem, the necessary and sufficient conditions in which W
is positive definite are
3k þ 2l [ 0; l [ 0; a [ 0; ð7:125Þ
In 1965, N. Sßandru [151, 152] stated theorems of reciprocity of the work both for
the static case (of Betti type) and for the dynamic case (of Graffi type).
Thus, considering two states of displacement and rotation and of stress and
couple-stress of an linearly elastic body of Cosserat type, marked by (0 ) and (00 ),
corresponding to two distinct systems of external loads, one may write, in the static
case,
ZZZ ZZ n
0 00 0 00
0 00
n
0 00
Fi ui þ Mi Ui dV þ pi ui þ mi Ui dS
V S
ZZZ ZZ n
00 0 00 0
00 0
n
00 0
¼ Fi ui þ Mi Ui dV þ pi ui þ mi Ui dS: ð7:126Þ
V S
In the dynamic case, we introduce the lost forces and the lost moments, so that
the corresponding relation of reciprocity reads
ZZZ ZZ n
0
0 00
0 0
00 0 00
n
0 00
Fi q€ €
ui ui þ Mi I Ui Ui dV þ pi ui þ mi Ui dS
V S
ZZZ ZZ n
00
00 0
00 00
0 00 0
n
00 0
¼ Fi q€ €
ui ui þ Mi I Ui Ui dV þ pi ui þ mi Ui dS:
V S
ð7:127Þ
7.2 Formulations of the Static and Dynamic Problems. General Theorems 347
References
A. Books
1. Boltzmann, L.: Populäre Schriften, 3rd edn. Barth Verlag, Leipzig (1925)
2. Cosserat, E., Cosserat, F.: Théorie des corps déformables. A. Herman et Fils, Paris (1909)
3. Grioli, G.: Mathematical Theory of Elastic Equilibrium (Recent Results) (Erg. Angew.
Math.). Springer, Berlin (1962)
4. Hamel, G.: Die Axiome der Mechanik. Handbuch der Physik, vol. 5. Springer, Berlin (1927)
5. Hellinger, E.: Die allgemeinen Ansätze der Mechanik der Kontinua. In: Klein, F., Wagner,
K. (eds.) Enz. Math. Wiss., vol. 4. Springer, Berlin (1914)
6. Heun, K.: Ansätze und allgemeine Methoden der Systemmechanik. Enz. Math. Wiss., vol. 4
(1914)
7. Jaramillo, T.J.: A Generalization of the Energy Function of Elasticity Theory. Dissertation,
University of Chicago (1929)
8. Kelvin, L.: Mathematical and Physical Papers, vol. 1–3. 1882, 1884, 1890
9. Kr}oner, E.: Kontinuumstheorie der Versetzungen und Eigenspannungen (Erg. Angew.
Math.). Springer, Berlin (1958)
10. Neuber, H.: Kerbspannungslehre. Grundlagen f} ur genaue Festigkeitsberechnung mit
Ber}ucksichtigung von Konstruktionsform und Werkstoff, 2nd edn. Springer, Berlin (1958)
11. Nowacki, W.: Teoria niesymetrycznej spre_zistości (Theory of Asymmetric Elasticity).
Państ. Wydawn. Naukowe, Warszawa (1971)
12. Savin, G.N.: Osnovy ploskoı̆ padachi momentnoi teorii uprugosti (Fundamentals of the
Problems of the Moment Theory of Elasticity). T. G. Shevchenko University, Kiev (1965)
13. Schaefer, H.: Continui di Cosserat. Funzioni potenziali. Calcolo numerico delle piastre.
Univ. di Trieste. Ist. di Mecc., Lezione e conferenzes, vol. 7, Trieste (1965)
14. Teodorescu, P.P.: Probleme spatßiale în teoria elasticitătßii (Space Problems in the Theory of
Elasticity). Ed. Acad., Bucuresßti (1970)
15. Teodorescu, P.P.: Dynamics of Linear Elastic Bodies. Ed. Acad., Bucuresßti. Abacus Press,
Tunbridge Wells (1975)
16. Teodorescu, P.P. (ed.): Probleme actuale în mecanica solidelor (Present-day Problems in
Solid Mechanics), vol. 1. Ed. Acad., Bucuresßti (1975)
348 7 Introduction to the Theory of Cosserat Type Bodies
17. Truesdell, C., Noll, W.: The Non-Linear Field Theories of Mechanics. Encycl. Phys. vol.
III(3). Springer, Berlin (1965)
18. Truesdell, C., Toupin, R.A.: The Classical Field Theories. Encycl. Phys. vol. III(1).
Springer, Berlin (1960)
B. Papers
19. Adomeit, G.: Ausbreitung elastischer Wellen in einen Cosserat Kontinuum mit freien
Oberfläche. ZAMM 46, 158 (1966)
20. Adomeit, G.: Determination of elastic constants of a structured material. IUTAM Symp.
1967, Mech. Generalized Continua, p. 80 (1968)
21. Aero, E.L., Buligin, A.N., Kuvshinskii, E.V.: Asymmetrical hydrodynamics. Prikl. mat. e
mekh. 2, 297 (1965)
22. Aero, E.L., Kuvshinskii, E.V.: Fundamental equations of the theory of elastic media with
rotational interaction of the particles. Fiz. tviordogo tiela 2, 1399 (1960)
23. Alblas, J.B.: The Cosserat continuum with electronic Spin. Mechanics of Generalized
Continua, IUTAM Symp. 1967, p. 350 (1968)
24. Anthony, K., Essmann, U., Seeger, A., Träuble, H.: Disclinations and the Cosserat
continuum with incompatible rotations. IUTAM Symp. 1967, Mechanics of Generalized
Continua, p. 355 (1968)
25. Barański, K., Wilmański, K., Wo_zniak, Cz.: Mechanika ośrodków ciaglich typa Cosseratów
(mechanics of bodies of Cosserat type). Mech. teoret. i stos. 5, 215 (1967)
26. Bodaśzewski, S.: O niesymetrycznym stanie niapecia i o zastowaniach w mechanike
ósrodków ciagluch (On the asymmetric state of stress and couple stress in the mechanics of
continuous bodies). Arch. Mech. Stos. 5, 351 (1953)
27. Boggio, T.: Sull’integrazione di alcune equazioni lineari alle derivate parziali. Ann. Mat.
Pura Appl. 8(Ser. 3), 181 (1903)
28. Bressan, A.: Sui sistemi continui nel caso asimmetrico. Ann. Mat. 512(Ser. 4), 169 (1963)
29. Burzyński, W.: Skrecenie bez skrecania. Przegl. Mech. (1949)
30. Carlson, D.E.: The general solution of the stress equations of equilibrium of a Cosserat
continuum. Proc. Fifth U. S. Natl. Congr. Appl. Mech., p. 249 (1966)
31. Carlson, D.E.: Stress functions for couple and dipolar stresses. Appl. Math. 24, 29 (1966)
32. Carlson, D.E.: On G} unther’s stress functions for couple stresses. Appl. Math. 25, 139 (1967)
33. Cauchy, A.L.: Mémoire sur les systèmes isotropes de points matériels. Mém. Acad. Sci.
Paris 22, 605 (1850)
34. Cauchy, A.L.: Note sur l’équilibre et les mouvements vibratoires des corps solides. C. Rend.
hebd. des séances Acad. Sci. Paris 32, 326 (1851)
35. Clintock, F.A.Mc: Contribution of interface couples to the energy of dislocation. Acta
metallurgica 8, 127 (1960)
36. Clintock, F.A.Mc., André, P.A., Schwedt, K.R., Stoeckly, R.E.: Interface couples in
crystals. Nature 182, 652 (1958)
37. Corre, Y.Le.: Constantes élastiques et piézoélectriques cristallines. Bull. Soc. Franc. Minér.
Crist. 76, 464 (1953)
38. Corre, Y.Le: Étude de l’élasticité. Bull. Soc. Franc. Minér. Crist. I, II, 77, 1363, 1393
(1954); III, IV, 28, 33, 54 (1955)
39. Corre, Y.Le.: La dissymétrie du tenseur des efforts et ses conséquences. J. Phys. Radium 17,
934 (1956)
40. Corre, Y.Le.: Les densités de couple et les pseudo rotations dans la théorie de l’élasticité de
Lavel. J. Phys. Radium 19, 541 (1958)
41. Cosserat, E., Cosserat, F.: Sur la théorie de l’élasticité. Ann. Éc. norm. Toulouse 10, 1
(1896)
References 349
42. Cosserat, E., Cosserat, F.: Sur la mécanique générale. C. Rend. hebd. Acad. Sci., Paris 145,
1139 (1907)
43. Dixon, R.C., Eringen, A.C.: A dynamical theory of polar elastic dielectrics. Int. J. Eng. Sci.
3, 359 (1965)
44. Djurić, S.: Dynamique et petites vibrations du continuum de Cosserat. Thèse Droit.,
Belgrade (1964)
45. Ericksen, J.L.: Conservation laws for liquid crystals. Trans. Soc. Rheol. 5, 23 (1961)
46. Ericksen, J.L.: Hydrostatic theory of liquid crystals. Arch. Rat. Mech. Anal. 9, 374 (1962)
47. Ericksen, J.L.: Nilpotent energies in liquid crystals theory. Rat. Mech. Anal. 10, 189 (1962)
48. Ericksen, J.L.: Orientation induced by flow. Trans. Soc. Rheol. 6, 275 (1962)
49. Ericksen, J.L., Truesdell, C.: Exact theory of stress and strain in rods and shells. Arch. Rat.
Mech. Anal. 1, 295 (1958)
50. Eringen, A.C.: Simple microfluids. Int. J. Eng. Sci. 2, 205 (1964)
51. Eringen, A.C.: Mechanics of micromorphic continua. IUTAM Symp. 1967, Mechanics
Generalized Continua, p. 18 (1968)
52. Eringen, A.C., Suhubi, E.S.: Nonlinear theory of simple microelastic solids. Int. J. Eng. Sci.
2(I, II), 189, 389 (1964)
53. Galerkin, B.G.: On the problem of stresses and deformations in an isotropic elastic body.
Dokl. Akad. Nauk SSSR 14(Ser. A), 353 (1930)
54. Galletto, D.: Sulle equazioni in coordinate generali della statica dei continui con
caratteristiche di tensione asimmetriche. St. Anna University Hospital, Ferrara 10(Ser. 7),
33 (1962)
55. Galletto, D.: Nuove forme per le equazioni in coordinati generali della statica dei continui
con caratteristiche di tensione asimmetriche. Ann. Scuola Norm. Sup. Pisa, Sci. Fis. Mat.
17(Ser. 3), 297 (1963)
56. Galletto, D.: Contributo allo studio dei sistemi continui a transformazioni reversibili con
caratteristiche di tensione asimmetriche. Rend. Sem. Mat. Univ. Padova, 35, 299 (1965)
57. Galletto, D.: Sull’unicità in presenza di vincoli interni di una condizione cinematica
fondamentale nella teoria delle deformazione finite. Atti Ist. Veneto Sci. Lett. Arti, Cl. Sci.
Mat. Nat. 123, 197 (1965)
58. Galletto, D.: Sistemi incomprimibili a trasformazioni reversibili nel caso asimmetrico.
Rend. Sem. Mat. Univ. Padova I, II, 36, 243 (1966); 37, 18 (1967)
59. Gazis, D.C., Wallis, R.F.: Extensional waves in cubic crystal plates. Proceedings of 4th U.
S. National Congress of Applied Mechanics, p. 161 (1962)
60. Gazis, D.C., Wallis, R.F.: Surface vibrational waves in crystal lattices with complex
interatomic interactions. J. Math. Phys. 3, 190 (1962)
61. Gordon, V.A., Tolokonnikov, L.A.: Equations of equilibrium in the moment theory of
elasticity. Prikl. Mekh., Kiev 4, 118 (1968)
62. Gorskii, B.V.: On some problems of the theory of asymmetric elasticity. Prikl. mekh., Kiev
3, 74 (1967)
63. Graff, K.F., Pao, Y.H.: The effects of couple stresses on the propagation and reflection of
plane waves in an elastic half space. J. Sound Vibr. 6, 217 (1967)
64. Green, A.E.: Micro materials and multipolar continuum mechanics. Int. J. Eng. Sci. 3, 533
(1965)
65. Green, A.E., Naghdi, P.M.: The Cosserat surface. IUTAM Symp. 1967, Mechanics of
Generalized Continua, p. 36 (1968)
66. Green, A.E., Naghdi, P.M., Wainrght, W.L.: A general theory of Cosserat surface. Arch.
Rat. Mech. Anal. 20, 287 (1965)
67. Green, A.E., Rivlin, R.S.: Multipolar continuum mechanics. Arch. Rat. Mech. Anal. 17, 113
(1964)
68. Grioli, G.: Elasticità asimmetrica. Ann. Mat. Pura Appl. 50(Ser. 4), 389 (1960)
69. Grioli, G.: Sulla meccanica dei continui a transformazioni reversibili con caratteristiche di
tensioni asimmetriche. Sem. Ist. Alta Mat. 535 (1962–1963)
350 7 Introduction to the Theory of Cosserat Type Bodies
70. Grioli, G.: On the thermodynamic potential of Cosserat continua. IUTAM Symp. 1967,
Mechanics of Generalized Continua, p. 63 (1968)
71. G}unther, W.: Zur Statik und Kinematik des Cosseratschen Kontinuums. Abh. der
Braunschw. Wiss. Gesellschaft, 10, 195 (1958)
72. G}unther, W.: Analoge systeme von schalengleichungen. Ing. Archiv. 30, 160 (1961)
73. Gurtin, M.E.: A generalization of the beltrami stress functions in continuum mechanics.
Arch. Rat. Mech. Anal. 13, 321 (1963)
74. Hartranft, R.J., Sih, G.C.: The effect of couple stresses on the stress concentrations of a
circular inclusion. J. Appl. Mech. 32, 429 (1965)
75. Hehl, F.: Space time as generalized Cosserat continuum. IUTAM Symp. 1967, Mechanics
of Generalized Continua, p. 347 (1968)
76. Hehl, F., Kr}oner, E.: Zum Materialgesetz eines elastischen Mediums mit Moment-
spannungen. Z. Naturforschung 20, 336 (1965)
77. Hermann, G., Achenbach, J.D.: Applications of theories of generalized Cosserat continua to
the dynamics of composite materials. IUTAM Symp. 1967, Mech. Generalized Continua,
p. 69 (1968)
78. Hoffman, W.H.: On the bending of thin elastic plates in the presence of couple stresses.
J. Appl. Mech. 31, 706 (1964)
79. Hoffman, W.H., Shahman, F.O.F.: Physical model of a 3 constant isotropic elastic material.
J. Appl. Mech. 4, 837 (1965)
80. Iacovache, M.: O extindere a metodei lui Galerkin pentru sistemul ecuatßiilor elasticitătßii (A
generalization of galerkin’s method for the system of elasticity equations). Bul. Acad. I(Ser.
A), 592 (1949)
81. Ionescu Cazimir, V.: Problem of linear coupled thermoelasticity. I. Theorems on reciprocity
for the dynamic problem of coupled thermoelasticity. Bull. Acad. Pol. Sci. Sér. Sci. Techn
12, 473 (1964)
82. Ionescu Cazimir, V.: Problem of linear coupled thermoelasticity. II. Some applications of
the theorems of reciprocity for the dynamic problem of coupled thermoelasticity. Bull.
Acad. Pol. Sci. Sér. Sci. Techn12, 481 (1964)
83. Joel, N., Wooster, W.A.: Theories of crystal elasticity. Nature 180, 430 (1957)
84. Joel, N., Wooster, W.A.: Number of elastic constants requested in crystal elasticity. Nature
182, 1078 (1958)
85. Kaliski, S.: O pewym modelu ošrodka ciaglego z istotnie niesymetricznym tensorem napié
(on the model of bodies with asymmetric stress tensor). Biul. WAT 4, 104 (1962)
86. Kaliski, S.: On a model of the continuum with essentially non symmetric tensor of
mechanical stress. Arch. Mech. Stos. 15, 33 (1963)
87. Kaliski, S., Plochocki, D., Rogula, D.: Asymmetric stress tensor and angular momentum
conservation law in the equations of combined mechanical and electromagnetic field in a
continuous medium. Proc. Vibr. Probl. 3, 253 (1962)
88. Kelly, P.D.: A reacting continuum. Int. J. Eng. Sci. 2, 129 (1964)
89. Kessel, S.: Lineare elastizitätstheorie des anisotropen Cosserat kontinuums. Abh. der
Braunschw. Wiss. Gesellschaft 16, 1 (1964)
}
90. Kessel, S.: Uber die Verträglichkeitsbedingungen in einem Cosseratschen Kontinuums.
Abh. der Braunschw. Wiss. Gesellschaft 17, 51 (1965)
91. Kessel, S.: Die spannungsfunktionen des Cosserat kontinuums. ZAMM 47, 329 (1967)
92. Kessel, S.: Stress functions and loading singularities for the infinitely extended, linear,
elastic isotropic Cosserat continuum. IUTAM Symp. 1967, Mechanics of Generalized
Continua, p. 114 (1968)
93. Kirichenko, A.M.: Couple stresses in the Study of the axisymmetrical states of stress of an
elastic sphere. Prikl. mekh. Kiev 4, 47 (1968)
94. Klein, F.: Nachr. kgl. Ges. Wiss, Göttingen, 235 (1918)
95. Koiter, W.T.: Couple stresses in the theory of elasticity. I, II. Koninkl. Nederl. Akad. van
Wettenschappen, Proc. 67(Ser. B), 17, 30 (1964)
References 351
96. Kondo, K.: On the two main currents of the geometrical theory of imperfect continua.
IUTAM Symp. 1967, Mechanics of Generalized Continua, p. 200 (1968)
97. Krishnan, R.S., Rajagopal, E.S.: The atomistic and the continuum theories of crystal
elasticity. Ann. der Physik 8, 121 (1961)
98. Kr}oner, E.: On the physical reality of torque stresses in continuum mechanics. Int. J. Eng.
Sci. 1, 261 (1963)
99. Kr}oner, E.: Das physikalische problem der antisymmetrischen Spannungen und der
sogennanten Momentspannungen. Proceedings of the 10th International Congress of
Applied Mechanics, München, p. 143,1964 (1966)
100. Kr}oner, E.: Interrelations between variuos branches of continuum mechanics. IUTAM
Symp. 1967, Mechanics of Generalized Continua, p. 330 (1968)
101. Kubenko, V.D., Shul’ga, N.A.: Plane dynamic problem of the moment theory of elasticity
and viscoelasticity. Prikl. mekh. Kiev 3, 82 (1967)
102. Kuvshinskii, E.V., Aero, E.L.: Theory of continuum in the asymmetric elasticity,
considerations of the ‘‘internal’’ rotations. Fiz. tviord. tiela 5, 2591 (1963)
103. Kunin, I.A.: The theory of elastic media with microstructure and theory of dislocations.
IUTAM Symp. 1967, Mechanics of Generalized Continua, p. 321 (1968)
104. Laval, J.: L’élasticité du milieu cristallin. J. Phys. Radium 18(1–3), 217, 289, 369 (1957)
105. Lomakin, V.A., Savov, L.N.: Deformation problems of viscoelastic micro non
homogeneous bodies and the theory of moments in viscoelasticity. Mekh. polimerov 2
(1967)
106. Mindlin, R.D.: Influence of couple stresses on stress concentrations. Exper. Mech. 3, 1
(1963)
107. Mindlin, R.D.: Microstructure in linear elasticity. Arch. Rat. Mech. Anal. 16, 51 (1963)
108. Mindlin, R.D.: Representation of displacements and stresses in plane strain with couple
stresses. IUTAM Symp. Tbilisi, 1963, p. 256 (1964)
109. Mindlin, R.D.: On the equations of elastic materials with microstructure. Int. J. Solid Struct.
1, 73 (1965)
110. Mindlin, R.D.: Stress functions for a Cosserat continuum. Int. J. Solid Struct. 1, 265 (1965)
111. Mindlin, R.D.: Second gradient of strain and surface tension in linear elasticity. Int. J. Solid
Struct. 1, 417 (1965)
112. Mindlin, R.D.: Theories of elastic continua and crystal lattice theories. IUTAM Symp. 1967,
Mechanics of Generalized Continua, p. 312 (1968)
113. Mindlin, R.D., Tiersten, H.F.: Effect of couple stresses in linear elasticity. Arch. Rat. Mech.
Anal. 11, 415 (1962)
114. Misßicu, M.: Theory of viscoelasticity with couple stresses and some reductions to the two
dimensional problems. Rev. Roum. Sci. Techn., Méc. Appl. I, II, 8, 921 (1963); 9, 3 (1964)
115. Misßicu, M.: On a theory of asymmetric plastic and viscoelasticplastic solids. Rev. Roum.
Sci. Techn., Méc. Appl. 9, 477 (1964)
116. Misßicu, M.: A generalization of the Cosserat equations of the motion of deformable bodies
(with internal degrees of freedom). Rev. Roum. Sci. Techn. Méc. Appl. 9, 1351 (1964)
117. Misßicu, M.: On a general solution of the theory of singular dislocation of media with couple
stresses. Rev. Roum. Sci. Techn. Méc. Appl. 10, 34 (1965)
118. Misßicu, M.: A generalization of the motion of deformable bodies. Arch. Mech. Stos. 17, 183
(1965)
119. Misßicu, M.: The generalized dual continuum in elasticity and dislocation theory. IUTAM
Symp. 1967, Mechanics Generalized Continua, p. 141 (1968)
120. Muki, R., Sternberg, E.: The influence of the couple stresses on singular stress concentration
in elastic bodies. ZAMP 16, 611 (1965)
121. Naghdi, P.M.: On a variational theorem in elasticity and its applications to shell theory.
J. Appl. Mech. 31, 647 (1964)
122. Naghdi, P.M.: A static geometric analogue in the theory of couple stresses. Proc. Konicl.
Nederl. Akad. Wet. 68(Ser. B), 29 (1965)
352 7 Introduction to the Theory of Cosserat Type Bodies
123. Nemish, Iu.N.: Plane problem of the moment theory of elasticity in a domain with a circular
hole. Prikl. Mekh. Kiev I, 52 (1965)
124. Nemish, Iu.N.: Stress concentration in the vicinity of curvilinear holes in asymmetric theory
of elasticity. Prikl. Mekh. Kiev II, 85 (1966)
125. Nemish, Iu.N.: Reinforced circular hole in an elastic domain with an asymmetric stress
tensor. Prikl. Mekh. Kiev II, 43 (1966)
126. Neuber, H.: Ein neuer Ansatz zur Lösung rämlicher Probleme der Elastizitätstheorie. Der
Hohlkegel unter Einzellart als Beispiel. ZAMM 14, 203 (1934)
127. Neuber, H.: On the general solution of linear elastic problems in anisotropic and isotropic
Cosserat continua. Proceedings of the 11th International Congress on Applied Mechanics,
M}unchen, p. 153 (1964)
}
128. Neuber, H.: Uber Probleme der Spannungskonzentrationen in Cosserat Körper. Acta Mech.
II, 48 (1966)
129. Neuber, H.: Die schubbeanspruchte Kerbe in Cosserat Körper. ZAMM 47, 313 (1967)
130. Neuber, H.: On the effect of stress concentration in Cosserat continuum. IUTAM Symp.
1967, Mechanics of Generalized Continua, p. 109 (1968)
131. Noether, E.: Nachr, kgl. Ges. Wiss., Göttingen, p. 171 (1918)
132. Noll, W.: A mathematical theory of mechanical behaviour of continuous media. Arch. Rat.
Mech. Anal. 2, 195 (1958)
133. Nowacki, W.: Couple stresses in the theory of thermoelasticity. Bull. Acad. Pol. Sci. Sér.
Sci. Techn. 14(1–3), 97, 203, 505 (1966)
134. Nowacki, W.: Some theorems of asymmetric thermoelasticity. Bull. Acad. Pol. Sci. Sér. Sci.
Techn. 15, 289 (1967)
135. Nowacki, W.: Couple stresses in thermoelasticity. Prikl. mekh., Kiev 3, 3 (1967)
136. Nowacki, W.: Couple stresses in the theory of thermoelasticity. IUTAM Symp. 1966,
Irreversible Aspects of Continuum Mechanics (1968)
137. Oshima, N.: Asymmetrical stress tensor and its application to a granular medium. In:
Proceedings of 3rd Japanese National Congress of Applied Mechanics, 1953, p. 77 (1954)
138. Oshima, N.: Dynamics of Granular Media, vol. I, p. 563. Memory RAAG, Tokyo (1955)
139. Pal’mov, V.A.: Fundamental equations of the theory of asymmetric elasticity. Prikl. mat.
mekh. 28, 401 (1964)
140. Pal’mov, V.A.: Plane problem of the theory of asymmetric elasticity. Prikl. mat. mekh. 28,
1117 (1964)
141. Papkovich, P.F.: Solution générale des équations differentielles fondamentales d’élasticité
exprimée par trois fonctions harmoniques. C. Rend. hebd. séances Acad. Sci., Paris 195, 513
(1932)
142. Pietras, F., Wyrwiński, J.: Thermal stresses in a plane anisotropic Cosserat continuum.
Arch. Mech. Stos. 19, 627 (1967)
143. Podil’chuk, N., Kirichenko, A.M.: Influence of a spherical hole on the state of stress of an
elastic medium, taking into account couple stresses. Prikl. mekh. Kiev 3, 69 (1967)
144. Poisson, S.D.: Mém. Acad. Paris 18 (1842)
145. Rajagopal, E.S.: The existence of interfacial couples in infinitesimal elasticity. Ann. Physik
6, 192 (1960)
146. Reissner, E.: Note on the theorem of the symmetry of the stress tensor. J. Math. Phys. 23,
192 (1944)
147. Reissner, E., Wan, F.Y.M.: A note on G} unther’s analysis of couple stress. IUTAM Symp.
1967, Mechanics of Generalized Continua, p. 83 (1968)
148. Rivlin, R.S.: Generalized mechanics of continuous media. IUTAM Symp. 1967, Mechanics
Generalized Continua, p. 1 (1968)
149. Rymarz, G.: Surface waves in Cosserat’s medium. Bull. Acad. Pol. Sci. Sér. Sci. Techn. 15,
177 (1967)
150. Saint Venant, A.J.C.B.: Note sur les valeurs que prennent les pressions dans un solide
élastique isotrpe lorsque l’on tient compte des dériveés d’ordre supérieur des déplacements
References 353
trés petits que leurs points ont éprouvés. C. Rend. hebd. séances Acad. Sci., Paris 68, 569
(1869)
151. S
ßandru, N.: Le théorème de réciprocité du type de Betti dans l’élasticité asymétrique.
C. Rend. hebd. séances Acad. Sci. Paris 260, 3565 (1965)
152. S
ßandru, N.: Le théorème de réciprocité dans l’élasticité asymétrique (cas dynamique). Atti
Accad. Naz. Lincei, Ser. VIII, Rend., Cl. Sci. Fis., Mat. e Nat., 38, 78, (1965)
153. S
ßandru, N.: On some problems of the linear theory of the asymmetric elasticity. Int. J. Eng.
Sci. 4, 81 (1966)
154. Savin, G.N., Guz’, A.N.: On a method of solution of the plane problems of the moment
theory of elasticity for multiply connected domains. Prikl. mekh. Kiev 2, 3 (1966)
155. Savin, G.N., Shul’ga, N.A.: Plane dynamic problem of the moment theory in elasticity.
Prikl. mekh. Kiev. 3, 1 (1967)
156. Savov, L.N.: Plane problem of the moment theory of viscoelasticity on the stress
concentration in the proximity of a circular hole. Prikl. mekh. Kiev. 4, 6 (1968)
157. Schade, K.D.: Cosserat Fläche und Schalentheorie. Doktorthesis, Techn. Hochschule,
Darmstadt (1966)
158. Schaefer, H.: Die Spannungsfunktionen des dreidimensionalen Kontinuums und des
elastischen Körpers. ZAMM 33, 356 (1953)
159. Schaefer, H.: Die Spannungsfunktionen des dreidimensionalen Kontinuums; statische
Dentung und Randwerte. Ing. Archiv 28, 291 (1959)
160. Schaefer, H.: Versuch einer Elastizitätstheorie des zweidimensionalen ebenen Cosserat
Kontinuums. Misz. Angew. Mech. Festschrift W. Tollmien, p. 277 (1962)
161. Schaefer, H.: Analysis der Motorfelder in Cosserat Kontinuum. ZAMM 47, 319 (1967)
162. Schaefer, H.: Die Spannungsfunktionen eines Kontinuums mit Momentspannungen. Bull.
Acad. Pol. Sci. Sér. Sci. Techn. I, II. 15, 63, 69 (1967)
163. Schaefer, H.: Das Cosserat kontinuum. ZAMM 47, 485 (1967)
164. Schaefer, H.: The basic affine connection in a Cosserat continuum. IUTAM Symp. 1967,
Mechanics of Generalized Continua, p. 57 (1968)
165. Schijve, J.: Note on couple stresses. J. Mech. Solid 14, 113 (1966)
166. Sokolowski, M.: Couple stresses in problems of torsion of prismatical bars. Bull. Acad. Pol.
Sci. Sér. Sci. Techn. 13, 419 (1965)
167. Soós, E., Teodorescu, P.P.: Les constantes de matérieau dans le cas des modèles du type de
Cosserat des cristaux élastiques. Lett. Appl. Eng. Sci. 1, 209 (1973)
168. Stefaniak, J.: Concentrated loads as body forces. Rev. Roum. Math. Pures Appl. 14, 119
(1969)
169. Sternberg, E.: Couple stresses and singular stress concentrations in elastic solids. IUTAM
Symp. 1967, Mechanics of Generalized Continua, p. 95 (1968)
170. Sternberg, E., Eubanks, R.A.: On stress functions for elastokinetics and the integration of
the repeated wave equation. Appl. Math. 15, 149 (1957)
171. Sternberg, E., Muki, R.: The effect of couple stresses on the stress concentration around a
crack. Int. J. Solid Struct. 3, 69 (1967)
172. Stoianović, R.: Dislocation in the generalized elastic Cosserat continuum. IUTAM Symp.
1967, Mechanics of Generalized Continua, p. 152 (1968)
173. Stoianović, R., Djurić, S.: Le problème plan de la théorie des corps élastiques orientés. 8th
Congress on Young Mechanics, Split (1966)
174. Stoianović, R., Djurić, S., Vujośević, L.: Stress strain relations for the elastic Cosserat
continuum. Conference on Department Continuing Medical, P.A.N., Zakopane (1961)
175. Stoianović, R., Djurić, S., Vujośević, L.: Contribution à la dynamique des continuum de
Cosserat. Mat. Vesnik 1, 127 (1964)
176. Stoianović, R., Vujośević, L.: Couple stress in non euclidean continua. Publ. Inst. Math.
New Ser. 2, 71 (1962)
177. Teodorescu, P.P.: Schwingungen der elastischen Kontinua. III. Konf. Nichtlin. Schwing.,
1964, Abh. deut. Akad. Wiss., Berlin 29 (1965)
354 7 Introduction to the Theory of Cosserat Type Bodies
178. Teodorescu, P.P.: One hundred years of investigations in the plane problem of the theory of
elasticity. Appl. Mech. Surv. p. 245, Spartan Books, Washington, DC (1966)
179. Teodorescu, P.P.: Sur un certain caractère tensoriel des charges concentrées. Rend. Naz.
Lincei, Ser. 8, Cl. Sci. Fis. Mat. Nat. 150, 251 (1966)
180. Teodorescu, P.P.: Sur l’action des charges concentrées dans le problème plan de la
mécanique des solides déformables. Arch. Mech. Stos. 18, 567 (1966)
181. Teodorescu, P.P.: Sur la notion de moment massique dans le cas des corps du type de
Cosserat. Bull. Acad. Pol. Sci., Sér. Sci.Tech. 20, 57 (1967)
182. Teodorescu, P.P.: Problèmes bidimensionnels de la théorie de l’élasticité. I. Une tension
normale nulle. Atti Acad. Naz. Lincei, Ser. III, Rend. Cl. Sci. Fis. Mat. Nat. 154, 141 (1968)
183. Teodorescu, P.P.: Problémes bidimensionels de la théorie de l’élasticité. II. Deux tensions
tangentielles nulles. Atti. Accad. Naz. Lincei, Ser. VIII, Rend. Cl. Sci. Fis. Mat. Nat. 154,
232 (1968)
184. Teodorescu, P.P.: On the action of concentrated loads in the case of a Cosserat continuum.
IUTAM Symp. 1967. Mechanics of Generalized Continua, p. 120 (1968)
185. Teodorescu, P.P.: Sur les corps du type de Cosserat à l’élasticité linéaire. Ist. Naz. Alta Mat.,
Symp. Mat. 1, 375 (1969)
186. Teodorescu, P.P.: Sur l’introduction des fonction potentiels en élasticité linéaire. An. Univ.
Bucuresßti, Mat. Mec. 20, 131 (1971)
187. Teodorescu, P.P.: Sur une représentation du type de Papkovich Neuber dans le cas des corps
du type de Cosserat. An. Univ. Bucuresßti, Mat. Mec. 21, 115 (1972)
188. Teodorescu, P.P.: Sur le problème dynamique des corps du type de Cosserat. Rev. Roum.
Math. Pures Appl. 17, 1097 (1972)
189. Teodorescu, P.P.: On continuity equations and cesàro type formulae in case of Cosserat type
solids. Mech. Res. Comm. 4, 63 (1977)
190. Teodorescu, P.P.: Sur l’état de déplacement et de rotation dans le cas des corps du type de
Cosserat. Bull. Math. Soc. Sci. Math. Roum. 17(65), 309 (1973)
191. Teodorescu, P.P.: Sur la représentation de S ßandru dans le cas des corps du type de Cosserat.
An. Univ. Bucuresßti, Mat. Mec. 22, 155 (1973)
192. Teodorescu, P.P.: Sur le calcul des récipients soumis à des charges concentrées dans le cas
de l’élasticité symétrique ou asymetrique. Cons. naz. ric., Quaderni de ‘‘la ric. sci.’’ 78, 24
(1973)
193. Teodorescu, P.P., S ßandru, N.: Sur l’action des charges concentrée en élasticité asymétrique
plane. Rev. Roum. Math. Pures Appl. 12, 1399 (1967)
194. Teodosiu, C.: On the determination of internal stresses and couple stresses in the continuum
theory of dislocations. Bull. Acad. Pol. Sci., Sér. Sci. Tech. 12, 605 (1964)
195. Teodosiu, C.: The determination of stresses and couple stresses generated by dislocations in
isotropic media. Rev. Roum. Sci. Tech. Sér. Mec. Appl. 10, 1461 (1965)
196. Teodosiu, C.: Non linear theory of the materials of grad 2 with initial stresses and
hyperstresses. I. Basic geometric and static equations. Bull. Acad. Pol. Sci., Sér. Sci. Techn.
15, 95 (1967)
197. Teodosiu, C.: Non linear theory of the materials of grad 2 with initial stresses and
hyperstresses. II. Constitutive equations. Bull. Acad. Pol. Sci. Sér. Sci. Tech. 15, 193 (1967)
198. Teodosiu, C.: Contributions to the continuum theory of dislocations and initial stresses. Rev.
Roum. Sci. Techn., Sér. Méc. Appl. 12(1–3), 961, 1061, 1291 (1967)
199. Teodosiu, C.: Continuous distribution of dislocations in hyperelastic material of grade 2.
IUTAM Symp. 1967, Mechanics of Generalized Continua, p. 279 (1967)
200. Tiffen, R., Stevenson, A.C.: Elastic isotropy with body force couple. Q. J. Mech. Appl.
Math. 9, 306 (1956)
201. Toupin, R.A.: Elastic materials with couple stresses. Arch. Rat. Mech. Anal. 11, 385 (1962)
202. Toupin, R.A.: Theories of elasticity with couple stress. Arch. Rat. Mech. Anal. 17, 85
(1964)
203. Toupin, R.A.: Dislocated and oriented media. IUTAM Symp. 1967, Mechanics of
Generalized Continua, p. 126 (1968)
References 355
204. Truesdell, C.: The mechanical foundations of elasticity and fluid dynamics. Rat. Mech.
Anal. 5, 125 (1952)
205. Vinokurov, L.P., Derevianko, N.I.: Construction of fundamental equations for calculation
without torsion with consideration of couple stresses. Prikl. mekh., Kiev 2, 72 (1966)
206. Voigt, W.: Theoretische Studien über die Elastizitätsverhältnisse der Krystalle. I, II. Abh.
Königl. Ges. Wiss., Göttingen, p. 34 (1887)
207. Weistmann, Y.: Couple stresses effects on stress concentrations around a cylindrical
inclusion in a field of uniaxial tension. J. Appl. Mech. 32, 424 (1965)
208. Weselowski, Z.: On the couple stresses in an elastic continuum. Arch. Mech. Stos. 17, 219
(1965)
209. Woźniak, Cz.: Theory of fibrous media, I, II. Arch. Mech. Stos. 17, 651, 777 (1965)
_
210. Woźniak, Cz., Zukowski, M.: On a model of elastic subsoil carrying surface moments. Bull.
Acad. Pol. Sci., Sér. Sci. Techn. 14 (1966)
211. Wirwiński, J.: Green functions for a thermoelastic Cosserat medium. Bull. Acad. Pol. Sci.
Sér. Sci. Tech. 14, 145 (1966)
212. Yamamoto, Y.: An intrinsic theory of a Cosserat continuum. RAAG, 3rd ser
Chapter 8
Theory of Concentrated Loads
With the aid of concentrated forces we may construct an arbitrary distributed load,
the representation of which will be considered in the following, i.e.: systems of
concentrated forces, directed moments and dipoles of forces; then one may obtain
centres of rotation and centres of dilatation.
so that
Re ðxÞ ¼ Fie ; Me ¼ 0; ð8:30 Þ
where ie is the unit vector of the Ox2 -axis (we have assumed that the concentrated
force F acts at the point O in the direction of the Ox2 -axis and has been obtained as
the limit of loads acting in the Ox1 x2 -plane).
F
(0, 3 )
F
(0 , ε )
Qε (x1)
By the composition of concentrated forces of the type defined above we obtain forces
of the same type; thus, the forces F1 ; F2 ; . . .; Fn ; applied at the point Aðr0 Þ; are
equivalent to the force F1 þ F2 þ þ Fn ; applied at the very same point (Fig. 8.2).
Indeed, the n forces are equivalent to the loads
Qi ðrÞ ¼ Fi dðr r0 Þ; i ¼ 1; 2; . . .; n; ð8:4Þ
Fn
A ( x10, x 02 , x 30 )
r0
O
360 8 Theory of Concentrated Loads
Let us consider, e.g., two parallel, equal, bound forces ðF1 ¼ F2 ¼ F; in the sense
of free vectors), applied at the points A1 ða; 0Þ and A2 ða; 0Þ; a [ 0 (Fig. 8.3a). To
these parallel forces there correspond the equivalent vector fields
Q1 ðx1 ; x2 Þ ¼ F1 dðx1 þ a; x2 Þ ¼ Fdðx1 þ a; x2 Þ;
ð8:5Þ
Q2 ðx1 ; x2 Þ ¼ F2 dðx1 a; x2 Þ ¼ Fdðx1 a; x2 Þ:
The vector field equivalent to the system formed by the two bound forces reads
Qðx1 ; x2 Þ ¼ Q1 ðx1 ; x2 Þ þ Q2 ðx1 ; x2 Þ ¼ F½dðx1 þ a; x2 Þ þ dðx1 a; x2 Þ;
which is a rule of composition of two bound equipolent forces (equal in the sense
of free forces).
The fact of considering an Ox1 x2 -plane and points of application on the Ox1 -
axis is not essential to particularize the problem.
Let us consider now two bound forces F1 and F2 ; having the same modulus, the
same direction, but opposite sense, for which we may write F1 ¼ F and F2 ¼ F
(in the sense of free vectors), applied at the points A1 ða; 0Þ and A2 ða; 0Þ; a [ 0
respectively (Fig. 8.3b). The equivalent vector field is given by
Qðx1 ; x2 Þ ¼ F½dðx1 þ a; x2 Þ dðx1 a; x2 Þ;
wherefrom
(a) (b)
8.1 Case of Linearly Elastic Bodies 361
M ¼ Fd0 : ð8:9Þ
-F
362 8 Theory of Concentrated Loads
The loads Q0 and Q00 equivalent to the concentrated forces applied at the points
A and B, respectively, are
MF0 o
QðrÞ ¼ dðr r0 Þ: ð8:110 Þ
u F0 ou
u F0 6¼ 0 ð8:120 Þ
From the relation (8.110 ) we see that the directed concentrated moment is
characterized by: (i) the location Aðx01 ; x02 ; x03 Þ; (ii) the unit vector F0 of the forces
which generate the moment; (iii) the unit vector u of the direction in which the
passage to the limit is effected (one observes that different results are obtained,
depending on the choice of the location B on the line of action of the force F1);
(iv) the magnitude M of the moment.
The direction of the moment results from the above data and is specified by the
unit vector n of the normal to the plane in which the directed concentrated moment
is acting; the unit vector is chosen such that the scalar triple product ðu; F0 ; nÞ [ 0:
The directed concentrated moment thus defined is positive since the rotation, as
seen from the unit vector n, occurs in the positive direction.
An important particular case is that where the unit vectors u and F0 are normal
to each other ðu F0 ¼ 0Þ; the corresponding equivalent load is
8.1 Case of Linearly Elastic Bodies 363
-F
x1
o
QðrÞ ¼ MF0 dðr r0 Þ ¼ MF0 ½u graddðr r0 Þ: ð8:13Þ
ou
For example, if we take the origin of the co-ordinate axes as point of application
and F0 ¼ i1 ; u ¼ i2 : Where ij ; j ¼ 1; 2; 3; are the unit vectors of the co-ordinate
axes; we obtain a direct concentrated moment, which induces a positive rotation in
the Ox1 x2 -plane (Fig. 8.5) and is expressed by the equivalent load
QðrÞ ¼ Mi1 d;2 ðrÞ: ð8:130 Þ
The directed concentrated moment introduced above is a moment of first order.
Starting from a directed concentrated moment of (n-1)th order, the vectors
ðn1Þ
Qðn1Þ and Q1 of which are applied at the points Aðr0 Þ and BðqÞ; respec-
tively, one may obtain, analogically, a directed concentrated moment of nth order.
The directed concentrated moments of higher order may be useful for certain
generalized deformable media.
n n
Q2 Q1 P P
Qn
M
A ( x10, x02 , x30 )
A ( x10, x02 , x30 )
(a) (b)
Fig. 8.6 Rotational concentrated moment: a component directed moments; b centre of rotation
where
Mi F0i o
Qi ðrÞ ¼ dðr r0 Þ; i ¼ 1; 2; . . .; n: ð8:15Þ
ui F0i oui
Replacing (8.15) in (8.14) and complying with the condition that the element
Q should not depend on the unit vectors ui and F0i ; i ¼ 1; 2; . . .; n one obtains
X
3
o
QðrÞ ¼ aj dðr r0 Þ; ð8:16Þ
j¼1
oxj
whence
X
3
o
ðaj a0j Þ uðr0 Þ ¼ 0; ð8:1700 Þ
j¼1
oxj
8.1 Case of Linearly Elastic Bodies 365
a relation which, owing to its linearity with respect to the derivatives of the
arbitrary fundamental function uðrÞ; can occur only if
aj ¼ a0j ; j ¼ 1; 2; 3: ð8:18Þ
u1 u2 ¼ 0; u1 u2 ¼ n: ð8:1900 Þ
The equivalent load QðrÞ will be given, in this case, by
M 0 o M o
QðrÞ ¼ F1 dðr r0 Þ F02 dðr r0 Þ
2 ou1 2 ou2
M
¼ fu2 ½u1 graddðr r0 Þu1 ½u2 graddðr r0 Þg
2
M
¼ ðu1 u2 Þ graddðr r0 Þ;
2
where we took into account the formula (A.41) of the vector triple product; thus,
the equivalent load of the centre of rotation is expressed in the form
1
QðrÞ ¼ Mn graddðr r0 Þ: ð8:20Þ
2
We notice that the vectors
1
aj ¼ Mij n; j ¼ 1; 2; 3; ð8:21Þ
2
are coplanar and located in the plane P.
Thus we see that the rotational concentrated moment is characterized by: (i) the
location Aðr0 Þ; (ii) the plane P determined by the unit vector n of the normal to it
at the point A; (iii) the magnitude M. This moment thus defined is positive since
the rotation in the plane P, seen from n (Fig. 8.6b), is counter-clockwise. In this
way, the direction of the moment results from the data specified above and is
indicated by the unit vector n.
In particular, choosing the Ox1 x2 -plane as plane Pðn ¼ i3 Þ; one obtains the
equivalent load
366 8 Theory of Concentrated Loads
1
Qðx1 ; x2 Þ ¼ Mi3 graddðx1 x01 ; x2 x02 Þ; ð8:22Þ
2
of components [25]
1 o
Q1 ðx1 ; x2 Þ ¼ Mi1 dðx1 x01 ; x2 x02 Þ;
2 ox2
ð8:220 Þ
1 o
Q2 ðx1 ; x2 Þ ¼ Mi2 dðx1 x01 ; x2 x02 Þ:
2 ox1
Taking into account the above considerations, we may use as canonical rep-
resentation of a rotational concentrated moment the representation obtained by the
superposition of two directed concentrated moments of the same direction, same
magnitude, and the component forces of which are normal to each other.
Thus, the system of concentrated forces acting on the sides of a square
(Fig. 8.7a) leads to the superposition of two orthogonal directed concentrated
moments of the same magnitude M/2 and sign (Fig. 8.7b), hence to a canonical
representation of the rotational concentrated moment (Fig. 8.7c). We remark that
the result is the same for any position of a square of which centre is O. The fact
that we have considered the Ox1 x2 -plane particularizes in no way the problem
from a physical point of view. It should be mentioned too that this points out a
method of constructing (modelling) experimentally a rotational concentrated
moment.
It can be shown that starting from loads which are tangential to and uniformly
distributed over the sides of a regular polygon or from loads tangential to and
uniformly distributed over a circle, we obtain again a rotational concentrated
moment.
x2 x2 x2
M
M1 = 2 M
O x1 x1 O x1
O
M
M2= 2
Fig. 8.7 Rotational concentrated moment: a on the sides of a square; b two orthogonal directed
concentrated moments; c centre of rotation
8.1 Case of Linearly Elastic Bodies 367
In Sect. 8.1.1.4 we have defined a concentrated moment assuming that the unit
vectors F0 and u are not collinear ðu F0 6¼ 0Þ: For the case where they are such,
we introduce a new type of concentrated moment: the concentrated moment of
linear dipole type [12].
Let –F, of components Fi ; i ¼ 1; 2; 3; be a concentrated force applied at the
fixed point A of position vector rðx01 ; x02 ; x03 Þ and let F1 ¼ F (the equality is con-
sidered in the sense of free vectors) be another concentrated force applied at the
variable point B of position vector qðn1 ; n2 ; n3 Þ: Let u be the unit vector of
the force F (Fig. 8.8). We assume that the two concentrated forces have the same
!
line of action; hence, the unit vector of the vector AB will be again u. We set the
arm AB ¼ d and introduce the magnitude
D ¼ Fd: ð8:23Þ
By definition, the concentrated moment of dipole type (dipole of concentrated
forces) at the point A is the limit, in the sense of the theory of distributions, of the
set of concentrated forces ðF; F1 Þ; when the arm d of the couple tends to zero.
The point A is considered fixed and the point B variable; we assume that the two
concentrated forces have the same line of action and that the unit vector u and the
magnitude D of the dipole moment are constant.
The common line of action of the two concentrated forces is the line of action
of the dipole, which is considered to be positive if it tends to further separate the
points of application of the forces (Fig. 8.8) and negative in the opposite case.
The equivalent load of the set of two concentrated forces may be
Du
Qd ðrÞ ¼ Fu½dðr qÞ dðr r0 Þ ¼ ½dðr r0 Þ dðr qÞ; ð8:24Þ
d
passing to the limit for d ! þ0 in the sense of the theory of distributions, we
obtain the equivalent load of a concentrated moment of dipole type
o
QðrÞ ¼ Du dðr r0 Þ ¼ D½u graddðr r0 Þu; ð8:25Þ
ou
where we have introduced the directional derivative in the direction defined by the
unit vector u.
It results that the concentrated moment of dipole type is characterized by:
(i) the location Aðr0 Þ; (ii) the unit vector u of the forces which generate the
moment; (iii) the magnitude D of the dipole moment.
The direction of the dipole is specified by the unit vector F0 of the force F; if
F0 ¼ u; which is the case considered above, then the dipole is positive, while if
F0 ¼ u; then the dipole is negative and the sign of the relation (8.25) must be
changed.
An important particular case is that where the point of application is the origin
of the co-ordinate axes and u ¼ i1 ; where i1 is the unit vector of the Ox1 -axis; we
thus obtain a dipole of concentrated forces, which is expressed by the equivalent
load
Starting from a concentrated moment of linear dipole type, one can construct
concentrated moments of plane or spatial dipole type. Let thus be a plane
P passing through the point Aðr0 Þ and n be the unit vector of the normal to the
plane. We denote by Q the set of moments of linear dipole type Qi ðrÞ; applied at
the point A; with the unit vectors ui ; i ¼ 1; 2; . . .; n belonging to the plane
P (Fig. 8.9a).
By definition, the sum of two or more elements of the set Q is called a con-
centrated moment of plane dipole type (centre of plane dilatation), corresponding
to the point A and to the plane P, in the case where the sum does not depend on the
unit vectors ui :
Such a sum, corresponding to a concentrated moment of plane dipole type is
(Fig. 8.9b)
X
n
QðrÞ ¼ Qi ðrÞ; n 2; ð8:27Þ
i¼1
n n
Q2 P P
Q1
Qn A ( x10, x02 , x30 )
Dp
A ( x10, x02 , x30 )
(a) (b)
Fig. 8.9 Concentrated moment of plane dipole type: a component linear dipoles; b cen-
tre of plane dilatation
8.1 Case of Linearly Elastic Bodies 369
where
o
Qi ðrÞ ¼ Di ui dðr r0 Þ: ð8:270 Þ
oui
A study analogous to that in Sect. 8.1.1.5 concerning centres of rotation shows
that the representation is unique, the set being non-void; the corresponding vectors
aj are given by
1
aj ¼ Dp n ðn ij Þ; j ¼ 1; 2; 3: ð8:28Þ
2
Let thus be two positive dipoles of concentrated forces normal to each other,
applied at the point Aðr0 Þ and contained in the plane P. We may take
1
D1 ¼ D2 ¼ Dp ; ð8:29Þ
2
the unit vectors u1 and u2 will satisfy the relations
u1 u2 ¼ 0; u1 u2 ¼ n: ð8:290 Þ
The equivalent load QðrÞ is, in this case, given by
Dp
QðrÞ¼ fu1 ½u1 graddðr r0 Þ þ u2 ½u2 graddðr r0 Þg
2
Dp
¼ n fu1 ½u2 graddðr r0 Þ u2 ½u1 graddðr r0 Þg
2
Dp
¼ n ½ðu1 u2 Þ graddðr r0 Þ;
2
where we took into account the second relation (8.280 ) and the formula (A.1.41) of
the triple vector product; thus, the equivalent load of the plane centre of dilatation
is given by
1
QðrÞ ¼ Dp n ½n dðr r0 Þ: ð8:30Þ
2
The concentrated moment of plane dipole type is considered to be positive if it
is obtained from positive linear dipoles; in this case, the concentrated moment of
plane dipole type is a centre of plane dilatation. Conversely, if we use negative
linear dipoles, then we obtain a negative concentrated moment of plane dipole
type, i.e., a centre of plane contraction and the relation (8.30) takes a changed sign
(supposing that Dp [ 0Þ:
The concentrated moment of plane dipole type is characterized by: (1) the
location Aðr0 Þ; (2) the unit vector n of the normal to the plane whereon the plane
dipole is acting; (3) the magnitude Dp of the dipole.
370 8 Theory of Concentrated Loads
Taking into account the formula (8.20) which expresses the rotational con-
centrated moment and assuming that M ¼ Dp ; we may write
x2 x2 x2
Dp
D1 = 2
O x1 O x1 x1
O Dp
Dp
D2 = 2
Fig. 8.10 Concentrated moment of plane dipole type: a on the sides of a square; b two orthog-
onal linear dipoles; c centre of plane dilatation
8.1 Case of Linearly Elastic Bodies 371
Ds
3
Ds
3
O x2
Ds
x1 3
1
D1 ¼ D2 ¼ D3 ¼ Ds : ð8:34Þ
3
Let now be three linear dipoles of concentrated forces of the same sign, their
lines of action being specified by the unit vectors u1 ; u2 ; u3 ; orthogonal to each
other, which form a dextrorsum trihedron (Fig. 8.11).
The equivalent load is given by
Ds o o o
QðrÞ ¼ u1 dðr r0 Þ þ u2 dðr r0 Þ þ u3 dðr r0 Þ
3 ou1 ou2 ou3
Ds
¼ fu1 ½u1 graddðr r0 Þ þ u2 ½u2 graddðr r0 Þ
3
þu3 ½u3 graddðr r0 Þg;
wherefrom
1
QðrÞ ¼ Ds graddðr r0 Þ: ð8:35Þ
3
A concentrated moment of spatial dipole type is considered to be positive if it
has been obtained from positive linear dipoles; in this case, it is a centre of spatial
dilatation. If the component linear dipoles are negative, then we obtain a negative
concentrated moment of spatial dipole type; it represents a centre of spatial
contraction and the relation (8.35) takes a changed sign (supposing that Ds [ 0Þ:
The concentrated moment of spatial dipole type is characterized by: (i) the
location Aðr0 Þ; (ii) the magnitude Ds of the dipole.
A canonical representation of a concentrated moment of spatial dipole type
may be obtained by three concentrated moments of linear dipole type, orthogonal
to each other, of the same magnitude Ds =3 and sign.
We observe that one obtains the same result if one considers uniform distrib-
uted normal loads acting over the surface of a sphere.
It is interesting to remark that a spatial centre of dilatation introduces a
singularity of a different type than that of the plane centre of dilatation; indeed,
372 8 Theory of Concentrated Loads
the latter concentrated load may be obtained with the aid of d representative
sequences, while the first one necessities d0 representative sequences.
We mention that one can construct concentrated moments of sectorial dipole
type (of a circle or of a sphere).
We have seen in the previous section how one may obtain directed concentrated
moments and concentrated moments of linear dipole type, starting from the notion
of concentrated force. By the superposition of two directed concentrated moments,
which are orthogonal and of the same magnitude and sign (and for which the arm
of the couple is normal to the direction of the component concentrated forces), we
obtain a rotational concentrated moment (centre of rotation), which has lost its
directional effect and leads to a mechanical phenomenon with axial antisymmetry;
in fact, this is the essential property of the concentrated moment.
In the same way, we can introduce a quadripole of concentrated forces,
obtained by superposing the effect of two orthogonal concentrated moments of
linear dipole type, of the same magnitude and sign; thus, we obtain a concentrated
moment of plane dipole type (centre of plane dilatation) which, likewise, has no
directional effect any longer and leads to a mechanical phenomenon with axial
symmetry.
By the superposition of effects of three orthogonal concentrated moments of
linear dipole type of the same magnitude and sign, we obtain a concentrated
moment of spatial dipole type (centre of spatial dilatation), which leads to a
mechanical phenomenon with central symmetry.
Results of this kind have been pointed out as early as the beginning of the
twentieth century, but without any further and deeper investigation of the matter.
In the sequel we shall justify the use of these concentrated loads and emphasize
their tensorial aspect; this will aid in classifying the concentrated loads. Using also
previous considerations of E. Kröner [3], we shall use the results given by us [28,
29].
In order to emphasize the different tensorial properties of concentrated loads,
we shall consider a deformable solid on which these loads are acting, thus
inducting in the body a state of strain and stress. We shall refer to a body on which
we may apply the principle of local superposition of effect, e.g., to a linearly elastic
body.
In the case of an elastic space (or an elastic plane), the state of strain and the
state of stress do not show any particular features depending on the direction of the
line of action of an internal concentrated load (concentrated force, directed con-
centrated moment, concentrated moment of linear dipole type etc.). But in the case
of a domain whose boundary is, in some part at least, at a finite distance we have to
deal with favourite directions which depend on the boundary.
8.1 Case of Linearly Elastic Bodies 373
1
f ðds Þ ¼ ½f ðd11 Þ þ f ðd22 Þ þ f ðd33 Þ; ð8:39Þ
3
hence, the centres of spatial dilatation are concentrated loads of the zero order.
We give, in the following, solutions for the concentrated loads considered above
which act in an elastic space (infinite domain); we deal thus with the static case, as
well as with the dynamic one.
8.1 Case of Linearly Elastic Bodies 377
F 1 x21
u1 ¼ 3 4m þ 2 ;
16pð1 mÞG r r
F x1 x2
u2 ¼ ; ð8:49Þ
16pð1 mÞG r 3
F x1 x3
u3 ¼ :
16pð1 mÞG r 3
Corresponding to a formulation in stresses (eventually starting from the Kelvin-
Somigliana fundamental solution tensor), one obtains the fundamental solution
tensor (6.1380 ); for the concentrated force F acting at the origin one gets
1 1h xi xj xk i
rij ¼ 3
ð1 2mÞðxk Fk dij xi Fj xj Fi Þ 3Fk 2 ;
8pð1 mÞ r r
i; j ¼ 1; 2; 3; ð8:50Þ
F x1 x21
r11 ¼ 1 2m þ 3 2 ;
8pð1 mÞ r 3 r
F x1 x22
r22 ¼ 1 2m 3 2 ; ð8:51Þ
8pð1 mÞ r 3 r
F x1 x23
r33 ¼ 1 2m 3 2 ;
8pð1 mÞ r 3 r
378 8 Theory of Concentrated Loads
3F x1 x2 x3
r23 ¼ ;
8pð1 mÞ r 5
F x3 x21
r31 ¼ 1 2m þ 3 2 ; ð8:510 Þ
8pð1 mÞ r 3 r
F x2 x21
r12 ¼ 1 2m þ 3 2 ;
8pð1 mÞ r 3 r
o M x2 x21
u1 ðrÞ ¼ M u11 ðrÞ ¼ 3 4m þ 2 ;
ox2 16pð1 mÞG r 3 r
2
o M x1 x
u2 ðrÞ ¼ M u21 ðrÞ ¼ 1 3 22 ; ð8:55Þ
ox2 16pð1 mÞG r 3 r
o 3M x1 x2 x3
u3 ðrÞ ¼ M u31 ðrÞ ¼ :
ox2 16pð1 mÞG r 5
We remark that the derivative occuring in the above expressions may be
considered in the ordinary sense since the formal derivative of the components of
the tensor uij are locally integrable functions.
8.1 Case of Linearly Elastic Bodies 379
Let us consider now the elastic space acted upon at the point Aðr0 Þ by a
rotational concentrated moment (centre of rotation) of magnitude M, specified by
the unit vector n, normal to the plane of action. We introduce the equivalent load
1
Qi ðrÞ ¼ Mijk nj d;k ðr r0 Þ; i ¼ 1; 2; 3: ð8:56Þ
2
It results the state of displacement
1
ui ðrÞ ¼ Mjkl nk uij;l ðr r0 Þ; i ¼ 1; 2; 3: ð8:57Þ
2
In particular, if n ¼ i3 and r0 ¼ 0; one obtains a centre of rotation applied at the
origin O, which corresponds to a positive rotation in the Ox1 x2 -plane; there cor-
responds the equivalent load
1 1
Q1 ðrÞ ¼ Md;2 ðrÞ; Q2 ðrÞ ¼ Md;1 ðrÞ; Q2 ðrÞ ¼ 0 ð8:560 Þ
2 2
and the state of displacement reads
1 o o M x2
u1 ðrÞ ¼ M u11 ðrÞ u12 ðrÞ ¼ ;
2 ox2 ox1 8pG r 3
1 o o M x1
u2 ðrÞ ¼ M u21 ðrÞ u22 ðrÞ ¼ ; ð8:58Þ
2 ox2 ox1 8pG r 3
1 o o
u3 ðrÞ ¼ M u31 ðrÞ u32 ðrÞ ¼ 0:
2 ox2 ox1
In the case of a concentrated moment of linear dipole type acting at the origin
O along the Ox1 -axis, the magnitude of which is D, one obtains
D x1 x21
u1 ðrÞ ¼ 1 4m þ 3 2 ;
16pð1 mÞG r 3 r
D x2 x2
u2 ðrÞ ¼ 3
1 3 21 ; ð8:59Þ
16pð1 mÞG r r
D x3 x21
u3 ðrÞ ¼ 13 2 :
16pð1 mÞG r 3 r
Calculating, analogically, the state of displacement corresponding to a con-
centrated moment of linear dipole type acting at the origin O along the Ox2 -axis, of
magnitude D and of the same sign, and superposing the actions of both linear
dipoles, one obtains a concentrated moment of plane dipole type (centre of plane
dilatation) of magnitude D=2 (taking into account the magnitudes of the two linear
dipoles). Thus, the centre of dilatation which acts in the Ox1 x2 -plane leads to the
state of displacement
380 8 Theory of Concentrated Loads
D x1 x23
u1 ðrÞ ¼ 3 4m þ 3 ;
32pð1 mÞG r 3 r2
D x2 x23
u2 ðrÞ ¼ 1 4m 3 2 ; ð8:60Þ
32pð1 mÞG r 3 r
D x3 x2
u3 ðrÞ ¼ 3
1 3 23 :
32pð1 mÞG r r
Superposing, analogically, a directed moment of linear dipole type, acting at
O along the Ox3 -axis, of the same magnitude D and of the same sign, one obtains a
concentrated moment of spatial dipole type (centre of spatial dilatation) of mag-
nitude D=3 (with respect to the magnitudes of the three linear dipoles). The cor-
responding state of displacement reads
ð1 2mÞD x1 ð1 2mÞD x2
u1 ðrÞ ¼ ; u2 ðrÞ ¼ ;
24pð1 mÞG r 3 24pð1 mÞG r 3
ð8:61Þ
ð1 2mÞD x3
u3 ðrÞ ¼ :
24pð1 mÞG r 3
ui ðr; tÞ ¼ uij
Fj ; i ¼ 1; 2; 3; ð8:63Þ
where the product of convolution refers both to the space and time variables. We
also assume that the initial conditions are homogeneous (null).
In the case of a concentrated force
1 1 r
ui ðr; tÞ ¼ Fi t
4pG r c2
2 1 r r
þ c2 Fj ðtÞ
t t ; i ¼ 1; 2; 3: ð8:640 Þ
r c1 þ c2 þ ;ij
If we take
Fj0 1 r
ui ðr; tÞ ¼ d t dij
4pG r c2
1 r r
þ c22 t t ; i ¼ 1; 2; 3: ð8:650 Þ
r c1 þ c2 þ ;ij
at the origin O at the initial moment, and then maintain it constant in time, the
state of generalized displacement reads
(
Fj0 2 r
ui ðr; tÞ ¼ h t dij
8pG r c2
( "
#) )
2 1 r 2 r 2
þ c2 t t ; i ¼ 1; 2; 3: ð8:660 Þ
r c1 þ c2 þ
;ij
and
8 r
>
> 0; t ;
>
> c
>
2 1
2
2 >>
< r r r
r r t ; t ;
t t ¼ c c c ð8:670 Þ
c1 þ c2 þ >> 1 1 2
>
>
>
> c 1 c 2 c 1 þ c 2 r
>
: 2t r r; t :
c1 c2 c1 c2 c2
We shall now consider the elastic space acted on by a rotational concentrated
moment of magnitude MðtÞ; applied at the origin O in a plane of normal n; taking
into account the equivalent load (8.20), the generalized displacement vector reads
1 1 r
uðr; tÞ ¼ n grad M t ; ð8:68Þ
8pG r c2
where
If
while if
Let us now consider the elastic space acted on at the origin O by a centre of
spatial dilatation (concentrated moment of spatial dipole type) of magnitude Ds ðtÞ;
taking into account the equivalent load (8.35), the generalized displacement vector
reads
1 c21 1 r
uðr; tÞ ¼ grad Ds t
12pG c22 r c1
ð1 2mÞ 1 r
¼ grad Ds t ; ð8:73Þ
24pð1 mÞG r c1
then we have to deal with an initial shock and there follows the generalized
displacement vector
ð1 2mÞD0s 1 r
uðr; tÞ ¼ grad d t ; ð8:750 Þ
24pð1 mÞG r c1
while if
then the centre of spatial dilatation corresponds to a load applied suddenly and
maintained constant in time and the generalized displacement vector reads
ð1 2mÞD0s 1 r
uðr; tÞ ¼ grad h t : ð8:760 Þ
24pð1 mÞG r c1
the displacement and rotation vectors. Other concentrated loads have been con-
sidered by us [30–32] and then, independently, by S. Kessel [18].
We deal firstly with the case of a concentrated force; then we shall consider other
concentrated loads of second and zero order.
where
1 1
P0 ðrÞ ¼ F grad ; ð8:78Þ
4p r
1 1
PðrÞ ¼ F grad ; ð8:780 Þ
4p r
in case of a concentrated force FdðrÞ; the Eqs. (7.6100 ) and (7.6200 ) lead thus to the
equations
1 1
ðk þ 2lÞDK0 ¼ F grad ; ð8:79Þ
4p r
1 1
D½ðl þ aÞðl0 þ a0 ÞD 4laK ¼ F grad ; ð8:790 Þ
4p r
respectively. One obtains thus
1 rF
K0 ¼ ; ð8:80Þ
8pðk þ 2lÞ r
l 1 1
K¼ curl v þ ð1 ev Þ F ; ð8:800 Þ
16pla 2 v
Taking into account the representations (7.610 ), (7.620 ), the state of displace-
ment and rotation is given by
1 1 rF k 0 þ l0 1 v
u¼ ð3 4mÞF þ 2 r curl curl ð1 e ÞF ;
16pð1 mÞl r r 16pl r
ð8:82Þ
1 1
U¼ curl ð1 ev ÞF : ð8:820 Þ
8pl r
In the case of a body of Cosserat type with constraint rotations, the state of
displacement and rotation is given by the same formulae (8.82), (8.820 ), but the
non-dimensional constant v is replaced by
r
v0 ¼ 0 ; ð8:810 Þ
l
where l0 is given by (7.510 ).
For a force F ¼ F1 i1 acting at the origin O along the Ox1 -axis it results
F1 1h x1 i
u¼ ð3 4mÞi1 þ 2 r
16pð1 mÞl r r
ð8:83Þ
l02 F1 1 n
2 v
x1
o
2 v
þ 1 1 þ v þ v e i1 2 3 3 þ 3v þ v e r ;
4pl r 2 r
F1 1
U¼ ½1 ð1 þ vÞev ðx3 i2 x2 i3 Þ: ð8:830 Þ
8pl r 3
Taking into consideration also the force F1 i1 which acts at the point ð0; d; 0Þ
and passing to the limit in the sense of the theory of distributions, one obtains thus
the directed concentrated moment M12 ; the corresponding state of displacement
and rotation reads
M12 1h x1 x2 i
u¼ ð 3 4m Þx 2 i1 x 1 i 2 þ 3 r
16pð1 mÞl r 3 r2
l02 M12 1
5
3 3 þ 3v þ 2v2 þ v3 ev x2 i1
4pl r
þ 3 3 þ 3v þ v2 ev x1 i2
x1 x2
2
3 v
2 15 15 þ 15v þ 6v þ v e r ; ð8:84Þ
r
386 8 Theory of Concentrated Loads
M12 1
U¼ ½1 ð1 þ vÞev i3
8pl r 3
x2
2 v
þ 2 3 3 þ 3v þ v e ðx3 i2 x2 i3 Þ : ð8:840 Þ
r
In the case of constraint rotations, the above formulae remain valid if one
replaces v by v0 :
If we suppose that the force F1 i1 acts at the point ðd; 0; 0Þ and if we pass to
the limit in the sense of the theory of distributions, we obtain the directed con-
centrated moment of linear dipole type D11 ; in this case, the state of displacement
and rotation is given by
D11 1 x21
u¼ 2 ð 1 2m Þx i
1 1 1 3 r
16pð1 mÞl r 3 r2
l02 D11 1
þ 5
6 6 þ 6v þ 3v2 þ v3 ev x1 i1
4pl r
x2
þ 3 3 þ 3v þ v2 ev 21 15 15 þ 15v þ 6v2 þ v3 ev r ;
r
ð8:90Þ
D11 x1
U¼ 5
3 3 þ 3v þ v2 ev ðx3 i2 x2 i3 Þ: ð8:900 Þ
8pl r
In the following, we deal with concentrated loads obtained starting from the loads
considered in Sect. 8.2.1.2, i.e. with concentrated moments of plane and spatial
dipole type as well as with rotational concentrated moments. As well, we introduce
the influence of a concentrated volume moment.
Calculating, analogically to the dipole D11 ; the state of displacement and rotation
corresponding to the dipole D22 and superposing the effects, one obtains the state
of displacement and rotation corresponding to a concentrated moment of plane
dipole type of magnitude D3 =2 ¼ D11 ¼ D22 and direction i3 (centre of plane
dilatation)
8.2 Case of Linearly Elastic Cosserat Type Bodies 387
D3 1 x23
u¼ 2 ð 1 2m Þx i
3 3 3 4m 3 r
32pð1 mÞl r 3 r2
l02 D3 1
5
6 6 þ 6v þ 3v2 þ v3 ev x3 i3
8pl r
x2
þ 3 3 þ 3v þ v2 ev 23 15 15 þ 15v þ 6v2 þ v3 ev r ;
r
ð8:91Þ
D 3 x3
U¼ 5
3 3 þ v þ v2 ev ðx2 i1 x1 i2 Þ: ð8:910 Þ
16pl r
In the case of constrained rotations one replaces the constant v by the constant
v0 :
Superposing also the action of the linear dipole D33 and assuming that D=3 ¼
D11 ¼ D22 ¼ D33 ; one obtains a state of displacement and rotation with central
symmetry, corresponding to a centre of spatial dilatation, i.e.
ð1 2mÞD 1
u¼ r; ð8:92Þ
24pð1 mÞl r 3
U ¼ 0: ð8:920 Þ
One observes that the results thus obtained are the same as those in the classical
elasticity, the effect of mechanical asymmetry being lost.
We obtained, in Sect. 8.2.1.2, the state of displacement and rotation for a directed
concentrated moment M12 ; analogically, in the case of a directed concentrated
moment M21 ; it results
M21 1h x1 x2 i
u¼ x i
2 1 ð 3 4m Þx i
1 2 3 r
16pð1 mÞl r 3 r2
l02 M21 1 n
þ 5
3 3 þ 3v þ v2 ev x2 i1 þ 3 3 þ 3v þ 2v2 þ v3 ev x1 i2
4pl r
x1 x2
o
2 15 15 þ 15v þ 6v2 þ v3 ev r ; ð8:93Þ
r
M21 1 n
U¼ ½1 ð1 þ vÞev i3
8pl r 3
x1
o
þ 2 3 3 þ 3v þ v2 ev ðx3 i1 x1 i3 Þ : ð8:930 Þ
r
388 8 Theory of Concentrated Loads
Superposing the effects of the directed concentrated moments M12 and M21 ; one
obtains a rotational concentrated moment (centre of rotation) of magnitude
M3 =2 ¼ M12 ¼ M21 and direction i3 ; i.e.
M3 1 a v
u¼ 1 ð1 þ vÞe r3 ih ; ð8:94Þ
8pl r 3 lþa
M3 1
U¼ 3
1 1 þ v þ v2 ev i3
16pl r
x3
2 3 3 þ 3v þ v2 ev r ; ð8:940 Þ
r
where
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r3 ¼ x21 þ x22 ; ð8:95Þ
the unit vector ih (polar co-ordinates in the Ox1 x2 -plane) being given by
r3 ih ¼ x2 i1 x1 i2 : ð8:950 Þ
If we consider a concentrated volume moment M, acting at the origin O and put
MdðrÞ ¼ gradP0 þ curlP; ð8:96Þ
then, taking into account (7.630 ) and (7.640 ), we get
u ¼ 2acurl curlK ð8:97Þ
Taking into account the representation (8.97), (8.970 ), the state of displacement
and rotation is given by
1 1 ,
u¼ curl ð1 e ÞM ; ð8:101Þ
8pl r
1 1
U¼ grad M grad ð1 e, Þ
16pa r
lþa 1
þ curlcurl ð1 e, Þ M : ð8:1010 Þ
16pal r
In the case of the Cosserat type body with constraint rotations, the displacement
conserves the form (8.101), but the rotation becomes
1 1
U¼ curl curl M ; ð8:102Þ
16p r
because a ! 1; h ! 0; so that , ! 1:
The concentrated volume moment M ¼ Mi3 leads to
M 1
u¼ ½1 ð1 þ vÞev r3 ih ; ð8:103Þ
8pl r 3
M 1n , x3
o
2 ,
U¼ ½ 1 ð 1 þ , Þe i 3 3 3 þ 3, þ , e r
16pa r 3 r2
ðl þ aÞM 1 n
3
1 1 þ v þ v2 ev i3
16pla r
x3
o
2 3 3 þ 3v þ v2 ev r : ð8:1030 Þ
r
Comparing the axial antisymmetric states of displacement and rotation (8.94),
(8.940 ) and (8.103), (8.1030 ), one observes that the centre of rotation M3 and the
concentrated volume moment M lead to different results for M3 ¼ M: It is thus to
be noticed that in the case of Cosserat type bodies with free rotations the con-
centrated volume moment M represents a concentrated load which cannot be
constructed starting from the notion of concentrated force as in the case of the
centre of rotation M3 : Thus, from a quantitative point of view, one can state that, in
a Cosserat continuum, the concentrated volume moment M is a fundamental load
as well as the concentrated volume force F [32–34].
Starting from the concentrated volume moment, one can thus construct many
other concentrated loads. Evidently, one can construct also loads of a mixed
nature, using simultaneously concentrated volume forces and moments.
For M3 ¼ M in the formulae (8.94), (8.940 ), (8.103), (8.1030 ) one observes that
the differences uðMÞ uðM3 Þ and UðMÞ UðM3 Þ tend to zero only for a !
1ðl ¼ l0 ; h ¼ 0; , ! 1Þ: Thus, only in the case of constrained rotations, the
centre of rotation and the concentrated volume moment give the same results, i.e.
390 8 Theory of Concentrated Loads
M 1h 0 v0
i
u¼ 1 ð1 þ v Þe r3 ih ; ð8:104Þ
8pl r 3
(
M 1 h 0 02 v0
i
U¼ 1 ð1 þ v þ v Þe i3
16pl r 3
)
x3 h 0 02 v0
i
2 3 ð3 þ 3v þ v Þe r ; ð8:1040 Þ
r
References
A. Books
1. Kecs, W., Teodorescu, P.P.: Applications of the Theory of Distributions in Mechanics. Ed.
Acad., Bucuresßti; Abacus Press, Tunbridge Wells, Kent (1974)
2. Kecs, W., Teodorescu, P.P.: Vvedenie v teoriyu obobshchennykh funktsiı̆ s prilozheniyami v
tekhnike (Introduction to the Theory of Distributions with Applications in Technics). Izd.
‘‘Mir’’, Moskva (1978)
3. Kr}oner, E.: Kontinuumstheorie der Versetzungen und Eigenspannungen. Springer, Berlin
(1958)
4. Love, A.E.H.: A Treatise on the Mathematical Theory of Elasticity, IVth edn. Cambridge
University Press, London (1934)
5. Nowacki, W.: Teoria niesymetrycznej spre_zystości (Theory of Asymmetrical Elasticity).
Państ. Wydawn. Nauk., Warszawa (1971)
6. Teodorescu, P.P.: Probleme spatßiale în teoria elasticitătßii (Spatial Problems in the Theory of
Elasticity). Ed. Acad., Bucuresßti (1970)
7. Teodorescu, P.P.: Dynamics of Linear Elastic Bodies. Ed. Acad. Bucuresßti; Abacus Press,
Tunbridge Wells, Kent (1975)
8. Teodorescu, P.P. (ed.): Probleme actuale în mecanica solidelor (Present-day Problems in
Solid Mechanics). Ed. Acad., Bucuresßti (1975)
9. Teodorescu, P.P., Ille, V.: Teoria elasticitătßii ßsi introducere în mecanica solidelor deformabile
(Theory of Elasticity and Introduction in Mechanics of Deformable Solids). I. Dacia, Cluj
Napoca (1976)
B. Papers
10. Khan, S.M., Dhaliwal, R.S., Chowdhury, K.L.: Singular solutions and green’s method in
micropolar theory of elasticity. Appl. Sci. Res. 25, 65 (1971)
11. Kecs, W.: Reprezentarea câmpurilor vectoriale cu ajutorul functßiilor generalizate
(Representation of the vector fields with the aid of generalized functions). St. Cerc. Mat.
15, 265 (1964)
12. Kecs, W.: Représentation mathématique des moments du type dipôle. Rev. Roum. Math.
Pures Appl. 10, 1391 (1965)
13. Kecs, W.: Les solutions généralisées des problèmes concernant l’espace élastique infini. Rev.
Roum. Math. Pures Appl. 11, 27 (1966)
References 391
14. Kecs, W.: Représentantion mathématique de certaines limites de groupes de charges et leurs
applications dans la mécanique des corps déformables. I. Bul. Inst. Polit. Iasßi, ser. nouă,
14(18), 443 (1968)
15. Kecs, W.: Les solutions généralisées de l’espace élastique sous l’action de certaines charges
variables. Rev. Roum. Math. Pures Appl. 15, 7 (1970)
16. Kecs, W., Teodorescu, P.P.: On the plane problem of an elastic body acted upon by dynamic
loads. I. General Results. II. Applications. Bull. Acad. Pol. Sci., sér. Sci. Techn. XIX, 47; 57
(1971)
17. Kecs, W., Teodorescu, P.P.: On the action of dynamic concentrated loads in case of linear
elastic bodies. Lucr. II Conf. Vibr. 49 (1978)
18. Kessel, S.: Stress functions and loading singularities for the infinitely extended linear elastic
isotropic cosserat continuum. IUTAM-Symposium 1967, Mechanics of Generalized
Continua, p. 114 (1968)
19. Mindlin, R.D.: On the equations of elastic materials with microstructure. Int. J. Solids Struct.
1, 73 (1965)
20. Mindlin, R.D., Tiersten, H.F.: Effects of couple stresses in linear elasticity. Arch. Rat. Mech.
Anal. 11, 415 (1962)
21. Stefaniak, J.: Concentrated loads as body forces. Rev. Roum. Math. Pures Appl. XIV, 119
(1969)
22. S
ßandru, N.: Asupra actßiunii unei fortße concentrate în spatßiul elastic nemărginit (On the action
of a concentrated load in the infinite elastic space). Com. Acad. Rom. XIII, 1019 (1963)
23. S
ßandru, N.: O deistviı̆ peremenykh sil v neogranichennom prostransve (On the action of
variable forces in the infinite space). Bull. Acad. Pol. Sci., sér. Sci. Technol. 12, 45 (1964)
24. S
ßandru, N.: On some problems of the linear theory of the asymmetric elasticity. Int. J. Engng.
Sci. 4, 81 (1966)
25. Teodorescu, P.P.: Sur l’action des charges concentrées dans le problème plan de la théorie de
l’élasticité. Bull. Math. Soc. Sci. Math. Roum. 8(56), 243 (1964)
26. Teodorescu, P.P.: Consideratßii asupra modului de a defini sarcinile concentrate în problema
plană a teoriei elasticitătßii (considerations on the definition of concentrated loads in the plane
problem of the theory of elasticity). An Univ. Timisßoara, ser. ßst. mat. fiz. III, 287 (1965)
27. Teodorescu, P.P.: Asupra actßiunii sarcinilor concentrate în teoria elasticitătßii (on the action of
concentrated loads in the theory of elasticity). An. Univ. Bucuresßti, ser. ßst. nat., mat. mec.
XV, 15 (1966)
28. Teodorescu, P.P.: Sur un certain caractère tensoriel des charges concentrées. Atti Accad. Naz.
Lincei, ser. VIII, Rend. Cl. Sci. fis., mat. e nat. XL, 251 (1966)
29. Teodorescu, P.P.: Sur l’action des charges concentrées dans le problème plan de la théorie de
l’élasticité. Arch. Mech. Stos. 18, 567 (1966)
30. Teodorescu, P.P.: Sur la notion de moment massique dans le cas des corps du type de
Cosserat. Bull. Acad. Pol. Sci., Sér. Sci. Techn. XV, 57 (1967)
31. Teodorescu, P.P.: On the action of concentrated loads in the case of a cosserat continuum.
IUTAM Symposium 1967, Mechanics of Generalized Continua, p. 120 (1968)
32. Teodorescu, P.P.: Sur les corps du type de Cosserat à éleasticité linéaire. Ist. Naz. Alta Mat.,
Symp. Math. I, 375 (1969)
33. Teodorescu, P.P.: Sur le calcul des récipients soumis à des charges concentrées dans le cas de
l’élasticité symétrique ou asymétrique. Quaderni de ‘‘la ricerca sci.’’, Napoli 24 (1973)
34. Teodorescu, P.P., S ßandru, N.: Sur l’action des charges concentrées en élasticité asymétrique
plane. Rev. Roum. Math. Pures Appl. XII, 1399 (1967)
35. Thomson, W. (Lord Kelvin): Cambridge and Dublin Math. J. (1848)
36. Toupin, R.A.: Elastic materials with couple stresses. Arch. Rat. Mech. Anal. 11, 385 (1962)
37. Vodička, V.: Ein durch allgemeine Massenkräfte beanspruchtes unendliches Medium.
ZAMM 39, 2 (1959)
Chapter 9
Elastic Space. Elastic Half-Space
In the present chapter and in the two following ones we deal with the study of
elastic bodies which occupy a domain in the form of a parallelepiped, as well as
with the study of the bodies which occupy infinite domains, obtained from the
above mentioned one, by throwing to infinite one or several of its faces. Thus, we
shall consider all spatial domains bounded by planes parallel to the co-ordinate
planes, hence by planes parallel or orthogonal to each other.
If all the faces, of the parallelepiped are thrown to infinite, then we have to do
with the elastic space. The elastic half-space is obtained if five or the faces of the
parallelepiped are thrown to infinite. If four of the faces of the parallelepiped are
thrown to infinite, then there are two possibilities: the two faces at a finite distance
are parallel to each other, obtaining thus an elastic layer, or are orthogonal to each
other, corresponding thus the elastic quarter-space. If three of the faces of the
parallelepiped are at a finite distance, then one obtains the elastic eighth-space,
when all these faces are orthogonal one to each other, or the elastic half-layer,
when two of these faces are parallel and orthogonal to a third one. If only two of
the faces of the parallelepiped are thrown to infinite, then there are still two
possibilities: if the faces at a finite distance are parallel two by two, then one gets
the elastic strip, while if only two faces are parallel, the other two ones being
orthogonal to the later faces and orthogonal to each other, then one obtains the
elastic quarter-layer. The elastic half-strip corresponds to the case when only one
of the faces of the parallelepiped is thrown to infinite.
In the following we shall consider the elastic space acted by volume loads (forces
and moments), in particular concentrated loads; we deal with the static case.
The elastic half-space acted by arbitrary volume forces may be studied by the
method of Fourier transforms after I. N. Sneddon and D. S. Berry [10] and after V.
Vodička [44]. Another method (in fact, another aspect of the method mentioned
above) is that of using, for the volume loads, Fourier representations: triple Fourier
series, if the volume loads are periodic, or triple Fourier integrals, if the volume
loads act locally. W. Kecs [14–16] dealt with concentrated loads, by means of the
theory of distributions.
We observe that any case of loading may be decomposed in eight cases of loading
after the properties of symmetry or antisymmetry with respect to the three co-
ordinate axes. We shall consider a case antisymmetric with respect to the Ox2 x3 -
plane and symmetric with respect to the Ox3 x1 -plane and to the Ox1 x2 -plane (case
useful for the study of the action of a concentrated force along the Ox1 -axis); any
other case of loading may be studied analogically.
The components of the volume load can be expressed, in this case, in the form
Z 1Z 1Z 1
F1 ¼ a1 cos a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 ;
Z0 1 Z0 1 Z0 1
F2 ¼ a2 sin a1 x1 sin a2 x2 cos a3 x3 da1 da2 da3 ; ð9:1Þ
Z0 1 Z0 1 Z0 1
F3 ¼ a3 sin a1 x1 sin a2 x2 sin a3 x3 da1 da2 da3 ;
0 0 0
Replacing in the Eq. (5.1000 ) of Lamé, we get the equations which must be
verified by the unknown functions Ai ; i ¼ 1; 2; 3,
ai 2a1 di1 1
ak a k A i þ ðak Ak 2a1 A1 Þ ¼ ai ; i ¼ 1; 2; 3; ð9:3Þ
1 2m l
by solving the system (9.3), one obtains
1 ðai 2a1 di1 Þðak ak 2a1 a1 Þ
Ai ¼ ai ; i ¼ 1; 2; 3; ð9:4Þ
lak ak 2ð1 mÞak ak
the state of displacement being thus specified. The components (9.2) of the dis-
placement vector, being expressed by Fourier transforms, are regular functions at
infinite; hence, the particular integrals thus obtained are just the solutions of the
problem.
Considering the same volume forces (9.1) as in the precedent subsection, we may
choose the components of the stress tensor in the form.
Z 1Z 1Z 1
r11 ¼ B1 sin a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 ;
Z0 1 Z0 1 Z0 1
r22 ¼ B2 sin a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 ; ð9:5Þ
Z0 1 Z0 1 Z0 1
r33 ¼ B3 sin a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 ;
0 0 0
Z 1 Z 1 Z 1
r23 ¼ C1 sin a1 x1 sin a2 x2 sin a3 x3 da1 da2 da3 ;
Z 01 Z 01 Z 01
r31 ¼ C2 cos a1 x1 cos a2 x2 sin a3 x3 da1 da2 da3 ; ð9:50 Þ
0 0 0
Z 1Z 1Z 1
r12 ¼ C3 cos a1 x1 sin a2 x2 cos a3 x3 da1 da2 da3 ;
0 0 0
a1 B1 þ a2 C3 þ a3 C2 ¼ a1 ;
a1 C 3 þ a2 B 2 a3 C 1 ¼ a2 ; ð9:6Þ
a1 C 2 a2 C 1 þ a3 B 3 ¼ a3 ;
396 9 Elastic Space. Elastic Half-Space
a2
a21 þ a22 þ a23 B1 þ 1 ðB1 þ B2 þ B3 Þ
1þm
m
¼ ða1 a1 þ a2 a2 þ a3 a3 Þ þ 2a1 a1 ;
1m
2 a2
a1 þ a22 þ a23 B2 þ 2 ðB1 þ B2 þ B3 Þ
1þm ð9:7Þ
m
¼ ða1 a1 þ a2 a2 þ a3 a3 Þ 2a2 a2 ;
1m
2 a2
a1 þ a22 þ a23 B3 þ 3 ðB1 þ B2 þ B3 Þ
1þm
m
¼ ða1 a1 þ a2 a2 þ a3 a3 Þ 2a3 a3 ;
1m
2 a2 a3
a1 þ a22 þ a23 C1 ðB1 þ B2 þ B3 Þ ¼ ða3 a2 þ a2 a3 Þ;
1þm
2 a3 a1
a1 þ a22 þ a23 C2 þ ð B1 þ B2 þ B3 Þ ¼ a 1 a3 a 3 a 1 ; ð9:70 Þ
1þm
2 a1 a2
a1 þ a22 þ a23 C3 þ ð B1 þ B2 þ B3 Þ ¼ a 1 a2 a 2 a 1 :
1þm
This system of 9 equations, which implies 6 unknowns is compatible, because the
Beltrami-Michell equations (5.42) are just conditions of compatibility in stresses;
by solving it, we get
2
1 a1 þ mða21 þ a22 þ a23 Þ ða1 a1 þ a2 a2 þ a3 a3 Þ
B1 ¼ 2 2a1 a1 ;
a1 þ a22 þ a23 ð1 mÞða21 þ a22 þ a23 Þ
1 a22 þ mða21 þ a22 þ a23 Þ ða1 a1 þ a2 a2 þ a3 a3 Þ
B2 ¼ 2 þ 2a 2 a 2 ;
a1 þ a22 þ a23 ð1 mÞða21 þ a22 þ a23 Þ
2
1 a3 þ mða21 þ a22 þ a23 Þ ða1 a1 þ a2 a2 þ a3 a3 Þ
B3 ¼ 2 þ 2a 3 3 ;
a
a1 þ a22 þ a23 ð1 mÞða21 þ a22 þ a23 Þ
ð9:8Þ
1 a2 a3 ða1 a1 þ a2 a2 þ a3 a3 Þ
C1 ¼ ða3 a2 þ a2 a3 Þ ;
a21 2
þ a2 þ a3 2 ð1 mÞða21 þ a22 þ a23 Þ
1 a3 a1 ða1 a1 þ a2 a2 þ a3 a3 Þ
C2 ¼ 2 þ a1 a3 a3 a1 ; ð9:80 Þ
a1 þ a22 þ a23 ð1 mÞða21 þ a22 þ a23 Þ
1 a1 a2 ða1 a1 þ a2 a2 þ a3 a3 Þ
C3 ¼ 2 þ a 1 a 2 a 2 a 1 ;
a1 þ a22 þ a23 ð1 mÞða21 þ a22 þ a23 Þ
In what follows, we introduce first a concentrated force, dealing then with other
concentrated loads; a new concentrated load (the centre of torsion) will be intro-
duced too.
To study the case of a concentrated force, we start from the results obtained for the
volume loads. We take thus F2 ¼ F3 ¼ 0 and F1 ¼ p, the load p being uniformly
distributed in the interior of a parallelepiped of dimensions 2d1 , 2d2 and 2d3 ; the
volume load will be thus represented in the form (9.1), the function a1 ða1 ; a2 ; a3 Þ
being given by the formula (6.800 ) in the form
Z Z Z
8 d1 d2 d3
a1 ¼ 3 p cos a1 n1 cos a2 n2 cos a3 n3 dn1 dn2 dn3
p 0 0 0
8p sin a1 d1 sin a2 d2 sin a3 d3
¼ 3 : ð9:9Þ
p a1 a2 a3
Denoting F1 ¼ F ¼ 8pd1 d2 d3 and passing to the limit for p ! 1 and
d1 d2 d3 ! 0, we may represent the concentrated force, directed in the positive
sense of the Ox1 -axis, by the Fourier series
Z Z Z
F d1 d2 d3
cos a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 : ð9:90 Þ
p3 0 0 0
The formulae (9.2) and (9.4) allow to express the state of displacement in the
form
Z 1Z 1Z 1
F ð1 2mÞa21 þ 2ð1 mÞ a22 þ a23
u1 ¼ 3 2 2
2p ð1 mÞl 0 0 0 a þ a2 þ a2 1 2 3
cos a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 ;
Z 1Z 1Z 1
F a1 a2
u2 ¼ 3 2 2
2p ð1 mÞl 0 0 0 a þ a2 þ a2 ð9:10Þ
1 2 3
sin a1 x1 sin a2 x2 cos a3 x3 da1 da2 da3 ;
Z 1Z 1Z 1
F a1 a3
u3 ¼ 3 2 2
2p ð1 mÞl 0 0 0 a þ a2 þ a2
1 2 3
sin a1 x1 cos a2 x2 sin a3 x3 da1 da2 da3 ;
while the formulae (9.5), (9.50 ), (9.8), (9.80 ) lead to the state of stress
398 9 Elastic Space. Elastic Half-Space
Z 1 Z 1 Z 1
F a1 ð1 mÞa21 þ ð2 mÞ a22 þ a23
r11 ¼ 3 2 2
p ð1 mÞ 0 0 0 a1 þ a22 þ a23
sin a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 ;
Z 1Z 1Z 1
F a1 ð1 mÞa22 m a23 þ a21
r22 ¼ 3 2 2
p ð1 mÞ 0 0 0 a þ a2 þ a2 ð9:11Þ
1 2 3
sin a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 ;
Z 1Z 1Z 1
F a1 ð1 mÞa23 m a21 þ a22
r33 ¼ 3 2 2
p ð1 mÞ 0 0 0 a þ a2 þ a2 1 2 3
sin a1 x1 cos a2 x2 cos a3 x3 da1 da2 da3 ;
Z 1Z 1Z 1
F a1 a2 a3
r23 ¼ 3 2 2
p ð1 mÞ 0 0
2 2
0 ða1 þ a2 þ a3 Þ
Starting from the results obtained in the above subsection, one may obtain the
states of displacement for other concentrated loads applied at the origin O.
Thus, for a directed concentrated moment along the Ox1 -axis, the state of
displacement is given by the formulae (8.55), while for a centre of rotation applied
in the Ox1 x2 -plane, the state of displacement is given by (8.58), the displacement
u3 vanishing.
A concentrated moment of linear dipole type along the Ox1 -axis leads to the
state of displacement (8.59). Starting from this load, one obtains the state of
displacement (8.60) for a plane centre of dilatation in the Ox1 x2 -plane and the
state of displacement (8.61) for a spatial centre of dilatation.
Analogically, starting from two centres of rotation of inverse sense, one may
obtain, on the same way, by a process of passing to the limit, a linear dipole of
9.1 Elastic Space 399
Let be the elastic half-space x3 0 acted upon by a normal periodic (in two
directions) load pðx1 ; x2 Þ, symmetric with respect to the co-ordinate axes Ox1 and
Ox2 ; to fix the ideas, we represent this load by means of a double Fourier series,
even with respect to x1 and x2 , of the form
XX
p3 ðx1 ; x2 Þ ¼ b0 þ bnm cos an x1 cos bm x2 ; ð9:14Þ
n m
where we denoted
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cnm ¼ a2n þ b2m ð9:17Þ
and where the functions in two variables have been taken equal to zero.
9.2 Elastic Half-Space 401
As well, it results
1
K3 ¼ b0 : ð9:22Þ
2ð1 þ mÞ
Due to the terms which are not under the sign sum in the expressions of the stress
functions, one obtains
402 9 Elastic Space. Elastic Half-Space
1 1
u1 ¼ ½K1 mðK2 þ K3 Þ; u2 ¼ ½K2 mðK1 þ K3 Þ; ð9:200 Þ
l l
putting the conditions that the displacements u1 and u2 do not increase indefinitely
together with the variables x1 and x2 (the lateral deformation is stopped by the
medium) and taking into account (9.22), one gets
m
K1 ¼ K2 ¼ b0 : ð9:220 Þ
2ð1 m2 Þ
Thus, the stress functions read
m m
F1 b0 x21 ¼ F2 þ b0 x22
2ð1 m2 Þ 2ð1 m2 Þ
m X X a2 b 2
¼ x3 n m
bnm ecnm x3 cos an x1 cos bm x2 ; ð9:23Þ
2þm n m
c3nm
XX 1
F3 ¼ 2
ð1 þ 2m cnm x3 Þbnm ecnm x3 cos an x1 cos bm x2 :
n m
c nm
XX bm
r23 ¼ bnm U3 ðcnm x3 Þ cos an x1 sin bm x2 ;
n m
c nm
XX an
r31 ¼ bnm U3 ðcnm x3 Þ sin an x1 cos bm x2 ; ð9:240 Þ
n m
c nm
XX an b
r12 ¼ bnm 2 m U46 ðcnm x3 Þ sin an x1 sin bm x2
n m
cnm
where we have introduced the notations in Sect. A.4.1.1. Because of the symmetry
of the displacements with respect to the Ox1 and Ox2 axes, we have taken
the displacement of rigid body u03 being impossible to determinate; hence, the
displacement u3 is obtained only as a displacement relative to another point,
considered as fixed.
We notice that the mean loading b0 leads to a displacement which tends to
infinite for x3 ! 1; this result is due to the fact that we dealt with an infinite solid
(which can be considered as a prism of infinite length acted upon by a simple axial
load). But we can admit that the result thus obtained is valid for points at a finite
distance.
The convergence of the above series depends on the factor bnm . But, even in the
case of concentrated loads, the sum of the series may be approximated with a small
number of terms, because of the exponential factor. The series corresponding to
the displacements are convergent after the factor bnm =cnm ; hence, the sum of these
series may be approximated with a smaller number of terms that in the case of the
stresses.
Taking into account (9.14), we can write the stresses corresponding to the
separation plane in the form.
404 9 Elastic Space. Elastic Half-Space
b0
r11 ðx1 ; x2 ; 0Þ ¼ p3 ðx1 ; x2 Þ ð1 2mÞ
1m
X X
b 2
m
þ bnm cos an x1 cos bm x2 ;
n m
cnm
ð9:27Þ
b0
r22 ðx1 ; x2 ; 0Þ ¼ p3 ðx1 ; x2 Þ ð1 2mÞ
1m
X X
an 2
þ bnm cos an x1 cos bm x2 ;
n m
cnm
X X an b
m
r12 ðx1 ; x2 ; 0Þ ¼ ð1 2mÞ bnm sin an x1 sin bm x2 ; ð9:270 Þ
n m
c2nm
where we took into consideration the difference between the external loading and
the mean one. As well, the deformation of the separation plane is given by
1 m X X bnm
u3 ðx1 ; x2 ; 0Þ ¼ u03 cos an x1 cos bm x2 ð9:28Þ
l n m
cnm
These results may be used to calculate the pressure (the normal stress r33 ) and
the soil settlement (the displacement u3 ) in case if isolated formulations, having an
approximately uniform distribution, or in case of a net of foundation beams, for
which the ratio between the spans in a direction is approximately equal to unity. If
the loading is not rigorously periodic, then the approximation must be evaluated in
each case.
Analogically, one may consider also other cases of loading with a normal
loading having other properties of symmetry with respect to the co-ordinate axes.
Cases of periodic tangential loads can be studied in the same way.
where
Z 1 Z 1
4
a3 ða1 ; a2 Þ ¼ 2 p3 ðn1 ; n2 Þ cos a1 n1 cos a2 n2 dn1 dn2 ; ð9:290 Þ
p 0 0
acting on the plane x3 ¼ 0. One uses double Fourier integrals, even with respect to
the variables x1 ; x2 , which may represent, as well, distributions. The function
a3 ða1 ; a2 Þ, which characterizes the normal load, is even with respect to a1 and a2 .
The considered load is thus symmetric with respect to the Ox1 and Ox2 axes.
Taking
1
A1 ða2 ; a3 Þ ¼ A2 ða3 ; a1 Þ ¼ 0; A3 ða1 ; a2 Þ ¼ a3 ða2 ; a3 Þ; ð9:30Þ
c3
we observe that the system of integral Eq. (10.21) is verified; the first two
equations do not intervene (indeed, they may be considered as verified, because
a1 ða1 ; a2 Þ, a2 ða1 ; a2 Þ are arbitrary functions).
One obtains the state of stress.
Z 1Z 1 "
a1 2
r11 ¼ a3 ða1 ; a2 Þ U4 ðc3 x3 Þ
0 0 c3
2 #
a2
þm U6 ðc3 x3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
c3
Z 1Z 1 "
a2 2 ð9:31Þ
r22 ¼ a3 ða1 ; a2 Þ U4 ðc3 x3 Þ
0 0 c3
2 #
a1
þm U6 ðc3 x3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
c3
Z 1Z 1
r33 ¼ a3 ða1 ; a2 ÞU2 ðc3 x3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
0 0
Z 1 Z 1
a2
r23 ¼ a3 ða1 ; a2 Þ
U3 ða3 x3 Þ cos a1 x1 sin a2 x2 da1 da2 ;
c3
Z0 1 Z0 1
a1
r31 ¼ a3 ða1 ; a2 Þ U3 ða3 x3 Þ sin a1 x1 cos a2 x2 da1 da2 ; ð9:310 Þ
0 c3
Z 10 Z 1
a1 a2
r12 ¼ a3 ða1 ; a2 Þ 2 U46 ða3 x3 Þ sin a1 x1 sin a2 x2 da1 da2
0 0 c3
406 9 Elastic Space. Elastic Half-Space
where the displacement of rigid body u03 is non-zero, because of the symmetry
assumed for the load.
Taking into account the load (9.29), one gets the normal stresses
One observes that the system of integral equations (10.55), (10.550 ) is verified;
indeed, the first two equations of both systems do not intervene, because
a01 ða2 ; a3 Þ, a001 ða2 ; a3 Þ; a02 ða3 ; a1 Þ; a002 ða3 ; a1 Þ are arbitrary functions.
The formulae (10.62)–(10.630 ) allow thus to express the state of stress and
displacement. For instance, the stress normal to the separation plane x3 ¼ 0 is
given by
Z 1Z 1
0
r33 ¼ x3 a1 a3 ða1 ; a2 Þ
0 0
þa2 a003 ða1 ; a2 Þ ec3 x3 sin a1 x1 sin a2 x2 da1 da2 ; ð9:37Þ
9.2.3 Applications
We shall consider, in the following, some particular cases with periodic or local
normal loads, which may be studied starting from the previous general results.
Let us search the state of stress in the earth, considered to be an elastic half-space
acted upon by a normal periodic load resulting from uniformly distributed isolate
foundations.
We choose square isolate foundations (of side a), so that between their axes be
a distance equal to 2a in both directions; the ratio between the area covered by the
foundation and the total area is 1=4, hence we may consider that the foundations
are isolated. The foundations act with a uniform distributed load of compression p.
Taking the periods L1 ¼ L2 ¼ 2a, we represent the load in the form
p pX 1
p3 ðx1 ; x2 Þ ¼ ð1Þðn1Þ=2 cos an x1
4 p n n
pX 1
ð1Þðm1Þ=2 cos bm x2
p m m
4p X X 1
þ 2 ð1ÞðnþmÞ=2 cos an x1 cos bm x2 ; n; m ¼ 1; 3; 5; . . .
p n m nm
ð9:39Þ
The normal stress r33 at the depth a under the centre of a foundation, hence for
x1 ¼ x2 ¼ 0; x3 ¼ a is given by (one observes that the two simple series have the
same sum)
p X 1 þ np np
r33 ð0; 0; aÞ ¼ 2p ð1Þðn1Þ=2 e
4 n
n
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4p X X 1 þ p n2 þ m2 ppffiffiffiffiffiffiffiffiffiffi
þ 2 ð1ÞðnþmÞ=2 e n2 þm2
; n; m ¼ 1; 3; 5; . . .
p n m nm
ð9:40Þ
Taking one term in the expansions into series, it results
r33 ð0; 0; aÞ ffi 0:250p 0:114p 0:026p ¼ 0:390p: ð9:400 Þ
We obtain the same result if we apply, e.g., the method of the corner points, on a
longer and more complicate way, using the tables of pressure in the earth at the
corners of the rectangular foundations.
If one calculates the pressure under the corners of the rectangle, one obtains
9.2 Elastic Half-Space 409
a a
r33 ; ; x3 ffi 0:250p ð9:41Þ
2 2
at any depth.
We notice the rapidity of convergence of the series which give the soil set-
tlement, because of a supplementary increasing factor at the denominator.
Let us consider the case studied for the first time by J. Boussinesq [1]: the elastic
half-space acted upon by a normal concentrated force.
Let firstly be the case of a normal load, uniformly distributed on a rectangle
p for jx1 j\a1 ; jx2 j\a2 ;
p3 ðx1 ; x2 Þ ¼ ð9:42Þ
0 for jx1 j [ a1 ; jx2 j [ a2 ;
it results
4p
a3 ða1 ; a2 Þ ¼ sin a1 a1 sin a2 a2 : ð9:420 Þ
p2 a1 a2
Making p ! 1 and a1 ; a2 ! 0 with 4pa1 a2 ¼ P, one obtains the results corre-
sponding to a concentrated force; thus, the Fourier coefficient reads
P
a3 ða1 ; a2 Þ ¼ : ð9:4200 Þ
p2
The normal stress r33 becomes
Z Z
P 1 1
r33 ¼ 2 ð1 þ c3 x3 Þec3 x3 cos a1 x1 cos a2 x2 da1 da2 ; ð9:43Þ
p 0 0
3P x33
r33 ¼ ð9:430 Þ
2p r 5
and
1mP
u3 ðx1 ; x2 ; 0Þ ¼ u03 þ ; ð9:440 Þ
2pl r3
where
410 9 Elastic Space. Elastic Half-Space
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r3 ¼ x21 þ x22 : ð9:45Þ
Let be now the elastic half-space x3 0 acted upon by a normal load, symmetric
with respect to both co-ordinate axes Ox1 and Ox2 , of the form
p
p3 ðx1 ; x2 Þ ¼ : ð9:46Þ
r33
x1 x2
r12 ðx1 ; x2 ; 0Þ ¼ 3ð1 2mÞp ð9:490 Þ
r35
Let be the elastic half-space x3 0 acted upon by the external loads pj ðx1 ; x2 Þ,
j ¼ 1; 2; 3, on the separation plane x3 ¼ 0. We search the fundamental solution, in
the frame of the theory of distributions, for displacements, as well as for stresses.
We start from Lamé’s equations (5.10) in the absence of volume forces, i.e.
lDuj þ ðk þ lÞh;j ¼ 0; j ¼ 1; 2; 3; ð9:56Þ
We introduce the matrix (6.123), whose elements satisfy the systems of dif-
ferential equations
lDukj þ ðk þ lÞhj;k ¼ 0; j; k ¼ 1; 2; 3; ð9:59Þ
9.2 Elastic Half-Space 413
where
hj ¼ ukj;k ; j ¼ 1; 2; 3: ð9:590 Þ
The corresponding boundary conditions on the separation plane read
1
lim ðu31;1 þ u11;3 Þ ¼ dðx1 ; x2 Þ;
x3 !þ0 l
lim ðu21;3 þ u31;2 Þ ¼ 0; ð9:60Þ
x3 !þ0
Applying the double Fourier transform with respect to the variables x1 and x2
and considering x3 as a parameter, we may write the equations (3.70) with (3.700 )
for j ¼ 1 in the form
2
d 2
l c 3 F ½ u ðx ; x
k1 1 2 3 ; x Þ ðk þ lÞa k a1 F½u11 ðx1 ; x2 ; x3 Þ
dx23
d
þ a2 F½u21 ðx1 ; x2 ; x3 Þ þ i F½u31 ðx1 ; x2 ; x3 Þ ¼ 0; k ¼ 1; 2; ð9:62Þ
dx3
2
d 2 d
l c 3 F½u31 ðx1 ; x2 ; x3 Þ iðk þ lÞ ½a1 F½u11 ðx1 ; x2 ; x3 Þ
dx23 dx3
d
þ a2 F½u21 ðx1 ; x2 ; x3 Þ þ i F½u31 ðx1 ; x2 ; x3 Þ ¼ 0; ð9:620 Þ
dx3
where F½ ¼ Fx1 ½Fx2 ½ and ak ; k ¼ 1; 2, are complex variables in the space of
transformations; applying the Fourier transforms to the boundary conditions (9.60)
as well, we obtain
414 9 Elastic Space. Elastic Half-Space
d
l lim F½u11 ðx1 ; x2 ; x3 Þ ia1 F½u31 ðx1 ; x2 ; x3 Þ ¼ 1;
x3 !þ0 dx3
d
lim F½u21 ðx1 ; x2 ; x3 Þ ia2 F½u31 ðx1 ; x2 ; x3 Þ ¼ 0;
x3 !þ0 dx3
ð9:63Þ
d
2l lim F½u31 ðx1 ; x2 ; x3 Þ ki lim a1 F½u11 ðx1 ; x2 ; x3 Þ
x3 !þ0 dx3 x3 !þ0
d
þa2 F½u21 ðx1 ; x2 ; x3 Þ þ i F½u31 ðx1 ; x2 ; x3 Þ ¼ 0:
dx3
Multiplying the Eq. (9.62) by ak ; k ¼ 1; 2, respectively and adding them, we
have
d2
l 2
a1 F½u11 ðx1 ; x2 ; x3 Þ þ a2 F½u21 ðx1 ; x2 ; x3 Þ
dx3
ðk þ 2lÞc23 a1 F½u11 ðx1 ; x2 ; x3 Þ þ a2 F½u21 ðx1 ; x2 ; x3 Þ
d
¼ iðk þ lÞc23 F½u31 ðx1 ; x2 ; x3 Þ; ð9:64Þ
dx3
on the other hand, the Eq. (9.620 ) may be written
d
ðk þ lÞ ½a1 F½u11 ðx1 ; x2 ; x3 Þ þ a2 F½u21 ðx1 ; x2 ; x3 Þ
dx3
d2 2
¼ i ðk þ 2lÞ 2 F½u31 ðx1 ; x2 ; x3 Þ lc3 F½u31 ðx1 ; x2 ; x3 Þ : ð9:640 Þ
dx3
Thus, from the Eq. (9.64) and (9.640 ), one obtains the differential equations which
must be satisfied by the Fourier transform F½u31 ðx1 ; x2 ; x3 Þ, namely
d4 2 d
2
F½u 31 ðx 1 ; x 2 ; x3 Þ 2c 3 F½u31 ðx1 ; x2 ; x3 Þ
dx43 dx23
þ c43 F½u31 ðx1 ; x2 ; x3 Þ ¼ 0; ð9:65Þ
as well as the first two conditions (9.63), the Eq. (9.640 ) gives
for x3 ! þ0. Substituting the Fourier transform F½u31 ðx1 ; x2 ; x3 Þ given by (9.650 )
into the Eq. (9.640 ) and integrating, one obtains
Passing to the limit for x3 ! þ0 in the relation (9.67) and taking into account the
third boundary condition (9.63) we get
lB ¼ ðk þ lÞc3 A: ð9:660 Þ
1 k a21 1 a21 c3 x3
F½u11 ðx1 ; x2 ; x3 Þ ¼ 1 x 3 e ;
lc3 2ðk þ lÞ c23 2 c3
1 a1 a 2 k c3 x3
F½u21 ðx1 ; x2 ; x3 Þ ¼ þ x 3 e ; ð9:70Þ
2l c23 ðk þ lÞc3
i a1 1 1
F½u31 ðx1 ; x2 ; x3 Þ ¼ þ x3 ec3 x3 :
2 c3 ðk þ lÞc3 l
For the computation of the Fourier transforms we remark that
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
Fx1 ¼ 2K0 a1 x22 þ x23 ; ð9:71Þ
R
where K0 is the modified Bessel function of the second species and zero order;
then we have
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
Fx K0 a1 x22 þ x23 ¼ ec3 x3 ; Re a1 [ 0; ð9:710 Þ
p 2 c3
whence
1 1 1
F ¼ ec3 x3 ; x3 [ 0: ð9:7100 Þ
2p R c3
there follows
Z 1
Z 1
o 1 o 1 a1
F dx3 ¼ F dx3 ¼ 2pi 2 ec3 x3 : ð9:73Þ
x3 ox 1 R x3 ox 1 R c3
In the following, we shall express the inverse Fourier transforms with the aid of
the functions 1=R and lnðR þ x3 Þ; x3 [ 0, between which there exists the differ-
ential relation
o 1
lnðR þ x3 Þ ¼ ; ð9:74Þ
ox3 R
also, we may write
9.2 Elastic Half-Space 417
Z 1
1
dx3 ¼ lnðR þ x3 Þ; ð9:740 Þ
0 R
a relation to which the partial differentiation with respect to x1 and x2 can be
applied.
Thus, we may write the double Fourier transforms
2
1 o a2
F 2 lnðR þ x3 Þ ¼ 21 ec3 x3 ;
2p ox1 c3
2 ð9:75Þ
1 o a1 a2 c3 x3
F lnðR þ x3 Þ ¼ 2 e ;
2p ox1 ox2 c3
Finally, we obtain
Z 1 2
1 2 k o
u11 ðx1 ; x2 ; x3 Þ ¼ lnðR þ x3 Þdx3
4pl R k þ l x3 ox21
o2 1 1 x2
x3 2 lnðR þ x3 Þ ¼ 1 þ 12
ox1 4pl R R
1 2m x21
þ 1 ;
R þ x3 RðR þ x3 Þ
Z 1 2
1 k o
u21 ðx1 ; x2 ; x3 Þ ¼ lnðR þ x3 Þdx3 ð9:76Þ
4pl k þ l x3 ox1 x2
" #
o2 1 x1 x2 1 1 2m
þ x3 lnðR þ x3 Þ ¼ ;
ox1 x2 4pl R R2 ðR þ x3 Þ2
1 l o o 1
u31 ðx1 ; x2 ; x3 Þ ¼ lnðR þ x3 Þ x3
4pl k þ l ox1 ox1 R
1 x1 x3 1 2m
¼ þ :
4pl R R2 R þ x3
Likewise from the equations (9.59), (9.590 ) and the boundary conditions (9.600 ),
(9.6000 ), respectively, we obtain
418 9 Elastic Space. Elastic Half-Space
" #
1 x1 x2 1 1 2m
u12 ðx1 ; x2 ; x3 Þ ¼ ;
4pl R R2 ðR þ x3 Þ2
1 1 x21 1 2m x22
u22 ðx1 ; x2 ; x3 Þ ¼ 1þ 2 þ 1 ; ð9:760 Þ
4pl R R R þ x3 RðR þ x3 Þ
1 x2 x3 1 2m
u32 ðx1 ; x2 ; x3 Þ ¼ þ :
4pl R R2 R þ x3
1 x1 x3 1 2m
u13 ðx1 ; x2 ; x3 Þ ¼ ;
4pl R R2 R þ x3
ð9:7600 Þ
1 x2 x3 1 2m
u23 ðx1 ; x2 ; x3 Þ ¼ :
4pl R R2 R þ x3
1 1 x2
u33 ðx1 ; x2 ; x3 Þ ¼ 2ð1 mÞ þ 32 :
4pl R R
The fundamental solutions for stresses are given by the components of the matrix
(6.1320 ), which may be obtained starting from the components of the matrix U; by
means of the relations (6.138), one obtains.
(
)
1 x1 1 x21 1 2m x21 ð3R þ x3 Þ
s111 ðx1 ; x2 ; x3 Þ ¼ 1 2m 3 2 3 2 ;
2p R R2 R ðR þ x3 Þ2 R ðR þ x3 Þ
(
1 x2 1 x21
s112 ðx1 ; x2 ; x3 Þ ¼ 1 2m 3 : ð9:77Þ
2p R R2 R2
)
1 2m x21 ð3R þ x3 Þ
3 2 ;
ðR þ x3 Þ2 R ðR þ x3 Þ
(
)
1 x3 1 x21 1 2m x21 ð2R þ x3 Þ
s113 ðx1 ; x2 ; x3 Þ ¼ 1 2m 3 2 3 2 ;
2p R R2 R ðR þ x3 Þ2 R ðR þ x3 Þ
(
)
1 x1 1 x22 2ð1 2mÞ x22 ð3R þ x3 Þ
s221 ðx1 ; x2 ; x3 Þ ¼ 1 2m 2 3 2 ;
2p R R2 R ðR þ x3 Þ2 R ðR þ x3 Þ
9.2 Elastic Half-Space 419
(
1 x2 1 x22
s222 ðx1 ; x2 ; x3 Þ ¼ 1 2m 3 2
2p R R2 R
) ð9:770 Þ
1 2m x22 ð3R þ x3 Þ
3 2 ;
ðR þ x3 Þ2 R ðR þ x3 Þ
1 x3 1 x22 1 2m x22 ð2R þ x3 Þ
s223 ðx1 ; x2 ; x3 Þ ¼ 1 2m 3 2 1 2 ;
2p R R2 R x3 ðR þ x3 Þ R ðR þ x3 Þ
3 x23 xk
s33k ðx1 ; x2 ; x3 Þ ¼ ; k ¼ 1; 2; 3; ð9:7700 Þ
2p R5
3 x2 x3 xk
s23k ðx1 ; x2 ; x3 Þ ¼ ; k ¼ 1; 2; 3; ð9:78Þ
2p R5
3 x3 x1 xk
s31k ðx1 ; x2 ; x3 Þ ¼ ; k ¼ 1; 2; 3; ð9:780 Þ
2p R5
(
)
1 x2 1 x21 1 2m x21 ð3R þ x3 Þ
s121 ¼ 13 2 1 2 ;
2p R R2 R ðR þ x3 Þ2 R ðR þ x3 Þ
(
)
1 x1 1 x22 1 2m x22 ð3R þ x3 Þ
s122 ¼ 13 2 1 2 ; ð9:7800 Þ
2p R R2 R ðR þ x3 Þ2 R ðR þ x3 Þ
" #
1 x1 x2 x3 2R þ x3
s123 ¼ 3 2 ð1 2mÞ ;
2p R3 R ðR þ x3 Þ2
r ¼ S
p; ð9:81Þ
where the matrices r and S are given by (6.1320 ). In components one has
rij ðx1 ; x2 ; x3 Þ ¼ sijk ðx1 ; x2 ; x3 Þ
pk ðx1 ; x2 Þ; i; j ¼ 1; 2; 3: ð9:810 Þ
Let be now a concentrated force F, of components
acting upon the half-space x3 0 on the separation plane, at the point Aðx01 ; x02 ; 0Þ.
The state of displacement is given by
hence
hence
for x3 ¼ 0 one obtains the deformation of the separation plane, getting again the
formula (9.440 ) for u03 ¼ 0.
9.2 Elastic Half-Space 421
3P x33
r33 ðx1 ; x2 ; x3 Þ ¼ ; ð9:85Þ
2p R5
3P x23 xi
r3i ðx1 ; x2 ; x3 Þ ¼ ; i ¼ 1; 2;
" 2p R5 # ð9:850 Þ
P x3 2R þ x3
r12 ðx1 ; x2 ; x3 Þ ¼ 3 ð1 2mÞ :
2pR R2 ðR þ x3 Þ2
These results correspond to the problem of Boussinesq [1], i.e., the problem of a
normal concentrated force.
If F2 ¼ F3 ¼ 0; F1 ¼ P 6¼ 0, the state of displacement is given by
P 1 x21 1 2m x21
u1 ðx1 ; x2 ; x3 Þ ¼ 1þ 2 þ 1 ;
4pl R R R þ x3 RðR þ x3 Þ
" #
P x1 x2 1 1 2m
u2 ðx1 ; x2 ; x3 Þ ¼ ; ð9:86Þ
4pl R R2 ðR þ x3 Þ2
P x1 x3 1 2m
u3 ðx1 ; x2 ; x3 Þ ¼ þ ;
4pl R R2 R þ x3
3P x1 x23
r33 ðx1 ; x2 ; x3 Þ ¼ ; ð9:87Þ
2p R5
422 9 Elastic Space. Elastic Half-Space
3P x3 x1 xi
r3i ðx1 ; x2 ; x3 Þ ¼ ; i ¼ 1; 2;
(
2p R5
)
P x2 1 x21 1 2m x21 ð3R þ x3 Þ
r12 ðx1 ; x2 ; x3 Þ ¼ 13 2 1 2 :
2p R R2 R ðR þ x3 Þ2 R ðR þ x3 Þ
ð9:870 Þ
The only non-zero stress on the separation plane is
mP x2 x21
r12 ðx1 ; x2 ; 0Þ ¼ 13 2 : ð9:8700 Þ
p R3 R
These results correspond to the problem of Cerruti [12], i.e. the problem of a
tangential concentrated force.
Let us now consider an elastic half-space x3 0 acted upon at the point Aðx01 ; x02 ; 0Þ
by a directed concentrated moment of magnitude M, specified by the unit vectors u
and F0 ; the state of displacement is obtained in the form
M o
ui ðx1 ; x2 ; x3 Þ ¼ u ðF0 Þ;
0 ou i
i ¼ 1; 2; 3; ð9:88Þ
ju F j
where o=ou is the directional derivative and ui ðF0 Þ; i ¼ 1; 2; 3, are the dis-
placements corresponding to the action of the concentrated force F0 (of unit
magnitude), in accordance with the formulae (9.820 ).
In particular, if u ¼ i1 and F0 ¼ i3 , where i1 and i3 are the unit vectors of the
Ox1 and Ox3 axes, respectively, we obtain a directed concentrated moment
assumed to be applied at the origin; the moment will cause a positive rotation in
the Ox3 x1 -plane. The equivalent load is given by
o
pi ðx1 ; x2 Þ ¼ 0; i ¼ 1; 2; p3 ðx1 ; x2 Þ ¼ M dðx1 ; x2 Þ ð9:89Þ
ox
and the resulting state of displacement is
9.2 Elastic Half-Space 423
o
u1 ðx1 ; x2 ; x3 Þ ¼ M u13 ðx1 ; x2 ; x3 Þ
ox1
M 1 x3 x21 1 2m x21 ð2R þ x3 Þ
¼ 13 2 1 2 ;
4pl R R2 R R þ x3 R ðR þ x3 Þ
o
u2 ðx1 ; x2 ; x3 Þ ¼ M u23 ðx1 ; x2 ; x3 Þ
ox1
" # ð9:90Þ
M x1 x2 x3 2R þ x3
¼ 3 2 ð1 2mÞ ;
4pl R3 R ðR þ x3 Þ2
o
u3 ðx1 ; x2 ; x3 Þ ¼ M u33 ðx1 ; x2 ; x3 Þ
ox1
M x1 x23
¼ 2ð1 mÞ þ 3 :
4pl R3 R2
Likewise, let x3 0 be an elastic half-space acted upon, on the separation plane,
at the point Aðx01 ; x02 ; 0Þ, by a rotational concentrated moment of magnitude M; we
assume that the moment causes a positive rotation, may be written in the form
1 o
p1 ðx1 ; x2 Þ ¼ M dðx1 x01 ; x2 x02 Þ;
2 ox2
1 o ð9:91Þ
p2 ðx1 ; x2 Þ ¼ M dðx1 x01 ; x2 x02 Þ;
2 ox1
p3 ðx1 ; x2 Þ ¼ 0;
We shall now consider that the elastic half-space x3 0 is acted upon at the origin
O, on the separation plane, by a concentrated moment of linear dipole type, of
magnitude D, the support of which being in the direction of the Ox1 -axis. The
equivalent load is expressed by
o
p1 ðx1 ; x2 Þ ¼ D dðx1 ; x2 Þ; pi ðx1 ; x2 Þ ¼ 0; i ¼ 2; 3; ð9:93Þ
ox1
hence, the state of displacement reads.
o
ui ðx1 ; x2 ; x3 Þ ¼ Dui1 ðx1 ; x2 ; x3 Þ
ox1
(
)
D xi 1 x21 1 2m x21 ð3R þ x3 Þ
¼ 13 2 3 2 ; i ¼ 1; 2;
4pl R R2 R ðR þ x3 Þ2 R ðR þ x3 Þ
ð9:94Þ
o
u3 ðx1 ; x2 ; x3 Þ ¼ D u31 ðx1 ; x2 ; x3 Þ
ox1
D 1 x3 x21 1 2m x21 ð3R þ x3 Þ
¼ 1 3 þ 1 :
4pl R R2 R2 R þ x3 R2 ðR þ x3 Þ
Other cases of loading may be treated in a similar way. By applying the result
presented in Sect. 9.2.3.2, one may easily write, in addition, the state of stress in
each of the considered cases to.
References
A. Books
10. Sneddon, I.N., Berry, D.S.: The Classical Theory of Elasticity, in Handbuch der Physik VI.
Elastizität und Plastizität. Springer, Berlin (1958)
11. Teodorescu, P.P.: Probleme spatßiale în teoria elasticitătßii (Spatial Problems in the Theory of
Elasticity). Ed. Acad., Bucuresßti (1970)
B. Papers
12. Cerruti, V.: Ricerche intorno all’equilibrio dei corpi elastici isotropi. Atti R. Accad. Lincei,
Mem. Cl. Sci. Fis., Mat. e Nat. 13, 81 (1852)
}
13. Hertz, H.: Uber die Berührung fester elasticher Körper. J. Für Reine Angew. Math. 92, 156
(1882)
14. Kecs, W.: Sur les problèmes concernant le demi-espace élastique. Bull. Math. Soc. Sci. Math.
Roum. 6(54), 157 (1962)
15. Kecs, W.: Les solutions généralisées des problèmes concernant l’espace élastique infini. Rev.
Roum. Math. Pures Appl. 11, 27 (1966)
16. Kecs, W.: Les solutions généralisées de l’espace élastique sous l’action de certaines charges
variables. Rev. Roum. Math. Pures Appl. 15, 7 (1970)
17. Lai Pham The and Mandel, J.: Force concentrée agissant au sein d’un semiespace élastique,
homogène, isotrope, limité par un plan dont les points restent fixes. C. Rend. Acad. Sci., Paris
260, 4417 (1965)
18. Lamb, H.: Proc. London Math. Soc. 34, 276 (1902)
19. Love, A.-E.-H.: The stress produced in a semi-infinite solid by pressure of part of the
boundary. Phil. Trans. Roy. Soc. London Ser. A 228, 377 (1929)
20. Lur’e, A.I.: Nekotorye kontaktnye zadachi teorii uprugosti (Some contact problems of the
theory of elasticity). Prikl. Mat. Mekh. 5, 383 (1941)
21. Mindlin, R.D.: Physics 7, 195 (1936)
22. Schleicher, F.: Bauingenieur 7 (1926)
23. Schleicher, F.: Bauingenieur 14, 242 (1933)
24. S
ßandru, N.: O sosredotochennykh vnutrennikh nagruzkakh v uprugom poluprostranstve s
pakreplenymi krayami (On the concentrated loads in the interior of the elastic half-space with
a free separation plane). Bull. Math. Soc. Sci. Math. Phys. Roum. 5(53), 205 (1961)
25. S
ßandru, N.: Asupra actßiunii unei fortße concentrate în spatßiul elastic nemărginit (On the action
of a concentrated force in the elastic infinite space). Com. Acad. Rom. XIII, 1019 (1963)
26. S
ßandru, N.: O deistvii peremennykh sil v neogranichennom uprugom prostranstve (On the
action of a variable force in the elastic space). Bull. Acad. Pol. Sci., Ser. Sci. Techn. 12, 45
(1964)
27. Tanimoto, B.: J. Shinshu Univ., Nagano (1957)
28. Tedone, O.: Saggio di una teoria generale delle equazioni dell’equilibrio elastico per un
corpo isotropo. Ann. Mat. Pura Appl. Ser. III 10, 13 (1904)
29. Teodorescu, P.P.: Asupra unor probleme spatßiale ale teoriei elasticitătßii (On some spatial
problems of the theory of elasticity). St. Cerc. Mec. Apl. VIII. 1101 (1957)
}
30. Teodorescu, P.P.: Uber die Berechnung des elastichen Halbraumes unter örtlicher Belastung.
Bull. Math. Soc. Sci. Math. Phys. Roum. 2(50), 113 (1958)
31. Teodorescu, P P.: Uber} die Berechnung des periodisch belasteten elastischen Halbraumes.
Rev. Méc. Appl. IV, 141 (1959)
32. Teodorescu, P.P.: Fonctions de tension dans le problème tridimensionnel de la théorie de
l’élasticité. Bull. Math. Soc. Sci. Math. Phys. Roum. 3(51), 499 (1959)
}
33. Teodorescu, P.P.: Uber einige räumliche Probleme der Elastizitätstheorie. Apl. Mat., Praha
IV, 225 (1959)
426 9 Elastic Space. Elastic Half-Space
34. Teodorescu, P.P.: Trei probleme tridimensionale ale teoriei elasticitătßii (Three tridimensional
problems of the theory of elasticity). An. Univ. Bucuresßti, Ser. ßst. Nat., Mat.-Mec. XV, 1–17
(1966)
35. Teodorescu, P.P.: Asupra actßiunii sarcinilor concentrate în teoria elasticitătßii (On the action
of concentrated loads in the theory of elasticity). An. Univ. Bucuresßti, Ser. ßst. Nat., Mat.-Mec.
XV, 2–15 (1966)
36. Teodorescu, P.P.: Einige Teilprobleme der dreidimensionalen Elastizitätstheorie. Wiss.
Zeitschrift Techn. Univ. Dresden 16, 77 (1967)
37. Teodorescu, P.P.: Asupra utilizării unor familii de functßii-potentßial reale în teoria elasticitătßii
(On the use of certain families of real potential functions in the theory of elasticity). An.
Univ. Bucuresßti, Ser. ßst. Nat., Mat.-Mec. XVI, 59 (1967)
38. Teodorescu, P.P.: Sur l’application des fonctions potentiel dans la théorie de l’élasticité.
Rend. Sem. Mat. e Fis. di Milano XXXVIII, 231 (1968)
39. Teodorescu, P.P.: Considérations concernant l’introduction des fonctions potentiel de
déplacement dans les problèmes en espace de la théorie de l’élasticité (cas statique). An.
Univ. Bucuresßti, Ser. Mat.-Mec. XVIII, 127 (1969)
40. Teodorescu, P.P.: Consideratßii în legătură cu utilizarea functßiilor potential în teoria
elasticitătßii (Considerations concerning the use of potential functions in the theory of
elasticity). Lucr. ßst. Inst. Ped., Galatßi 5, 27 (1971)
41. Teodorescu, P.P.: Sur l’introduction des fonctions-potentiel en élasticité linéaire. An. Univ.
Bucuresßti, Mat.-Mec., 20, 131 (1971)
42. Terezawa, K.: J. Coll. Sci. Univ. Tokyo
43. Thomson, W.: (Lord Kelvin), Cambridge and Dublin Math. J. (1848)
44. Vodička, V.: Ein durch allgemeine Masseukräfte beauspruchtes unendliches medium.
ZAMM 39, 2 (1959)
45. Weitsman, Y.: Quart. Appl. Math. XXV, 213 (1967)
Chapter 10
Elastic Eighth-Space. Elastic
Quarter-Space
In what follows we deal with the study of the elastic eighth-space subjected to the
action of a normal or tangential load; the results thus obtained will then be par-
ticularized for the case of the elastic quarter-space. We remember that they have
been used in Sect. 9.2 for the elastic half-space. We will take advantage of the
stress functions introduced in [3, 4, 13, 16].
The problem of the state of strain and stress in the interior of an elastic eighth-
space has been only a few times studied. But the problem is of interest, because
one can thus obtain informations concerning such a local state around a solid angle
with three faces of an elastic solid; if the three plane faces are not orthogonal one
each other, then the problem may be studied similarly, passing to a system of
oblique Cartesian co-ordinate axes. Interesting results are known for the corre-
sponding plane case (the elastic quarter-plane [51], the elastic wedge). A particular
case of loading with an internal concentrated force has been considered by
W. A. Hijab [2].
The elastic eighth-space, acted upon by a periodic load on one of its faces, may
practically constitute the end beam in the case of a continuous wall-beam of
infinite height, having an infinity of equal spans and identically acted upon. The
results corresponding to a local loading may be as well used for the study of the
local effect (corner effect) in the support zone of a wall-beam with a single span.
We will consider the elastic eighth-space xi 0; i ¼ 1; 2; 3, acted upon by
normal loads, using the results given in [6, 8] or by tangential loads, using the
results given in [7]. We will present also a particular case of loading, which leads
to results in finite form, as it has been shown in [10, 14].
Z
1 X X
bnm 2 2
cnm x3
a n ð1 c nm x 3 Þ þ 2mb m e dx3
0
n m
c2nm
(" 2 Z 1 #
XX an 2 b
jbnm j þ2m m x3 ecnm x3 dx3
n m
c nm c nm 0
2 Z 1 )
an
þcnm ecnm x3 dx3 ð10:2Þ
cnm 0
( " 2 # 2 )
XX 1 an 2 bm 1 an
¼ jbnm j þ2m þ 2 ¼ finite;
n m
cnm cnm cnm cnm cnm
Z 1 X X
bnm 2 2
cnm x3
b n ð1 c nm x 3 Þ þ 2ma n e dx3 ¼ finite:
0 n m
c2nm
We mention that these conditions are verified for any case of distributed loading.
10.1.1.2 Discussion
where
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cm ¼ a2 þ b2m : ð10:40 Þ
One obtains thus normal stresses periodic on one direction and local on the
other one on the faces xi ; i ¼ 1; 2. By an analogous reasoning, we conclude that
one must study the elastic quarter-space xi 0; i ¼ 1; 3, or the elastic quarter-
space xi 0; i ¼ 2; 3, acted upon by local loads on both directions.
In conclusion, the problem of the elastic eighth-space acted upon by an arbi-
trary normal periodic load may be reduced, using the principle of superposition of
effects, to the study of the same domain, acted upon by local loads, and to the
study of an elastic quarter-space, acted upon in an analogous manner. Thus, it is
sufficient to consider only the case of a local load.
Z Z
1 1 1 0
F11 ¼ A ða2 ; a3 Þ
4 1 1 1
þc1 x1 A02 ða2 ; a3 Þ ec1 x1 cos a2 x2 cos a3 x3 da2 da3
Z Z
1 1 1 0
þ B ða3 ; a1 Þ
4 1 1 1
þc2 x2 B02 ða3 ; a1 Þ ec2 x2 cos a3 x3 cos a1 x1 da3 da1
Z Z
1 1 1 0
þ C ða1 ; a2 Þ
4 1 1 1
þc3 x3 C20 ða1 ; a2 Þ ec3 x3 cos a1 x1 cos a2 x2 da1 da2 ; . . .; ð10:7Þ
and where the integrals are considered in the sense used in the theory of regu-
larization of the Fourier transforms in the theory of distributions.
The integrable functions A0i ða2 ; a3 Þ; B0i ða3 ; a1 Þ; Ci0 ða1 ; a2 Þ; A00i ða2 ; a3 Þ; . . .,
000
Ci ða1 ; a2 Þ; a1 ; a2 ; a3 2 ð1; 1Þ; i ¼ 1; 2, must be specified by the boundary
conditions (10.6). These functions are even with respect to the variables
a1 ; a2 ; a3 . The conditions at infinite are automatically verified, the method used
corresponding to Fourier transformations.
Using the results given at the above mentioned subsection, we obtain the
normal stresses
Z Z
1 1 1 2
r11 ¼ c A1 þ 2ð2 þ mÞA02
4 1 1 1
ð2 c1 x1 ÞA 2 ec1 x1 cos a2 x2 cos a3 x3 da2 da3
Z Z
1 1 1 2
þ a B1 2ð2 þ mÞa22 B02
4 1 1 1
2 c2 x2 ec2 x2 cos a3 x3 cos a1 x1 da3 da1
þ a21 B
Z Z
1 1 1 2
þ a C1 2ð2 þ mÞa23 C20
4 1 1 1
2 c3 x3 ec3 x3 cos a1 x1 cos a2 x2 da1 da2 ; . . .;
þ a21 C ð10:8Þ
432 10 Elastic Eighth-Space. Elastic Quarter-Space
Z Z
1 1 1 1 2 2
r23 ¼ a2 a3 A1 þ c41 A02 c21 ð2a23 ma22 ÞA002
4 1 1 a2 a3
c x
þ ð2a22 ma23 ÞA000 2 2
2 þ a 2 a 3 A2 c 1 x 1 e
1 1
sin a2 x2 sin a3 x3 da2 da3
Z 1Z 1
1 c2 2
a B1 ða23 ma21 ÞB02 þ ð1 þ mÞa23 þ ma21 B002
4 1 1 a3 3
c x
ð3 þ mÞa23 þ 2a21 B000 2
2 þ a 3 B2 c 2 x 2 e
2 2
sin a3 x3 cos a1 x1 da3 da1
Z 1Z 1
1
c3 2
a2 C1 ða22 ma21 ÞC20 ð3 þ mÞa22 þ 2a21 C200
4 1 1 a2
þ ð1 þ mÞa22 þ a21 C2000 þ a22 C 2 c3 x3 ec3 x3 cos a1 x1 sin a2 x2 da1 da2 ; . . .;
ð10:80 Þ
Putting the boundary conditions for the tangential stresses and taking into
account (10.80 ), one obtains
1 þ ð1 þ mÞa2 þ a2 A0 ð3 þ mÞa2 þ 2a2 A00 ða2 ma2 ÞA000 ¼ 0; . . .;
a22 A 2 3 2 2 3 2 2 3 2
1 þ ð1 þ mÞa2 þ a2 A0 ða2 ma2 ÞA00 ð3 þ mÞa2 þ 2a2 A000 ¼ 0; . . .;
a23 A 3 2 2 3 2 2 2 2 2
wherefrom
1 ¼ 2 þ m a2 þ 2ma2 A00 a2 þ 2ma2 A000 ; . . .;
A 3 2 2 2 3 2
a23 a22
ð10:10Þ
1
A02 ¼
2 2
2a23 þ ma22 A002 2a22 þ ma23 A000
2 ; . . .;
m a3 a 2
we may write
2 ¼
2 þ m a2 A00 a2 A000 ; . . .
A ð10:100 Þ
2 2 3 2 2 2
m a3 a2
if results
Z 1 2
2 a1
ð1 c1 x1 Þec1 x1 ¼ c1 cos a1 x1 da1 ; . . .;
p a4
Z1
1 ð10:1300 Þ
2 1
ð1 c1 x1 Þec1 x1 ¼ c31 4
cos a1 x1 da1 ; . . .
p 1 a
Taking into account the relations (10.13), (10.1300 ), the representation (10.12)
leads to
Z Z
1 1 1
r11 ¼ c C1 ð1 þ c1 x1 Þ ec1 x1 cos a2 x2 cos a3 x3 da2 da3
4 1 1 1
Z Z Z 1 2
1 1 1 a2
þ C2 a21 4
cos a2 x2 da2
2p 1 1 1 a
Z 1
1
þma23 2
cos a2 x2 da2 cos a3 x3 cos a1 x1 da3 da1
1 a
Z Z Z 1 2
1 1 1 a1
þ C3 a21 4
cos a3 x3 da3
2p 1 1 1 a
Z 1
1
þma22 2
cos a3 x3 da3 cos a1 x1 cos a2 x2 da1 da2 ; . . . ð10:1400 Þ
1 a
and
i.e.,
un1 ða2 ; a3 ; x1 Þ ! u01 ða2 ; a3 ; x1 Þ; . . .; ð10:180 Þ
Taking into account the isomorphism of the distributions space and that of the
Fourier transforms, one obtains the equations
Z
2 1 a21 2
c1 C1 ða2 ; a3 Þ þ 4
a2 C2 ða3 ; a1 Þ þ a23 C3 ða1 ; a2 Þ da1
p 1 a
Z
2m 1 1 2
þ 2
a3 C2 ða3 ; a1 Þ þ a23 C3 ða1 ; a2 Þ da1 ¼ a1 ða2 ; a3 Þ; . . . ð10:190 Þ
p 1 a
If there exists a system of even bounded functions C1 ða2 ; a3 Þ, C2 ða3 ; a1 Þ,
C3 ða1 ; a2 Þ which verifies the system of integral equations (10.190 ), then the state of
stress in the elastic eighth-space is given by the formulae (10.12); as it can be seen,
the boundary conditions are verified in the sense of the uniform convergence of
distributions.
We remark that the system (10.190 ), considered as homogeneous, is indepen-
dent on the loading of the elastic eighth-space, this one appearing only in the right-
hand member of the equations.
In what follows we deal only with distributions, i.e., the case in which
Ci ; i ¼ 1; 2; 3, are even, bounded, locally integrable functions. As in the cor-
responding plane case [5], one can show that the system of equations (10.190 ) has a
unique solution in the class of distributions.
Z
2 1
c1 C2 ða2 ; a3 Þ þ Kða3 ; a1 ; a2 ÞC3 ða3 ; a1 Þda1
p 1
Z 1
2
þ Kða2 ; a3 ; a1 ÞC1 ða1 ; a2 Þda1 ¼ a2 ða2 ; a3 Þ;
p 1
Z ð10:210 Þ
2 1
c1 C3 ða2 ; a3 Þ þ Kða3 ; a1 ; a2 ÞC1 ða3 ; a1 Þda1
p 1
Z
2 1
þ Kða2 ; a3 ; a1 ÞC2 ða1 ; a2 Þda1 ¼ a3 ða2 ; a3 Þ:
p 1
Subtracting the Eqs. (10.21), written in the form (10.210 ), two by two, and taking
into account (10.22), one obtains the system
Z
2 1
c1 ½C1 ða2 ; a3 Þ C2 ða3 ; a2 Þ þ Kða3 ; a1 ; a2 Þ½C2 ða3 ; a1 Þ
p 1
Z 1
2
C1 ða1 ; a3 Þda1 þ Kða2 ; a3 ; a1 Þ½C3 ða1 ; a2 Þ C3 ða2 ; a1 Þda1 ¼ 0;
p 1
Z ð10:26Þ
2 1
c2 ½C3 ða1 ; a2 Þ C3 ða2 ; a1 Þ þ Kða2 ; a3 ; a1 Þ½C1 ða2 ; a3 Þ
p 1
Z
2 1
C2 ða3 ; a2 Þda3 þ Kða1 ; a2 ; a3 Þ½C2 ða3 ; a1 Þ C1 ða1 ; a3 Þda3 ¼ 0;
p 1
which leads to
C1 ða2 ; a3 Þ ¼ C2 ða3 ; a2 Þ; C3 ða1 ; a2 Þ ¼ C3 ða2 ; a1 Þ: ð10:260 Þ
This result represents the mathematical consequence of certain mechanical
properties; one remarks thus that, for an element of the bisecting plane x1 ¼ x2 one
may write
1
r ¼ ðr11 þ r22 Þ r12 ; ð10:27Þ
2
pffiffiffi
2 1
rkx3 ¼ ðr23 r31 Þ; r? x3 ¼ ðr11 r22 Þ; ð10:270 Þ
2 2
where the normal stress and the tangential stresses parallel and normal to the Ox3 -
axis, respectively, are put in evidence; the formulae (2.2.1) and (2.2.4) have been
used. For these tangential stresses, one obtains
pffiffiffi Z 1
2 a2 a3
rkx3 ¼ ½ð1 2m c1 x1 Þec1 x1 C1 ða2 ; a3 Þ sin a2 x2
8 1 c1
ð1 2m c1 x2 Þec1 x2 C2 ða3 ; a2 Þ sin a2 x1 sin a3 x3 da2 da3
pffiffiffi Z 1 Z 1
2
a3 c1 ½x1 ec1 x1 C1 ða2 ; a3 Þ cos a2 x2 x2 ec1 x2 C2 ða3 ; a2 Þ cos a2 x1
8 1 1
sin a3 x3 da2 da3
pffiffiffi Z 1 Z 1
2
a1 c3 x3 ec3 x3 ½C3 ða1 ; a2 Þ sin a1 x1 cos a2 x2
8 1 1
C3 ða2 ; a1 Þ cos a2 x1 sin a1 x2 da1 da1 ; ð10:28Þ
10.1 Elastic Eighth-Space 439
Z Z ("
1 1 1 a3 2
r?x3 ¼ c1 ð12mc1 x1 Þ
8 1 1 c1
2 # "
a2 c1 x1 a3 2
þ2 c1 x 1 e C1 ða2 ;a3 Þcosa2 x2 ð12mþc1 x2 Þ
c1 c1
2 # )
a2
þ2 c1 x2 ec1 x2 C2 ða3 ;a2 Þcosa2 x1 cosa3 x3 da2 da3
c1
Z Z " 2 #
1 1 1 a1 2 a2
c3 ð1c3 x3 Þþ2m ec3 x3 ½C3 ða1 ;a2 Þcosa1 x1 cosa2 x2
8 1 1 c3 c3
C3 ða2 ;a1 Þcosa2 x1 cosa1 x2 da1 da2 ð10:280 Þ
These tangential stresses must vanish for a symmetric loading (one makes
x1 ¼ x2 ), which leads necessarily to the relations (10.260 ). This must constitute, at
the same time, a demonstration of the uniqueness of the solution of the system
(10.26).
In particular, if the bisecting plane x2 ¼ x3 is, as well, a symmetry plane for the
loading, i.e., if one has the relations
a2 ða3 ; a1 Þ ¼ a3 ða1 ; a3 Þ; a1 ða2 ; a3 Þ ¼ a1 ða3 ; a2 Þ ð10:29Þ
too, then the third bisecting plane x3 ¼ x1 must also be a plane of symmetry for the
loading; indeed, there result the relations
a3 ða1 ; a2 Þ ¼ a1 ða1 ; a1 Þ; a2 ða3 ; a1 Þ ¼ a2 ða1 ; a3 Þ; ð10:290 Þ
Taking into account the relations (10.30), the equations allow to write
440 10 Elastic Eighth-Space. Elastic Quarter-Space
Z
2 1
c1 ½C1 ða2 ; a3 Þ þ C2 ða3 ; a2 Þ þ Kða3 ; a1 ; a2 Þ½C1 ða1 ; a3 Þ
p 1
Z 1
2
þ C2 ða3 ; a1 Þda1 þ Kða2 ; a3 ; a1 Þ½C3 ða1 ; a2 Þ þ C3 ða2 ; a1 Þda1 ¼ 0;
p 1
Z
2 1
c3 ½C3 ða1 ; a2 Þ þ C3 ða2 ; a1 Þ þ Kða2 ; a3 ; a1 Þ½C1 ða2 ; a3 Þ
p 1
Z
2 1
þ C2 ða3 ; a2 Þda3 þ Kða1 ; a2 ; a3 Þ½C1 ða1 ; a3 Þ þ C2 ða3 ; a1 Þda3 ¼ 0;
p 1
ð10:31Þ
which leads to
C1 ða2 ; a3 Þ ¼ C2 ða3 ; a2 Þ; C3 ða1 ; a2 Þ ¼ C3 ða2 ; a1 Þ: ð10:310 Þ
The normal stress (10.27) on the bisecting plane x1 ¼ x2 will be of the form
Z Z ("
1 1 1
a3 2
r¼ c1 ð1 þ 2m þ a1 x1 Þ
8 1 1 c1
2 # "
a2 a3 2
þ2 ec1 x1 C1 ða2 ; a3 Þ cos a2 x2 þ ð1 þ 2m þ c1 x2 Þ
c1 c1
2 # )
a2 c1 x2
þ2 e C2 ða3 ; a2 Þ cos a2 x1 cos a3 x3 da2 da3
c1
Z 1Z 1
1
a2 c1 ½x1 ec1 x1 C1 ða2 ; a3 Þ sin a2 x2 þ x2 ec1 x2 C2 ða3 ; a2 Þ sin a2 x1 cos a3 x3 da2 da3
8 1 1
Z Z " 2 #
1 1 1 a1 2 a2
þ c ð1 c3 x3 Þþ2m ec3 x3 ½C3 ða1 ; a2 Þ cos a1 x1 cos a2 x2
8 1 1 3 c3 c3
þ C3 ða2 ; a1 Þ cos a2 x1 cos a1 x2 da1 da2
Z Z
1 1 1 a1 a2
þ ð1 2m c3 x3 Þec3 x3 ½C3 ða1 ; a2 Þ sin a1 x1 sin a2 x2
8 1 1 c3
þ C3 ða2 ; a1 Þ sin a2 x1 sin a1 x2 da1 da2 :
ð10:3100 Þ
the axis x1 ¼ x2 ¼ x3 being thus a symmetry axis of third order. This case is
contained in the general one considered above.
If the functions a1 ða2 ; a3 Þ; a2 ða3 ; a1 Þ; a3 ða1 ; a2 Þ are even with respect to the
variables a2 ; a3 or a3 ; a1 or a1 ; a2 , respectively, the system (10.21) shows that the
functions C1 ða2 ; a3 Þ; C2 ða3 ; a1 Þ; C3 ða1 ; a2 Þ have the same property. Thus, the
system (10.21) may be written in the form
Z
4 1
c1 C1 ða2 ; a3 Þ þ K ða3 ; a1 ; a2 ÞC2 ða3 ; a1 Þda1
p 0
Z 1
4
þ K ða2 ; a3 ; a1 ÞC3 ða1 ; a2 Þda1 ¼ a1 ða2 ; a3 Þ; . . . ð10:33Þ
p 0
In the case of a symmetry with respect to the axis x1 ¼ x2 ¼ x3 , we are led to
the integral equation
Z
4 1
c 1 C 1 ð a2 ; a3 Þ þ K ða3 ; a1 ; a2 ÞC1 ða3 ; a1 Þda1
p 0
Z
4 1
þ K ða2 ; a3 ; a1 ÞC1 ða1 ; a2 Þda1 ¼ a1 ða2 ; a3 Þ: ð10:34Þ
p 0
If, in this last case, we decompose the loading with respect to the properties of
symmetry or of antisymmetry with respect to the bisecting planes x1 ¼ x2 ;
x2 ¼ x3 ; x3 ¼ x1 , one can replace the Eq. (10.34) by the equations
Z
4 1
c 1 C 1 ð a2 ; a3 Þ K ða3 ; a1 ; a2 ÞC1 ða1 ; a3 Þda1
p 0
Z
4 1
K ða2 ; a3 ; a1 ÞC1 ða2 ; a1 Þda1 ¼ a1 ða2 ; a3 Þ; ð10:35Þ
p 0
the upper signs corresponding to the symmetry, while the lower ones correspond to
the antisymmetry.
In general, an arbitrary case of loading may be decomposed in two cases of
loading if one considers the properties of symmetry and of antisymmetry with
respect to the plane x1 ¼ x2 ; we have studied above these two cases of loading.
Using the relations (10.260 ) and (10.310 ), one may replace the system (10.33) by
two subsystems, having each one two integral equations
Z
4 1
c 1 C 1 ð a2 ; a 3 Þ K ða3 ; a1 ; a2 ÞC1 ða1 ; a3 Þda1
p 0
Z 1
4
þ K ða2 ; a3 ; a1 ÞC3 ða1 ; a2 Þda1 ¼ a1 ða2 ; a3 Þ;
p 0
Z ð10:36Þ
4 1
c 3 C 3 ð a1 ; a 2 Þ þ K ða2 ; a3 ; a1 ÞC1 ða2 ; a3 Þda3
p 0
Z
4 1
K ða1 ; a2 ; a3 ÞC1 ða1 ; a3 Þda3 ¼ a3 ða1 ; a2 Þ:
p 0
442 10 Elastic Eighth-Space. Elastic Quarter-Space
The system with the upper signs corresponds to the symmetric loading, while
that with lower signs corresponds to the antisymmetric loading. Although one must
integrate two systems of integral equations, this procedure presents the advantage
that any subsystem has only two equations, possessing a form of the same degree
of difficulty as the Eqs. (10.33).
In the case of a total symmetry (symmetry with respect to all bisecting planes)
or of a total antisymmetry (with respect to the same planes), the system (10.36) is
reduced to the Eqs. (10.35).
The Eqs. (10.36) allow to express C3 ða1 ; a2 Þ with respect to C1 ða2 ; a3 Þ in the
form
Z
1 4 1
C3 ða1 ; a2 Þ ¼ a3 ð a1 ; a2 Þ K ða2 ; a3 ; a1 ÞC1 ða2 ; a3 Þda3
c3 p 0
Z 1
4
K ða1 ; a2 ; a3 ÞC1 ða1 ; a3 Þda3 : ð10:37Þ
p 0
Eliminating C3 ða1 ; a2 Þ between the Eqs. (10.36), one obtains two integral equa-
tions of the form
Z 1 Z 1
C1 ða2 ; a3 Þ K1 ða2 ; a3 ; a1 ÞC1 ða1 ; a3 Þda1 þ K2 ða2 ; a3 ; bÞC1 ða2 ; bÞdb
Z 1 Z 10 0
1
Kða2 ; a3 ; a1 ; bÞC1 ða1 ; bÞda1 db ¼ a1 ða2 ; a3 Þ Aða2 ; a3 Þ;
0 0 c 1
ð10:38Þ
with the notations
4 4 a21 a22 ma23
K1 ða2 ; a3 ; a1 Þ ¼ K ð a3 ; a1 ; a2 Þ ¼ þ 2 ; ð10:39Þ
pc1 pc1 a4 a
" Z Z 1 #
16ma22 2 1 a21 da1 2 da1
K2 ða2 ;a3 ;bÞ¼ 2 a
þma2
2
p c1 3 0 c a4 c2 þb2 2 0 c3 a4 c2 þb2
3 3 3
" Z Z 1 #
16b2 2 1 a41 da1 a 2
da 1
2 a
þma22
1
2
p c1 3 0 c3 a4 c23 þb2 0 c3 a4 c2 þb2
3
( "
8 a23 c21 þb2 ðb2 þa22 Þ a23 b2
c þa2
¼ 2
2 2 2 þ
3
c1 4a22 þa23 þ3b2 ln 1
p c1 a3 b a2 a23 b2 a3
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi #
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a22 þb2 þa2
2 2 2 2 2
a2 þb 4a2 þ3a3 þb ln
b
10.1 Elastic Eighth-Space 443
(
ma2 1 c þa2
þ 2 2 2ð1mÞa22 þa23 ln 1
a3 b c1 a3
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi ))
2ð1mÞa22 þb2 a22 þb2 þa2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffi ln ð10:390 Þ
b
a22 þb2
16
Kða2 ; a3 ; a1 ; bÞ ¼ ða21 a23
pa4 c1 c3 ðc21 þ b2 Þ
þ ma22 a2 Þ ma21 ðc21 þ b2 Þ þ a22 b2 ; ð10:3900 Þ
Z 1 2
1 a1
Aða2 ; a3 Þ ¼ a2 a1 ða2 ; a1 Þda1
pc1 3 0 c3 a4
Z 1
2 1
þma2 a ða ; a Þda1 :
4 1 2 1
ð10:39000 Þ
0 c3 a
We reduced thus the problem to the integration of two integral equations having
ðiÞ
each one only one unknown function. The sequence of functions C1 ða2 ; a3 Þ; i ¼
0; 1; 2; . . . with
ð0Þ 1
C 1 ð a2 ; a3 Þ ¼ a 1 ð a 2 ; a 3 Þ Að a 2 ; a 3 Þ ð10:40Þ
c1
and
Z 1
ðiþ1Þ 1 ðiÞ
C1 ða 2 ; a 3 Þ ¼ a1 ða2 ; a3 Þ Aða2 ; a3 Þ K1 ða2 ; a3 ; a1 ÞC1 ða1 ; a3 Þda1
c1 0
Z 1 Z 1Z 1
ðiÞ ðiÞ
K2 ða2 ; a3 ; bÞC1 ða2 ; bÞdb K ða2 ; a3 ; a1 ; bÞC1 ða1 ; bÞda1 db
0 0 0
ð10:400 Þ
may approximate the solution of the equations (10.38).
If a1 ða2 ; a3 Þ; a2 ða3 ; a1 Þ; a3 ða1 ; a2 Þ are even functions with respect to the variables
a2 ; a3 or a3 ; a1 or a1 ; a2 , respectively, then one may write the normal loads (10.5)
in the form
Z 1Z 1
p1 ð x 2 ; x 3 Þ ¼ a1 ða2 ; a3 Þ cos a2 x2 cos a3 x3 da2 da3 ; . . . ð10:41Þ
0 0
By means of the U functions introduced in Sect. A.4.1.1, the state of stress reads
444 10 Elastic Eighth-Space. Elastic Quarter-Space
Z 1 Z 1
r11 ¼ c1 C1 ða2 ; a3 ÞU2 ðc1 x1 Þ cos a2 x2 cos a3 x3 da2 da3
0 0
Z 1Z 1 " 2 #
a1 2 a3
þ c2 C2 ða3 ; a1 Þ U 4 ð c 2 x2 Þ þ m U6 ðc2 x2 Þ cos a3 x3 cos a1 x1 da3 da1
0 0 c2 c2
Z 1Z 1 " 2
a1
þ c3 C3 ða1 ; a2 Þ U 4 ð c 3 x3 Þ
0 0 c3
2 #
a2
þm U4 ðc3 x3 Þ cos a1 x1 cos a2 x2 da1 da2 ; . . .
c3
ð10:42Þ
Z 1Z 1
a 2 a3
r23 ¼ C1 ða2 ; a3 ÞU46 ðc1 x1 Þ sin a2 x2 sin a3 x3 da2 da3
c1
Z0 1 Z0 1
þ a3 C2 ða3 ; a1 ÞU3 ðc2 x2 Þ sin a3 x3 cos a1 x1 da3 da1
0 0
Z 1 Z 1
þ a2 C3 ða1 ; a2 ÞU3 ðc3 x3 Þ cos a1 x1 sin a2 x2 da1 da2 ; . . . ð10:420 Þ
0 0
where the motion of rigid body may be specified by certain conditions of fixity. For
instance, if the displacement and the rigid body rotation of the elastic eighth-space
vanish, then the rigid body rotations vanish too, while the rigid body displacements
are given by
Z Z
0 1m 1 1
u1 ¼ C1 ða2 ; a3 Þ cos a2 x2 cos a3 x3 da2 da3 ; . . . ð10:430 Þ
l 0 0
One obtains remarkable results for the separation plans. Thus, the normal
stresses are given by
10.1 Elastic Eighth-Space 445
Z Z "
1 1
a2 2
r11 ðx1 ; x2 ; 0Þ ¼ p1 ðx1 ; x2 Þ þ c1 C1 ða2 ; a3 Þ ð1 2m þ c1 x1 Þ
0 0 c1
2 #
a3
þ2 c1 x1 ec1 x1 cos a2 x2 da2 da3
c1
Z 1Z 1 " 2 #
a3
þ c2 C2 ða3 ; a1 Þ 1 2 ð1 2m c2 x2 Þec2 x3 cos a1 x1 da3 da1
0 0 c1
Z 1Z 1 2
a2
ð1 2mÞ c3 C3 ða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2 ;
0
Z 1Z 1
0 "c3 #
a2 2
r22 ðx1 ; x2 ; 0Þ ¼ p1 ðx1 ; x2 Þ þ c1 C1 ða2 ; a3 Þ 1 2 ð1 2m
0 0 c1
Z 1Z 1 "
a3 2
c1 x1 Þec1 x1 cos a2 x2 da2 da3 þ c2 C2 ða3 ; a1 Þ ð1 2m þ a2 x2 Þ
0 0 c1
2 #
a1
þ2 a2 x2 ea2 x2 cos a1 x1 da3 da1
c2
Z 1Z 1 2
a1
ð1 2mÞ c3 C3 ða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2
0 0 c3
ð10:44Þ
in the separation plane x3 ¼ 0, while the tangential stress reads
Z 1Z 1
r23 ðx1 ; x2 ; 0Þ ¼ x1 a2 c1 C1 ða2 ; a3 Þec1 x1 sin a2 x2 da2 da3
0 Z 0 Z
1 1
þ x2 a1 c2 C2 ða3 ; a1 Þec2 x2 sin a1 x1 da3 da1
0 0Z
1Z 1
a1 a2
ð1 2mÞ C3 ða1 ; a2 Þ sin a1 x1 sin a2 x2 da1 da2 :
0 0 c3
ð10:440 Þ
What concerns the state of strain, the displacements in the plane x3 ¼ 0 are given by
Z Z
1 1 1
0
u1 ðx1 ; x2 ; 0Þ ¼ u1 C1 ða2 ; a3 Þ½2ð1 mÞ þ c1 x1 ec1 x1 cos a2 x2 da2 da3
2l 0 0
Z Z
1 1 1 a1
þ C2 ða3 ; a1 Þð1 2m c2 x2 Þec2 x2 sin a1 x1 da3 da1
2l 0 c
Z 0 Z2
1 2m 1 1 a1
þ C3 ða1 ; a2 Þ sin a1 x1 cos a2 x2 da1 da2 ;
2l c3
Z 10 Z 10
1 a2
u2 ðx1 ; x2 ; 0Þ ¼ u02 þ C1 ða2 ; a3 Þð1 2m c1 x1 Þec1 x1 sin a2 x2 da2 da3
2l 0 0 c1
Z Z
1 1 1
C2 ða3 ; a1 Þ½2ð1 mÞ þ c2 x2 ec2 x2 cos a1 x1 da3 da1
2l 0 0
Z Z
1 2m 1 1 a2
C3 ða1 ; a2 Þ cos a1 x1 sin a2 x2 da1 da2 ; ð10:45Þ
2l 0 0 c3
446 10 Elastic Eighth-Space. Elastic Quarter-Space
with the notations (10.70 ). The stress functions in two variables are taken equal to
zero. The unknown parameters are integrable functions, which must be specified
by boundary conditions; the conditions at infinite are automatically verified,
because we deal with Fourier transforms.
Calculating the normal and tangential stresses, as in the preceding subsection,
and putting the boundary conditions for the normal stresses, we get the relations
1 þ 2 ð1 þ mÞA0 A00 þ A000 ¼ 0; . . .;
A ð10:49Þ
2 2 2
1 c41
C10 ða2 ; a3 Þ ¼ 2
A2 ;
ð 2 þ m Þ a2 a3
1 c42
C20 ða3 ; a1 Þ ¼ B2 ; ð10:50Þ
ð2 þ mÞ2 a3 a1
1 c43
C30 ða1 ; a2 Þ ¼ 2
C2 ;
ð 2 þ m Þ a1 a2
1 c21
2 00
C100 ða2 ; a3 Þ ¼ a A a22 A000
2 ;
2 þ m a2 a3 3 2
1 c22
2 000
C200 ða3 ; a1 Þ ¼ a1 B2 a23 B02 ; ð10:500 Þ
2 þ m a3 a1
1 c23
2 0
C300 ða1 ; a2 Þ ¼ a2 C2 a21 C200 ;
2 þ m a1 a2
the components of the stress tensor read
Z Z
1 1 1 a2 a3 0
r11 ¼ C ða2 ; a3 Þc1 x1 ec1 x1 sin a2 x2 sin a3 x3 da2 da3
4 1 1 c21 1
Z Z (
1 1 1 a3 a1 a1 2
þ ½2ð1 þ mÞ c2 x2 C20 ða3 ; a1 Þ
4 1 1 c22 c22
)
2C200 ða3 ; a1 Þ ec2 x2 sin a3 x3 sin a1 x1 da3 da1
Z Z (
1 1 1
a1 a2 a1 2
2
½2ð1 þ mÞ c3 x3 C30 ða1 ; a2 Þ
4 1 1 c3 c3
)
þ 2C3000 ða1 ; a2 Þ ec3 x3 sin a1 x1 sin a2 x2 da1 da2 ; . . .; ð10:51Þ
448 10 Elastic Eighth-Space. Elastic Quarter-Space
Z Z ( ( 2 )
1 1
1 a2 a3
r23 ¼ C10 ða2 ; a3 Þ 1 ½ 2ð 1 þ m Þ c 1 x 1
4 1 1 c21
a22 a23 00
þ C1 ða2 ; a3 Þ ec1 x1 cos a2 x2 cos a3 x3 da2 da3
c21
Z Z (" #
1 1 1 a1 a3 2 a23 a21 0
þ ð1 c2 x2 Þ þ m C2 ða3 ; a1 Þ
4 1 1 c2 c2 c22
)
þ C200 ða3 ; a1 Þ ec2 x2 sin a3 x3 sin a1 x1 da3 da1
Z Z (" #
1 1 1
a1 a2 2 a21 a22 0
þ ð1 c3 x3 Þ m C3 ða1 ; a2 Þ
4 1 1 c1 c3 c23
)
C300 ða1 ; a2 Þ ec3 x3 sin a1 x1 cos a2 x2 da1 da2 ; . . . ð10:510 Þ
Taking into account (10.13), one can use the representations (10.1300 ), by means
of even functions or the representations
Z
1 1 a1
ec1 x1 ¼ sin a1 x1 da1 ; . . .;
p 1 a2
Z 1 ð10:13000 Þ
c1 x1 2 a1
x1 e ¼ c1 4
sin a1 x1 da1 ; . . .;
p 1 a
ð1 þ mÞa22 ma23 0 1
C1 ða2 ; a3 Þ þ C100 ða2 ; a3 Þ
c31 c1
Z 2 2
1 1 1 a1 a2
2 2
0
þ 2 4 m a3 a1 C2 ða3 ; a1 Þda1
p 1 c22 a2 a
Z 1
1 1 00
C ða3 ; a1 Þda1
p 1 a2 2
Z 1
1 1 a21 a22 a23
þ m 2 4 m þ 2 C30 ða1 ; a2 Þda1
p 1 a2 c3 a
Z 1 2 2
1 a1 a2 00 1
þ 2 2
C3 ða1 ; a2 Þda1 ¼ a01 ða2 ; a3 Þ; . . . ð10:52Þ
p 1 c3 a a3
10.1 Elastic Eighth-Space 449
ð1 þ mÞa23 ma22 0 1
3
C1 ða2 ; a3 Þ C100 ða2 ; a3 Þ
c1 c1
Z 1
Z
1 1 2 2
a1 a3 a22 0 1 1 a23 a21 00
þ m 2 m þ C ða
2 3 1 ; a Þda 1 þ C2 ða3 ; a1 Þda1
p 1 a2 c42 a2 p 1 c22 a2
Z 2 2
1 1 1 a1 a3
2 2
0
þ 2 2 þ m a1 a2 C3 ða1 ; a2 Þda1
p 1 a2 c23 a
Z 1
1 1 00 1
þ C ða1 ; a2 Þda1 ¼ a001 ða2 ; a3 Þ; . . . ð10:520 Þ
p 1 a2 3 a2
If there exists a system of bounded symmetric functions C10 ða2 ; a3 Þ;
C100 ða2 ; a3 Þ; . . .;
C300 ða1 ; a2 Þ, which verifies the system of integral equations (10.52),
0
(10.52 ), then the state of stress in the elastic eighth-space is given by (10.51),
(10.510 ); the boundary conditions are verified in the sense of the uniform con-
vergence of the distributions.
We remark that the system (10.52), (10.520 ), considered as a homogeneous one,
is independent on the loading of the elastic eighth-space with tangential loads,
because these ones intervene only in the right-hand member.
In what follows we suppose to be in the class of generalized functions, i.e. in
the case of which C10 ða2 ; a3 Þ; C20 ða2 ; a3 Þ; . . .; C300 ða1 ; a2 Þ are locally integrable
bounded even functions. As in the previous case, the system (10.52), (10.520 ) has a
unique solution in the class of distributions.
If the functions a01 ða2 ; a3 Þ; a001 ða2 ; a3 Þ; . . .; a003 ða1 ; a2 Þ are even with respect to
the variable a2 and odd with respect to the variable a3 or with respect to a3 ; a1 or to
a1 ; a2 , respectively, then the system (10.52), (10.520 ) shows that the functions
C10 ða2 ; a3 Þ; C100 ða2 ; a3 Þ; . . .; C300 ða1 ; a2 Þ are even with respect to the variables a2 ; a3
or a3 ; a1 or a1 ; a2 , respectively.
Summing and subtracting the corresponding equations of the system (10.52),
(10.520 ) and taking into account the above remarks, one obtains the equivalent
system
Z Z
1 0 4 1 a41 a22 0 4 1 a21 00
C1 ða2 ; a3 Þ þ m þ C ða 3 ; a 1 Þda 1 C ða3 ; a1 Þda1
c1 p 0 c42 a2 a2 2 p 0 c22 a2 2
Z Z
4 1 a41 a23 0 4 1 a21 00
þ m þ C ða 1 ; a 2 Þda 1 þ C ða1 ; a2 Þda1
p 0 c43 a2 a2 3 p 0 c23 a2 3
1 1
¼ a01 ða2 ; a3 Þ þ a001 ða2 ; a3 Þ; ð10:53Þ
a3 a2
450 10 Elastic Eighth-Space. Elastic Quarter-Space
a22 a23 0 2
ð1 þ 2mÞ 2
C1 ða2 ; a3 Þ þ C100 ða2 ; a3 Þ
c1 k
Z 1 2 2
4
1 a1 a 2 2 2
4 0
þ a 1 þ 2a 3 ma 3 C2 ða3 ; a1 Þda1
p 0 c42 a2 a2
Z Z
4 1 a23 00 4 1 1 a21 a23
2
C ða3 ; a1 Þda1 a1 þ 2a2 ma2 C30 ða1 ; a2 Þda1
2 4
p 0 c22 a2 2 p 0 c43 a2 a2
Z
4 1 a22 00
C ða1 ; a2 Þda1
p 0 c23 a2 3
1 1
¼ a01 ða2 ; a3 Þ a001 ða2 ; a3 Þ; . . . ð10:530 Þ
a3 a2
Let be, e.g., the case of a loading with an axial symmetry of third order with
respect to the x1 ¼ x2 ¼ x3 axis; this symmetry is characterized by the relations
the upper signs corresponding to the symmetric case, while the lower ones cor-
respond to the antisymmetric case. The remarks of mechanical order which can be
made, analogous to those in the case of a normal loading, for an element of the
452 10 Elastic Eighth-Space. Elastic Quarter-Space
Using the notations in Sect. A.4.1.1, we may express the state of stress in the form
Z 1Z 1
a2 a3 0
r11 ¼ C ða2 ; a3 ÞU3 ðc1 x1 Þ sin a2 x2 sin a3 x3 da2 da3
0 0 c21 1
Z 1Z 1 "
a1 a3 a1 2 0
C2 ða3 ; a1 ÞU56 ðc2 x2 Þ
0 0 c22 c2
#
þ C200 ða3 ; a1 ÞU6 ða2 x2 Þ sin a3 x3 sin a1 x1 da3 da1
Z Z "
1 1
a1 a2 a1 2 0
C3 ða1 ; a2 ÞU56 ðc3 x3 Þ
0 0 c23 c3
#
C300 ða1 ; a2 ÞU6 ða3 x3 Þ sin a1 x1 sin a2 x2 da1 da2 ; ð10:60Þ
Z Z (
1 1
a2 a3 2 0 1
r23 ¼ C3 ða2 ; a3 ÞU56 ðc1 x1 Þ þ mC10 ða2 ; a3 Þ
0 0 c21 2
)
a2 a2
þ 2 2 3 C100 ða2 ; a3 Þ U6 ðc1 x1 Þ cos a2 x2 cos a3 x3 da2 da3
c1
Z 1 Z 1 ( 2
a1 a3 0 1 a23 a21 0
þ C2 ða3 ; a1 ÞU4 ðc2 x2 Þ þ m C2 ða3 ; a1 Þ
0 0 c2 c2 2 c22
)
þ C200 ða3 ; a1 Þ U6 ðc2 x2 Þ cos a3 x3 sin a1 x1 da3 da1
Z Z (
1 1
a1 a2 2 0 1 a21 a22 0
þ C3 ða1 ; a2 ÞU4 ðc3 x3 Þ m C3 ða1 ; a2 Þ
0 0 c3 c3 2 c23
)
þ C300 ða1 ; a2 Þ U6 ðc3 x3 Þ sin a1 x1 cos a2 x2 da1 da2 ; . . .; ð10:600 Þ
Z Z
1 1 1 a2 a3 0
u1 ¼u01 x2 x03 þx3 x02 þ C ða2 ;a3 ÞU26 ðc1 x1 Þsina2 x2 sina3 x3 da2 da3
2l 0 0 c31 1
Z Z ( 2
1 1 1 a3 a1
þ C20 ða3 ;a1 ÞU56 ðc2 x2 Þþ mC20 ða3 ;a1 Þ
2l 0 0 c22 c2
)
00
þC2 ða3 ;a1 Þ U6 ðc2 x2 Þ sina3 x3 cosa1 x1 da3 da1
Z (
1Z 1
1 a2 a1 2 0
þ C3 ða1 ;a2 ÞU56 ðc3 x3 Þ þ mC30 ða1 ;a2 Þ
2l 0 0 c23 c3
)
C300 ða1 ;a2 Þ U6 ðc3 x3 Þ cosa1 x1 sina2 x2 da1 da2 ;...; ð10:61Þ
where the motion of rigid body must be specified by certain conditions of fixity.
For instance, if the displacement and the rigid body rotation of the elastic eighth-
space vanish, then the rigid body displacements vanish too, while the rigid body
local rotations are given by
Z Z
1 1 1 a22 a23 0
x01 ¼ 2
mC1 ða2 ; a3 Þ C100 ða2 ; a3 Þ da2 da3 ; . . . ð10:610 Þ
2l 0 0 c 1
One obtains
Z Z " 2
1 1
a 1 a2 a1
r11 ðx1 ; x2 ; 0Þ ¼ 2 ð1 þ m Þ C30 ða1 ; a2 Þ
0 0 c23 c3
#
þ C300 ða1 ; a2 Þ sin a1 x1 sin a2 x2 da1 da2 ;
Z Z " 2 ð10:62Þ
1 1
a 1 a2 a2
r22 ðx1 ; x2 ; 0Þ ¼ 2 2
ð1 þ m Þ C30 ða1 ; a2 Þ
0 0 c3 c3
#
C300 ða1 ; a2 Þ sin a1 x1 sin a2 x2 da1 da2 ;
for the normal stresses in the plane x3 ¼ 0 (one of the separation planes), while the
corresponding tangential stress reads
Z Z (" #
1 1 1 a1 a2 2 0
r12 ðx1 ; x2 ; 0Þ ¼ m 2ð 1 þ m Þ C1 ða1 ; a2 Þ
m 0 0 c23
)
a21 a22 00
þ C3 ða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2 : ð10:620 Þ
c23
10.1 Elastic Eighth-Space 455
What concerns the state of strain, the displacements in the plane x3 ¼ 0 are
given by
u1 ðx1 ; x2 ; 0Þ ¼ x2 x03
Z Z
1 1 1 a2 ma22 a21 0 00
þ C3 ða 1 ; a2 Þ 2C 3 ð a1 ; a 2 Þ cos a1 x1 sin a2 x2 da1 da2 ;
l 0 0 c23 c23
u2 ðx1 ; x2 ; 0Þ ¼ x1 x03
Z Z
1 1 1 a1 ma21 a22 0 00
þ C3 ða 1 ; a2 Þ þ 2C 3 ð a1 ; a 2 Þ sin a1 x1 cos a2 x2 da1 da2 ;
l 0 0 c23 c23
ð10:63Þ
while the displacement normal to this plane may be written in the form
Z Z
1 2m 1 1 a1 a2 0
u3 ðx1 ; x2 ; 0Þ ¼ x1 x02 þ x2 x01 ¼ C ða1 ; a2 Þ sin a1 x1 sin a2 x2 da1 da2
2l 0 0 c23 3
Z Z ( " 2 2 #
1 1 1 a2 a2 a3
þ 2m ð2 c1 x1 Þ C10 ða2 ; a3 Þ
2l 1 1 c21 c1 c1
) Z Z ("
00 c1 x1 1 1 1 a3 a1 2
þ 2C1 ða2 ; a3 Þ e sin a2 x2 da2 da3 þ 2m
2l 1 1 c22 c2
2 # )
a3
ð2 c2 x2 Þ C20 ða3 ; a2 Þ 2C200 ða3 ; a1 Þ ec2 x2 sin a1 x1 da3 da1 :
c2
ð10:630 Þ
planes (neglecting the rigid body rotation x03 ). One may thus solve the mixed
boundary value problem in which on the separation planes are given the tangential
displacements, the normal stresses vanishing.
We consider two particular cases which are not contained in the above general
study; we deal than with an interesting application.
t t
u1 ¼ x 3 ; u3 ¼ x1 : ð10:650 Þ
2l 2l
10.1.4.2 Application
We consider now the elastic eighth-space acted upon by a normal load symmetric
with respect to the axis x1 ¼ x2 ¼ x3 and with respect to the bisecting planes
x2 ¼ x3 ; x1 ¼ x2 ; x2 ¼ x3
p
p1 ð x 2 ; x 3 Þ ¼ ; . . .; ð10:66Þ
r13
and negative in the exterior of this cone; the stresses r33 vanish on the conical
surface. One can make analogous remarks for the stresses r11 and r22 . In partic-
ular, all three normal stresses vanish on the axis x1 ¼ x2 ¼ x3 :
458 10 Elastic Eighth-Space. Elastic Quarter-Space
The maximum normal stresses (positive) are obtained on the Ox3 -axis; for
x3 ¼ x03 one has
rffiffiffi
max 12 3 p p
r33 ¼
ffi 0:372
0 3 ð10:74Þ
25 5 x0 3 x
3 3
on the cylinder r3 ¼ 2x03 . Along the Ox3 -axis, the normal stress has a hyperbolical
variation
p
r33 ð0; 0; x3 Þ ¼ 2 : ð10:7400 Þ
x33
1. for r3 ¼ 0 a maximum
p
rmax
33 ¼ 2 ; ð10:75Þ
ðx03 Þ3
and positive in the exterior of it; obviously, it vanishes on it. We remark that the
cone C0 is interior to the cone C. Along the Ox3 -axis ðr3 ¼ 0Þ, the normal stress
has a hyperbolical variation
p
rr3 r3 ¼ : ð10:76Þ
R3
For x3 ¼ x03 one obtains a minimum (negative)
16 pffiffiffi p p
rmin ¼ 5 0 3 ffi 0:286 0 3 ; ð10:760 Þ
r r
3 3 125 ðx3 Þ ðx3 Þ
on the cylinder 2r3 ¼ x03 . The maximum normal stresses are obtained in the
exterior of the cone C 0 ; for x3 ¼ x03 , one obtains
rffiffiffi
max 8 2 p p
rr3 r3 ¼ ffi 0:202 0 3 ð10:7600 Þ
25 5 ðx03 Þ3 ðx3 Þ
1. for r3 ¼ 0 a minimum
p
rmin
r3 r3 ¼ ; ð10:77Þ
ðx03 Þ3
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffi
2. for r3 ¼ ð1=2Þ 6 30x03 ffi 0:36x03 a point of inflection
pffiffiffiffiffi
8 30 4 p p
rr3 r3 ¼ pffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffi 0 3 ffi 0:545 0 3 ; ð10:770 Þ
5 ð13 2 30Þ 10 30 ðx3 Þ ðx3 Þ
pffiffiffi
3. for r3 ¼ 2=2x03 ffi 0:71x03 the normal stress vanishes;
pffiffiffiffiffiffiffiffi 0
4. for r3 ¼ 3=2x3 ffi 1:22x03 a maximum
rffiffiffi
max 8 2 p p
rr3 r3 ¼ ffi 0:202 0 3 ; ð10:7700 Þ
25 5 ðx03 Þ3 ðx3 Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffi 0
5. r3 ¼ ð1=2Þ 6 þ 30x3 ffi 1:69x03 a point of inflection
pffiffiffiffiffi
8 4 þ 30 p p
rmin
r3 r3 ¼ pffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffi ffi 0:161 0 3 ; ð10:77000 Þ
5 ð13 þ 2 30Þ 10 þ 30 ðx03 Þ3 ðx3 Þ
Concerning the tangential stress r12 which acts in the plane x3 ¼ x03 we remark
that one obtains a maximum at the intersection of the common tangent of the cones
r1 ¼ 2x1 and r2 ¼ 2x2 with this plane, hence at the point x1 ¼ x2
pffiffiffi
¼ ð 3=3Þx03 ffi 0:58x03 ; it results
rffiffiffi
max 9 3 p p
r12 ¼ 3
ffi 0:279 0 3 : ð10:78Þ
25 5 ðx3 Þ
0
ðx3 Þ
Varying x3 , one obtains a maximum
x1 x2
rmax
12 ¼ 3p : ð10:780 Þ
r15
Analogically, one can consider the stress r31 in the plane x3 ¼ x03 ; one observes
that at the point x1 ¼ x03 =2; x2 ¼ 0 one obtains
48 pffiffiffi p p
rmax
31 ¼ 5 0 3 ffi 0:859 0 3 : ð10:79Þ
125 ðx3 Þ ðx3 Þ
A variation along the variable x3 , we have
48 pffiffiffi p p
rmax
31 ¼ 5 0 3 ffi 0:859 0 3 ð10:790 Þ
125 ðx1 Þ ðx1 Þ
rffiffiffi
48 3 p p
r3r3 ¼ ffi 0:758 0 3 ; ð10:800 Þ
49 5 ðx03 Þ3 ðx3 Þ
4. for r3 ! 1 the tangential stress tends asymptotically to zero.
The relations (10.72), (10.720 ) of the stresses in spherical co-ordinates allow to
specify the principal directions of the stress tensor. The principal direction which
leads to a positive maximum for rR is given by the radius R; any other direction
normal to the radius R is a principal direction (leading to a minimum ruu ¼ rhh ),
the quadric of Cauchy being a two sheet hyperboloid of rotation. The relations
(10.7800 ) of the displacement show that the displacement vector is directed along
the vector radius R.
We also remark that the vertex of the elastic eighth-space is a singular point. It
is thus useful to eliminate this point, by cutting up an eighth-sphere of radius R0 ,
having the centre at this point; this sphere will be acted upon only by the normal
stresses rRR ¼ rR0 . Projecting these stresses on the Ox3 -axis and summing for the
eighth-sphere, one obtains the resultant
Z
p p
rR0 cos udA0 ¼ ; ð10:81Þ
A0 2 R0
in the negative sense of Ox3 , taking into account that the elementary area is given
by dA0 ¼ R20 sin ududh. As well, the total load on x3 ¼ 0 is given by
Z 1 Z p=2
p p p
RdRdh ¼ ; ð10:810 Þ
R0 0 R3 2 R0
being directed in the positive sense of Ox3 . A relation of global equilibrium is thus
obtained.
Projecting all the stresses which act on the eighth-sphere on the straight-line L
of equation x1 ¼ x2 ¼ x3 and summing, one obtains the resultant
Z pffiffiffi
p 3 p
rR0 cosðR; LÞd A0 ¼ ; ð10:82Þ
A0 2 R0
Z 1 Z p=2
p
r0r3 r3 r30 sin hdx3 dh ¼ ; ð10:84Þ
0 0 r30
in the negative sense of this axis; as well, the total load on x1 ¼ 0 is given by
Z 1Z 1
p p
dx dx ¼ 0 ;
3 2 3
ð10:840 Þ
0 r 3 r3
0 r3
in the positive sense of the Ox1 -axis. A relation of global equilibrium is thus put in
evidence. The resultant of the load along the bisectrix of the solid angle x1 ¼
x2 ¼ 0 is
pffiffiffi
p 2 p
2 0 ffi 1:41 0 : ð10:85Þ
r3 2 r3
Throwing to infinite one of the faces of the elastic eighth-space, i.e. the face
x3 ¼ 0, one obtains the elastic quarter-space x1 0; x2 0. We study the action of
normal or tangential loads, particularizing the results obtained in the previous
section.
As well, one can adapt to this domain the results obtained in Sect. 10.1.1.1
concerning the action of a periodic load. We conclude that the problem of the
elastic quarter-space acted upon by an arbitrary load, periodic on two directions,
may be reduced, by means of the principle of superposition of effects, to the study
of the same domain acted upon by a load, local on the semiinfinite direction and
periodic on the infinite one, and to the study of an elastic half-space, analogically
acted upon. This case of loading of an elastic quarter-space cannot be reduced to a
simpler one. We also mention the case of the elastic quarter-space acted upon by a
load, local on the infinite direction and periodic on the infinite one, which may be
reduced to the study of an elastic half-space, analogically loaded. In the following,
we will consider only the case of local loads, using the results given by us for
normal loads [8, 10], for tangential loads [11. 14] and for a certain particular case
of loading [9, 13].
Z 1 Z 1
p1 ðx2 ; x3 Þ ¼ a1 ða2 ; a3 Þ cos a2 x2 cos a3 x3 dx2 dx3 ;
Z0 1 Z0 1 ð10:86Þ
p2 ðx3 ; x1 Þ ¼ a2 ða3 ; a1 Þ cos a3 x3 cos a1 x1 dx3 dx1
0 0
(10.11), we have C3 ða1 ; a2 Þ ¼ 0 too. We may thus obtain all the results for the
elastic quarter-space, starting from the results in Sect. 10.1.2.
Z 1 Z 1
a2 a3
r23 ¼ C1 ða2 ; a3 ÞU46 ðc1 x1 Þ sin a2 x2 sin a3 x3 da2 da3
0 0 c1
Z 1 Z 1
þ a3 C2 ða3 ; a1 ÞU3 ðc2 x2 Þ sin a3 x3 cos a1 x1 da3 da1 ;
0 0
Z 1Z 1
r31 ¼ a3 C1 ða2 ; a3 ÞU3 ðc1 x1 Þ cos a2 x2 sin a3 x3 da2 da3
00
Z 1Z 1 ð10:880 Þ
a3 a1
C2 ða3 ; a1 ÞU46 ðc2 x2 Þ sin a3 x3 sin a1 x1 da3 da1 ;
0 0 c2
Z 1Z 1
r12 ¼ a2 C1 ða2 ; a3 ÞU3 ðc1 x1 Þ sin a2 x2 cos a3 x3 da2 da3
0 0
Z 1Z 1
þ a1 C2 ða3 ; a1 ÞU3 ðc2 x2 Þ cos a3 x3 sin a1 x1 da3 da1 ;
0 0
Z 1 Z 1 2
a3
r22 ð0; x2 ; x3 Þ ¼ p1 ðx2 ; x3 Þ ð1 2mÞ c1 C1 ða2 ; a3 Þ cos a2 x2 cos a3 x3 da2 da3
0 c1
Z Z "0
1 1
a3 2
þ c2 C2 ða3 ; a1 Þ ð1 2m þ c2 x2 Þ
0 0 c2
2 #
a1
þ2 c2 x2 ec2 x2 cos a3 x3 da3 da1 ; ð10:90Þ
c2
Z 1Z 1 2
a2
r33 ð0; x2 ; x3 Þ ¼ p1 ðx2 ; x3 Þ ð1 2mÞ c1 C1 ða2 ; a3 Þ cos a2 x2 cos a3 x3 da2 da3
0 0 c1
Z 1Z 1 " #
a3 2
þ c2 C2 ða3 ; a1 Þ 1 2 ð1 2m c2 x2 Þec2 x2 cos a3 x3 da3 da1
0 0 c2
We also remark that the unknown functions C1 ða2 ; a3 Þ; C2 ða3 ; a1 Þ are in direct
proportion to the Fourier coefficients of the displacements normal to the separation
planes u1 ð0; x2 ; x3 Þ; u2 ðx1 ; 0; x3 Þ, respectively (neglecting the displacements of
rigid body u01 ; u02 ), which confers them a certain physical significance. On the other
hand, this allows to solve the mixed fundamental problem in which, on the sep-
aration planes are given the normal stresses, the tangential stresses vanishing.
466 10 Elastic Eighth-Space. Elastic Quarter-Space
By means of the kernel (10.20), the system of integral equations which is verified
by the unknown functions Ci ; i ¼ 1; 2, may be written in the form
Z
4 1
c1 C1 ða2 ; a3 Þ þ Kða3 ; a1 ; a2 ÞC2 ða3 ; a1 Þda1 ¼ a1 ða2 ; a3 Þ;
p 0
Z 1 ð10:92Þ
4
c2 C2 ða3 ; a1 Þ þ Kða3 ; a1 ; a2 ÞC1 ða2 ; a3 Þda2 ¼ a2 ða3 ; a1 Þ:
p 0
We observe that the third integral equation (10.33) does not intervene; as a matter
of fact, it may be considered as verified, because the function a3 ða1 ; a2 Þ may be
considered as arbitrary.
One remarks that the system (10.92) has some properties of symmetry, leading
to a simplification of the computation. In general, an arbitrary loading can be
decomposed in two cases of loading, corresponding to the properties of symmetry
or antisymmetry with respect to the plane x1 ¼ x2 , which are characterized by the
relations
a1 ða2 ; a3 Þ ¼ a2 ða3 ; a2 Þ; ð10:93Þ
one obtains thus
C1 ða2 ; a3 Þ ¼ C2 ða3 ; a2 Þ; ð10:930 Þ
the upper signs corresponding to the symmetric case, while the lower ones to the
antisymmetric one. The system (10.92) may be thus replaced by two integral
equations
Z
4 1
c1 C1 ða2 ; a3 Þ Kða3 ; a1 ; a2 ÞC1 ða1 ; a3 Þda1 ¼ a1 ða2 ; a3 Þ; ð10:94Þ
p 0
each one of a single unknown.
The Eqs. (10.92) allow to express C2 ða3 ; a1 Þ with regard to C1 ða2 ; a3 Þ in the
form
Z
1 4 1
C2 ða3 ; a1 Þ ¼ a2 ða3 ; a1 Þ Kða3 ; a1 ; a2 ÞC1 ða2 ; a3 Þda2 : ð10:95Þ
c2 p 0
Eliminating C2 ða3 ; a1 Þ between the integral equations (10.92), we get an inte-
gral equation of the form
Z 1
1
C1 ða2 ; a3 Þ þ K 0 ða2 ; a3 ; bÞC1 ðb; a3 Þdb ¼ a1 ða2 ; a3 Þ þ A0 ða2 ; a3 Þ; ð10:96Þ
0 c 1
Z 1 2 Z 1
4 a1 a2 ða3 ; a1 Þ a2 ða3 ; a1 Þ
A0 ða2 ; a3 Þ ¼ a22 4
da 1 þ ma 2
3 2
da 0
1 : ð10:97 Þ
pc1 0 c 2 a 0 c 2 a
Thus, we reduce the problem to the integration of a single integral equation with
ðiÞ
only one unknown function. The sequence of functions C1 ða2 ; a3 Þ; i ¼ 0; 1; 2; . . .,
with
1 ð0Þ
a1 ða2 ; a3 Þ þ A0 ða2 ; a3 Þ;
C1 ¼ ð10:98Þ
c1
Z 1
ðiþ1Þ 1 0 ðiÞ
C1 ða2 ; a3 Þ ¼ a1 ða2 ; a3 Þ þ A ða2 ; a3 Þ K 0 ða2 ; a3 ; bÞC1 ðb; a3 Þdb;
c1 0
ð10:980 Þ
Using the results obtained in Sect. 10.1.3, we get the state of stress
Z 1Z 1
a2 a3 0
r11 ¼ C ða2 ; a3 ÞU3 ðc1 x1 Þ sin a2 x2 sin a3 x3 da2 da3
0 0 c21 1
Z 1Z 1 "
a3 a1 a1 2 0
C2 ða3 ; a1 ÞU56 ðc2 x2 Þ
0 0 c22 c2
#
þ C200 ða3 ; a1 ÞU6 ðc2 x2 Þ sin a3 x3 sin a1 x1 da3 da1 ;
Z Z "
1 1
a2 a3 a2 2 0
r22 ¼ C1 ða2 ; a3 ÞU56 ðc1 x1 Þ
0 0 c21 c1
#
C100 ða2 ; a3 ÞU6 ðc1 x1 Þ sin a2 x2 sin a3 x3 da2 da3
Z 1 Z 1 ð10:102Þ
a3 a1 0
þ C ða3 ; a1 ÞU3 ðc2 x2 Þ sin a3 x3 sin a1 x1 da3 da1 ;
0 0 c22 2
Z 1Z 1 "
a2 a3 a3 2 0
r33 ¼ C1 ða2 ; a3 ÞU56 ðc1 x1 Þ
0 0 c21 c1
#
þ C100 ða2 ; a3 ÞU6 ðc1 x1 Þ sin a2 x2 sin a3 x3 da2 da3
Z Z "
1 1
a3 a1 a3 2 0
C2 ða3 ; a1 ÞU56 ðc2 x2 Þ
0 0 c22 c2
#
C200 ða3 ; a1 ÞU6 ðc2 x2 Þ sin a3 x3 sin a1 x1 da3 da1
10.2 Elastic Quarter-Space 469
Z Z ( "
1 1
a2 a3 2 0 1
r23 ¼ C1 ða2 ; a3 ÞU56 ðc1 x1 Þ þ mC10 ða2 ; a3 Þ
0 0 c21 2
# )
a22 a23 00
þ C2 ða2 ; a3 Þ U6 ðc1 x1 Þ cos a2 x2 cos a3 x3 da2 da3
c21
Z 1 Z 1 ( 2 "
a1 a3 1 a2 a2
þ C20 ða3 ; a1 ÞU4 ðc2 x2 Þ þ m 3 2 1 C20 ða3 ; a1 Þ
0 0 c2 c2 2 c2
# )
þ C200 ða3 ; a1 Þ U6 ðc2 x2 Þ cos a3 x3 sin a1 x1 da3 da1 ;
Z Z ( "
1 1
a2 a3 2 0 1 a22 a23 0
r31 ¼ C1 ða2 ; a3 ÞU4 ðc1 x1 Þ m C1 ða2 ; a3 Þ
0 0 c1 c1 2 c21
# )
þ C100 ða2 ; a3 Þ U6 ðc1 x1 Þ sin a2 x2 cos a3 x3 da2 da3 ð10:1020 Þ
Z Z ( "
1 1
a3 a1 2 0 1
þ C2 ða3 ; a1 ÞU56 ðc2 x2 Þ þ mC20 ða3 ; a1 Þ
0 0 c22 2
# )
a23
a21 00
þ C2 ða3 ; a1 Þ U6 ðc2 x2 Þ cos a3 x3 cos a1 x1 da3 da1 ;
c22
Z 1 Z 1 ( 2 "
a3 a2 0 1 a22 a23 0
r12 ¼ C1 ða2 ; a3 ÞU4 ðc1 x1 Þ þ m C1 ða2 ; a3 Þ
0 0 c1 c1 2 c21
# )
þ C100 ða2 ; a3 Þ U6 ðc1 x1 Þ cos a2 x2 sin a3 x3 da2 da3
Z Z ( "
1 1
a3 a1 2 0 1 a23 a21 0
þ C2 ða3 ; a1 ÞU4 ðc2 x2 Þ m C2 ða3 ; a1 Þ
0 0 c2 c2 2 c22
# )
þ C200 ða3 ; a1 Þ U6 ðc2 x2 Þ sin a3 x3 cos a1 x1 da3 da1
Z 1 Z 1
1 a2 a3 0
u1 ¼ x3 x02 þ C ða2 ; a3 ÞU26 ðc1 x1 Þ sin a2 x2 sin a3 x3 da2 da3
0 2l
0 c31 1
Z Z ( 2
1 1 1 a3 a1
þ 2
C20 ða3 ; a1 ÞU56 ðc2 x2 Þ þ ½mC20 ða3 ; a1 Þ
2l 0 0 c 2 c2
)
þ C200 ða3 ; a1 ÞU6 ðc2 x2 Þ sin a3 x3 cos a1 x1 da3 da1 ;
Z (
Z
1 1
a2 2 0
1
a3
u2 ¼ x3 x01 þ C1 ða2 ; a3 ÞU56 ðc1 x1 Þ
2l 0 0 c1 c21
)
0 00
þ mC1 ða2 ; a3 Þ C1 ða2 ; a3 Þ U6 ðc1 x1 Þ cos a2 x2 sin a3 x3 da2 da3
Z 1 Z 1
a3 a1 0
þ C ða3 ; a1 ÞU26 ðc2 x2 Þ sin a3 x3 sin a1 x1 da3 da1 ; ð10:103Þ
0 0 c32 2
Z Z (
0 0 1 1 1 a2 a3 2 0
u3 ¼ x 2 x 1 x 1 x 2 þ 2
C1 ða2 ; a3 ÞU56 ðc1 x1 Þ
2l 0 0 c1 c1
)
0 00
þ mC1 ða2 ; a3 Þ þ C1 ða2 ; a3 Þ U6 ðc1 x1 Þ sin a2 x2 cos a3 x3 da2 da3
Z Z (
1 1 1
a3 2 0
a1
þ C2 ða3 ; a1 ÞU56 ðc2 x2 Þ
2l 0 0 c2c22
)
0 00
þ mC2 ða3 ; a1 Þ C2 ða3 ; a1 Þ U6 ðc2 x2 Þ cos a3 x3 sin a1 x1 da3 da1 ;
where the motion of rigid body was specified (only x01 ; x02 6¼ 0).
The normal stresses corresponding to the plane x1 ¼ 0 are
Z 1Z 1 " 2
a2 a3 a2
r11 ð0; x2 ; x3 Þ ¼ 2 2
ð1 þ mÞ C10 ða2 ; a3 Þ
0 0 c1 c1
#
C100 ða2 ; a3 Þ sin a2 x2 sin a3 x3 da2 da3 ;
Z Z " 2 ð10:104Þ
1 1
a2 a3 a3
r22 ð0; x2 ; x3 Þ ¼ 2 2
ð1 þ mÞ C10 ða2 ; a3 Þ
0 0 c1 c1
#
C100 ða2 ; a3 Þ sin a2 x2 sin a3 x3 da2 da3 ;
10.2 Elastic Quarter-Space 471
We notice also here that the unknown functions C10 ða2 ; a3 Þ; C100 ða2 ; a3 Þ;
C20 ða3 ; a1 Þ;
C200 ða3 ; a1 Þ are in direct proportion to the displacements tangential to
the separation planes (neglecting the rigid rotations x01 ; x02 ). This allows the
solving of the mixed fundamental problem in which on the separation planes are
given the tangential displacements, the normal stresses vanishing.
By means of the kernels (10.54), one may write the system of integral equations
for the unknown functions C10 ða2 ; a3 Þ; C100 ða2 ; a3 Þ; C20 ða3 ; a1 Þ; C200 ða3 ; a1 Þ in the
form
472 10 Elastic Eighth-Space. Elastic Quarter-Space
Z Z
1 0 4 1 4 1
C1 ða2 ; a3 Þ þ K1 ða1 ; a2 ; a3 ÞC20 ða3 ; a2 Þda1 K3 ða1 ; a2 ; a3 ÞC200 ða3 ; a1 Þda1
c1 p 0 p 0
1 1
¼ a01 ða2 ; a3 Þ þ a001 ða2 ; a3 Þ; ð10:106Þ
a3 a
Z 12 Z 1
1 0 4 4
C ða3 ; a1 Þ þ K1 ða2 ; a1 ; a3 ÞC10 ða2 ; a3 Þda2 þ K3 ða2 ; a1 ; a3 ÞC100 ða2 ; a3 Þda2
c2 2 p 0 p 0
1 1
¼ a02 ða3 ; a1 Þ þ a002 ða3 ; a1 Þ;
a1 a3
Z
a2 a 2 2 4 1
ð1 þ 2mÞ 2 2 3 C10 ða2 ; a3 Þ þ C100 ða2 ; a3 Þ þ K2 ða1 ; a2 ; a3 ÞC20 ða3 ; a1 Þda1
c1 c1 p 0
Z
4 1 1 1
K3 ða3 ; a2 ; a1 ÞC200 ða3 ; a1 Þda1 ¼ a01 ða2 ; a3 Þ a001 ða2 ; a3 Þ; ð10:1060 Þ
p 0 a3 a2
Z
a23 a21 0 2 00 4 1
ð1 þ 2mÞ C2 ða3 ; a1 Þ þ C1 ða2 ; a3 Þ K2 ða2 ; a1 ; a3 ÞC10 ða2 ; a3 Þda2
c22 c2 p 0
Z
4 1 1 1
K3 ða3 ; a1 ; a2 ÞC100 ða2 ; a3 Þda2 ¼ a02 ða3 ; a1 Þ a002 ða3 ; a1 Þ:
p 0 a1 a3
This system of four integral equations has a unique solution in the class of
generalized functions.
An arbitrary case of loading may be decomposed in two cases of loading after
the properties of symmetry or of antisymmetry with respect to the plane x1 ¼ x2 ,
which are characterized by the relations
a01 ða2 ; a3 Þ ¼ a002 ða3 ; a2 Þ; a001 ða2 ; a3 Þ ¼ a02 ða3 ; a2 Þ: ð10:107Þ
It results
C10 ða2 ; a3 Þ ¼ C200 ða3 ; a2 Þ; C100 ða2 ; a3 Þ ¼ C200 ða3 ; a2 Þ; ð10:1070 Þ
the upper signs corresponding to the symmetric case, while the lower ones the
antisymmetric one. In this case, the system of equations (10.106), (10.1060 ) may be
replaced by two subsystems
Z Z
1 0 4 1 0 4 1
C ða2 ;a3 Þ K1 ða1 ;a2 ;a3 ÞC1 ða1 ;a3 Þda1 K3 ða1 ;a2 ;a3 ÞC100 ða1 ;a3 Þda1
c1 1 p 0 p 0
1 1
¼ a01 ða2 ;a3 Þ þ a001 ða2 ;a3 Þ; ð10:108Þ
a3 a2
Z
a2 a2 2 4 1
ð1 þ 2mÞ 2 3 3 C10 ða2 ;a3 Þ þ C100 ða2 ;a3 Þ K2 ða1 ;a2 ;a3 ÞC10 ða1 ;a3 Þda1
c1 c1 p 0
Z
4 1 1 1
K3 ða3 ;a2 ;a1 ÞC100 ða1 ;a3 Þda1 ¼ a01 ða2 ;a3 Þ a001 ða2 ;a3 Þ;
p 0 a3 a2
having each one only two unknown functions, which, obviously, is an advantage
for the computation; the upper signs correspond to the symmetric case and the
10.2 Elastic Quarter-Space 473
lower ones to the antisymmetric one. Eliminating, e. g., C20 ða3 ; a1 Þ and
C200 ða3 ; a1 Þbetween the Eqs. (10.106), (10.1060 ), one obtains a system of two
integral equations having C10 ða2 ; a3 Þ and C100 ða2 ; a3 Þ as unknown functions; this
system may be integrated by successive approximations.
The case of a loading symmetric with respect to the axes Ox1 and Ox2 , of the
form
Z 1Z 1
t12 ðx2 ; x3 Þ ¼ a01 ða2 ; a3 Þ cos a2 x2 cos a3 x3 da2 da3 ;
0 0
Z 1Z 1 ð10:109Þ
00
t13 ðx2 ; x3 Þ ¼ a1 ða2 ; a3 Þ sin a2 x2 sin a3 x3 da2 da3 ;
0 0
Z 1 Z 1
t21 ðx3 ; x1 Þ ¼ a02 ða3 ; a1 Þ sin a3 x3 sin a1 x1 da3 da1 ;
0 0
Z 1 Z 1 ð10:1090 Þ
t23 ðx3 ; x1 Þ ¼ a002 ða3 ; a1 Þ cos a3 x3 cos a1 x1 da3 da1 ;
0 0
Let us deal with two particular cases which are not contained in the above general
theory. As well we shell consider two applications.
10.2.3.2 Applications
We will deal with two cases of normal loading, where we put in evidence prop-
erties of symmetry or antisymmetry with respect to the bisecting plane x1 ¼ x2 ,
obtaining results in a finite form.
The symmetric load
p
p1 ðx2 ; x3 Þ ¼ ð3 2mÞx22 þ 4mx23 ;
r15
p ð10:114Þ
p2 ðx3 ; x1 Þ ¼ 5 ð3 2mÞx21 þ 4mx23
r2
leads to
2p 2
a1 ða2 ; a3 Þ ¼ 2a2 þ ð1 þ 2mÞa23 ;
pc1
ð10:1140 Þ
2p 2
a2 ða3 ; a1 Þ ¼ 2a1 þ ð1 þ 2mÞa23 ;
pc2
the Eq. (10.94) with the upper sign leads to
2
C1 ða2 ; a3 Þ ¼ C2 ða3 ; a1 Þ ¼ p: ð10:115Þ
p
One obtains the state of stress
x 2 x 2 x 2
p 2m 3 1 3
r11 ¼ 3 3 1 ð1 2mÞ 35 ;
R 3 R R R
x 2 x 2 x 2
p 2m 3 2 3
r22 ¼ 3 3 1 ð1 2mÞ 35 ; ð10:116Þ
R 3 R R R
x 2 x 4
p 3 3
r33 ¼ 3 1 þ 2m 6ð2 þ mÞ þ15 ;
R R R
10.2 Elastic Quarter-Space 475
x 2
x2 x3 3
r23 ¼ 3p 3 þ 2m 5 ;
R5 R
x 2
x3 x1 3
r31 ¼ 3p 5 3 þ 2m 5 ; ð10:1160 Þ
R R
x 2
x1 x2 3
r12 ¼ 3p 5 3 5
R R
3p x2 x3 x21 x22
r23 ¼ 2m þ 5 ;
1 2m R5 R2
3p x3 x1 x21 x22 ð10:1260 Þ
r31 ¼ 2m 5 ;
1 2m R5 R2
15p x1 x2
2
r12 ¼ 7
x1 x22
1 2m R
and the state of displacement
p x1 3 x21 x22
u1 ¼ 2þ ;
2l R3 1 2m R2
p x2 3 x21 x22
u2 ¼ 2 ; ð10:127Þ
2l R3 1 2m R2
3p x3
2
u3 ¼ x x22 ;
2ð1 2mÞl R5 1
In cylindrical co-ordinates one obtains the normal stresses
z 2 z 2
p 3 r3 2
rr3 r3 ¼ 3 1 3 þ 35 cos 2h;
R R 1 2m R R
z 2
p
rhh ¼ 3 1 3 cos 2h; ð10:128Þ
R R
r32 5 z 4
rzz ¼ 3p 5 1 cos 2h
R 1 2m R
References
A. Books
1. Teodorescu, P.P.: Probleme spatßiale în teoria elasticitătßii (Spatial Problems in the Theory of
Elasticity). Ed. Acad. Rom., Bucuresßti, (1970)
B. Papers
2. Hijab, W.A.: Concentrated force in eighth-space with mixed boundary conditions. J. für reine
angew. Math. 225, 85 (1967)
3. Teodorescu, P.P.: Fonctions de tension dans le problème tridimensionnel de la théorie de
l’élasticité. Bull. Math. Soc. Sci. Math. Phys. Roum. 3(51), 499 (1959)
References 479
}
4. Teodorescu, P.P.: Uber einige räumliche Probleme der Elastizitätstheorie. Aplik. Mat. Praha.
4, 225 (1959)
5. Teodorescu, P.P.: Sur le problème du quart de plan élastique. Aplik. Mat. Praha 6, 359 (1961)
6. Teodorescu, P.P.: Sur le problème du demi–quart d’espace èlastique. Bull. Acad. Pol. Sci.
Sér. Sci. Techn. 12, 599 (1964)
7. Teodorescu, P.P.: Sur le problème du demi–quart d’espace élastique soumis à l’action d’une
charge tangentielle. Rev. Roum. Sci. Techn. Sér.. Méc. Appl. 10, 81 (1965)
8. Teodorescu, P.P.: Sur l’effet de coin en élasticité tridimensionnelle. Arch. Mech. Stos. 17,
249 (1962)
9. Teodorescu, P.P.: Sur le calcul du quart d’espace élastique. Rev. Roum. Sci. Techn., sér.
Méc. Appl. 11, 1141 (1966)
10. Teodorescu, P.P.: Trei probleme tridimensionale ale teoriei elasticitătßii (Three tridimensional
problems of the theory of elasticity). An. Univ. Bucuresßti, ser. ßst. nat., mat.–mec. 15, 17
(1966)
11. Teodorescu, P.P.: Sur le problème du quart d’espace élastique. I. Charges normales. Bull.
Acad. Pol. Sci., sér. Sci. Techn. 14, 487 (1966)
12. Teodorescu, P.P.: Sur le problème du quart d’espace élastique. II. Charges trangentielles.
Bull. Acad. Pol. Sci., sér. Sci. Techn. 14, 497 (1966)
13. Teodorescu, P.P.: Asupra utilizării unor familii de functßii potentßial reale în teoria elasticitătßii
(On the use of certain families of real potential functions in the theory of elasticity). An.
Univ. Bucuresßti, ser. ßst. nat., mat.–mec. 16, 59 (1967)
14. Teodorescu, P.P.: Einige Teilprobleme der dreidimensionalen Elastizitätstheorie. Wiss.
Zeitschsift Techn. Univ. Dresden 16, 77 (1967)
15. Teodorescu, P.P.: Sur le problème du demi–quart d’espace élastique soumis à l’action d’une
charge tangentielle. In: Problems of Fluid Flow Machines, Panst. wgdaw. nauk., Warszawa,
pp. 883 (1968)
16. Teodorescu, P.P.: Sur l’application des fonctions potentiel dans la théorie de l’élasticité.
Rend. Sem. Mat. Fis., Milano. 38, 231 (1968)
Chapter 11
Elastic Parallelepiped. Elastic Strip.
Elastic Layer. Thick Plate
The first fundamental problem for an elastic parallelepiped has been formulated, in
1852, by G. Lamé [4], who considered it to be very important both from the
theoretical and the practical point of view, emphasizing—at the same time—the
difficulties which appear in its solving. In his book, G. Lamé stimulates the young
researchers to deal with this problem, for which he only sketches some results. At
the same time, twice (in 1846 and 1858), the Academy of Sciences in Paris
proposed this problem, endowing it with a prize; it became a classical problem of
the theory of elasticity.
This problem has not been sufficiently studied. There have been various con-
siderations: theoretical, as well experimental; from the last point of view, the
researchers dealt with the problem of compression of the concrete cubes. We
mention thus the classical treatise of A. Föppl and L. Föppl [2].
The Russian school of elasticity tackled this problem by various approximate
methods of computation. We mention thus the papers of M. M. Filonenko-Borodich
[14–18], based on a variational method of Castigliano and on a choice of a system
of functions which have been previous considered by the same author [1]; one used
triple Fourier series, obtaining results in the first or in the second approximation for
some cases of loading. A. I. Meshkov [28, 29] takes again this methods of com-
putation for the oblique parallelepiped, considering loads which lead to torsion too.
The same method of computation has been applied by E. S. Kononenko [24, 25] to
study the state of stress of concrete cubes subjected to compression. V. P. Netrebko
[32, 33] deals with the torsion of the elastic parallelepiped acted upon by tangential
loads. Ē. N. Baı̆da [9–11], using a method of computation analogous to that indi-
cated by Boussinesq and Galerkin, makes some considerations concerning the
isotropic and the anisotropic elastic parallelepiped. M. Mishonov [31] deals with
the elastic parallelepiped acted upon by arbitrary volume forces, using triple
Fourier series; by a certain procedure, one also obtains results for superficial loads
as a limit case.
In what follows, we will use the results given by us [38] and presented at the
1960 IUTAM Congress.
To determine the state of strain and stress in the elastic parallelepiped thus acted
upon, we will introduce stress functions of the form (where the simple series which
intervene, as well as the free terms are put in evidence)
X
F11 ¼ L01 x21 þ L02 x22 þ L03 x23 þ A01m0 cosh bm x1 þ bm x1 A02m0 sinh bm x1 cos bm x2
m
X
þ A010n cosh cn x1 þ cn x1 A020n sinh cn x1 cos cn x3
n
XX
þ A01mn cosh kmn x1 þ kmn x1 A02mn sinh kmn x1 cos bm x2 cos cn x3
m n
X
þ B01n0 cosh cn x2 þ cn x2 B02n0 sinh cn x2 cos cn x3
n
X
þ B010l cosh al x2 þ al x2 B020l sinh al x2 cos al x1
l
XX
þ B01nl cosh lnl x2 þ lnl x2 B02nl sinh lnl x2 cos cn x3 cos al x1
n l
X
0 0
þ C1l0 cosh al x3 þ al x3 C2l0 sinh al x3 cos al x1
l
11.1 Elastic Parallelepiped 483
X
0 0
þ C10m cosh bm x3 þ bm x3 C20m sinh bm x3 cos bm x2
m X
X
0 0
þ C1lm cosh mlm x3 þ mlm x3 C2lm sinh mlm x3 cos al x1 cos bm x2 ; . . .;
l lm
where ð11:1Þ
lp mp np
al ¼ ; bm ¼ ; cn ¼ ; l; m; n ¼ 1; 2; 3; . . .; ð11:2Þ
a1 a2 a3
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
kmn ¼ b2m þ c2n ; lnl ¼ c2n þ a2l ;
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð11:20 Þ
mlm ¼ a2l þ b2m ; l; m; n ¼ 1; 2; 3; . . .;
X
þlnl x2 a2l B2nl sinh lnl x2 cos cn x3 cos al x1 þ a2l f½C1l0
0
l
þ2ð2 þ mÞC2l0 cosh al x3 þ al x3 C2l0 sinh al x3 cos al x1
X
0
þ2ð2 þ mÞ b2m C20m cosh bm x2
X X m
0
þ a2l C1lm þ 2ð2 þ mÞm2lm C2lm cosh mlm x3
l m
þmlm x3 a2l C2lm sinh mlm x3 cos al x1 cos bm x2 ;. .. ð11:5Þ
and the tangential stresses
X
r23 ¼ x3 b3m mðA02m0 þ A002m0 Þ A000
2m0 coshbm x1 sinbm x2
m
X X X 1
þ x2 c2n mðA020n þ A000 00
20n Þ 2A20n coshcn x1 sincn x3 b2m c2n A1mn
m m n
bm cn
mk4mn A02mn þ k2mn ð2c2n mb2m ÞA002mn þ k2mn ð2b2m mc2n ÞA000
2mn coshkmn x1
X
þkmn x1 b2m c2n A2mn sinhkmn x1 sinbm x2 sincn x3 þ c2n B1n0 þ B02n0
n
ð1 þ mÞB002n0 þð3 þ mÞB000 2n0 sinhcn x2 þ cn x2 B2n0 coshcn x2 sincn x3
X X X l
x3 a2l mðB020l þ B0020l Þ B000
20l sinhal x2 cosal x1 þ
nl
c2n B1nl
l n l
cn
þðc2n ma2l ÞB02nl ½ð1þ mÞc2n þ ma2l B002nl þ ð3 þ mÞc2n þ 2a2l B000
2nl sinhlnl x2
X
0 000
þlnl x2 c2n B2nl coshlnl x2 sincn x3 cosal x1 x2 a2l mðCl0 þ Cl0 Þ
l
X
00 0 00
2Cl0 sinhal x3 cosal x1 þ b2m C10m þ C20m þ ð3þ mÞC20m
m
000
ð1þ mÞC20m sinhbm x3 þ bm x3 C20m coshbm x3 sinbm x2
X X mlm
0 00
þ b2m C1lm þ ðb2m ma2l ÞC2lm þ ½ð3 þ mÞb2m þ 2a2l C2lm
m n
b m
000
½ð1þ mÞb2m þ ma2l C2lm sinhmlm x3
2
þmlm x3 bm C2lm coshmlm x3 cosal x1 sinbm x2 ;...; ð11:50 Þ
Concerning the free terms, the boundary conditions for the normal stresses lead to
as well
1
A2m0 ¼ ð2 þ mÞðA02m0 þ A002m0 Þ;
2
1
A20n ¼ ð2 þ mÞðA020n þ A000
20n Þ; ð11:11Þ
2
1 c2 A00 b2m A000 2mn Þ
A2mn ¼ ð2 þ mÞ n 2mn ;...
m c2n b2m
We introduce the notations
b2m c2n
A0m0 ¼ A000 0
2m0 ; A0n ¼ A000 ; m; n ¼ 1; 2; 3; . . . ð11:120 Þ
mð2 þ mÞ mð2 þ mÞ 20n
Thus, the state of stress may be written in the form
X
r11 ¼ a000 þ A0m0 ½ð1 þ bm a1 coth bm a1 Þ cosh bm x1 bm x1 sinh bm x1 cos bm x2
m
X
þ A00n ½ð1 þ cn a1 coth cn a1 Þ cosh cn x1 cn x1 sinh cn x1 cos cn x3
n
XX
þ A0mn ½ð1 þ kmn a1 coth kmn a1 Þ cosh kmn x1
m n
X
kmn x1 sinh kmn x1 cos bm x2 cos cn x3 þ 2l B0n0 cosh cn x2 cos cn x3
n
X
þ B00l ½ð1 al a2 coth al a2 Þ cosh al x2 þ al x2 sinh al x2 cos al x1
l
(
2
XX al
þ B0nl ½ð1 lnl a2 coth lnl a2 Þ cosh lnl x2 þ lnl x2 sinh lnl x2
n l
lnl
2 )
c X
0
þ2m n cosh lnl x2 cos cn x3 cos al x1 þ Cl0 ½ð1 al a3 coth al a3 Þ cosh al x3
lnl l
X
0
þ al x3 sinh al x3 cos al x1 þ 2l C0m cosh bm x3 cos bm x2
m
(
2
XX al
0
þ Clm ½ð1 mlm a3 cothmlm a3 Þ cosh mlm x3 þ mlm x3 sinh mlm x3
l m
mlm
2 )
b
þ2m m cosh mlm x3 cos al x1 cos bm x2 ; . . .;
mlm
ð11:13Þ
11.1 Elastic Parallelepiped 487
XX bm cn
r23 ¼ A0mn
½ð1 kmn a1 coth kmn a1 Þ cosh kmn x1 þ kmn x1 sinh kmn x1
m n k2mn
X
2m cosh kmn x1 sin bm x2 sin cn x3 B0n0 ðcn a2 coth cn a2 sinh cn x2
n
XXcn
cn x2 cosh cn x2 Þ sin cn x3 B0nl
ðl a2 coth lnl a2 sinh lnl x2
n l
lnl nl
X
0
lnl x2 cosh lnl x2 Þ sin cn x3 cos al x1 C0m ðbm a3 coth bm a3 sinh bm x3
m
XX bm
0
bm a3 cosh bm x3 Þ sin bm x2 Clm ðmlm a3 coth mlm a3 sinh mlm x3
l m
mlm
mlm x3 cosh mlm x3 Þ cos al x1 sin bm x2 ; . . .
ð11:130 Þ
Putting the boundary conditions for the normal stresses too, one obtains fol-
lowing relations which must be verified by the three double sequences of
parameters.
X
X
0 bm a1 0 c n a1
Am0 cosh bm a1 þ cos bm x2 þ A0n ðcosh cn a1 þ cos cn x3
m
sinh bm a1 n
sinh cn a1
XX
kmn a1
þ A0mn cosh kmn a1 þ cos bm x2 cos cn x3
m n
sinh kmn a1
X X
þ 2m B0n0 cosh cn x2 cos cn x3 þ B0nl cos lp½ð1 al a2 coth al a2 Þcosh al x2
n l
(
2
XX al
þ al x2 sinh al x2 þ 2l B0nl cos lp ½ð1 lnl a2 coth lnl a2 Þcosh lnl x2
n l
lnl
2 )
cn
þlnl x2 sinh lnl x2 þ 2l cosh lnl x2 cos cn x3
lnl
X
0
þ Cl0 cos lp½ð1 al a3 coth al a3 Þ cosh al x3 þ al x3 sinh al x3
l
X
0
þ 2m C0m coshbm x3 cos bm x2
m
(
2
XX al
0
þ Clm cos lp ½ð1 mlm a3 coth mlm a3 Þ cosh mlm x3 þ mlm x3 sinh mlm x3
l m
mlm
2 )
b
þ2m m cosh mlm x3 cos bm x2 ¼ p1 ðx2 ; x3 Þ;.. .
mlm
ð11:14Þ
488 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
1 2b X cos lp cos al x1
cosh bm x1 ¼ sinh bm a1 þ m sinh bm a1 ; ð11:15Þ
bm a 1 a1 l a2l þ b2m
1 2c X cos lp cos al x1
cosh cn x1 ¼ sinh cn a1 þ n sinh cn a1 ; . . .;
c n a1 a1 l
a2l þ c2n
1
kmn x1 sinh kmn x1 ¼ ðkmn a1 cosh kmn a1 sinh kmn a1 Þ
kmn a1
" #
kmn X cos lpcos al x1 a2l ðb2m þ c2n Þ
þ2 sinh kmn a1 2
kmn a1 coth kmn a1 þ 2 ;
a1 2 2
l al þ b m þ c n al þ b2m þ c2n
1
bm x1 sinh bm x1 ¼ ðb a1 cosh bm a1 sinh bm a1 Þ
b m a1 m
!
bm X cos lp cos al x1 a2l b2m
þ 2 sinh bm a1 bm a1 coth bm a1 þ 2 ;
a1 l a2l þ b2m al þ b2m
1
cn x1 sinh cn x1 ¼ ðc a1 cosh cn a1 sinh cn a1 Þ
c n a1 n
c X cos lp cos al x1 a2l c2n
kmn a1 cos mp 0 c
A0mn cosh kmn a1 þ þ 4m Bn0 sinh cn a2 2 n
sinhkmn a1 a2 bm þ c2n
"
4 cos mp X 0 al 2 b2m
þ Bnl lnl cos lp sinh lnl a2 2 2
a2 l
lnl al þ b2m þ c2n
2 #
c 1 4m cos mp 0 b
þm n þ C0m sinh bm a3 2 m
lnl a2l þ b2m þ c2n a3 bm þ c2n
11.1 Elastic Parallelepiped 489
"
2
4 cos np X 0 al c2n
þ Clm mlm cos lp sinh mlm a3 2 2
a3 l
mlm al þ b2m þ c2n
2 #
bm 1
þm ¼ a0mn ;...; ð11:17Þ
mlm a2l þ b2m þ c2n
4b2m cos mp X 0
bm a1 1
A0m0 cosh bm a1 þ þ B0l al cos lp sinh al a2 2
sinh bm a1 a2 l al þ b2m
2
X
2
0 sinh bm a3 0 b sinh mlm a3
þ 2mC0m þ 2m Clm cos lp m ¼ a0m0 ; . . .;
bm a3 l
m lm m lm a 3
c n a1 sinh cn a2
A00n cosh cn a1 þ þ 2mB0n0 ð11:170 Þ
sinh cn a1 c n a2
X
2
c sinh lnl a2
þ 2m B0nl cos lp n
l
lnl lnl a2
2
4c cos np X 1
0 0
þ n Cl0 al cos lp sinh al x3 2 ¼ a0n ; . . .
a3 l
2
al þ c2n
Using the above results, one may obtain a solution of the problem, reducing it to
the solving of an infinite system of linear algebraic equations with constant
coefficients.
Taking into account the notations (11.16), one may write the relations (11.17),
(11.170 ) in the form
X
kmn a1 a2i b2m þ mc2n ða2i þ b2m þ c2n Þ
Amn coth kmn a1 þ k a
mn 1 þ 4 B ni 2 2
sinh2 kmn a1 i a i þ b2 þ c 2 m n
X a2i c2n þ mb2m ða2i þ b2m þ c2n Þ
þ4 Cim 2 2 ¼ ð1Þmþn a0mn ; m; n ¼ 1; 2; 3; . . .;
i ai þ b2m þ c2n
ð11:18Þ
490 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
X
bm a1 a2 b2
Am0 coth bm a1 þ 2
bm a 1 þ 4 Bni i m 2
sinh bm a1 i a2i þ b2 m
X b2
þ 2m Cim 2 m 2 ¼ ð1Þm a0m0 ; m ¼ 1; 2; 3; . . .;
i a i þ bm
ð11:180 Þ
c n a1 X c2
A0n coth cn a1 þ 2
cn a1 þ 2m Bni 2 n 2
sinh cn a1 i
ai þ c n
X a2 c 2
þ4 Ci0 i n 2 ¼ ð1Þn a00n ; n ¼ 1; 2; 3; . . .;
i a2i þ c2n
with m; n ¼ 0; 1; 2; 3; . . .; l ¼ 0; 1; 2; 3; . . . and
2la2l
fmnl ¼ ; ð11:190 Þ
a2l þ b2m þ c2n
with m ¼ 0; n ¼ 0; 1; 2; 3; . . . or n ¼ 0; m ¼ 0; 1; 2; 3; . . . and l ¼ 0; 1; 2; 3; . . .; by
circular permutations one obtains the coefficients gnlm and hlmn .
One observes that, introducing the above coefficients and taking into account
that
km0 ¼ bm ; k0n ¼ cn ; . . .; ð11:20Þ
one may write each group of relations as only one relation; the corresponding
indices may vanish.
Introducing the notations given in Sect. A.4.1.2, we remark that the sequences
of parameters Amn ; Bmn ; Cmn ; are given by a system of linear algebraic equations
with a triple infinity of unknowns, of the form
X X
vðkmn a1 ÞAmn þ himn Bni þ gnim Cim ¼ ð1Þmþn a0mn ;
i i
X X
hljn Ajn þ vðlnl a2 ÞBnl þ fjnl Clj ¼ ð1Þnþl a00nl ; ð11:21Þ
j j
X X
gklm Amk þ fmkl Bkl þ vðmlm a3 ÞClm ¼ ð1Þlþm a000
lm ;
k k
case, considered by S. Kaliski [22, 23]. The system may be studied, e.g., using the
Banach–Schauder theorem of fixed point, as in [22].
As a matter of fact, the results thus obtained generalize thus given by us [6] for
the elastic rectangle. The functions and notations which have been introduced
correspond to those used in the mentioned plane case; these functions have been
plotted into diagrams, which leads to a simplification of the computation.
From a practical point of view, one can proceed otherwise too. Eliminating the
sequence of coefficients Amn , one obtains a linear system with a double infinity of
unknowns
X XX
anli Bni þ bnlij Cij ¼ a00nl ;
i i j
XX X ð11:22Þ
bkilm Bkl þ cilm Cim ¼ a000
im ;
k l l
where
X hijn hljn
anli ¼ ; anli ¼ anil ; i 6¼ l;
j
vðkjn a1 Þ
ð11:23Þ
X h2ljn
anll ¼ vðlnl a2 Þ ;
j
vðkjn a1 Þ
gnij hljn
bnlij ¼ ; i 6¼ l;
vðkjn a1 Þ
ð11:230 Þ
gnij hljn
bnllj ¼ fjnl ;
vðkjn a1 Þ
X gklm gkim
cilm ¼ ; cilm ¼ clim ; i 6¼ l;
k
vðkmk a1 Þ
X ð11:2300 Þ
g2klm
cllm ¼ vðmlm a3 Þ ;
k
vðkmk a1 Þ
X hljn
a00nl ¼ ð1Þnþl a00nl
ð1Þjþn a0jn
;
j
vðkjn a1 Þ
X ð11:23000 Þ
gklm
a000
lm ¼ ð1Þlþm a000
lm ð1Þmþk a0mk :
k
vðk mk a1 Þ
Taking into account the relations (11.13), (11.130 ), (11.20), the notations (11.16) and
the functions introduced in Sect. A.4.1.2, one can write the state of stress in the form
XX
r11 ¼ a000 þ ð1Þmþn Amn W2 ðkmn ; x1 ; a1 Þ cos bm x2 cos cn x3
m n "
XX nþl al 2
þ ð1Þ Bnl W4 ðlnl ; x2 ; a2 Þ
n
lnl
2
l #
c
þ m n W5 ðlnl ; x2 ; a2 Þ cos cn x3 cos al x1
lnl
"
XX lþm al 2
þ ð1Þ Clm W4 ðmlm ; x3 ; a3 Þ
m
mlm
l
2 #
bm
þm W5 ðmlm ; x3 ; a3 Þ cos al x1 cos bm x2 ; . . .; ð11:24Þ
mlm
XX bm cn
r23 ¼ ð1Þmþn Amn W45 ðkmn ; x1 ; a1 Þ sin bm x2 sin cn x3
m n k2mn
XX cn
þ ð1Þnþl Bnl W3 ðlnl ; x2 ; a2 Þ sin cn x3 cos al x1
n l
lnl
XX bm
þ ð1Þlþm Clm W3 ðmlm ; x3 ; a3 Þ cos al x1 cos bm x2 ; . . .; ð11:240 Þ
l m
mlm
the sign sum being extended for the case in which one of the indices vanishes (the
two indices cannot vanish simultaneously).
For the state of displacement one obtains, similarly,
1 0
u1 ¼ a00 mða0000 þ a000
00 Þ x1
2ð1 þ mÞl
XX Amn
þ ð1Þmþn W16 ðkmn ; x1 ; a1 Þ cos bm x2 cos cn x3
m n
kmn
XX Bnl al
þ ð1Þnþl W45 ðlnl ; x2 ; a2 Þ cos cn x3 sin al x1
n l
lnl lnl
XX Clm al
þ ð1Þlþm W45 ðmlm ; x3 ; a3 Þ sin al x1 cos bm x2 ; . . .; ð11:25Þ
l m
mlm mlm
where the displacements and the rotations of rigid body vanish, because of sym-
metry reasons with respect to the co-ordinate planes.
In use of the action of volume forces which has analogous properties of sym-
metry, one may use a representation by means of triple Fourier series
11.1 Elastic Parallelepiped 493
XXX
F1 ¼ a0lmn sinal x1 sin bm x2 sin cn x3 ; . . . ð11:26Þ
l m n
For the components of the stress tensor, one obtains easily a system of particular
integrals of the same form, which verifies the equilibrium and the Beltrami–Michell
equations. Putting the boundary conditions, the triple Fourier series become double
Fourier series and one has to solve a problem analogous to that considered above.
We deal, in what follows, with the study of other two infinite bodies: the elastic
strip and the elastic layer, using analogous methods of computations.
Throwing to infinite two opposite faces of the elastic parallelepiped, one obtains
the elastic strip jxi j ai ; i ¼ 1; 2; which we consider to be acted upon by the
periodic loads (11.5).
To solve the problem, one may use the representation (11.1)–(11.20 ), taking
L03 ¼ L003 ¼ L000 0 0 000
3 ¼ 0; C1lm ¼ C2lm ¼ . . . ¼ C2lm ¼ 0. The sequences of parameters
to be determined are obtained by putting the boundary conditions
x1 ¼ a1 : r11 ¼ p1 ðx2 ; x3 Þ; r12 ¼ r13 ¼ 0;
ð11:27Þ
x2 ¼ a2 : r22 ¼ p2 ðx3 ; x1 Þ; r21 ¼ r23 ¼ 0:
The study of the problem leads thus to the study of a system of linear algebraic
equations with a double infinity of unknowns of the form
X
vðkmn a1 ÞAmn þ himn Bni ¼ ð1Þmþn a0mn ;
i
X ð11:28Þ
hljn Ajn þ vðlnl a2 ÞBnl ¼ ð1Þnþl a00nl ;
j
X
a00nl ;
anli Bni ¼ ð11:29Þ
i
XX nþl al 2
þ ð1Þ Bnl W4 ðlnl ; x2 ; a2 Þ
n
lnl
2
l #
c
þ m n W5 ðlnl ; x2 ; a2 Þ cos cn x3 cos al x1 ;
lnl
"
XX mþn bm 2
00
r22 ¼ a00 þ ð1Þ Amn W4 ðkmn ; x1 ; a1 Þ
m n
kmn
2 #
cn
þm W5 ðkmn ; x1 ; a1 Þ cos bm x2 cos cn x3
kmn
XX ð11:30Þ
þ ð1Þnþl Bnl W2 ðlnl ; x2 ; a2 Þ cos cn x3 cos al x1 ;
n l "
XX mþn cn 2
0 00
r33 ¼ mða00 þ a00 Þ þ ð1Þ Amn W4 ðkmn ; x1 ; a1 Þ
m n
kmn
2 #
bm
þm W5 ðkmn ; x1 ; a1 Þ cos bm x2 cos cn x3
kmn
"
XX nþl cn 2
þ ð1Þ Bnl W4 ðlnl ; x2 ; a2 Þ
n
lnl
2
l #
al
þm W5 ðlnl ; x2 ; a2 Þ cos cn x3 cos al x1 ;
lnl
11.2 Elastic Strip. Elastic Layer 495
XX bm cn
r23 ¼ ð1Þmþn Amn W45 ðkmn ; x1 ; a1 Þ sin bm x2 sin cn x3
m n k2mn
XX cn
þ ð1Þnþl Bnl W3 ðlnl ; x2 ; a2 Þ sin cn x3 cos al x1
n l
l nl
XX c
r31 ¼ ð1Þmþn Amn n W5 ðkmn ; x1 ; a1 Þ cos bm x2 sin cn x3
m n
kmn
XX c al ð11:300 Þ
ð1Þnþl Bnl n 2 W45 ðlnl ; x2 ; a2 Þ sin cn x3 sin al x1 ;
n l
lnl
XX b
r12 ¼ ð1Þmþn Amn m W3 ðkmn ; x1 ; a1 Þ sin bm x2 cos cn x3
m n
kmn
XX al
þ ð1Þnþl Bnl W3 ðlnl ; x2 ; a2 Þ sin cn x3 sin al x1
n l
lnl
we took into account that the displacement must be finite in the direction of the
Ox3 -axis (this fact has an influence on the normal stresses too).
496 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
Throwing to infinite two opposite faces of an elastic strip, one obtains the elastic
layer jx3 j a3 ; we shall consider the action of a periodic load, as well as that of a
local load.
We shall deal with the elastic layer subjected to the action of a periodic normal
load, symmetric with respect to the Ox1 an Ox2 co-ordinate axes; any other case of
normal or tangential load may be studied similarly. In what follows, we shall use
the results obtained by us in [38], results which generalize those known for the
wall-beams acted similarly [6].
We will consider both a case of a symmetric load with respect to the Ox1 x2 -
plane and the case of an antisymmetric load with respect to the same plane.
Let thus be the elastic layer jx3 j a3 ; acted upon by a normal load of the form
(the simple Fourier series are contained in the double Fourier series).
XX
p3 ðx1 ; x2 Þ ¼ a000
00 þ a000
lm cos al x1 cos bm x2 : ð11:32Þ
l m
The third Eq. (11.21) is thus verified; what concerns the first two equations, they
are replaced by conditions of global equilibrium on the planes x1 ¼ a1 and
x2 ¼ a2 .
It results the state of stress
11.2 Elastic Strip. Elastic Layer 497
"
m X X a000 al 2
000 lm
r11 ¼ a00 þ W4 ðmlm ; x3 ; a3 Þ
1m l m
vðmlm a3 Þ mlm
2 #
bm
þm W5 ðmlm ; x3 ; a3 Þ cos al x1 cos bm x2 ;
mlm
"
m X X a000 bm 2
000 lm ð11:36Þ
r22 ¼ a00 þ W4 ðmlm ; x3 ; a3 Þ
1m l m
vðmlm a3 Þ mlm
2 #
al
þm W5 ðmlm ; x3 ; a3 Þ cos al x1 cos bm x2 ;
mlm
X X a000
r33 ¼ a000
00 þ
lm
W2 ðmlm ; x3 ; a3 Þ cos al x1 cos bm x2 ;
l m
vðm lm a3 Þ
XX a000 bm
lm
r23 ¼ W3 ðmlm ; x3 ; a3 Þ cos al x1 sin bm x2
l m
vðmlm a3 Þ mlm
XX a000 al
lm
r31 ¼ W3 ðmlm ; x3 ; a3 Þ sin al x1 cos bm x2 ; ð11:360 Þ
l m
vðmlm a3 Þ mlm
X X a000 al b
lm m
r12 ¼ W45 ðmlm ; x3 ; a3 Þ sin al x1 sin bm x2
l m
vðmlm a3 Þ m2lm
we took into account that the displacements in the direction of the Ox1 and Ox2
axes must be finite and that they have some properties of symmetry with respect to
these axes.
The convergence of these series depend on the factor a000lm ; even in the case of
concentrated loads, the sum of the series may be approximated with a small
number of terms, due to the hyperbolic lines at the denominator. If we approach
the separation planes x3 ¼ a3 ; then the expansion in series becomes slowly
convergent. Taking into account 11.32, we may write the stresses corresponding to
these planes in the form
498 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
1 2m 000
r11 ðx1 ; x2 ; a3 Þ ¼ p3 ðx1 ; x2 Þ a
1 m 00
X X b
2
ð1 2mÞ m
a000
lm cos al x1 cos bm x2
l m
m lm
X X a2 þ mb2 4mlm a3
l m
a000 cos al x1 cos bm x2 ;
l m m2lm 2mlm a3 þ sinh 2mlm a3 lm
1 2m 000
r22 ðx1 ; x2 ; a3 Þ ¼ p3 ðx1 ; x2 Þ a
1 m 00
X X al
2
ð1 2mÞ a000
lm cos al x1 cos bm x2
l m
m lm
X X b2 þ ma2 4mlm a3
m
2
l
a000
lm cos al x1 cos bm x2 ;
l m m lm 2m lm a 3 þ sinh 2m lm a 3
ð11:38Þ
X X al b
r12 ðx1 ; x2 ; a3 Þ ¼ ð1 2mÞ m
a000
lm sin al x1 sin bm x2
l m m2lm
XX 4mlm a3 al bm 000
þ ð1 mÞ a sin al x1 sin bm x2 ;
l m
2mlm a3 þ sinh 2mlm a3 m2lm lm
ð11:380 Þ
where intervenes the difference between the external load and the mean one. Thus,
the computation is easy, the above expansions being rapid convergent.
The series which correspond to the displacements are convergent after the
factor a000
lm =mlm ; hence they may be calculated with a smaller number of terms than
the stresses. As well, the deformation of the separation planes is given by
1 2m 000
u3 ðx1 ; x2 ; a3 Þ ¼ a
2ð1 mÞl 00
1 XX 2 sinh2 mlm a3 a000
lm
ð1 mÞ sin al x1 sin bm x2 :
l l m
sinh 2m lm a 3 þ 2m lm a 3 m lm
ð11:39Þ
In case of a loading symmetric with respect to Ox1 and antisymmetric with
respect to Ox2
XX
p3 ðx1 ; x2 Þ ¼ a000
lm sin al x1 cos bm x2 ; ð11:40Þ
l m
XX
0 0
F11 ¼ ðC1lm cosh mlm x3 þ mlm x3 C2lm sinh mlm x3 Þ sin al x1 cos bm x2 ; . . .
l m
ð11:400 Þ
For a load antisymmetric with respect to Ox1 and symmetric with respect to Ox2
XX
p3 ðx1 ; x2 Þ ¼ a000
lm cos al x1 sin bm x2 ; ð11:41Þ
l m
X X a000 al 2 0
lm
r11 ¼ w4 ðmlm ; x3 ; a3 Þ
l m
v0 ðmlm a3 Þ mlm
2 #
bm 0
þm w5 ðmlm ; x3 ; a3 Þ cos al x1 cos bm x2 ;
mlm
"
X X a000 bm 2 0
r22 ¼ lm
w4 ðmlm ; x3 ; a3 Þ ð11:43Þ
v 0 ðm a Þ m
l m lm 3 lm
2 #
al 0
þm w5 ðmlm ; x3 ; a3 Þ cos al x1 cos bm x2 ;
mlm
X X a000
r33 ¼ lm
w0 ðm ; x ; a Þ cos al x1 cos bm x2 ;
0 ðm a Þ 2 lm 3 3
l m
v lm 3
500 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
XX a000 bm 0
lm
r23 ¼ W3 ðmlm ; x3 ; a3 Þ cos al x1 sin bm x2 ;
l m
v0 ðm lm a3 Þ mlm
XX a000 al 0
lm
r31 ¼ W3 ðmlm ; x3 ; a3 Þ sin al x1 cos bm x2 ; ð11:430 Þ
l m
v0 ðm a Þ
lm 3 lm m
XX a000 al bm 0
lm
r12 ¼ W ðmlm ; x3 ; a3 Þ sin al x1 sin bm x2 ;
l m
v0 ðmlm a3 Þ m2lm 45
1 X X a000 1 0
u3 ¼ u03 þ lm
W ðmlm ; x3 ; a3 Þ cos al x1 sin bm x2 ;
2l l m v ðmlm a3 Þ mlm 16
0
X X al
2
r22 ðx1 ; x2 ; a3 Þ ¼ p3 ðx1 ; x2 Þ ð1 2mÞ a000
lm cos al x1 cos bm x2
l m
mlm
X X b2 þ ma2 4mlm a3
m
2
l
a000
lm cos al x1 cos bm x2 ;
l m m lm sinh 2m lm a 3 2m lm a 3
ð11:45Þ
X X al b
r23 ðx1 ; x2 ; a3 Þ ¼ ð1 2mÞ m
a000
lm sin al x1 sin bm x2
l m m2lm
XX 4mlm a3 al bm 000
ð1 mÞ a sin al x1 sin bm x2 :
l m
sinh 2mlm a3 2mlm a3 m2lm lm
ð11:450 Þ
As well, the deformation of the separation planes is given by
11.2 Elastic Strip. Elastic Layer 501
ð11:46Þ
In the case of an elastic layer subjected to the action of a normal load one obtains
results analogous to thus presented in the preceding subsection. We will use the
results obtained by us [37].
Let thus be the elastic layer jx3 j a3 ; acted upon by a normal load of the form
Z 1Z 1
p3 ðx1 ; x2 Þ ¼ a3 ða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2 ; ð11:47Þ
0 0
symmetric with respect to the co-ordinate axes Ox1 and Ox2 . We introduce the
stress functions
Z 1Z 1
F11 ¼ ½C 0 ða1 ; a2 Þ cosh m3 x3
0 0
þm3 x3 C 00 ða1 ; a2 Þ sinh m3 x3 cos a1 x1 cos a2 x2 da1 da2 ; . . .; ð11:48Þ
the variable parameters which are introduced must verify boundary conditions of
the form (11.34). We have denoted
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
m3 ¼ a21 þ a22 : ð11:49Þ
a3 ða1 ; a2 Þ a1 2
r11 ¼ W4 ðm3 ; x3 ; a3 Þ
0 0 vðm3 a3 Þ m3
2 #
a2
þm W5 ðm3 ; x3 ; a3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
m3
Z 1Z 1 "
a3 ða1 ; a2 Þ a2 2 ð11:50Þ
r22 ¼ W4 ðm3 ; x3 ; a3 Þ
0 0 vðm3 a3 Þ m3
2 #
a1
þm W5 ðm3 ; x3 ; a3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
m3
Z 1Z 1
a3 ða1 ; a2 Þ
r33 ¼ W2 ðm3 ; x3 ; a3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
0 0 vðm3 a3 Þ
502 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
Z 1 Z 1
a3 ða1 ; a2 Þ a2
r23 ¼ W3 ðm3 ; x3 ; a3 Þ cos a1 x1 sin a2 x2 da1 da2 ;
0 0 vðm3 a3 Þ m3
Z 1Z 1
a3 ða1 ; a2 Þ a1
r31 ¼ W3 ðm3 ; x3 ; a3 Þ sin a1 x1 cos a2 x2 da1 da2 ; ð11:500 Þ
0 0 vðm3 a3 Þ m3
Z 1Z 1
a3 ða1 ; a2 Þ a1 a2
r12 ¼ W45 ðm3 ; x3 ; a3 Þ sin a1 x1 sin a2 x2 da1 da2
0 0 vðm3 a3 Þ m23
the displacements and rotations of rigid body are taken equal to zero because of the
symmetry of the load.
For the separation planes, there correspond the stresses
r11 ðx1 ; x2 ; a3 Þ ¼ pðx1 ; x2 Þ
Z 1 Z 1
2
a2
ð1 2mÞ a3 ða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2
0 0 m3
Z 1Z 1 2
a1 þ ma22 4m3 a3
a3 ða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2 ;
0 0 m23 sinh 2m3 a3 þ 2m3 a3
r22 ðx1 ; x2 ; a3 Þ ¼ pðx1 ; x2 Þ
Z 1 Z 1
2
a1
ð1 2mÞ a3 ða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2
0 0 m3
Z 1 Z 1
a22 þ ma21 4m3 a3
a3 ða1 ; a2 Þ cos al x1 cos a2 x2 da1 da2 ; ð11:52Þ
0 0 m23 sinh 2m3 a3 þ 2m3 a3
Z 1Z 1
a1 a2
r12 ðx1 ; x2 ; a3 Þ ¼ ð1 2mÞ a3 ða1 ; a2 Þ sin a1 x1 sin a2 x2 da1 da2
0 0 m23
Z 1Z 1
4m3 a3 a1 a2
þ ð1 mÞ 2
a3 ða1 ; a2 Þ sin a1 x1 sin a2 x2 da1 da2 ;
0 0 sinh 2m3 a3 þ 2m3 a3 m3
ð11:520 Þ
while the deformations of the separation planes are given by
11.2 Elastic Strip. Elastic Layer 503
Z1 Z1
1 2 sinh2 m3 a3 a3 ða1 ; a2 Þ
u3 ðx1 ; x2 ; a3 Þ ¼ ð1 mÞ cos a1 x1 cos a2 x2 da1 da2 :
l sinh 2m3 a3 þ 2m3 a3 m3
0 0
ð11:53Þ
For a load symmetric with respect to the Ox1 -axis and antisymmetric with
respect to the Ox2 -axis
Z 1Z 1
p3 ðx1 ; x2 Þ ¼ a3 ða1 ; a2 Þ sin a1 x1 cos a2 x2 da1 da2 ð11:54Þ
0 0
If the load is antisymmetric with respect to the Ox1 -axis and symmetric with
respect to the Ox2 -axis
Z 1Z 1
p3 ðx1 ; x2 Þ ¼ a3 ða1 ; a2 Þ cos a1 x1 sin a2 x2 da1 da2 ; ð11:55Þ
0 0
We can make analogous considerations for the elastic layer jx3 j a3 acted upon
by the normal load (11.47), antisymmetric with respect to the Ox1 x2 -plane.
We obtain thus the state of stress
Z 1Z 1 "
a3 ða1 ; a2 Þ a1 2 0
r11 ¼ W4 ðm3 ; x3 ; a3 Þ
0 0 v0 ðm3 a3 Þ m3
2 #
a2 0
þm W5 ðm3 ; x3 ; a3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
m3
Z 1Z 1 "
a3 ða1 ; a2 Þ a2 2 0
r22 ¼ W4 ðm3 ; x3 ; a3 Þ ð11:57Þ
0 0 v0 ðm3 a3 Þ m3
2 #
a1 0
þm W5 ðm3 ; x3 ; a3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
m3
Z 1Z 1
a3 ða1 ; a2 Þ 0
r33 ¼ W ðm3 ; x3 ; a3 Þ cos a1 x1 cos a2 x2 da1 da2 ;
0 0 v0 ðm3 a3 Þ 2
Z 1Z 1
a3 ða1 ; a2 Þ a2 0
r23 ¼ W ðm3 ; x3 ; a3 Þ cos a1 x1 sin a2 x2 da1 da2 ;
0 0 v0 ðm3 a3 Þ m3 3
Z 1Z 1
a3 ða1 ; a2 Þ a1 0
r31 ¼ W ðm3 ; x3 ; a3 Þ sin a1 x1 cos a2 x2 da1 da2 ; ð11:570 Þ
0 0 v0 ðm3 a3 Þ m3 3
Z 1Z 1
a3 ða1 ; a2 Þ a1 a2 0
r12 ¼ W ðm3 ; x3 ; a3 Þ sin a1 x1 sin a2 x2 da1 da2
0 0 v0 ðm3 a3 Þ m23 45
Z 1Z 1
a1 a2
r12 ðx1 ; x2 ; a3 Þ ¼ ð1 2mÞ a3 ða1 ; a2 Þ sin a1 x1 sin a2 x2 da1 da2
0 0 m23
Z 1Z 1
4m3 a3 a1 a2
ð1 mÞ 2
a3 ða1 ; a2 Þ sin a1 x1 sin a2 x2 da1 da2
0 0 sinh 2m a
3 3 2m 3 3 m3
a
ð11:590 Þ
The first approximation which can be made concerning the thick plate is that of
considering it as infinite, i.e. as an elastic layer. The first researches of the elastic
layer subjected to the action of local loads are due to Dougall [13] and L. Orlando
[34], who use the method of reflection. We mention an interesting paper of
J. N. Sneddon [35] too. We mention also a presentation of those problems made by
A. I. Lur’e [5] in his treatise, which contains the studies [26, 27] made by himself
in this direction too.
In the case of finite plates one uses usual the hypothesis of linear element
(normal to the middle plane, which remains linear, normal to the deformed middle
plane, without linear deformations) of G. R. Kirchhoff [3]; the solving of the
problem, on the basis of this supplementary hypothesis (of the nature of the
hypotheses of strength of materials), is reduced to the integration of the Lagrange-
Sophie Germain partial differential equation, with certain boundary conditions,
corresponding to the supports of the plate.
506 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
a3
M22 T22
N21 O x1
M21
a3
N 22
The thick plate for which this hypothesis is no more verified has been less
studied; but such plates can intervene in engineering. We mention thus the study of
B. G. Galerkin [19], concerning the simply supported circular thick plate, pub-
lished in 1932, and the study, published in 1935 by the same author [20], con-
cerning the thick rectangular plate. One gives both qualitative and quantitative
results, with some displacement functions; but these functions have a form more
complicated as necessary. A direct solution, in displacements too, of the problem
has given, in 1933, S. Woinowski-Krieger [41], searching solutions in form of
trigonometric series expansions, which verify Lamé’s equations. Other studies in
this direction have been made by E. Almansi [8]. In 1955, L. Bonneau [12] dealt
again with the problem, especially with the integration of the equations of equi-
librium and of those of continuity of Beltrami for the thick plate, making various
considerations of practical order. As well, M. Misßicu [30] presented some results
with a general character, as an application to the representation of the solution in
displacements of the spatial problem of the theory of elasticity by means of some
monogenic functions.
M. Haimovici [21] used Galerkin’s method, starting from functions of the form
X
n
Ak ðx1 ; x2 ÞPk ðx3 Þ; ð11:61Þ
k¼1
Let be a rectangular thick plate of dimensions 2a1 and 2a2 and of thickness 2a3 ; we
denote the ratios a1 ¼ a1 =a3 ; a2 ¼ a2 =a3 and a ¼ a1 =a2 ¼ a1 =a2 : Our study is, in
general, valid in the frame of Saint-Venant’s principle, hence we shall assume that
a1 ; a2 2.
In the computation, one uses, usually, the sectional efforts along the thickness of
the plate (Fig. 11.1).
We introduce thus the bending moments
Z a3
M11 ðx1 ; x2 Þ ¼ r11 ðx1 ; x2 ; x3 Þx3 dx3 ;
a
Z a33 ð11:62Þ
M22 ðx1 ; x2 Þ ¼ r22 ðx1 ; x2 ; x3 Þx3 dx3
a3
0
T11 ðx1 ; x2 Þ ¼ T11 ðx1 ; x2 Þ þ M12;2 ðx1 ; x2 Þ;
0 ð11:64Þ
T22 ðx1 ; x2 Þ ¼ T22 ðx1 ; x2 Þ þ M21;1 ðx1 ; x2 Þ;
where the comma indicates the differentiation with respect to the variable at the
right of it.
The concentrated reactions at the vertices of the plate are given by
V ¼ 2M12 ða1 ; a2 Þ: ð11:65Þ
and in an antisymmetric one with respect to the same plane. The last case remains
to be studied.
We choose the stress functions
1 1
þ A7 x21 x33 þ A8 x22 x33 A1 þ A5 þ A6 þ A7 þ A8 x53 ; . . .; ð11:67Þ
5 3
using the biharmonic polynomials considered in Sect. 5.2.1.3. One must thus
determine the constants Ai ; Bi ; Ci ; i ¼ 1; 2; . . .; 8:
The functions F23 and F31 contain x1 and x2 , respectively, at the zeroth power,
so that they cannot be odd with respect to these variables. But r23 and r31 are odd
functions with respect to x2 and to x1 , respectively, because of the conditions of
symmetry. Hence, we must take
F23 ¼ F31 ¼ 0: ð11:68Þ
The harmonic function F12 must be odd with respect to both variables; we may
take
F12 ¼ E1 x1 x2 þ E2 ðx31 x2 x1 x32 Þ: ð11:680 Þ
The normal stresses are given by
11.3 Thick Plate 509
where L and M are other constants which must be determined and where we have
used the notation
Di ¼ Ai þ Bi þ Ci ; i ¼ 1; 2; . . .; 8: ð11:70Þ
510 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
x3 3 p 3 p 3ν
p α ξ α ξ α αξ ξ
8 1 1 8 2 2 4 1 2 1 2
σS
σ11E
0.448a3
11
2+ν p
a3
20 5
O x1, x2
0.775a3
a3
p
2
To specify the parameters which have been introduced, we put the boundary
conditions
By the aid of the coefficients determined as it has been shown above, the state of
strain and stress in the considered thick plate may be specified.
where we have denoted by S the results obtained by the methods of the strength of
materials and by E the corrections given by the theory of elasticity.
We introduce the reduced co-ordinates
xi
ni ¼ ; i ¼ 1; 2; 3: ð11:73Þ
ai
11.3 Thick Plate 511
1 3 p
rE11 ¼ rE22 ¼ ð2 þ mÞp n23 n3 ; rE33 ¼ ð1 þ n3 Þ2 ð2 n3 Þ ð11:740 Þ
8 5 4
for the normal stresses, where we took into account (11.66) too; for the tangential
stresses, one may write
3 3
rS23 ¼ a2 pð1 n23 Þn2 ; rS31 ¼ a1 pð1 n23 Þn1 ;
8 8 ð11:75Þ
S 3m
r12 ¼ a1 a2 pn1 n2 n3
4
as well as
as well
Hence, the maximal difference is less than 7 %, whatever be m, so that the dia-
grams of the stresses r11 and r22 remain very near to the linear ones.
The normal stress r33 is equal to zero in the frame of the strength of materials,
but in the frame of the theory of elasticity has a variation after a parabola of third
order, being equal to p=2 in the middle plane.
The tangential stresses have variatiations analogous to those obtained by the
methods of the strength of materials.
1
M12 ðx1 ; x2 Þ ¼ M21 ðx1 ; x2 Þ ¼ mpa1 a2 n1 n2 ; ð11:780 Þ
2
1
Tii ðx1 ; x2 Þ ¼ pai ni ; i ¼ 1; 2; ð11:79Þ
2
1
Tii0 ðx1 ; x2 Þ ¼ ð1 mÞpai ni ; i ¼ 1; 2: ð11:790 Þ
2
The reactions of the four sides read
0 1 0 1
T11 ða1 ; x2 Þ ¼ ð1 mÞa1 ; T22 ðx1 ; a2 Þ ¼ ð1 mÞa2 ð11:80Þ
2 2
and the reactions concentrated at the vertices become
V ¼ mpa1 a2 : ð11:800 Þ
The global equilibrium of the plate may by easily verified.
Taking into account the conditions of symmetry in the case of the motion of rigid
body, it results the state of displacement in the form
u1 1
E ¼ ð1 þ mÞp½a21 ð3 n21 Þ 3ma22 ð1 n22 Þ þ ð2 mÞn23 n3 n1
a1 8
3 m
þ ð2 þ 9m m2 Þpn3 n1 þ pn1 ;
40 2 ð11:81Þ
u2 1
E ¼ ð1 þ mÞp½a2 ð3 n2 Þ 3ma21 ð1 n21 Þ þ ð2 mÞn23 n3 n2
2 2
a2 8
3 m
þ ð2 þ 9m m2 Þpn3 n2 þ pn2 ;
40 2
u3 u03 1 6
E ¼ ð1 þ mÞp 3m½a21 ð1 n21 Þ þ a22 ð1 n22 Þ ð1 þ mÞ n23 n23
a3 16 5
p 3 3 1
1 þ n3 n3 þ mð1 þ mÞa21 a22 pð1 n21 Þð1 n22 Þ þ ð1 þ mÞp½a41 ð6 n21 Þn21
2 5 16 32
3
þ a22 ð6 n22 Þn22 þ ð8 þ m þ m2 Þpða21 n21 þ a22 n22 Þ: ð11:810 Þ
80
u3 ðx1 ; x2 ; 0Þ u03 3
E ¼ mð1 þ mÞa21 a22 pð1 n21 Þð1 n22 Þ
a3 16
1
þ ð1 þ mÞp½a41 ð6 n21 Þn21 þ a42 ð6 n22 Þn22
32
3
þ ð8 þ m þ m2 Þpða21 n21 þ a22 n22 Þ: ð11:82Þ
80
The conditions of simple support at the vertices
n1 ¼ 1; n2 ¼ 1; n3 ¼ 0 : u3 ¼ 0 ð11:83Þ
lead to
pa3
u03 ¼ ½25ð1 þ mÞða41 þ a42 Þ þ 6ð8 þ m þ m2 Þpða21 þ a22 Þ; ð11:830 Þ
160E
the displacement of rigid body being thus completely determined.
References
A. Books
B. Papers
8. Almansi, E.: Sulle deformazioni delle piastre elastiche. Atti R. Accad. dei Lincei, Rend., Cl.
Sci. fis., mat. nat., ser. 6, 17, 12 (1933)
9. Baı̆da, Ē.N.: Obshchee reshenie uravneniı̆ ravnoveciya i padacha o parallelipipede dlya
anizotropnogo i izotropnogo tela (General solution of the equilibrium equations and the
problem of the parallelepiped for anisotropic and isotropic bodies). Sb. Dokl. Leningr. inzh.-
stroit. inst. 611 (1958)
514 11 Elastic Parallelepiped. Elastic Strip. Elastic Layer. Thick Plate
Taking into account (12.1), the results given in the above-mentioned subsections
lead to the following relations between strains and displacements
our ur ouz
err ¼ ; ehh ¼ ; ezz ¼ ; ð12:2Þ
or r oz
ouz our
czr ¼ þ ; ð12:20 Þ
or oz
to the cubical dilatation
our ur ouz 1 o ouz
h¼ þ þ ¼ ðrur Þ þ ; ð12:3Þ
or r oz r or oz
to the equations of motion
orrr orzr 1
þ þ ðrrr rhh Þ þ Fr ¼ q€ur ;
or oz r
ð12:4Þ
orrz orzz 1
þ þ rzr þ Fz ¼ q€uz
or oz r
and to Hooke’s law (4.1.110), to which is added
1
czr ¼ rzr : ð12:5Þ
G
By eliminating the strains and the stress, we find Lamé’s equations
1 oh 1 oh
h02 uR þ ¼ 0; h02 uz þ ¼ 0; ð12:6Þ
1 2m or 1 2m oz
in the absence of the volume forces; here, beside the classical d’Alembert’s
operators, we introduced the operators
1 o2
h0k ¼ M0 ; k ¼ 1; 2; ð12:7Þ
c2k ot2
with
1
M0 ¼ M ; ð12:8Þ
r2
where Laplace’s operator is given by
1o o o2 o2 1 o o2
M¼ r þ 2¼ 2þ þ 2 ð12:80 Þ
r or or oz or r or oz
12.1 Axisymmetrical Problems 519
n being the unit vector of the external normal to the area element considered.
At the time t ¼ t0 , we can consider initial conditions for the whole body, in
displacements and displacements velocities, in the form
u_ r ¼ u_ 0r ; u_ z ¼ u_ 0z ð12:100 Þ
We can thus formulate the three fundamental problems of elastodynamics for the
problem with axial symmetry with two space variables.
h02 F1 ¼ 0; h2 F2 ¼ 0: ð12:130 Þ
The functions F1 ; F2 , of class C2, will be supplied by the Eq. (12.130 ). while the
functions U, of class C3, will result from the Eq. (12.13); thus, the formulae
(12.12) give a representation by displacement potentials of the solution of the
Lamé equations in case of the axially symmetrical problem (A Schaefer type
representation).
If the functions F1 ; F2 are written in the form
F1 ¼ h01 Cr ; F2 ¼ h1 Cz ; ð12:14Þ
the Eq. (12.13) allows, in this case, to choose the function U in the form
1 oCr Cr oCz
U¼ þ þ : ð12:140 Þ
1 m or r oz
with the help of two potential functions Cr ¼ Cr ðr; z; tÞ; Cz ¼ Cz ðr; z; tÞ that verify
two double waves equations; these functions must be of class C4.
The same results can be obtained by particularization, starting from the general
three-dimensional case.
We shall now show that the state of displacement can be expressed with the help of
a single potential function; to this end, we shall introduce the function K, given by
o2 K
¼ Cr : ð12:16Þ
oroz
We see that
o2 o2
M0 ¼ M;
oroz oroz
the first Eq. (12.5) leads to
o2
h1 h2 K ¼ 0;
oroz
from which we deduce
h1 h2 K ¼ g1 þ g2 ;
where g1 ¼ g1 ðr; tÞ; g2 ¼ g2 ðz; tÞ are arbitrary functions. The general integral of
this equation will assume the form
K ¼ K þ h1 þ h2 ;
where h1 ¼ h1 ðr; tÞ; h2 ¼ h2 ðz; tÞ are particular integrals corresponding to the
functions g1 and g2, respectively, while K ¼ K ðr; z; tÞ is the general integral of
the homogeneous equation
h1 h2 K ¼ 0: ð12:160 Þ
In this case, the notation (12.16) leads to
o2 K
Cr ¼ ; ð12:17Þ
oroz
by introducing (12.17) in the representation (12.15), we observe that the potential
functions K and Cz may be grouped as follows
l o2 K ð1 2mÞl
v¼ Cz 2 h2 K : ð12:18Þ
1m oz 1m
522 12 Dynamical Problems of Elastic Bodies
Thus, with the help of the function v ¼ vðr; z; tÞ, we can write Predeleanu’s
representation in the form
1 o2 v
ur ¼ ;
2l oroz
ð12:19Þ
1m 1 o2 v
uz ¼ h1 v ;
l 2l oz2
the state of stress is given by
o o2 v
rrr ¼ mh2 v 2 ;
oz or
ð12:20Þ
o 1 ov
rhh ¼ mh2 v ;
oz r or
o o2 v
rzz ¼ ð1 mÞh2 v þ Mv 2 ;
oz oz
2
o ov
rrz ¼ ð1 mÞh1 v 2 : ð12:200 Þ
oz oz
The potential function v must belong to the class C4 and verify the equation
h1 h2 v ¼ 0: ð12:21Þ
We mention that this representation is complete.
We remark moreover that we can apply Boggio’s theorem which allows to
break down the waves into longitudinal and transverse ones.
With respect to a given fixed frame of reference, we shall consider the three
equations of motion (5.1.6), the six relations between strain and displacements
(5.1.2) and the constitutive law (4.1.56). This system of 15 equations is associated
with boundary conditions and initial conditions.
Hereafter, we shall deal with waves propagation and elastic vibrations.
12.2.1 Introduction
The tensor Hijkl has the properties (4.1.57), (4.1.570 ) of symmetry, hence the body
under study is a hyperelastic one.
12.2 Progressive Waves. Free and Characteristic Vibrations 523
the tensor H is called positive definite if, for any symmetric tensor e 6¼ 0,
Hijkl eij ekl [ 0: ð12:220 Þ
If H is positive definite, it is also strictly elliptic; the converse is not true.
Indeed, let H be positive definite; then 8x; y 2 E3 ; x; y 6¼ 0, the dydic product
x y admits the decomposition xi xj ¼ xði yjÞ þ x½i yj . It follows that
Hijkl xi yj xk yl ¼ Hijkl xði yjÞ xðk ylÞ [ 0; thus, the property is proved.
For a homogeneous (H is constant), isotropic elastic body, Hooke’s tensor H is
symmetric, positive definite, hence strictly elliptic too, with the components
Hijkl ¼ kdij dkl þ lðdik djl þ dil djk Þ; ð12:23Þ
Let m 2 E3 be a unit vector and let A(m) be the tensor of components Aik ðmÞ,
given by the contraction of H with m, according to the rule
1
Aik ðmÞ ¼ Hijkl mj ml ; ð12:24Þ
q
A(m) is called the acoustic tensor associated with the given elastic body, which
represents a second order tensor field with the properties:
(i) A(m) is symmetric for any m if and only if H is symmetric; indeed, we have
(ii) A(m) is positive definite for any m if and only if H is strictly elliptic; indeed,
8y 2 E3 ; y 6¼ 0, we have
qAik ðmÞ ¼ kdij dkl mj ml þ lðdik djl þ dil djk Þmj ml ¼ kmi mk
þ lðdik mj mj þ mi mk Þ ¼ ðk þ 2lÞmi mk þ lðdik mi mk Þ:
Let us note c21 ¼ ðk þ 2lÞ=q; c22 ¼ l=q, introducing thus the propagation
velocities (5.3.4), (5.3.40 ). Thus, it follows that, for an isotropic elastic body, the
acoustic tensor has the form
Let us show that c21 and c22 represent the eigenvalues of the tensor A(m) for any
m and that, for a given fixed m, the corresponding eigenvectors are m and m? ,
respectively; we have noted by m? a vector normal to m. Hence, if m1 and m2 are
two vectors normal to m and if m1 m2 ¼ 0, then the characteristic couples of
A(m) are ðc21 ; mÞ; ðc22 ; m1 Þ; ðc22 ; m2 Þ.
It follows easily that the characteristic equation of A(m), i.e.
det½Aij kdij ¼ 0, can be put in the form
m2 þ d m1 m2 m1 m3
1
m2 m1 m2 þ d m2 m3 ¼ 0; ð12:26Þ
2
m3 m1 m3 m2 m2 þ d
3
The progressive wave represents a special type of wave, which illlustrates the
properties of a given elastic medium concerning the propagation of waves in
general.
12.2 Progressive Waves. Free and Characteristic Vibrations 525
where a and m are unit vectors called direction of motion and direction of
propagation, respectively, and U is a real function of class C2, with second
derivatives not identically zero, i.e. the mapping R
s ! UðsÞ 2 R has the
property
d2 UðsÞ
U 00 ðsÞ ¼ 6¼ 0; ð12:28Þ
ds2
c is a real constant, called velocity of propagation.
If a and m are linearly dependent, which is equivalent to say (the two being unit
vectors) that a ¼ m, the wave is called longitudinal; if a is normal to m, then
the wave is called transverse.
One can see that u and a are collinear at any moment and at any point. Since
u represents the displacement vector, it follows that a, permanently collinear with
u, represents the direction of motion. Let us consider a 2 R; a ¼ const: Let us also
discuss the plane P fx 2 E3 jx m ct ¼ ag. From (12.27) it follows that u is
constant on P, having the value uðx; tÞ ¼ aUðaÞ. Hence, all points of this plane, at
every moment t, have the same displacement of direction a and magnitude UðaÞ.
The normal vector to the plane P is m; as a function of t, P moves in the direction
m with the velocity c. These considerations justify the terms: direction of motion,
direction of propagation, velocity of propagation, longitudinal wave and transverse
wave.
Gradu ¼ ru ¼ U 0 a m; u
€ ¼ c2 U 00 a: ð12:29Þ
AðmÞa ¼ c2 a; ð12:32Þ
a relation called the Fresnel-Hadamard propagation condition (for elastic pro-
gressive waves).
This condition expresses the following physical fact: in order that an elastic
progressive wave, with direction of propagation m, should be propagated, it is
necessary and sufficient that the direction of motion a be an eigenvector for the
acoustic tensor A(m) and the square of the velocity of propagation be the corre-
sponding eigenvalue.
If, H is symmetric, then A(m) is symmetric too (see Sect. 12.2.1.1, point (i));
hence, there exist at least three eigenvectors for any m. But, in general, the
eigenvalues that give the squares of the corresponding velocities of propagation
are not always positive, hence it is possible that the velocities of propagation be
not real.
If in addition, H is strictly elliptic or positive definite, then A is positive definite
and the velocities of propagation are real.
In conclusion, using also the general results of Sect. 1.2.1.1, if H is symmetric
and strictly elliptic, then for any direction of propagation m there exist three
mutually perpendicular directions of propagation and, correspondingly, three real
direction velocities of propagation.
If a is a possible direction of motion corresponding to the direction of propa-
gation m, the velocity of propagation is given by the relation
c2 ¼ a ðAðmÞaÞ; ð12:33Þ
Since H is symmetric with respect to the last two indices, one may write
r ¼ Hru: ð12:34Þ
This relation allows us to reduce the system of 15 equations of elastodynamics to
Lamé’s system of three scalar equations
DivðHruÞ þ F ¼ q€
u; ð12:35Þ
let us denote by L the operator (mapping) L : D ! L2 ðXÞ (L2 ðXÞ represents the
space of square integrable functions on X) given by the relation
Lu ¼ DivðHruÞ; ð12:37Þ
L is called Lamé’s operator.
Thus, Lamé’s equations can be written as
Lu þ q€
u ¼ F; ð12:350 Þ
it is easy to conclude that, if the operator L is symmetric and negative definite, then
Lamé’s equations are hyperbolic in the dynamic case and elliptic in the static case.
The problem of free vibrations of an elastic body is a study of the motions of the
form
528 12 Dynamical Problems of Elastic Bodies
Lu ¼ qx2 u: ð12:39Þ
Let us discuss the analogy between the relation (12.39) and the Fresnel-Had-
amard propagation condition.
Let S1, S2 represent a partition of the boundary oX of the body, i.e.
S1 \ S2 ¼ [, S1 [ S2 ¼ oX. The problem of free vibrations can be formulated as
the problem of finding possible combinations of the circular frequency x and of
the amplitude u. such that (12.39) be satisfied for bodies X fixed on the part S1 of
the boundary and free otherwise.
In a mathematical formulation, this means the determination of the so-called
characteristic solutions of the problem of free vibrations, which we shall now
explain.
By characteristic solution we mean an ordered couple ðk; uÞ; k 2 R; u 2 D,
such that
Lu ¼ ku; jjujj ¼ 1; ð12:40Þ
ujS1 ¼ 0; ð12:400 Þ
n
rðuÞjS2 ¼ 0; ð12:4000 Þ
Z Z
½u; v ¼ Hijkl ui;j vk;l dX ¼ Hijkl eij ðuÞekl ðvÞdX
ZX ZX ð12:41Þ
¼ rij ðvÞeij ðuÞdX ¼ rkl ðuÞekl ðvÞdX
X X
represents the energetic scalar product and the norm (juj2 ¼ ½u; u), respectively.
Supposing that H is symmetric, positive definite and of class C0 ðXÞ, we have
12.2 Progressive Waves. Free and Characteristic Vibrations 529
R
denoted by hu; vi ¼ Xu vdX; jjujj2 hu; ui, the scalar product and the norm in
n
L2 ðXÞ, respectively. If rðuÞ represents the stress vector on the surface element of
external normal n, then
n
rðuÞ ¼ rn ¼ ðHruÞn; ð12:42Þ
using the above notations, one may prove (using the Gauss-Ostrogradski formula)
Betti’s relation, which states that
Z
n
hLu; vi ¼ ½u; v ¼ rðuÞ vdS; 8u; v 2 D: ð12:43Þ
oX
(ii) k ¼ 0 if and only if u is a rigid displacement of the body; moreover, there exist
characteristic solutions (k, u) with k ¼ 0 if and only if the body is free, in
which case any rigid displacement u with jjujj ¼ 1 is a characteristic dis-
placement corresponding to the eigenvalue k ¼ 0.
With this notation, one may state a minimum principle in following form.
Let us consider u1 2 D \ K such that
ju1 j jvj; 8v 2 K; ð12:46Þ
The limit problems of elastodynamics (the study of the action of external loads,
variable in time, the study of free and forced vibrations etc.) have made the object
of many researches, even in the last two centuries.
Thus, a problem qualitatively different from that of the elastic space, considered
in Sect. 6.2.4.3, is the problem of the elastic half-space; in this field, Lord Ray-
leigh [10, 29] discovered in 1887 the surface waves, called nowadays by his name,
L. Knopoff [20] gave numerical results concerning the propagation velocity of
Rayleigh’s waves. In the case of a stratified body, formed by an elastic layer,
superposed on an elastic half-space with different mechanical properties, we find
the waves of A. E. H. Love [9].
Another important related problem is that of a point or a linear load, variable in
time and acting upon the separation plane of the elastic half-space; the first study
in this direction was made by H. Lamb [21], whose name was given to this
problem. Interesting results were given, in a preliminary form, in this direction, by
F. Santer [30]; these results have been numerically estimated by K. B. Broberg
[12]. The case of a load, varying in time and applied within the elastic half-space,
was considered by E. Pinney [27], C. C. Pekeris [25] and by C. C. Pekeris and
H. Lifson [26], who later gave numerical results. Other researches into the elastic
half-space were made by G. Eason [16, 17], L. M. Flitman [18], M. Hayes and
R. S. Rivlin [19], H. A. Lang [22] and C. C. Chao, H. H. Bleich and J. Sackmann
[14]; some of the results, especially those about the case of a state of plane strain,
are due to J. W. C. Sherwood [31]. C. C. Chao [13] used the technique of inversion
of the integral transforms, given by L. Cagniard [3], to obtain results in case of a
concentrated tangential force, suddenly applied to the separating plane of the
elastic half-space and maintained constant in time. P. P. Teodorescu [11] showed
how to obtain waves of Rayleigh type, corresponding to a three-dimensional state
of strain and stress (in the classical case, these waves correspond to a plane state of
strain). He indicated moreover how to study systematically, by means of the same
family of stress functions, the elastic parallelepiped and all the infinite domains
derived from it by throwing to infinite one or several of its faces, from the
standpoint of both free and forced vibrations [35, 36].
In connection with the waves propagation, the monographs of L. M. Ber-
ekhovskikh [2], R. M. Davies [4], W. M. Ewing, W. S. Jardetzky and F. Press [5],
H. Kolsky [8] and W. M. Ewing and F. Press [6]. As well, we mention
the monographs of W. Kecs and P. P. Teodorescu [7], I. Beju et al. [1] and
P. P. Teodorescu [11]. We mention moreover the synthesis studies of R. M. Davies
[15] and J. Miklowitz [23], as well as the syntheses study presented by Teodorescu
[34] at a Conference on non-linear vibrations in Berlin.
Hereafter, we shall deal with the problem of free and forced vibrations in the
case of the elastic half-space and give some indications concerning other infinite
spatial domains. We use the results given in [34], as well as the methods of the
theory of distributions as in [7].
532 12 Dynamical Problems of Elastic Bodies
The influence of the volume forces on the results given hereafter can be
introduced with the help of some particular integrals. Thus, can be considered
volume forces such as Fi ðx1 ; x2 ; x3 ; tÞ ¼ F i ðx1 ; x2 ; x3 Þ sin xt; i ¼ 1; 2; 3; if for
i ðx1 ; x2 ; x3 Þ we use representations made with the help of Fourier series and
F
integrals, then we can easily compute the corresponding particular integrals in the
same form; we remark that the boundary conditions are of the same form as those
used in the absence of the volume forces. Thus, the problem where non-vanishing
volume forces appear is reduced to the similar problem without volume forces, but
with conventional boundary conditions.
In the steady-state problems it is not necessary to consider initial conditions, the
only limit conditions to be verified being the boundary conditions.
Hereafter, we shall give some results developed for the elastic half-space; we shall
then supply a few computing hints concerning other space domains.
Let be the elastic half-space x3 0, on the separation plane of which acts the
normal periodic load
pðx1 ; x2 ; tÞ ¼ pðx1 ; x2 Þ sin xt; ð12:48Þ
for simplifying the computation, we took a only one term from the expansion with
respect to time. The results thus obtained can help, by superposition of effects, in
the case of a Fourier series or of a Fourier integral with respect to time.
We shall admit that the load is periodic with respect to the variables x1 and x2
and antisymmetric with respect to both axes of co-ordinates Ox1 and Ox2
XX
pðx1 ; x2 Þ ¼ anm sin an x1 sin bm x2 ; ð12:49Þ
n m
with
np mp
an ¼ ;b ¼ ; n; m ¼ 1; 2; 3; . . .; ð12:490 Þ
a1 m a2
where Li ¼ 2ai ; i ¼ 1; 2, are the periods.
The loading cases with other symmetry or antisymmetry properties with respect
to the co-ordinate axes, as well as the cases of tangential loads, can be similarly
studied.
We can choose the stress functions, odd with respect to x1 and x2 , in the form
12.3 Forced and Free Vibrations 533
XX
Fi ðx1 ; x2 ; x3 ; tÞ ¼ sin xt ðA0nm ecnm x3
n m
þ B0nm e dnm x3
Þ sin an x1 sin bm x2 ; i ¼ 1; 2; 3; ð12:50Þ
XX
1
r23 ¼ sin xt cnm dnm b2m Anm ecnm x3 þ d2nm b2m
n m
d nm b m
1 2 2 1
þ mk ðdnm b2m Þ B0nm þ d2nm b2m k2 ð2d2nm þ mb2m Þ B00nm
2 2
1
þ d2nm b2m þ k2 ðmd2nm þ 2b2m Þ B000 nm e dnm x3
sin an x1 sin bm x2 ;
2
XX 1
r31 ¼ sin xt cnm dnm a2n Anm ecnm x3 þ d2nm a2n
n m
dnm an
1 2 2 1 2 2
k ð2dnm man Þ Bnm þ dnm an þ mk ðdnm an Þ B00nm
2 0 2 2 2 ð12:530 Þ
2 2
2 2 1 2 2 2 000 dnm x3
þ dnm an þ k ðmdnm þ 2an Þ Bnm e sin an x1 cos bm x2 ;
2
XX 1
r12 ¼ sin xt a2n b2m Anm ecnm x3 þ a2n b2m
n m
an bm
1 1
þ k2 ðma2n 2b2m Þ B0nm þ a2n b2m k2 ð2a2n mb2m Þ B00nm
2 2
1
þ a2n b2m þ mk2 ða2n þ b2m Þ B000 nm e dnm x3
cos an x1 cos bm x2
2
by subtracting the last two equations from the first one, we get
ða2n þ b2m Þ½ð1 mÞc2nm 2cnm dnm þ md2nm Anm d2nm ½a2n þ b2m
ð2 mÞðc2nm d2nm ÞðB0nm þ B00nm Þ d2nm ½a2n þ b2m
þ 2mðc2nm d2nm Þ ¼ ða2n þ b2m Þanm : ð12:57Þ
Taking into account the relation (12.5100 ), we can write the first Eqs. (12.56) and
the Eq. (12.57) in the form
with
likewise, we obtain
where, for symmetry’s sake, there remains only to be determined the rotation of
rigid body x03 .
If the displacements vanish at t ¼ 0 or if we impose the condition that the
displacements should be periodic functions, we shall have x03 ¼ 0.
The distorted form of the separation plane is supplied by
qx2 X X anm
u3 ðx1 ; x2 ; 0; tÞ ¼ sin xt c sinan x1 sin bm x2 : ð12:64Þ
4l2 n m
Dnm nm
The above series are uniformly convergent, while the series (12.64) is rapidly
convergent.
In case of a local load, we can make use of Fourier integrals with regard to the
variables x1 and x2 .
where the parameters cnm ; dnm are related by (12.51). The functions Fij ; i 6¼ j;
i; j ¼ 1; 2; 3, are taken equal to zero.
The sequences of coefficients A0nm ; B0nm ; . . .; B000
nm will be defined by the boundary
conditions
x3 ¼ a3 : r33 ¼ pðx1 ; x2 ; tÞ; r31 ¼ r32 ¼ 0; ð12:66Þ
while the state of stress and the state of displacement will be expressed by means
of uniformly convergent series.
In case of the elastic quarter-space x1 0; x3 0, we shall admit that one
single local normal load in the form
Z 1Z 1
pðx1 ; x2 Þ ¼ aða1 ; a2 Þ cos a1 x1 cos a2 x2 da1 da2 ; ð12:67Þ
0 0
with
Z 1 Z 1
4
aða1 ; a2 Þ ¼ 2 pðn1 ; n2 Þ cos a1 n1 cos a2 n2 dn1 dn2 ; ð12:670 Þ
p 0 0
538 12 Dynamical Problems of Elastic Bodies
and the problem is reduced to the integration of a system of three integral equa-
tions of the second kind of Fredholm type.
In case of an elastic half-layer and of an elastic quarter-layer, we shall use
stress functions built up of Fourier series, along a direction normal to the parallel
separation planes and of Fourier integrals in the other directions.
In the case of an elastic strip jx2 j a2 ; jx3 j a3 undergoing normal loads such
as (12.48), with
XX
pðx1 ; x2 Þ ¼ alm sin al x1 sin bm x2 ð12:73Þ
l m
on the separation planes x3 ¼ a3 , we shall use stress functions of the form
XX
F1 ðx1 ; x2 ; x3 ; tÞ ¼ sin xt B01nl sinh l1nl x2
n l
þB02nl sinh l2nl x2 sin cn x3 sin al x3
XX
0 0
þ sin xt C1lm sinh m1lm x3 þ C2lm sinh m2lm x3 sin al x1 sin bm x2 ; . . .;
l m
ð12:74Þ
XX
Fi ðx1 ; x2 ; x3 ; tÞ ¼ sin xt ðA01mn sinhk1mn x1
m n
þ A02mn sinhk2mn x1 Þ sin bm x2 sin cn x3
XX
þ sin xt ðB01nl sinhl1nl x2 þ B02nl sinhl2nl x2 Þ sin cn x3 sin al x1
n
l
XX
0 0
þ sin xt ðC1lm sinhm1lm x3 þ C2lm sinhm2lm x3 Þ sin al x1 sin bm x2 ; . . .;
l m
ð12:77Þ
with the relations (12.75), as well as with the relations
1 2 1
b2m þ c2n ¼ k21mn þ x ¼ k22mn þ 2 x2 : ð12:750 Þ
c21 c2
The sequences of coefficients A01mn ; A02mn ; . . .; C2lm
000
will be determined by the
boundary conditions
Hereafter, we shall give results and computational hints about the free vibrations
of the domains considered in the previous section, by emphasizing chiefly the
Rayleigh waves,
1 2m q 2 q
a2n þ b2m ¼ c2nm þ x ¼ d2nm þ x2nm : ð12:80Þ
2ð1 mÞ l nm l
The components of the stress tensor tend to zero when x3 ! 1 and the
sequence of coefficients A0nm ; B0nm ; . . .; B000
nm are defined by the boundary conditions
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffi 1 2m
4v v 1 v ¼ ð2v 1Þ2 ; ð12:87Þ
2ð1 mÞ
If we remark that
anm 2Anm
¼ 2 ð12:92Þ
Dnm dnm þ a2n þ b2m
cnm ðtÞ ¼ Anm ða2n þ b2m ÞðCnm sin xnm t þ Dnm sin xnm tÞ; ð12:93Þ
XX cnm bm cnm x3
r23 ¼ unm ðtÞ e ednm x3 sin an x1 sin bm x2 ;
a 2 þ b2
n m n m
XX c an
r31 ¼ unm ðtÞ nm 2 ecnm x3 ednm x3 cos an x1 sin bm x2 ; ð12:940 Þ
2
an þ bm
n m
XX
an bm cnm x3 1 dnm x3
r12 ¼ unm ðtÞ e 1 e cos an x1 sin bm x2 ;
n m a2n þ b2m 2v
where the rotation of rigid body x03 must be determined by an additional condition
of fixed point of the elastic half-space.
On the separation plane, appear the normal stresses
1 XX m
r11 ðx1 ; x2 ; 0; tÞ ¼ u ðtÞ
2v n m nm 1m
!
a2n
þ sin an x1 sin bm x2 ;
a2n þ b2m
ð12:96Þ
1 XX m
r22 ðx1 ; x2 ; 0; tÞ ¼ u ðtÞ
2v n m nm 1m
!
b2m
þ sin an x1 sin bm x2
a2n þ b2m
If the stress functions (12.77) are expressed in cos an x1 and cos bm x2 we shall
obtain similar results and the same sequence of eigenvalues.
We come thus to a state of strain and stress corresponding to the Rayleigh
waves and generalizing the classical results suppplied in the case of a state of plane
strain.
In the case of: the elastic layer, the elastic half-layer or the elastic quarter-layer,
the elastic quarter-space or the elastic eights-space, the elastic strip or the elastic
half-strip, as in the case of the elastic parallelepiped, the study of the free
vibrations can be made, as in the previous subsection, by using the results obtained
in the case of forced vibrations.
References
A. Books
1. Beju, I., Soós, E., Teodorescu, P.P.: Euclidiean Tensor Calculus with Applications. Ed.
Tehnică, Bucuresßti, Abacus Press, Tunbridge Wells, Kent (1983)
2. Brekhovskikh, L.M.: Waves Layered Media. Appl. Math. Mech. 6, Academic Press, New
York (1960)
3. Cagniard, L.: Reflexion et refraction des ondes seismiques progressives. Gauthier-Villars,
Paris (1935)
4. Davies, R.M.: Stress Waves in Solids. Surv. Mech. Cambridge University Press, Cambridge
(1956)
5. Ewing, W.M., Jardetzky, W.S., Press, F.: Elastic Waves in Layered Media. McGraw-Hill
Book Co., Inc., New York (1957)
6. Ewing, W.M., Press, F.: Surface Waves and Guided Waves. Enc. Phys. 47, 75. Springer,
Berlin (1956)
7. Kecs, W., Teodorescu, P.P.: Applications of the Theory of Distributions in Mechanics. Ed.
Acad., Bucuresßti, Abacus Press, Tunbridge Wells, Kent (1974)
546 12 Dynamical Problems of Elastic Bodies
8. Kolsky, H.: Stress Waves in Solids. Oxford University Press, London (1953)
9. Love, A.-E.-H.: Treatise on the Mathematical Theory of Elasticity, 4th edn. Cambridge
University Press, London (1934)
10. Rayleigh, L., Strutt, J.W.: The Theory of Sound, 2nd edn. Dover Publishing, New York
(1945)
11. Teodorescu, P.P.: Dynamics of Linear Elastic Bodies. Ed. Acad., Bucuresßti, Abacus Press,
Tunbridge Wells, Kent (1975)
B. Papers
12. Broberg, K.B.: Shock waves in elastic and elastic-plastic media. Kungl. Fort. Befast,
Stockholm, 12 (1956)
13. Chao, C.C.: Dynamic response of an elastic half-space in tangential surface loadings. J. Appl.
Mech. 27, 559 (1960)
14. Chao, C.C., Bleich, H.H., Sackman, J.: Surface waves in an elastic half-space. ASME-Trans.
J. Appl. Mech. 300 (1961)
15. Davies, R.M.: Stress waves in solids. Appl. Mech. Rev. 6, 1 (1953)
16. Eason, G.: On the torsional impulsive loading of an elastic half-space. Quart. J. Mech. Appl.
Math 17, 279 (1964)
17. Eason, G.: On the torsional impulsive body force within an elastic half-space. Math 2, 75
(1964)
18. Flitman, L.M.: Dynamic problem of the die on an elastic half-space. J. Appl. Math. Mech 23,
997 (1959)
19. Hayes, M., Rivlin, R.S.: Surface waves in deformed elastic materials. Arch. Rat. Mech. Anal
8, 358 (1961)
20. Knopoff, L.: On rayleigh wave velocities. Bull. Seism. Soc. Amer 42, 307 (1952)
21. Lamb, H.: On the propagation of tremors over the surface of an elastic solid. Phil. Trans. Roy.
Soc. London Ser. A, 203, 1 (1904)
22. Lang, H.A.: Surface displacements in an elastic half-space. ZAMM 41, 141 (1961)
23. Miklowitz, J.: Recent developments in elastic wave propagation. Appl. Mech. Rev. 13, 12
(1960)
24. Noll, W.: Verschiebungsfunktionen für elastische Schwingungprobleme. ZAMM 37, 81
(1957)
25. Pekeris, C.C.: The seismic surface pulse. Proc. Nat. Acad. Sci 41, 469 (1955)
26. Pekeris, C.C., Lifson, H.: Motion on the surface of a uniform elastic medium. Bull. Seism.
Soc. Am 29, 1233 (1957)
27. Pinney, E.: Surface motion due to a point in a semi-infinite elastic medium. Bull. Seism. Soc.
Am 44, 571 (1954)
}
28. Predeleanu, M.: Uber die Verschiebungsfunktionen für das achsensymmetrische problem der
Elastodynamik. ZAMM 38, 402 (1958)
29. Rayleigh, L., Strutt, J.W.: On waves propagated along the plane surface of an elastic solid.
Proc. London Math. Soc. 17, 4 (1887)
30. Santer, F.: Der elastische Halbraun bei einer mechanischen Beeinflussung seiner Oberfläche.
ZAMM 30, 203 (1950)
31. Sherwood, J.W.C.: Elastic wave propagaton in a semi-infinite medium. Proc. Phys. Soc.
London 71, 207 (1958)
32. Teodorescu, P.P.: Schwingungen der elastischen Kontinua. Abh. der deutschen Akad. Wiss.,
Berlin, Kl. Math. Phys. Techn. 29 (1965)
}
33. Teodorescu, P.P.: Uber das dreidimensionale problem der elastokinetik. ZAMM 45, 513
(1965)
34. Teodorescu, P.P.: Sur quelques problèmes dynamiques de la théorie de l’élasticité. Rev.
Roum. Math. Pures Appl. 11, 773 (1966)
Chapter 13
Particular Cases of States of Strain
and Stress
Hereafter we shall put conditions for normal or for tangential stresses, obtaining
general stress functions for the states of stress thus put in evidence.
First of all, we deal with the general problem of elastostatics which results by
putting the condition that one normal stress vanishes, namely
r33 ¼ 0: ð13:1Þ
Taking into account the condition (13.1), in the absence of volume forces, the third
equation of equilibrium (3.62) leads to
r31 ¼ K2 ; r23 ¼ K;1 ; ð13:2Þ
wherefrom
r11 ¼ A;2 ; r22 ¼ B;1 ;
ð13:20 Þ
r12 þ K;3 ¼ A;1 ; r12 K;3 ¼ B;2 ;
;3 ;
A¼A ;3 ;
B¼B ð13:4Þ
¼ Aðx
with A 1 ; x2 ; x3 Þ; ¼ Bðx
B 1 ; x2 ; x3 Þ; so that the relation (3.30 ) becomes
1
K ¼ ðA ;1 B;2 Þ þ C; ð13:3000 Þ
2
where C ¼ Cðx1 ; x2 Þ is an arbitrary function in two variables only.
Taking into account (13.30 ) and (13.4), the relations (13.2), (13.20 ) and (13.3)
allow us to represent the state of stress in the form
;23 ;
r11 ¼ A ;31 ;
r22 ¼ B ð13:5Þ
1 1
r23 ¼ ðA ;11 B;12 Þ C;1 ; r31 ¼ ðA ;22 B;22 Þ þ C;2 ; ð13:50 Þ
2 2
1
r;12 ¼ ðA ;31 þB;23 Þ: ð13:500 Þ
2
Taking now into account the condition (13.1), the third equation of Beltrami (5.43)
leads to
H;33 ¼ 0; ð13:6Þ
A ;1 ¼ 1 x23 H0 þ x3 H0 þ w;
;2 þB ð13:10Þ
2
where w ¼ wðx1 ; x2 Þ is an arbitrary function. Similarly, starting from (13.5),
(13.50 ) and introducing in the first two equations (5.43) and in the third Eq. (5.430 ),
one may write
DA ;23 þ 1 x3 H0;11 þ H0;11 ¼ 0; DB ;13 þ 1 x3 H0;22 þ H0;22 ¼ 0;
1þm 1þm
ð13:11Þ
1
;23 ¼ 1
DA;13 þDB x3 H0;12 þ H0;12 : ð13:110 Þ
2 1þm
Taking into account (13.70 ), the Eqs. (13.11) become
DA ;23 ¼ 1 x3 H0;22 þ H0;22 ; DB ;13 ¼ 1 x3 H0;11 þ H0;11 ;
1þm 1þm
integrating, it results
1 1 2
¼
DA x3 H0;2 þ x3 H0;2 þ f1 þ f2;2 ;
1þm 2
ð13:12Þ
¼ 1
DB
1 2
x3 H0;1 þ x3 H0;1 þ g1 þ g2;1 :
1þm 2
To specify the function f1 ¼ f1 ðx1 ; x3 Þ and g1 ¼ g1 ðx2 ; x3 Þ; we remark that the
Eq. (13.110 ) must be verified too; one has
f1;13 þ g1;23 ¼ 0:
f1 ¼ x1
h þ h1 þ h3 ; g1 ¼ x 2
h þ h2 þ h4 : ð13:13Þ
Applying Laplace’s operator to the relation (13.3000 ) and taking into account
(13.800 ), (13.12) and (13.13), it results that
1 1 1
DK ¼
h ðf2 g2 Þ;12 ðh3;1 h4;2 Þ þ DC ¼ Du þ h:
2 2 1þm
13.1 Conditions for Stresses 551
u¼u 0;12 ; C ¼ C
;12 ; H0 ¼ H ;1122 ; ð13:140 Þ
Taking into account the theorem of Boggio (see A.1.2.7), the functions u and H0
differ from a biharmonic and a harmonic function, respectively, only by a sum of
two functions of only one variable. Introducing in (13.14), one observes that the
latter functions may be neglected; one may thus suppose that the functions u
and
H0 verify the equations
u0 ¼ 0;
DD 0 ¼ 0
DH ð13:15Þ
and that the Cauchy-Riemann equations
0;2 ; D
u;1 ¼ H
D 0;1
u;2 ¼ H ð13:150 Þ
H0 ¼ H
0
;12 ð13:18Þ
0 ¼ H
has been used, the function H 0 ðx1 ; x2 Þ being harmonic (similar consider-
ations to those above)
0 ¼ 0:
DH ð13:19Þ
Denoting further
;
w¼w ð13:180 Þ
;12
¼ wðx
where w 1 ; x2 Þ is an arbitrary function, the relation (13.10) becomes
;1 ¼ 1 x23 H
;2 þ B
A 0 þ x3 H
0 þ w : ð13:20Þ
2 ;12
Let be
¼ U;1 ;
A ¼ W;2 :
B ð13:200 Þ
The formulae (13.5)–(13.500 ) allow to write the state of stress in the form
r11 ¼ U;123 ; r22 ¼ W;123 ; ð13:21Þ
1
;1122 W;22 ; 1
;1122 W;22 ; ð13:210 Þ
r23 ¼ U;11 2C ;1
r31 ¼ U;11 2C
2 2
1
r12 ¼ U;11 þ W;22 ;3 ; ð13:2100 Þ
2
where (13.140 ) has been taken into account too.
Introducing (13.200 ) in (13.17) and in (13.20) and integrating, one obtains
1 1 2 0 1 1
DU ¼ x3 H0 þ x3 H þ DC þ x þ H0 þ a1 x21 h þ x1 h1 ;
1þm 2 ;22 1 þ m 2
ð13:22Þ
1 1 2 1 1
DW ¼ x H0 þ x3 H0 þx
DC þ H0 þ a2 þ x22 h þ x2 h2 ;
1þm 2 3 ;11 1þm 2
1 0
U þ W ¼ x23 H 0 þ x 3 H þ w þ b 1 þ b2 ; ð13:220 Þ
2
where one took into account (13.150 ) and where one has introduced the functions
of two variables a1 ¼ a1 ðx2 ; x3 Þ; a2 ¼ a2 ðx1 ; x3 Þ; b1 ¼ b1 ðx2 ; x3 Þ; b2 ¼ b2 ðx1 ; x3 Þ:
Applying the operator D to the relation (13.220 ) and taking into account (13.22),
(13.15) and (13.19), it results
13.1 Conditions for Stresses 553
2 1 þ a1 þ a2
DðU þ WÞ ¼ H0 þ Dx ðo11 o22 ÞDC
1þm 2
1 þ Dðb þ b Þ;
0 þ Dw
ðx21 x22 Þh þ x1 h1 þ x2 h2 ¼ H 1 2
2
which leads to
1
a1 ¼ Db1 x22 h x2 h2 þ c2 þ c3 ;
2 ð13:23Þ
1
a2 ¼ Db2 þ x21 h x1 h1 þ c1 c3 ;
2
1 1m
D x w ðo11 o22 ÞC ¼ H 0 c1 c2 ; ð13:230 Þ
2 1þm
x¼x
þw 1 mH
þ 1 ðo11 o22 ÞC ;
0
2 1þm
¼ xðx
where x 1 ; x2 Þ is a harmonic function
¼0
Dx ð13:24Þ
¼H
and H ðx ; x Þ is a particular integral of the equation
0 0 1 2
¼H
DH 0: ð13:240 Þ
0
The relation (13.220 ) allows to express the functions U and W in the form
1 1 2 þb þb ;
0 þ w
U¼vþ x3 H0 þ x3 H 1 2
2 2
ð13:25Þ
1 1 2 þb þb ;
0 þ w
W ¼ v þ x3 H0 þ x3 H 1 2
2 2
ð13:250 Þ
554 13 Particular Cases of States of Strain and Stress
where (13.250 ), (13.15), (13.19), (13.24), (13.240 ) have been taken into consider-
ation too.
One remarks that the function v appears only by its derivative with respect to
x3 . Thus, in the right member of the Eq. (13.250 ) one may neglect the functions
which depend only on the variables x1 and x2 ; indeed, these functions lead to
particular integrals which may depend only on the variables x1 and x2 ; which
disappear by differentiation.
Taking into account (13.15), (13.19), the Eq. (13.250 ) allows to write
1 1 0 1
Dv ¼ ðo11 o22 Þ x23 H 0 þ x 3 H þ Dðb1 b2 Þ
2ð1 þ mÞ 2 2
1 2
x1 þ x22 h þ x1 h1 x2 h2 þ c3 ;
2
wherefrom
1 1 0 1
v ¼
v ðx1 o1 x2 o2 Þ x23 H 0 þ x 3 H þ ðb1 b2 Þ
4ð1 þ mÞ 2 2
Z Z Z Z Z Z
1 2
x1 þ x22 dx3 hdx3 þ 2 dx3 dx3 dx3 hdx3
2
Z Z Z Z Z Z
þ x1 dx 3 h 1 dx3 x2 dx 3 h 2 dx3 þ dx3 c3 dx3 ;
13.1 Conditions for Stresses 555
v¼
where vðx1 ; x2 ; x3 Þ is a harmonic function
v ¼ 0;
D ð13:27Þ
the relation (A.102) has been taken into account too.
It results that
Z
1
v;3 ¼ X ð x 1 o1 x 2 o2 Þ ð x 3 H 0 Þ þ 1 ðb1 b2 Þ 1 x2 þ x2
0 þH hdx3
;3 1 2
4ð1 þ mÞ 2 2
Z Z Z Z Z Z
þ 2 dx3 dx3 hdx3 þ x1 h1 dx3 x2 h2 dx3 þ c3 dx3 ;
ð13:28Þ
with the notation
X¼
v;3 :
The conditions (13.27) allows to state that also the function X ¼ Xðx1 ; x2 ; x3 Þ is
harmonic
DX ¼ 0: ð13:29Þ
Introducing (13.28) in (13.26), one obtains
1 0 Þ þ 1 ðx3 H
0 þ H 0 þ H
0Þ ;
r11 ¼ X;12 ðx1 o112 x2 o122 Þðx3 H ;12
4ð1 þ mÞ 2
ð13:30Þ
1 0 þ H 0 Þ þ 1 ðx3 H
0 þ H
0Þ :
r22 ¼ X;12 þ ðx1 o112 x2 o122 Þðx3 H ;12
4ð1 þ mÞ 2
1 1 1 2 0
r23 ¼ X;13 þ ðo11 o22 Þ m x3 H0 þ x3 H þ ð1 mÞH0
2 4ð1 þ mÞ 2 ;1
1 1
0;1 ðo11 o22 Þx
þ ð x 1 o1 x 2 o2 Þ H ;1 ;
8ð1 þ mÞ 4
1 1 1 2 0
r31 ¼ X;23 ðo11 o22 Þ m x3 H0 þ x3 H þ ð1 mÞH0
2 4ð1 þ mÞ 2 ;2
1 1
0;2 þ ðo11 o22 Þx
ð x 1 o1 x 2 o2 Þ H ;2 :
8ð1 þ mÞ 4
ð13:3000 Þ
Taking into account (13.240 ), one may write
Z Z
1 1 0;1 þ 2H 0
ðo11 o22 ÞH0 ¼ ðo11 o22 Þ x1 H0 dx1 þ x2 H0 dx2 ¼ x1 H
4 4
Z Z
1 0;
þ x2 H0;11 dx2 x2 H0;2 þ 2H0 þ x1 H0;22 dx1 þ ð x 1 o1 x 2 o2 Þ H
2
Du0 ¼ 0:
Thus, the tangential stresses (13.3000 ) become
1 m 1 2 0
r23 ¼ X;13 þ ðo11 o22 Þ x3 H0 þ x3 H
2 4ð1 þ mÞ 2 ;1
2m 1
u0 ;1 ðo11 o22 Þx
u ;1 ;
2ð1 þ mÞ ;12 4
1 m 1 2 0
r31 ¼ X;23 ðo11 o22 Þ x3 H0 þ x3 H
2 4ð1 þ mÞ 2 ;2
2m 1
þ u0 ;2 þ ðo11 o22 Þx
u ;2 :
2ð1 þ mÞ ;12 4
13.1 Conditions for Stresses 557
1 1
r12 ¼ ðo11 o22 ÞX ðx1 o2 x2 o1 ÞH: ð13:3100 Þ
2 4ð1 þ mÞ
The function X ¼ Xðx1 ; x2 ; x3 Þ is harmonic (Eq. (13.29)). The sum of the
normal stresses is a linear with respect to x3 function, given by (13.7); the func-
tions H0 ¼ H0 ðx1 ; x2 Þ and H0 ¼ H0 ðx1 ; x2 Þ are harmonic (Eq. (13.70 )). The bi-
harmonic function u ¼ uðx1 ; x2 Þ (Eq. (13.9)) is connected by its Laplacian to the
function H0 (Cauchy-Riemann Eqs. (13.800 )).
One also observes that the function X must be of class C4 ; as well as the
function u; it is sufficient that the functions H0 and H0 be of class C3 :
Moreover, one remarks that the given representation depends only on three
arbitrary functions, i.e.: the biharmonic function X (in three variables), the har-
monic function H0 and the biharmonic function u (the two latter functions in two
variables). Because one must determine a function in three variables too, the
problem is, in general, a quasi-bidimensional problem.
Starting from the equations of the problem, one arrived step by step to the final
representation; hence, this representation is complete (any state of stress may be
represented in this form).
As it has been shown in Sect. 4.1.3.7, if m ¼ 1=2 one has to do with an elastic
incompressible body; as well, if H ¼ 0 (without having m ¼ 1=2), one has a state
558 13 Particular Cases of States of Strain and Stress
H0 ¼ H0 ¼ 0: ð13:32Þ
the last term of which appears because the function must be harmonic.
One can put C ¼ 0 and the state of stress takes the form
r11 ¼ r22 ¼ X;12 ; ð13:34Þ
1 1
r23 ¼ X;13 f;1 ¼ X;3 þ f ;
2 2 ;1
ð13:340 Þ
1 1
r31 ¼ X;23 þ f;2 ¼ X;3 þ f ;
2 2 ;2
1
r12 ¼ ðo11 o22 ÞX: ð13:3400 Þ
2
Hence, the stress function X ¼ Xðx1 ; x2 ; x3 Þ corresponds to an incompressible state
of deformation. The function f must be of class C3 ; in the compressible case, it is
included in the function u:
13.1 Conditions for Stresses 559
Taking into account (2.43) and introducing the notation (13.140 ) and (13.18), with
¼ x3 H
H 0;
0 þ H ð13:36Þ
¼ X;2
1 þ 1m H
ðx1 o1 x2 o2 ÞH ;2 ;
4ð1 þ mÞ 2ð1 þ mÞ
ð13:37Þ
2l½u2 ðx3 x01 þ x1 x03 þ u02 Þ
¼ X;1 þ
1 þ 1m H
ðx1 o1 x2 o2 ÞH ;1 ;
4ð1 þ mÞ 2ð1 þ mÞ
0 0 0 1 2
2l½u3 ðx1 x2 þ x2 x1 þ u3 Þ ¼ m x3 H0 x3 H þ g; ð13:370 Þ
2
These two problems may be put in connection with the study of plates of mean
thickness, acted upon by loads normal to the middle plane in the case of a state of
stress symmetric or antisimmetric with regard to the middle plane, respectively.
In the first case, the sum of the normal stresses is given by
1
H ¼ x23 DH0 þ x3 H0 þ H0 ð13:41Þ
2
and the normal stress r33 takes the form
1
r33 ¼ H0 þ #; ð13:410 Þ
1þm
where the function H0 ¼ H0 ðx1 ; x2 Þ is biharmonic
DDH0 ¼ 0 ð13:42Þ
Starting from the results obtained above for the case of a zero normal stress, one
may put supplementary conditions to the other components of the stress tensor,
obtaining results which can be interesting in various problems.
a ¼ a0 x 3 þ a0 ; b ¼ b0 x 3 þ b0 ; c ¼ c 0 x 3 þ c 0 ; d ¼ d0 x 3 þ d 0 ; ð13:490 Þ
a0 ; a0 ; b0 ; . . .; d0 being arbitrary constants.
The relation (13.47) becomes
1 1
X;12 ¼ ðax1 x2 þ bx1 þ cx2 þ dÞ þ ðbx1 cx2 Þ;
2 4ð1 þ mÞ
13.1 Conditions for Stresses 563
hence
1 1 2 2
X¼ ax1 x2 þ bx21 x2 þ cx1 x22 þ 2dx1 x2
4 2
1
þ ðbx21 x2 cx1 x22 Þ þ u1 þ u2 ; ð13:50Þ
8ð1 þ mÞ
where u1 ¼ u1 ðx1 ; x3 Þ; u2 ¼ u2 ðx2 ; x3 Þ.
The condition (13.29) leads to
1 2
1
a x1 þ x22 þ 2ðbx2 þ cx1 Þ þ ðbx2 cx1 Þ þ Du1 þ Du2 ¼ 0;
4 4ð1 þ mÞ
hence
1 2
1
a x1 þ x22 þ 2ðbx2 þ cx1 Þ þ ðbx2 cx1 Þ ¼ d1 d2 ;
4 4ð1 þ mÞ
Du1 ¼ d3;33 d1 ; Du2 ¼ d2 d3;33 :
with d1 ¼ d1 ðx1 Þ;
d2 ¼ d2 ðx2 Þ; d3 ¼ d3 ðx3 Þ; it results
1 2 1
d1 ¼ ax1 þ 2cx1 cx1 þ e0 ; d2
4 4ð1 þ mÞ
1 1
¼ ax22 þ 2bx2 bx2 þ e0 ;
4 4ð1 þ mÞ
e0 being an arbitrary constant.
One may also write
1 1 4 1 1
u¼u 1 3
ax þ cx1 þ cx3 e0 x2 þ d3 ;
12 4 1 24ð1 þ mÞ 1 2 1
1 1 4 1 1
u¼u 2 3
ax þ bx2 bx3 þ e0 x2 d3 ;
12 4 2 24ð1 þ mÞ 2 2 2
1 ¼ u
where u 1 ðx1 ; x3 Þ; u
2 ¼ u
2 ðx2 ; x3 Þ are harmonic functions in two variables
u1 ¼ D
D u2 ¼ 0:
Introducing in (13.50), one obtains
1 4
X¼ a x1 6x21 x22 þ x42 þ 4b x32 3x21 x2 þ 4c x31 3x1 x22 24dx1 x2
48
1 3
þ24e0 x21 x22 b x2 3x21 x2 c x31 3x1 x22 þ u1 þ u 2:
24ð1 þ mÞ
Taking into account (13.49), (13.490 ), the Eqs. (13.8000 ) become
Du;1 ¼ a0 x1 þ c0 ; Du;2 ¼ a0 x2 b0 ;
564 13 Particular Cases of States of Strain and Stress
it results
1
Du ¼ a0 x21 x22 þ c0 x1 b0 x2 þ e0
2
and
1 4 1 1
u ¼ u0 þ a0 x1 x42 þ c0 x31 b0 x32 þ e0 x21 þ x22 ;
24 6 4
where u0 ¼ u0 ðx1 ; x2 Þ is a harmonic function
Du0 ¼ 0
and e0 is an arbitrary constant.
We introduce the functions in two variables F23 ¼ F23 ðx2 ; x3 Þ; F31 ¼
F31 ðx3 ; x1 Þ by means of the Cauchy-Riemann equations
1;1 ;3 ¼ 2F31;1 ; u
u 1;1 ;1 ¼ 2F31;1 ;
2;2 ;3 ¼ 2F23;2 ; u
u 2;2 ;2 ¼ 2F23;3 :
m 2
r23 ¼ ðF12 F31 Þ;1 x3 ða0 x1 þ c0 Þ þ 2x3 ða0 x1 þ c0 Þ
4ð1 þ mÞ
2m 1 3 2 1
a0 x1 þ c0 x1 þ e0 x1 2b0 x1 x2 þ c0 x21 x22
4ð1 þ mÞ 3 16ð1 þ mÞ
1 3
þ a0 x1 3x1 x2 6b0 x1 x2 þ 3e0 x21 x22 6d0 x2 ;
2
24
m 2
r31 ¼ ðF23 F12 Þ;2 x3 ða0 x2 þ b0 Þ þ 2x3 ða0 x2 þ b0 Þ
4ð1 þ mÞ
2m 1 3 2 1 2
a0 x 2 þ b0 x 2 e 0 x 2 þ b0 x1 x22 2c0 x1 x2
4ð1 þ mÞ 3 16ð1 þ mÞ
1 3
a0 x2 3x21 x2 3b0 x21 x22 6e0 x1 x2 6d0 x1 ;
24
1 2
r12 ¼ ðF31 F23 Þ;3 a x1 þ x22 þ 2ðbx2 þ cx1 Þ
4ð1 þ mÞ
1
þ a x21 x22 2ðbx2 cx1 Þ þ 4e0 :
4
introducing some terms of the tangential stresses in these functions, one may—
without losing the generality—express these stresses in the form
1
r23 ¼ ðF12 F31 Þ;1 ða0 x1 þ 3c0 Þx21 ; ð13:510 Þ
6ð1 þ mÞ
566 13 Particular Cases of States of Strain and Stress
m
r31 ¼ ðF23 F12 Þ;2 x2 ða0 x2 þ b0 Þ þ 2x3 ða0 x2 þ b0 Þ:
4ð1 þ mÞ 3
1 1 1
x22 ða0 x2 þ 3b0 Þ a0 x2 x22 þ 3x23 þ x3 ðb0 x3 þ b0 Þ ;
3 2ð1 þ mÞ 6
ð13:5100 Þ
m
r12 ¼ ðF31 F23 Þ;3 x2 ða0 x3 þ a0 Þ þ 2x2 ðb0 x3 þ b0 Þ:
4ð1 þ mÞ 2
1 2 0 1 1 2 2
0 0
x3 ða0 x3 þ 3a Þ a0 x3 x3 þ 3x2 þ x2 ða x2 þ b Þ ;
3 2ð1 þ mÞ 6
E½u1 ðx2 x03 þ x3 x02 þ u01 Þ ¼ 2ð1 þ mÞF 23 þ 1 a0 x2 x3 ½3x21 ðx22 þ x23 Þ
6
1 0 1
þ a x2 ð3x21 2x22 Þ þ b0 x3 ð3x21 2x23 Þ
6 6
1 0 2 2 2
þ b ½x1 ðx2 þ x3 Þ
2
þ c0 x1 x2 x3 þ c0 x1 x2 þ d0 x3 x1 þ d0 x1 ;
ð13:52Þ
E½u2 ðx3 x01 þ x1 x03 þ u02 Þ ¼ 2ð1 þ mÞF 31 1 a0 x1 x3 ½x21 þ mð3x22 x23 Þ
6
1 0
a x1 ½x1 þ 3mðx22 x23 Þ mb0 x1 x2 x3
2
6
1
mb0 x1 x2 c0 x3 ½3x21 þ mð3x22 x23 Þ
6
1 0 2
c ½x1 þ mðx22 x23 Þ md0 x2 x3 md 0 x2 ;
2
0 0 0
E½u3 ðx1 x2 þ x2 x1 þ u3 Þ ¼ 2ð1 þ mÞF 12 1 a0 x1 x2 ½x21 þ mðx22 3x23 Þ
6
0 1
ma x1 x2 x3 b0 x1 ½x21 3mðx22 x23 Þ
6
0 1
mb x1 x3 c0 x2 ½3x21 mðx22 3x23 Þ
6
1
mc0 x2 x3 ½x21 mðx22 x23 Þ md0 x3 ;
2
ð13:520 Þ
23 ¼ F
where the functions in two variables F 23 ðx2 ; x3 Þ; F
31 ¼ F
31 ðx3 ; x1 Þ; F
12 ¼
F12 ðx1 ; x2 Þ are harmonic
13.1 Conditions for Stresses 567
23 ¼ 0; DF
DF 31 ¼ 0; DF
12 ¼ 0 ð13:53Þ
and verify Cauchy-Riemann equations
23;3 ; F23;3 ¼ F
F23;2 ¼ F 23;2 ; ð13:54Þ
31;1 ; F31;1 ¼ F
F31;3 ¼ F 31;3 ; ð13:540 Þ
12;2 ; F12;2 ¼ F
F12;1 ¼ F 12;1 : ð13:5400 Þ
r11 ¼ b0 x3 x1 þ a0 x1 x2 þ b0 x1 þ c0 x2 þ d0 x3 þ d0 ; ð13:56Þ
568 13 Particular Cases of States of Strain and Stress
1 nm
r31 ¼ F23;2 þ b0 x22 x23 2x3 ða0 x2 þ b0 Þ
2ð1 þ mÞ 2
o
x 3 ð b0 x 3 þ b0 Þ þ k 1 x 2 ;
nm
ð13:560 Þ
1
r12 ¼ F23;3 a0 x22 x23 þ 2x2 ðb0 x3 þ b0 Þ
2ð1 þ mÞ 2
o
þ x 2 ð a0 x 2 þ b0 Þ k 1 x 3 :
1
E½u2 ðx3 x01 þ x1 x03 þ u02 Þ ¼ a0 x1 ½x21 þ 3mðx22 x23 Þ þ mb0 x1 x2 x3
6
1
mb0 x1 x2 c0 ½x21 þ mðx22 x23 Þ
2
md0 x2 x3 md 0 x2 2ð1 þ mÞk1 x3 x1 ;
1
E½u3 ðx1 x02 þ x2 x01 þ u03 Þ ¼ ma0 x1 x2 x3 b0 x1 ½x21 3mðx22 x23 Þ ð13:570 Þ
6
1
mb x3 x1 mc0 x2 x3 d0 ½x21 mðx22 x23 Þ
0
2
md0 x3 þ 2ð1 þ mÞk1 x1 x2 :
a0 ¼ b0 ¼ a0 ¼ 0:
13.1 Conditions for Stresses 569
where e3 ¼ e3 ðx1 Þ and e4 ¼ e4 ðx1 Þ are arbitrary functions. Observing that the above
functions are harmonic and taking into account the other relations in which they
are involved, one is lead to
1
F23 ¼ c1 x22 x23 þ c4 x2 þ c5 x3 þ c6 ;
4
1 1
F31 ¼ c1 x21 x23 þ ðc2 þ c3 Þx1 þ c5 x3 þ c7 ;
4 2
1 2 2
1
F12 ¼ c1 x1 x2 þ ðc2 c3 Þx1 þ c4 x2 þ c8 ;
4 2
where c1 ; c2 ; . . .; c8 are arbitrary constants.
Taking into account (13.510 ), (13.5100 ) and (13.52), (13.520 ), one obtains the
state of stress
r11 ¼ c0 x2 x3 þ c0 x2 þ d0 x3 þ d 0 ð13:59Þ
1
r23 ¼ c0 x2 þ c1 x1 þ c2 ð13:590 Þ
2ð1 þ mÞ 1
and the state of displacement
1
E½u2 ðx3 x01 þ x1 x03 þ u02 Þ ¼ c0 x3 ½3x21 þ mð3x22 x23 Þ
6
1
c0 ½x21 þ mðx22 x23 Þ md0 x2 x3
2
1
md 0 x2 þ ð1 þ mÞc1 x3 x1 þ ð1 þ mÞc2 x3 ;
2 ð13:600 Þ
0 0 0 1 2 2 2 0
E½u3 ðx1 x2 þ x2 x1 þ u3 Þ ¼ c0 x2 ½3x1 þ mðx2 3x3 Þ mc x1 x2 x3
6
1
d0 ½x21 mðx22 x23 Þ md0 x3
2
1
þ ð1 þ mÞc1 x1 x2 þ ð1 þ mÞc2 x2 :
2
r11 ¼ c0 x2 þ d0 x3 þ d 0 ð13:62Þ
c0 ¼ c0 ¼ d0 ¼ d0 ¼ 0:
13.1 Conditions for Stresses 571
1
u2 ¼ x3 ðc1 x1 þ 2c2 Þ x3 x01 þ x1 x03 þ u02 ;
4l
ð13:660 Þ
1
u3 ¼ x2 ðc1 x1 þ 2c2 Þ x1 x02 þ x2 x01 þ u03 :
4l
One obtains a state of pure shear if all the normal stresses vanish
r11 ¼ r22 ¼ r33 ¼ 0: ð13:67Þ
If one puts the first condition to (13.51) and one takes into account (13.510 )–
(13.520 ), one obtains the state of stress
r23 ¼ ðF12 F31 Þ;1 ; r31 ¼ ðF23 F12 Þ;2 ; r12 ¼ ðF31 F23 Þ;3 ; ð13:68Þ
Taking into account the harmonicity of the functions F23 ; F31 ; one may write
1
F23 ¼ kðx22 x23 Þ þ d1 x3 þ d2 x2 þ d3 ;
2
1
F31 ¼ kðx21 x23 Þ þ d1 x3 þ d4 x1 þ d5 ;
2
where k; d1 ; d2 ; . . .; d5 are arbitrary constants. Observing that one may use a stress
function of the form
F12 ¼ d4 x1 d2 x2
and introducing the terms which depend on the constants d3 ; d5 in the function
F12 ; it results the state of stress
r23 ¼ F12;1 þ kx1 ; r31 ¼ F12;2 kx2 ð13:71Þ
Applying Laplace’s operator and observing that the function X is harmonic, one
obtains
H;12 ¼ 0;
where the relation (A.102) has been taken into account; it results
H ¼ ðf0 þ g0 Þx3 þ f 0 þ g0 ;
the functions of a single variable f0 ¼ f0 ðx1 Þ; g0 ¼ g0 ðx2 Þ; f 0 ¼ f 0 ðx1 Þ; g0 ¼ g0 ðx2 Þ
being specified by the harmonicity condition of the function H: One obtains
13.1 Conditions for Stresses 573
1 1
H¼ b1 ðx21 x22 Þ þ b2 x1 þ b3 x2 þ b4 x3 þ b01 ðx21 x22 Þ þ b02 x1 þ b03 x2 þ b04 ;
2 2
Applying Laplace’s operator and taking into account (13.7), (13.70 ), (13.9),
(13.29), it results (equations of Cauchy-Riemann type)
mH0;2 þ ð1 mÞDu;1 ¼ 0; mH0;1 ð1 mÞDu;2 ¼ 0;
and
Du;1 ¼ Du;2 ¼ 0;
hence
H0 ¼ k; Du ¼ C;
where k and C are arbitrary constants.
The relations (13.76) become
1 h i
X;13 ¼ mx3 H0;2 þ ð2 mÞu;1 ;
1þm
1 h i
X;23 ¼ mx3 H0;1 ð2 mÞu;2 :
1þm
Integrating with respect to x3 ; one obtains
1 1 2 0
X;1 ¼ mx H þ ð2 mÞx3 u;1 þ p;1 ;
1 þ m 2 3 ;2
ð13:77Þ
1 1 2 0
X;2 ¼ mx H ð2 mÞx3 u;2 þ p;2 ;
1 þ m 2 3 ;1
where the function p ¼ pðx1 ; x2 Þ has been introduced, so that Schwartz’s theorem
be verified for mixed derivatives of second order of the function X:
One applies once more Laplace’s operator to the relations (13.77); there result
the Cauchy-Riemann equations
m m
Dp;1 ¼ H0 ; Dp;2 ¼ H0 ; ð13:78Þ
1 þ m ;2 1 þ m ;1
hence, the function p is biharmonic
DDp ¼ 0: ð13:780 Þ
Taking into account (13.78), one can integrate the relations (13.77); it results
1 2m
X ¼ x23 Dp x3 u þ p; ð13:79Þ
2 1þm
where an arbitrary function of x3 has been neglected; indeed, the function X
appears in the expressions of the stresses only by its derivatives with respect to x1
and x2 . One may easily verify that the function (13.79) is harmonic.
Replacing in the representation (13.31), (13.3100 ), one can write
13.1 Conditions for Stresses 575
1 2m 1 1
r11 ¼ x23 Dp;12 x3 u;12 þ p;12 ðx1 o1 x2 o2 ÞH0 þ ðkx3 þ H0 Þ;
2 1þm 4ð1 þ mÞ 2
1 2 2m 1 1
r22 ¼ x3 Dp;12 þ x1 u;12 p;12 þ ðx1 o1 x2 o2 ÞH0 þ ðkx3 þ H0 Þ;
2 1þm 4ð1 þ mÞ 2
1 2 2m
r12 ¼ x3 ðo11 o22 ÞDp þ x3 ðo11 o22 Þu
4 2ð1 þ mÞ
1 1
ðo11 o22 Þp ðx1 o2 þ x2 o1 ÞH0 :
2 4ð1 þ mÞ
DF0 ¼ H0 ; ð13:81Þ
0 ¼ 0
DF
1 1
D p;12 ðx1 o1 x2 o2 ÞH0 þ H0 ¼ H0;22 ;
4ð1 þ mÞ 2
1 1 0
D p;12 ðx1 o1 x2 o2 ÞH H ¼ H0;11 ;
0
4ð1 þ mÞ 2
1 1
D ðo11 o22 Þp þ ðx1 o2 þ x2 o1 ÞH ¼ H0;12 :
0
2 4ð1 þ mÞ
The above results lead to
1 1
p;12 ðx1 o1 x2 o2 ÞH0 þ H0 ¼ F0;22 0
;
4ð1 þ mÞ 2
1 1
p;12 ðx1 o1 x2 o2 ÞH0 H0 ¼ F0;11 0
;
4ð1 þ mÞ 2
1 1
ðo11 o22 Þp þ ðx1 o2 x2 o1 ÞH0 ¼F0;12
0
;
2 4ð1 þ mÞ
In the case of a state of incompressible deformation one has the relation (13.38),
which leads to
r11 þ r22 ¼ 0: ð13:86Þ
Taking into account the representation (13.83) and applying the operator of
Laplace to the function (13.820 ), one obtains the conditions
DF0 ¼ 0; DF1 ¼ 0; k ¼ 0: ð13:87Þ
The stress function F ¼ Fðx1 ; x2 ; x3 Þ is thus harmonic and of the form
F ¼ x 3 F1 þ F0 : ð13:88Þ
r33;3 ¼ 0: ð13:900 Þ
The Eq. (13.90) lead to the representation of Airy
r11 ¼ F;22 ; r22 ¼ F;11 ; r12 ¼ F;12 ð13:91Þ
r11 þ r22 ¼ D0 F;
where D0 ¼ o11 þ o22 is Laplace’s operator in two variables. The Eq. (13.900 ) leads
to
r33 ¼ h; ð13:910 Þ
where h ¼ hðx1 ; x2 Þ is an arbitrary function too.
The Beltrami equations (5.420 ) become
1 1
Dr11 þ H;11 ¼ 0; Dr22 þ H;22 ¼ 0; ð13:92Þ
1þm 1þm
578 13 Particular Cases of States of Strain and Stress
1
D0 r33 þ H;33 ¼ 0; ð13:920 Þ
1þm
H;23 ¼ 0; H;31 ¼ 0; ð13:9200 Þ
1
Dr12 þ H;12 ¼ 0: ð13:92000 Þ
1þm
The Eq. (13.920 ) leads to
H;3 ¼ f;3 ;
H ¼ D0 F þ h ¼ f þ g þ h; ð13:93Þ
hence
13.1 Conditions for Stresses 579
1
g þ mCðx21 þ x22 Þ;
g¼
4
g¼
where gðx1 ; x2 Þ is a harmonic function
g ¼ 0:
D
By means of the representation (13.90), (13.900 ) and of the formula (13.93), the
Eqs. (13.92), (13.92000 ) lead to
1 1
F;2233 þ g;22 þ ðg þ hÞ;11 ¼ 0; F;1133 þ g;11 þ ðg þ hÞ;22 ¼ 0;
1þm 1þm
ð13:95Þ
1
F;1233 g;12 þ ðg þ hÞ;12 ¼ 0: ð13:950 Þ
1þm
Integrating the Eq. (13.950 ) with respect to x1 and x2 ; one gets
1
F;33 ¼ ðg þ hÞ g þ u þ w; ð13:96Þ
1þm
where u ¼ uðx1 ; x3 Þ; w ¼ wðx2 ; x3 Þ are arbitrary functions. One has
1
DF ¼ f þ ðg þ hÞ þ u þ w ð13:97Þ
1þm
too, by tooking into account (13.93).
Introducing (13.96) in (13.93), one may write
u;11 þ C ¼ 0; w;22 þ C ¼ 0;
wherefrom
1 1
u ¼ Cx21 þ x1 f1 þ f2 ; w ¼ Cx22 þ x2 f3 þ f4 ;
2 2
the functions fi ¼ fi ðx3 Þ; i ¼ 1; 2; 3; 4, being arbitrary. The relation (13.97)
becomes
1 1 1
DF ¼ ðg þ F2 Þ Cðx21 þ x22 Þ ð1 þ mÞCx23
1þm 4 2
þ x1 f1 þ x2 f3 þ f2 þ f4 þ kx3 þ C1 ; ð13:98Þ
hence
Z Z
1 1 1
þ
F ¼ F x2 ðg þ F2 Þ Cðx21 þ x22 Þ2 ð1 þ mÞCx43 þ x1 dx3 f1 dx3
2ð1 þ mÞ 3 64 24
Z Z Z Z
1 1
þ x2 dx3 f3 dx3 þ dx3 ðf2 þ f4 Þdx3 þ kx23 þ C1 x23 : ð13:99Þ
6 2
580 13 Particular Cases of States of Strain and Stress
¼ Fðx
where F 1 ; x2 ; x3 Þ is a harmonic function
¼ 0:
DF ð13:980 Þ
Differentiating the function F with respect to x3 ; one obtains
;33 þ 1 1
F;33 ¼ F g þ F2 Þ ð1 þ mÞCx23 þ x1 f1 þ x2 f3 þ f2 þ f4 þ kx3 þ C1 :
ð
1þm 2
Subtracting the latter relation from (13.97), one obtains
1 Cðx21 þ x22 Þ:
D0 F ¼ D0 F
4
Taking into account (13.980 ), it follows that
1
v2 ¼ F 0 þ ð1 þ mÞCðx21 þ x22 Þ2 ;
64
with F0 ¼ F0 ðx1 ; x2 Þ biharmonic function
DDF0 ;
it also results
g ¼ DF0 C1 :
ð13:1010 Þ
The function (13.101) becomes
which leads to
1 1
DF0 þ F2 þ ð1 þ mÞCðx21 þ x22 Þ ð1 þ mÞCx23 þ kx3 ¼ 0:
4 2
The latter relation must be verified for any value of x3 ; one obtains
k ¼ C ¼ 0;
as well as
F2 þ DF0 ¼ 0:
The stress function (13.103) will be of the form
1
F ¼ x23 DF0 þ x3 F1 þ F0 ð13:104Þ
2
and the stress (13.9100 ) will be given by
r33 ¼ DF0 : ð13:1040 Þ
One remarks that, in this case, the stress function F is harmonic
DF ¼ 0: ð13:105Þ
Starting from the results obtained above for the case of two zero tangential
stresses, one can put supplementary conditions to the other components of the
stress tensor, obtaining results which can be interesting for various problems.
calculating now the mixed derivative of second order of the formula (13.103) and
observing that the expression must vanish for any x3 ; one obtains
1
ðF2 mDF0 Þ;12 ¼ 0; F1;12 ¼ 0; F0 þ mCx1 x2 ¼ 0: ð13:1080 Þ
8 ;12
Now, it results
F1 ¼ u1 þ w1 : F2 ¼ u2 þ w2 :
where the functions of a single variable u1 ¼ u1 ðx1 Þ; u2 ¼ u2 ðx1 Þ; w1 ¼ w1 ðx2 Þ;
w2 ¼ w2 ðx2 Þ will be determined by the harmonicity condition of the functions F1
and F2 . It results
1
F1 ¼ Kðx21 x22 Þ þ K1 x1 þ K2 x2 þ K3 ;
2
1 2 2
F2 ¼ Kðx 1 x2 Þ þ K1 x1 þ K2 x2 þ K3 ;
2
. . .; K
where K; K; 3 are arbitrary constants.
The function F0 is given by
1
F0 ¼ mCðx21 þ x22 Þ þ u3 þ w3 ;
64
where u3 ¼ u3 ðx1 Þ and w3 ¼ w3 ðx2 Þ; the condition of biharmonicity of this
function leads to
1 1 x41 1 mCx21 x22 þ 1 1 mC K x42
F0 ¼ mC þ K
24 8 32 24 8
1 3 3 1 2 2 5 x1 þ K
6 x2 þ K
7;
þ ðK1 x1 þ K2 x2 Þ þ ðK3 x1 þ K4 x2 Þ þ K
6 2
584 13 Particular Cases of States of Strain and Stress
K
with K; 1 ; . . .; K
7 arbitrary constants.
One obtains the stress function
1 2 x2 Þx2 1 ð1 þ mÞCðx2 þ x2 Þx2
mKÞðx
F¼ ðK 1 2 3 1 2 3
4ð1 þ mÞ 8
1 1 x2 þ 1 mC K x2 þ 1 Kðx2 x2 Þx
þ mC þ K 1 2 1 2 3
24 2 2 2
1 1 3 3 1 2 2
þ kðx21 þ x22 Þx3 þ ðK 1 x1 þ K 2 x3 Þ þ ðK 3 x1 þ K 4 x2 Þ;
4 6 2
where a polynomial of first degree in x1 and x2 ; as well as a function which
depends only on x3 ; which lead to a zero state of stress have been neglected.
The representation (13.83) leads to
1 2 1 ð1 þ mÞCx2 þ 1 1 mC K
mKÞx x2
r11 ¼ ðK 3 3 2
2ð1 þ mÞ 4 2 2
1 x þK ;
Kx3 þ kx3 þ K 2 2 4
2
1 2 1 ð1 þ mÞCx2 þ 1 1 mC þ K
mKÞx x2
r22 ¼ ðK 3 3 1
2ð1 þ mÞ 4 2 2
1 x þK ;
þ Kx3 þ kx3 þ K 1 1 3
2
and the formula (13.9100 ) allows to write
1 2 2 1 2 2
r33 ¼ Kðx 1 x2 Þ þ Cðx1 þ x2 Þ þ K1 x1 þ K2 x2 þ K3 :
2 4
Introducing the notations
1 mKÞ 1 C; k ¼ 1 þ 1 C;
mKÞ
k1 ¼ ðK 2 ðK
2ð1 m2 Þ 4 2ð1 m2 Þ 4
1
mKÞ;
k3 ¼ ðK
2ð1 m2 Þ
; a ¼ 1 k K; a ¼ K
a1 ¼ K ; b ¼ 1 k þ K; b ¼ K ;b ¼K ;
2 2 3 4 1 2 1 3 3
2 2
1 ; c2 ¼ K
c1 ¼ K 2 ; c3 ¼ K
3;
so that
1
k1 ¼ k2 ¼ k3 ¼ K;
1m
a2 ¼ b1 ; b2 ¼ c1 ; c2 ¼ a1 ; a3 ¼ b c; b3 ¼ c a; c3 ¼ a b;
1
E½u3 ðx1 x02 þ x2 x01 þ u03 Þ ¼ mð1 þ mÞcðx21 x22 Þx3 ½ða2 mb1 Þx21
2
þ ðb1 ma2 Þx2 þ mða2 þ b1 Þx23
2
hence,
c ¼ 0; a1 ¼ b2 ¼ 0; a2 ¼ b1 ¼ a; a3 ¼ b3 ¼ b;
1
2l½u3 ðx1 x02 þ x2 x01 þ u03 Þ ¼ aðx21 x22 Þ: ð13:1170 Þ
2
2m 3 1
r12 ¼ x3 H0;12 þ ðx1 o2 þ x2 o1 Þ½ð1 mÞx3 H0 H0 :
6ð1 þ mÞ 4ð1 þ mÞ
ð13:11900 Þ
The harmonic part of the biharmonic function u ¼ uðx1 ; x2 Þ which, as a matter
of fact, is arbitrary in the formulae (13.3100 ) will be taken equal to the harmonic
conjugate of the function a2 H0 =ð2 mÞ: Putting H0 ¼ 0 and introducing the
notation
1m 0;
w¼ ðx1 o2 þ x2 o1 ÞH
4ð1 þ mÞ
which leads to
1m
Dw ¼ H0 ;
1þm
one obtains the state of stress
2m 3 1
r11 ¼ x H0;22 þ x3 H0 þ x3 w;22 ;
6ð1 þ mÞ 3 1þm
ð13:120Þ
2m 2 1
r22 ¼ x3 H0;11 þ x3 H0 þ x3 w;11 ;
6ð1 þ mÞ 1þm
1
r23 ¼ ðx2 a2 ÞH0;2 ;
2ð1 þ mÞ 3
ð13:1200 Þ
1
r31 ¼ ðx2 a2 ÞH0;1 ;
2ð1 þ mÞ 3
2m 3
r12 ¼ x H0;12 x3 w;12 ; ð13:12000 Þ
6ð1 þ mÞ 3
given by J. H. Michell [6] and A.-E. H. Love [5] for a straight cylinder of finite
length 2a (or a thick plate of thickness 2a), the end sections x3 ¼ a of which are
free of loads (r33 ¼ r31 ¼ r32 ¼ 0 for x3 ¼ a).
One obtains thus the must important classic results. Moreover, one observes
that the formulae (13.31)–(13.3100 ) are much more simple that the previous ones
and have a larger field of applicability (as a matter of fact, it is a complete
representation, but the above representation has not this property). For example,
2ð1 þ mÞ
F12 ðx1 ; x2 Þ; DF;12 ¼ 0;
2m
is the harmonic part of the biharmonic function u; if one puts
2
H0 ¼ H0 ¼ 0; X ¼ cx33 cx3 ðx21 þ x22 Þ;
3
13.1 Conditions for Stresses 589
Let be now
X ¼ Xþ þ X ; ð13:123Þ
13.1.5.2 Discussion
One has presented above the state of stress and the state of displacement for two
bidimensional (eventually, quasi-bidimensional) fundamental problems (the case
of a zero normal stress and the case of two zero tangential stresses), as well as for
many particular cases, by means of real stress potential functions. One remarks
thus that one can make a unitary study of a lot of problems the solutions of which
are known, as well as of new ones. The use of functions in two variables has a
great importance for the formulations of boundary value problems and each par-
ticular case may be put in connection to such a problem.
The case r33 ¼ 0 corresponds to the approximate study of a plate, the sepa-
ration planes (x3 ¼ const) of which are acted upon by tangential loads; in this case,
this stress may be neglected with respect to the other normal stresses, even for
plates of mean thickness. But G. R. Kirchhoff’s hypothesis must no more be used.
As well, at a point of the boundary one may put exact conditions on the contour of
the plate (three conditions at a point instead of two conditions as in the classic
theory of thin plates). The planes x3 ¼ const are sliding planes.
The case r31 ¼ r32 ¼ 0 corresponds to the study of superposed plates (stratified
body), the sliding being free between the faces x3 ¼ const: No other simplifying
hypothesis is introduced (linear element, zero normal stress r33 etc.). Such a case
13.1 Conditions for Stresses 591
may also be used for an initial study of the ‘‘sandwich’’ plates. The normal
directions to these plates are principal directions.
The representation in which one of the stress functions (e.g., the function H)
has a specified physical significance allows easily to obtain the results concerning
an incompressible state of deformation.
The case of a plane state of stress (Sects. 13.1.2.9 and 13.1.4.1) corresponds to a
plate free of loads on the parallel faces and acted upon on the contour by loads
parallel to the middle plane. Eventually, one must apply the principle of Saint-
Venant along the contour, obtaining a generalized plane state of stress.
The case of an antiplane state of stress (Sect. 13.22) corresponds to a cylindrical
body of finite length acted upon only on the end faces by tangential stresses and by
normal stresses which have a linear variation on these faces. This problem may be
considered as a complementary one to the preceding one. It corresponds to the
pure torsion superposed to the pure bending of the straight bars; it is the must
general case where the shear force does not pass through the centre of torsion.
The case in which two tangential stresses are non-zero (Sect. 13.1.2.7) corre-
sponds to the pure torsion of the straight bars; the shear force passes through the
centre of torsion and no bending effect appears.
The case of a state of stress without shear (Sect. 13.1.4.2) corresponds to a
triaxial state of stress; the principal directions are the same for any point of the
elastic body. In particular, if the normal stresses are equal, then one obtains a
uniform state of stress.
The case of a state of pure shear (Sect. 13.1.2.6) corresponds to the case of three
triorthogonal families of sliding planes.
The other considered particular cases have an interest for itself, either in what
concerns the possibility to specify sliding planes or principal directions for all the
body. For example, in the case in which one has r11 6¼ 0 and r23 6¼ 0 (Sect. 13.1.2.3)
the straight lines x1 ¼ const; x2 ¼ const are principal directions, while in the
planes parallel to these lines one has a sliding along the directions normal to them.
The given demonstrations allow to affirm that the representations thus obtained
are complete; it means that any state of stress which verifies the imposed condi-
tions may be expressed in this form. Moreover, one must remark that in these
representations one has used the minimal number of necessary stress functions.
In the case in which one has r33 ¼ 0; the stress vector on the planes x3 ¼ const
is tangent to these planes; it is a family of sliding planes (parallel planes). In the
case in which one has r31 ¼ r32 ¼ 0; the stress vector on the planes x3 ¼ const is
normal to these planes; it is a family of principal planes (parallel planes). The
directions normal to these planes are principal directions.
Proceeding analogically for the stresses in arbitrary curvilinear co-ordinates one
may obtain families of sliding surfaces or of principal surfaces.
592 13 Particular Cases of States of Strain and Stress
Hereafter we shall put conditions for linear or for angular strains, obtaining general
potential functions for the states of strain and stress. The corresponding problems
are bidimensional problems, as in the preceding paragraph.
We deal, at the beginning, with the general problem of elastostatics which results
by putting the condition that one linear strain vanishes, namely
r33 ¼ 0: ð13:126Þ
1þm
rkk ¼ ð1 þ mÞðr11 þ r22 Þ ¼ r33 ;
m
so that
1 1
e11 ¼ ðr11 m0 r22 Þ ¼ ðr11 r33 Þ;
E0 2l
ð13:127Þ
1 1
e22 ¼ ðr22 m0 r11 Þ ¼ ðr22 r33 Þ;
E0 2l
where one has introduced the generalized elastic constants
E m E E0
E0 ¼ ; m0 ¼ ;l¼ ¼ : ð13:1270 Þ
1 m2 1m 2ð1 þ mÞ 2ð1 þ m0 Þ
The equations of Cauchy (2.28) lead to
u3;3 ¼ 0;
hence
2lu3 ¼ f;12 ;
13.2 Conditions for Strains 593
g¼
where gðx1 ; x2 Þ is an arbitrary function; it results that
1
2lu1;1 ¼ U;12 þ ð1 2mÞðx3 Df þ DgÞ;12 ;
2
1
2lu2;2 ¼ U;12 þ ð1 2mÞðx3 Df þ DgÞ;12 ;
2
hence
1
2lu1 ¼U;2 þ ð1 2mÞðx3 Df þ DgÞ;2 þ h1;2 ;
2 ð13:129Þ
1
2lu2 ¼ U;1 þ ð1 2mÞðx3 Df þ DgÞ;1 þ h2;1 ;
2
with U ¼ Uðx1 ; x2 ; x3 Þ; h1 ¼ h1 ðx2 ; x3 Þ; h2 ¼ h2 ðx1 ; x3 Þ:
The harmonic Eq. (5.15) leads to
hence
DDf;12 ¼ 0; DD
g;12 ¼ 0;
wherefrom
f ¼ f þ f1 þ f2 ;
g ¼ g þ g1 þ g2 ; ð13:130Þ
1
DU;2 ð1 2mÞðx3 f2 þ g2 Þ;22222 þ Dh1;2 þ Dðx3 f þ gÞ;112 ¼ 0;
2
1
DU;1 þ ð1 2mÞðx3 f1 þ g1 Þ;11111 Dh2;1 Dðx3 f þ gÞ;122 ¼ 0;
2
wherefrom
1
DU ¼ ð1 2mÞðx3 f2 þ g2 Þ;2222 Dh1 Dðx3 f þ gÞ;11 þ Dw2 ;
2
1
DU ¼ ð1 2mÞðx3 f1 þ g1 Þ;1111 þ Dh2 þ Dðx3 f þ gÞ;22 Dw1 ;
2
with w1 ¼ w1 ðx2 ; x3 Þ; w2 ¼ w2 ðx1 ; x3 Þ; one observes that one must have
1 h i
ð1 2mÞ ðx3 f1 þ g1 Þ;1111 þ ðx3 f2 þ g2 Þ;2222 þ Dðw1 h1 Þ þ Dðw2 h2 Þ ¼ 0
2
or
1
Dðw1 h1 Þ þ ð1 2mÞðx3 f2 þ g2 Þ;2222 ¼ w3;33 ;
2
1
Dðw2 h2 Þ þ ð1 2mÞðx3 f1 þ g1 Þ;1111 ¼ w3;33 ;
2
with w3 ¼ w3 ðx3 Þ: Hence,
1
Dw1 ¼ Dh1 þ w3;33 ð1 2mÞðx3 f2 þ g2 Þ;2222 ;
2
1
Dw2 ¼ Dh2 w3;33 ð1 2mÞðx3 f1 þ g1 Þ;1111
2
and one obtains
1 h i
DU ¼ ð1 2mÞ ðx3 f1 þ g1 Þ;1111 ðx3 f2 þ g2 Þ;2222
2
1
Dðh1 h2 Þ w3;33 ðo11 o22 ÞDðx3 f þ gÞ:
2
The function U will thus be of the form
1
U ¼ u ðo11 o22 Þ½ð1 2mÞðx3 f1 þ g1 Þ ðx3 f þ gÞ ðh1 h2 Þ w3 ;
2
where the function u ¼ uðx1 ; x2 ; x3 Þ must be harmonic
Du ¼ 0: ð13:131Þ
Introducing this function in (13.129) and taking into account (13.130), one
obtains
13.2 Conditions for Strains 595
2l½u1 ðx2 x03 þ x3 x02 þ u01 Þ ¼ u;2 ½o11 ð1 mÞo22 ðx3 f þ gÞ;2 ;
2l½u2 ðx3 x01 þ x1 x03 þ u02 Þ ¼ u;1 þ ½ð1 mÞo11 mo22 ðx3 f þ gÞ;1 ;
ð13:132Þ
Hence, one may express the state of displacement which verifies the condition
(13.131) by means of the representation (13.132), (13.1310 ), where u ¼
uðx1 ; x2 ; x3 Þ is a harmonic function of class C3 and where f ¼ f ðx1 ; x2 Þ; g ¼
gðx1 ; x2 Þ are biharmonic functions of class C5 in two variables; the corresponding
problem is thus a quasi-bidimensional problem.
By means of Cauchy relations (2.28), one obtains the state of strain
1 1
2le23 ¼ u;13 þ ½ð1 mÞo11 ð1 þ mÞo22 f;1 ;
2 2 ð13:1330 Þ
1 1
2le31 ¼ u;23 ½ð1 þ mÞo11 ð1 mÞo22 f;2 ;
2 2
1
2le12 ¼ ðo11 o22 Þu ðx3 f þ gÞ;1122 ð13:13300 Þ
2
and the volume strain
2lem ¼ ð1 2mÞDðx3 f þ gÞ;12 : ð13:133000 Þ
1 1
r23 ¼ u;13 þ ½ð1 mÞo11 ð1 þ mÞo22 f;1 ;
2 2 ð13:13400 Þ
1 1
r31 ¼ u;23 ½ð1 þ mÞo11 ð1 þ mÞo22 f;2 ;
2 2
1
r12 ¼ ðo11 o22 Þu ðx3 f þ gÞ;1122 : ð13:134000 Þ
2
596 13 Particular Cases of States of Strain and Stress
so that
þw þw ;
f ¼w g ¼ v þ v1 þ v2 ;
1 2
¼ wðx
with w1 ¼ w1 ðx1 Þ; w2 ¼ w2 ðx2 Þ; v1 ¼ v1 ðx1 Þ; v2 ¼ v2 ðx2 Þ and w 1 ; x2 Þ;
v ¼ vðx1 ; x2 Þ; the latter functions being harmonic ones
¼ 0;
Dw Dv ¼ 0:
But the functions f and g are biharmonic, wherefrom it results
ðo1111 þ o2222 Þðw1 þ w2 Þ ¼ 0; ðo1111 þ o2222 Þðv1 þ v2 Þ ¼ 0;
wherefrom
1 þ d1 ;
w1;111 ¼ dx1 þ d1 ; v1;111 ¼ dx
w2;222 ¼ dx2 þ d2 ; v2;222 ¼ dx 2 þ d2 :
1
ðo11 o22 Þðx3 w þ vÞ ð1 mÞ x3 1 d 3 x2 þ x2 2x2
1 2 3
2 6
1 2 2 2
þ d1 x1 d2 x2 þ d x1 þ x2 2x3 þ d1 x1 d2 x2
2
13.2 Conditions for Strains 597
to the arbitrary function u; obtaining thus a new arbitrary harmonic function; one
can thus neglect the polynomial terms in the above representation, as well as the
term in w and v in the representation of u1 and u2 ; without losing the generality.
With the notation
¼ w
w ;12
1 1
r23 ¼ ðu;13 w;2 Þ; r31 ¼ ðu;23 w;1 Þ; ð13:1370 Þ
2 2
1
r12 ¼ ðo11 o22 Þu: ð13:13700 Þ
2
hence
1 1
,1 ¼ cx21 þ a1 x1 þ b1 ; ,2 ¼ cx22 þ a2 x2 þ b2 :
2 2
One obtains thus
1 1
Df ¼ cðx21 x22 Þ ða1 x1 a2 x2 þ b1 b2 Þ;
4m 2m
where
1 1 1
f ¼ f0 cðx4 x42 Þ ða1 x31 a2 x32 Þ ðb1 x21 b2 x22 Þ;
48m 1 12m 4m
the function f 0 ¼ f 0 ðx1 ; x2 Þ being harmonic
Df 0 ¼ 0: ð13:1310 Þ
Returning to the function u; it results
0 1 1 1 2 2
u;3 ¼ ðo11 o22 Þf þ v3;3 þ 1 cðx þ x2 Þ þ a1 x1 þ a2 x2 þ b1 þ b2 ;
2 m 2 1
13.2 Conditions for Strains 599
hence
0 1 1 1 2
u ¼ x3 ðo11 o22 Þ f þ v3 þ 1 cðx þ x22 Þ
2 m 2 1
þa1 x1 þ a2 x2 þ b1 þ b2 x3 þ v0 ;
with v0 ¼ v0 ðx1 ; x2 Þ: Taking into account the conditions (13.131), (13.1310 ), one
may write
1
Dv0 þ D,3 þ 1 cx3 ¼ 0;
m
wherefrom
1 1
,3 ¼ 1 cx33 þ c2 x23 þ c2 x3 þ c3 :
6 m
Finally,
Dv0 þ 2c1 ¼ 0;
so that
1
v0 ¼
v c1 ðx21 þ x22 Þ;
2
v¼
the function vðx1 ; x2 Þ satisfying the harmonic equation
v ¼ 0:
D
It results
0 1 1 1 2
u ¼ x3 ðo11 o22 Þf
vþ 1 c 3ðx1 þ x22 Þ 2x23
2 m 6
1
þa1 x1 þ a2 x2 þ b1 þ b2 x3 c1 ðx21 þ x22 2x23 Þ þ c2 x3 þ c3 :
2
We introduce the notations
0
f;12 ¼ F0 ;
g;12 ¼ F;
DF 0 ¼ 0; ¼ 0:
DDF ð13:1380 Þ
But, using Almansi’s formula (A.100), the arbitrary harmonic function v may be
without any
included in the harmonic part of an arbitrary biharmonic function F;
lose of generality of the above representation (one may take v ¼ 0). With the
function
F ¼ x3 F 0 þ F; ð13:139Þ
which is biharmonic
DDF ¼ 0; ð13:1390 Þ
one obtains the formulae of Airy type
r11 ¼ F;22 ; r22 ¼ F;11 ; r33 ¼ mDF; ð13:140Þ
where DF ¼ DF depends only on the variables x1 and x2 : Observing that the zero
sum (because of the biharmonicity of the function F and of the theorem of Fubini)
Z Z Z Z
dx2 DF;22 dx1 þ dx1 DF;11 dx2
differs of the sum between parentheses in the Eq. (13.142) only by a sum of two
functions of a single variable, it results that the latter equation is of the form
f1;1 þ f2;2 þ F1 þ F2 ¼ 0; ð13:1420 Þ
where F1 ¼ F1 ðx1 Þ; F2 ¼ F2 ðx2 Þ are known functions. The Eq. (13.142) is thus
with separate variables and one may take
Z Z
f1 ¼ F1 dx1 ; f2 ¼ F2 dx2 : ð13:14200 Þ
One supposed that f1 and f2 do not depend on the variable x3 ; indeed, this
hypothesis is justified by the form of the representations (13.16), (13.1410 ). Indeed,
assuming such a dependence and taking into account the relations
u1;2 þ u2;1 ¼ 0; u1;3 þ u3;1 ¼ 0;
DF ¼ 0; ð13:143Þ
DF 0 ¼ 0; ¼ 0:
DF ð13:145Þ
0
What concerns the state of displacement (13.141), (13.141 ), taking into
account (13.143), the functions f1 and f2 correspond to a motion of rigid body and
one obtains.
neglecting a motion of rigid body. The two harmonic functions of class C3 which
appear in this representation are both in two variables, the problem being a
bidimensional one.
too. But it is just the case considered in Sect. 13.3. The state of stress is given by
Airy’s type formulae of the form (13.91), (13.9100 ), where the stress is given by the
formula (13.103).
It is obvious that particular cases may easily obtained starting from these
results.
The same considerations can be made for an incompressible state of strain.
13.2 Conditions for Strains 603
13.2.2.2 Discussion
All the representations given above, both for bidimensional and quasi-bidimen-
sional problems, are obtained by a demonstration step by step, with any lose of
generality. Hence, these representations are complete (any state of stress, any state
of strain and any state of displacement which verify the imposed conditions may
be represented in the given form) for a simply connected domain; in the case of a
multiply connected domain, one introduced certain supplementary terms, corre-
sponding to singularities.
What concerns an antiplane state of strain, excepting the normal stresses
r11 ; r22 which are non-zero, the results which one obtains are similar to those
obtained in the case of an antiplane state of stress (r11 ¼ r22 ¼ r12 ¼ 0).
These representations are useful for the study of a bar in the form of a straight
cylinder.
First of all, we give some general results concerning the plane and antiplane states
of stress, problems which complete one another.
We remark that the stress tensor can be decomposed in the sum of two tensors, i.e.
Tr ¼ Tpr þ Tar ð13:149Þ
2 3
r11 r12 0
Tpr 4 r21 r22 05 ð13:14900 Þ
0 0 0
In the static case, the stresses depend on the three variables x1 ; x2 ; x3 ; while in the
dynamic case, one takes into account the time variable t too.
The plane state of stress corresponds to a plate acted upon by loads parallel to
the middle plane. If the loads are uniformly distributed on the thickness at the
frontier of the plate, than the problem is a bidimensional problem; otherwise, the
problem is a quasi-bidimensional one.
In the static case, one uses the equations of equilibrium
r11;1 þ r12;2 ¼ 0; r21;1 þ r22;2 ¼ 0; ð13:150Þ
which lead to a representation of the Airy type for the stresses, in the absence of
volume forces.
In the dynamic case, inertia forces are added, and the equations of motion read
r11;1 þ r12;2 ¼ q€
u1 ; r21;1 þ r22;2 ¼ q€u2 ; ð13:1500 Þ
Hereafter we give some general results concerning plane and antiplane states of
strain; then we deal with representation of such states of strain in the dynamic case.
As the stress tensor, the strain tensor may also be decomposed in the sum of two
tensors, i.e.
1
e12 ¼ x3 u;12 þ ðw1;2 þ w2;1 Þ: ð13:1550 Þ
2
Saint-Venant’s equations of continuity are satisfied by means of Lamé’s equations
1
h2 ui þ h;i ¼ 0; i ¼ 1; 2; 3; h ¼ ei;i ¼ x3 Du þ w1;1 þ w2;2 ; ð13:156Þ
1 2m
where the hyperbolic operator of d’Alembert has been introduced (see Sect. 5.3.1),
One obtains the equations
1
x3 h2 u;1 þ h2 w1 þ ðx3 Du;1 þ w1;11 þ w2;12 Þ ¼ 0;
1 2m
1
x3 h2 u;2 þ h2 w2 þ ðx3 Du;2 þ w1;12 þ w2;22 Þ ¼ 0;
1 2m
1
h2 u þ ðDuÞ ¼ 0;
1 2m
so that
ð1 2mÞh2 u;1 þ Du;1 ¼ 0; ð1 2mÞh2 u;2 þ Du;2 ¼ 0; ð13:1560 Þ
ð1 2mÞh2 u Du ¼ 0 ð13:15600 Þ
and
ð1 2mÞh2 w1 þ ðw1;1 þ w2;2 Þ;1 ¼ 0; ð1 2mÞh2 w2 þ ðw1;1 þ w2;2 Þ;2 ¼ 0:
ð1 2mÞh2 u þ Du ¼ 2€f ;
where f ¼ f ðtÞ; using also the Eq. (13.15600 ), it follows that
ð1 2mÞh2 u ¼ Du ¼ €f :
One obtains thus
2m 2 € m 2€
€¼
u c2 f ¼ c f;
1 2m 1m 1
so that
m 2
u¼ c f þu
1t þ u
2;
1m 1
13.3 Plane and Antiplane Problems 607
i ¼ u
where u i ðx1 ; x2 Þ; i ¼ 1; 2; by applying the operator D; we have
u2 ¼ €f ;
u1 þ D
tD
so that
ui ¼ ai ;
D i ¼ 1; 2;
and
1 1
f ¼ a1 t 3 þ a2 t 2 þ a3 t þ a4 ; ð13:157Þ
6 2
where ak ; k ¼ 1; 2; 3; 4; are constants. One observes that
1
i ¼ ui þ ai ðx21 þ x22 Þ;
u i ¼ 1; 2: ð13:158Þ
4
The above formulae lead to the representation of the state of displacement in
the form
1 1
u1 ¼ u1;1 þ a1 x1 t þ u2;1 þ a2 x1 x3 þ w1 ;
2 2
ð13:159Þ
1 1
u2 ¼ u1;2 þ a1 x2 t þ u2;2 þ a2 x2 x3 þ w2 ;
2 2
m 2 1 3 1 2
u3 ¼ c 1 a1 t þ a2 t þ a3 t þ a4
1m 6 2
1 1
þ u1 þ a1 ðx21 þ x22 Þ t þ u2 þ a2 ðx21 þ x22 Þ; ð13:1590 Þ
4 4
By the aid the formulae (13.155), (13.1550 ), one obtains the state of strain and
then the state of stress. It results
r31 ¼ 0; r32 ¼ 0 ð13:162Þ
and
2lm
r33 ¼ h2 w;2 x3 ða1 t þ a2 Þ ; ð13:163Þ
1 2m
in case of the representation (13.161), or
2lm
r33 ¼ h2 w;1 x3 ða1 t þ a2 Þ ; ð13:1630 Þ
1 2m
in case of the representation (13.1610 ).
Starting from the representation (13.155), (13.1550 ), which is polynomial in
t (excepting the functions w1 and w2 ), we obtain the static case if all the coeffi-
cients of t vanish; hence, we must have u1;1 ¼ u1;2 ¼ u1 ¼ 0; a1 ¼ a2 ¼ a3 ¼ 0:
At the same time, the velocity c1 vanishes. In this case, we use the formulae
(13.164), where the function u ¼ uðx1 ; x2 Þ of class C3 is a harmonic function
Du ¼ 0: ð13:164Þ
The functions w1 ; w2 are given by
1 1
w1 ¼ w ; w2 ¼ Dw þ w ð13:165Þ
1 2m ;12 1 2m ;11
or by
1 1
w1 ¼ Dw þ w ; w2 ¼ w ; ð13:1650 Þ
1 2m ;22 1 2m ;12
where the function w ¼ wðx1 ; x2 Þ of class C4 is biharmonic
DDw ¼ 0: ð13:1640 Þ
We obtain the state of stress in an analogous manner. We have
r31 ¼ r32 ¼ 0 ð13:166Þ
and
2Gm
r33 ¼ Dw;2 ; ð13:167Þ
1 2m
in case of the representation (13.156), or
2Gm
r33 ¼ Dw;1 ; ð13:1670 Þ
1 2m
in case of the representation (13.1560 ).
13.3 Plane and Antiplane Problems 609
Observing that the above representations (in the dynamic case, as well as in the
static one) have been obtained by a demonstration step by step, the respective
representations being thus complete.
The study puts into evidence the approximation made assuming that in the
plane state of strain the displacements depend only on the two variables x1 and x2 :
hence,
u1 ¼ x 2 u þ u
1;3 ; u2 ¼ x1 u þ u
2;3 ; ð13:169Þ
u3;31 ¼ ð1 2mÞðx2 h2 u þ h2 u
1;3 Þ;
u3;32 ¼ ð1 2mÞðx1 h2 u þ h2 u 2;3 Þ;
1 2m 1
u3;33 ¼ u3;11 þ u3;22 2 €u3 :
2ð1 mÞ c3
u;3 ¼ u)
Integrating with respect to x3 ; it results (
Finally, we get
1 2m
h 1 u3 ¼ w þ w2;2 ; ð13:171Þ
2ð1 mÞ 1;1
so that
w1;2 w2;1 ¼ 2ð1 2mÞh2 u
¼ 2ð1 2mÞh;
¼w
with the condition w ; hence, w ¼W ;1 ; w2 ¼ W
;2 with W
¼ Wðx
1 ; x2 ; tÞ:
1;2 2;1 1
Introducing in (13.170) and taking into account
¼ 0;
h2 u ð13:172Þ
we get
;1 ð1 2mÞh2 u
u3;1 ¼ W ;2 ð1 2mÞh2 u
1 ; u3;2 ¼ W 2;
so that
ð1 2mÞðx1 h2 u
u3 ¼ W 1 þ x2 h2 u
2 þ UÞ; ð13:1690 Þ
¼ Uðx
with U 3 ; tÞ:
The Eq. (13.171) leads to
ð1 2mÞðx1 h1 h2 u
h1 u3 ¼ h1 W 1 þ x2 h 1 h 2 u
2 þ h1 UÞ
1 2m Þ ¼ 1 2m DW:
¼ ðw1;1 þ w2;2
2ð1 mÞ 2ð1 mÞ
Differentiating with respect to x3 ; one obtains
1;3 þ x2 h1 h2 u
x1 h1 h2 u ;3 ¼ 0;
2;3 þ h1 U
1;3 ¼ 0; h1 h2 u
h1 h2 u ;3 ¼ 0;
2;3 ¼ 0; h1 U
wherefrom
1 ¼ h1 ; h 1 h 2 u
h1 h2 u ¼ H;
2 ¼ h2 ; h 1 U ð13:173Þ
h1 U ¼ 0; ð13:1750 Þ
ð13:1740 Þ
where the function W ¼ Wðx1 ; x2 ; tÞ verifies the equation
h2 W ¼ 0: ð13:176Þ
Differentiating the Eq. (14.135) with respect to x3 ; it results
h2 u ¼ 0: ð13:1760 Þ
Introducing the functions thus obtained in the relations (13.169) and (13.1690 )
and taking into account (13.174) and (13.1740 ), one may express the state of
displacement in the form
u1 ¼ x2 u þ u1;3 ; u2 ¼ x1 u þ u2;3 ; ð13:177Þ
2m 2m
r31 ¼ l W;1 þ x2 u;3 2 u
€ ; r32 ¼ l W;2 x1 u;3 2 u
€ ; ð13:1780 Þ
c2 1 c2 2
obviously, r12 ¼ 0:
In the corresponding static case, if we start from the representation (13.1770 )
and equate to zero the dependence of the potential functions on the time variable t,
then we take the functions h1 ; h2 and H as constants. The Eq. (14.3.25) show that
the functions ui ¼ u
i ðx3 Þ; i ¼ 1; 2; are polynomials of fourth degree, the func-
tion / ¼ /ðx3 Þ being a polynomial of second degree; as well, the particular
integral of (13.1730 ) will be a polynomial of third degree, while the function
W ¼ Wðx1 ; x2 Þ must be harmonic
DW ¼ 0: ð13:179Þ
We obtain thus the state of displacement
E½u1 ðx2 x03 þ x3 x02 þ u01 Þ ¼ ½ða1 x3 þ 3b1 Þx3 þ ax2 x3 ;
ð13:180Þ
E½u2 ðx3 x01 þ x1 x03 þ u02 Þ ¼ ½ða2 x3 þ 3b2 Þx3 ax1 x3 ;
E½u3 ðx1 x02 þ x2 x01 þ u03 Þ ¼ W þ ð1 mÞ½2ða1 x31 þ a2 x32 Þ 3bðx21 þ x22 Þ
3ð1 2mÞ½ða1 x3 þ 2b1 Þx1
þ ða2 x3 þ 2b2 Þx2 ðbx3 þ cÞx3 ; ð13:1800 Þ
References
A. Books
B. Papers
4. Almansi, E.: Sulle deformationi delle piastre elastiche. Rend R. Accad. Lincei, Cl. Sci. fis.
mat. e nat. ser. 6. 17, 12 (1933)
5. Davidescu-Moisil, A.: Asupra unui caz de echilibru al corpurilor elastice izotrope (On a Case
of Equilibrium of Isotropic Elastic Bodies). Stud. cerc. mat. 15, 225 (1964)
6. Michell, J.H.: On the direct determination of stress in an elastic solid with application to the
theory of plates. Proc. Lond. Math. Soc. 31, 100 (1900)
7. Moisil, Gr.C.: Integrale ale ecuatßiilor echilibrului elastic. I. Integrale caracterizate prin
conditßii geometrice privind deplasările (Integrals of the equations of elastic equilibrium. I.
Integrals characterized by conditions concerning displacements). Bul. ßst. Acad. ser. mat. fiz.
chim. 2, 283 (1950)
8. Supino, G.: I sistemi elastici in due dimensioni e le loro relazioni con la deformazione
spaziale. Rend. R. Accad. Lincei Cl. Sci. fis. mat. e nat. ser. 6. 1, 116 (1925)
9. Supino, G.: Sul problema di Clebsch. Rend. R. Accad. Lincei Cl. Sci. fis. mat. e nat. ser. 6.
15, 366 (1932)
10. Supino, G.: Sopra la deformazione delle lastre. Rend. R. Accad. Lincei Cl. Sci. fis. mat. e nat.
ser. 6. 15, 448 (1932)
11. Supino, G.: Il problema elastico piano e la sua interpretazione nello spazio. Rend. R. Accad.
Lincei Cl. Sci. fis. mat. e nat. ser. 6. 22, 522, 581 (1935)
12. Supino, G.: Sopra la teoria delle lastre elastiche. Ann. mat. pura ed appl. ser. IV. 24, 39
(1945)
13. Supino, G.: Calcolo approxximato delle piatre elastiche. Atti V. Congresso Unione Mat.
Italiana, Pavia-Torino, 154 (1955)
14. Teodorescu, P.P.: Sur l’approximation de calcul bidimensionnel dans le cas d’un état de
tension plane. Mathematica (Cluj) 1(24), 345 (1959)
15. Teodorescu, P.P.: Considérations concernant la formulation mathématique du problème plane
de la théorie de l’élasticité. Rev. Roum. Math. Pures Appl. 9, 317 (1964)
16. Teodorescu, P.P.: Quelques problèmes bidimensionnels de la théorie de l’élasticité. Bull.
Math. Soc. Sci. Math. Roum. 10(58), 345 (1966)
17. Teodorescu, P.P.: Problèmes bidimensionnels de la théorie de l’élasticité. I. Une tension
normale nulle. Atti. Accad. Naz. Lincei. ser. VIII Rend. Cl. Sci. fis. mat. e nat. XLIV, 201
(1968)
18. Teodorescu, P.P.: Problèmes bidimensionnels de la théorie de l’élasticité. II. Deux tensions
tangentielles nulles. Atti. Accad. Naz. Lincei. ser. VIII, Rend. Cl. Sci. fis. mat. e nat. XLIV,
370 (1968)
19. Teodorescu, P.P. : Sur les problèmes bidimensionnels de la théorie de l’élasticité. Rend. Sem.
Matem. Torino, 27, 87 (1967–1968)
20. Teodorescu, P.P.: Quelques considérations sur les problèmes bidimensionnels de la théorie de
l’élasticité. Rev. Roum. Math. Pures Appl. XIII, 1467 (1968)
21. Teodorescu, P.P.: Quelques considérations sur le problème antiplan de la théorie de
l’élasticité. Rev. Roum. Sci. Techn. Méc. Appl. 22, 907 (1977)
22. Teodorescu, P.P.: La représentation par fonctions potentiel d’un état de déformation
antiplane. Rev. Roum. Sci. Techn., Méc. Appl. 23, 687 (1978)
23. Teodorescu, P.P.: Sur quelques problèmes bidimensionnels de la théorie de l’élasticité. Rev.
Roum. Sci. Techn. Méc. Appl. 25, 401 (1980)
24. Teodorescu, P.P.: On some quasibidimensional problems of the theory of elasticity.
SiSOM’99, C39 (1999)
25. Teodorescu, P.P.: On the Plane State of Strain of Elastodynamics. Bul. S ßt. Univ. Transilvania,
Brasßov, ser. A, Mec. 79 (2001)
26. Teodorescu, P.P.: Asupra stării de deformatßie antiplană în elastodinamică (On the Antiplane
State of Deformation in Elastodynamics). Bul. ßstiintß. Conf. natß. mec. sol. Brăila, 191 (2002)
Chapter 14
Anisotropic and Non-homogeneous Bodies
We have introduced in Sect. 4.1.3.1 Hooke’s tensor Hijkl , putting in evidence the
conditions which must be verified so that the quadratic form of the elastic potential
(4.54) be positive defined, using the criterion of Sylvester. Considering succes-
sively bodies with a plane of elastic symmetry (Sect. 4.1.3.2) and orthotropic
bodies (Sect. 4.1.3.3), one obtained Hooke’s law for an isotropic body (Sects.
4.1.3.4 and 4.1.3.5), which has been used in the study made till now.
Besides the bodies with intrinsic anisotropic properties, we may consider
structures which can be modelled as anisotropic ones, e.g., beams with transverse
ribs used for bridges and modelled as orthotropic plates. Other modelling as
anisotropic bodies is that of composite bodies, e.g., fiber-reinforced composites,
which—theoretically—are studied as such ones (see, for instance, the monograph
of Cristescu, Crăciun and Soós [3]).
Hereafter we shall consider various cases of anisotropy and we shall present
some elements of crystallography.
Starting from the general form of Hooke’s tensor (i.e., of Hooke’s law) we shall
give mechanical interpretations to the elastic coefficients which are involved, in
the general case as well as in particular cases. Thermoelastic and dynamic prob-
lems will be considered too.
e11 ¼ a11 r11 þ a12 r22 þ a13 r33 þ a14 r23 þ a15 r31 þ a16 r12 þ e011 ; . . .; ð14:1Þ
c23 ¼ a41 r11 þ a42 r22 þ a43 r33 þ a44 r23 þ a45 r31 þ a46 r12 þ c023 ; . . .; ð14:10 Þ
r23 ¼ A41 e11 þ A42 e22 þ A43 e33 þ A44 c23 þ A45 c31 þ A46 c12 ; . . .; ð14:20 Þ
V.
g1;23 g23;1 g2;31 g31;2 g3;12 g12;3
a14 ¼ ¼ ; a25 ¼ ¼ ; a36 ¼ ¼ ; ð14:60 Þ
G23 E1 G31 E2 G12 E3
VI.
g2;23 g23;2 g3;31 g31;3 g1;31 g31;1
a24 ¼ ¼ ; a35 ¼ ¼ ; a15 ¼ ¼ ;
G23 E2 G31 E3 G31 E1
g2;12 g12;2 g3;23 g23;3 g1;12 g12;1 ð14:600 Þ
a26 ¼ ¼ ; a34 ¼ ¼ ; a16 ¼ ¼ :
G12 E2 G23 E3 G12 E1
We call body with transverse isotropy that one for which the planes normal to a
certain axis are isotropic planes (in an arbitrary point of the body one has the same
properties in all the directions contained in such a plane). This axis (called by
P. Bekhterev axis of monotropy) is an axis of total symmetry (of order n ¼ 1, as it
has been shown by S. G. Lekhnitski); this axis is identical to that with an order of
symmetry n ¼ 6.
Starting from the general relations (7.4), (6.1.40 ), one may write the generalized
Hooke law in the form (see [25])
1 m0 1 m0
e11 ¼ ðr11 mr22 Þ 0 r33 ; e22 ¼ ðr22 mr11 Þ 0 r33 ; ð14:7Þ
E E E E
m00 1
e33 ¼ ðr11 þ r22 Þ þ 0 r33 ; ð14:70 Þ
E E
1 1 1
c23 ¼ r23 ; c31 ¼ 0 r31 ; c12 ¼ 0 r12 ; ð14:700 Þ
G0 G G
where we have chosen the Ox3 -axis as axis of monotropy. Here, E and E0 are
moduli of longitudinal elasticity, G and G0 are moduli of transverse elasticity and
m; m0 and m00 are coefficients of transverse contraction of Poisson type.
One has thus 7 elastic constants, which verify the relations
E E
G¼ ; m00 ¼ 0 m0 ; ð14:8Þ
2ð1 þ mÞ E
620 14 Anisotropic and Non-homogeneous Bodies
we are lead to the same relations (14.10), (14.100 ), where we replace the elastic
constants E and m by the generalized elastic constants
E m þ m0 m00
E0 ¼ 0 00
; m0 ¼ ; ð14:11Þ
1mm 1 m0 m00
because the deformation is stopped in the direction of the Ox3 -axis, appears the
normal stress
Excepting the determination of e33 in the case of a plane state of stress and of r33
in the case of a plane state of strain, the problem is identical, from the mathe-
matical point of view, in both cases; as well, the problem is identical to that of an
isotropic body.
The second important case is that in which the axis of monotropy is contained
in the plane in which appear the stresses and the strains. Some results in this
direction have been given by A. Moisil [17] and M. Iacovache [13], as well as
by M. Cristea [10].
We assume that the plane mentioned above is normal to the Ox2 -axis. In the
case of a plane state of stress one has
r22 ¼ 0; r21 ¼ 0; r23 ¼ 0: ð14:13Þ
14.1 Anisotropic Elastic Bodies 621
obtaining the same relations (14.14), (14.1400 ), where we replace the elastic con-
stants E; E0 ; m0 ; m00 by the generalized elastic constants
E E0
E0 ¼ ; E00 ¼ ;
1m 2 1 m0 m00 ð14:16Þ
1 þ m 0 00 m00
m00 ¼ 0 00
m ; m0 ¼ :
1mm 1m
There appears also the normal stress
Taking into account the relations (14.14), (14.140 ), we may write the equation of
continuity in stresses
1 1 1 m00 m0
r11;33 þ 0 r33;11 ¼ r13;13 ; ð14:1800 Þ
E E G0 E E 0
622 14 Anisotropic and Non-homogeneous Bodies
the stress function F ¼ Fðx1 ; x3 Þ must satisfy the partial differential equation
1 1 m00 m0 1
0
F;1111 þ 0 F;1133 þ F;3333 ¼ 0: ð14:20Þ
E G E E E
This equation may be written also in the form
ðs1 o1 þ io3 Þðs1 o1 io3 Þðs2 o1 þ io3 Þðs2 o1 io3 ÞF ¼ 0; ð14:200 Þ
where
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s ffi
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 2
u E E E
t
s1;2 ¼ m00 m00 0
2G0 2G0 E
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffiffiffi
1 E E 1 E E
¼ 2m00 þ 2 2m00 2 ð14:2000 Þ
2 G0 E0 2 G0 E0
and where we used the relations (14.8). Thus, the general integral of this equation
may be represented, by means of certain functions of complex variable, in the form
One may obtain then the state of strain and the state of displacement.
We call orthotropic body that one through any point of which pass three planes of
elastic symmetry (see [26]).
Starting from the general results given in Sect. 14.1.1.1, we obtain the gen-
eralized Hooke law
1 m12 m13
e11 ¼ r11 r22 r33 ;
E1 E2 E3
m21 1 m23
e22 ¼ r11 þ r22 r33 ; ð14:22Þ
E1 E2 E3
m31 m32 1
e33 ¼ r11 r22 þ r33 ;
E1 E2 E3
1 1 1
c23 ¼ r23 ; c31 ¼ r31 ; c12 ¼ r12 ; ð14:220 Þ
G23 G31 G12
14.1 Anisotropic Elastic Bodies 623
there are involved 12 elastic constants of the material, between which take place
the relations
m12 m21 m23 m32 m31 m13
¼ ; ¼ ; ¼ ; ð14:23Þ
E2 E1 E3 E 2 E1 E3
so that only 9 elastic constants are independent.
In the case of a plane state of stress, we put the conditions
r33 ¼ 0; r31 ¼ r32 ¼ 0
and the relations (14.22), (14.220 ) take the form
1 m12 m21 1
e11 ¼ r11 r22 ; e22 ¼ r11 þ r22 ; ð14:24Þ
E1 E2 E1 E2
1
c12 ¼ r12 ; ð14:240 Þ
G12
to which one adds the linear strain
m31 m32
e33 ¼ r11 r22 ; ð14:2400 Þ
E1 E2
one has thus only 6 distinct elastic constants.
In the case of a state of plane strain one has e33 ¼ 0; c31 ¼ c32 ¼ 0; one is thus
led to the same relations (14.24), (14.240 ), where the elastic constants
E1 ; E2 ; m12 ; m21 are replaced by the generalized elastic constants
E1 E2
E10 ¼ ; E20 ¼ ;
1 m13 m31 1 m23 m32
ð14:25Þ
m12 þ m13 m32 0 m21 þ m23 m31
m012 ¼ ; m21 ¼ :
1 m23 m32 1 m13 m31
There appears also the stress
r33 ¼ m13 r11 þ m23 r22 ; ð14:26Þ
because the deformation in the Ox3 -direction vanishes. There appear only 6 elastic
constants too.
Excepting the determination of e33 for the state of plane stress and of r33 for the
state of plane strain, the problem is identical from the mathematical point of view,
involving only 4 distinct elastic constants (the other 2 constants are used to
determine the quantities mentioned above).
An interesting study is due to H. A. Lang [14] who, by an affine transformation,
replaces an orthotropic problem by an isotropic one. We remark also the results
given by E. Reissner [19] for some particular cases, especially if some constants
tend to zero. I .S. Sokolnikoff [23] gave an interesting approximate method of
624 14 Anisotropic and Non-homogeneous Bodies
calculation; starting from the corresponding isotropic problem, one introduces the
method of perturbations (a parameter of deviation from the isotropic solution to
the anisotropic one).
To the equations of equilibrium (without volume forces)
r11;1 þ r12;2 ¼ 0; r12;1 þ r22;2 ¼ 0
Using the relations (14.24), (14.240 ), the equations of continuity in stresses take
the form
1 1 1 m21 m12
r11;22 þ r22;11 ¼ r12;12 : ð14:27Þ
E1 E2 G12 E1 E2
The general integral of this equation may be represented by means of the functions
of complex variable in the form
Solving the Eqs. (14.24), (14.240 ) with respect to stresses, one can write
E1 E2
r11 ¼ ðe11 þ m12 e22 Þ; r22 ¼ ðe22 þ m21 e11 Þ; ð14:30Þ
1 m12 m21 1 m12 m21
r12 ¼ G12 c12 : ð14:300 Þ
Replacing these expressions in the equations of equilibrium and taking into
account Cauchy’s relations, one obtains the equations which must be satisfied by
the displacements
14.1 Anisotropic Elastic Bodies 625
E1
ðu1;11 þ m12 u2;12 Þ þ G12 ðu1;11 þ u2;12 Þ ¼ 0;
1 m12 m21
ð14:31Þ
E2
G12 ðu1;12 þ u2;11 Þ þ ðu2;22 þ m21 u1;12 Þ ¼ 0:
1 m12 m21
Because the differential operators which act upon the functions u1 and u2 are
prime between them, it results that one may represent the displacements by means
of only one stress function F ¼ F ðx1 ; x2 Þ in the form
E1 m12 E1
u1 ¼ þ G12 F ; 12 ; u2 ¼ F ; 11 þ G12 F ; 22 ; ð14:32Þ
1 m12 m21 1 m12 m21
or in the form
E2 E1 m12
u1 ¼ F ; 22 þ G12 F ; 11 ; u2 ¼ þ G12 F ; 12 : ð14:320 Þ
1 m12 m21 1 m12 m21
If we put the condition that each of these representations, given by one of the
Eq. (14.31), verifies the other one, we observe that the function F must verify the
same Eq. (14.28) of elliptic type. This representation is analogue to that given by
K. Marguerre [15] for the plane isotropic problem. The solution is useful for the
first fundamental problem.
We call body with a plane of elastic symmetry that one for which there exists a plane
with the property that two directions symmetric with respect to it are equivalent from
the point of view of the elastic properties (see [27]). Starting from the general results
in Sect. 14.1.1.1, one can write Hooke’s generalized law in the form
1 m12 m13 g1;12
e11 ¼ r11 r22 r33 þ r12 ;
E1 E2 E3 G12
m21 1 m23 g2;12
e22 ¼ r11 þ r22 r33 þ r12 ; ð14:33Þ
E1 E2 E3 G12
m31 m32 1 g3;12
e33 ¼ r11 r22 þ r33 þ r12 ;
E1 E2 E3 G12
1 m23;31
c23 ¼ r23 þ r31 ;
G23 G31
ð14:330 Þ
m31;23 1
c31 ¼ r33 þ r31 ;
G23 G31
g12;1 g12;2 g12;3 1
c12 ¼ r11 þ r22 þ r33 þ r12 ; ð14:3300 Þ
E1 E2 E3 G12
where the Ox3 -axis is normal to the plane of elastic symmetry.
626 14 Anisotropic and Non-homogeneous Bodies
In these relations are involved 20 constants between which take place the
relations
m21 m12 m32 m23 m13 m31
¼ ; ¼ ; ¼ ; ð14:34Þ
E1 E2 E2 E 3 E3 E1
g12;1 g1;12 g12;2 g2;12 g12;3 g3;12
¼ ; ¼ ; ¼ ; ð14:340 Þ
E1 G12 E2 G12 E3 G12
m31;23 m23;31
¼ ; ð14:3400 Þ
G23 G31
there remain thus only 13 independent elastic constants.
The conditions
r33 ¼ 0; r31 ¼ 0; r32 ¼ 0
lead to a state of plane stress, for which Hooke’s law takes the form
1 m12 g1;12
e11 ¼ r11 r22 þ r12 ;
E1 E2 G12
ð14:35Þ
m21 1 g2;12
e22 ¼ r11 þ r22 þ r12 ;
E1 E2 G12
g12;1 g12;2 1
c12 ¼ r11 þ r22 þ r12 ð14:350 Þ
E1 E2 G12
to which one adds the linear strain
m31 m32 g3;12
e33 ¼ r11 r22 þ r12 ; ð14:3500 Þ
E1 E2 G12
there remain 9 distinct elastic constants.
If we assume that c33 ¼ 0; c31 ¼ c32 ¼ 0, then, for the case of a plane state of
strain, one can use the same constitutive law (14.35), (14.350 ), where the elastic
constants are replaced by the generalized elastic constants
E1 E2 G12
E10 ¼ ; E20 ¼ ; G012 ¼ ; ð14:36Þ
1 m13 m31 1 m23 m32 1 g12;3 g3;12
Because the deformation is stopped in the direction Ox3 . there appears the normal
stress
14.1 Anisotropic Elastic Bodies 627
We say that a body has an axis of symmetry of order n (an n-gonal axis) if two
arbitrary directions are superposing after a rotation of 2p=n about that axis are
equivalent from the point of view of the elastic properties. It has been stated that
one can have only n ¼ 2 (digonal axis), n ¼ 3 (trigonal axis), n ¼ 4 (tetragonal
axis), n ¼ 6 or 1 (hexagonal axis) (see, e.g., [6]); this corresponds to the law of
rational indices. The cases n ¼ 2 (plane of elastic symmetry) and n ¼ 6 (trans-
verse isotropy) have been considered above (see [28, 29]).
628 14 Anisotropic and Non-homogeneous Bodies
In the case of a trigonal axis, the generalized Hooke law takes the form
1 m12 m13 g1;23 g1;31
e11 ¼ r11 r22 þ r33 þ r23 þ r31 ;
E1 E2 E3 G23 G31
ð14:40Þ
m21 1 m23 g2;23 g2;31
e22 ¼ r11 r22 r33 þ r23 þ r31 ;
E1 E2 E3 G23 G31
m31 m32 1
e33 ¼ r11 r22 þ r33 ; ð14:400 Þ
E1 E2 E3
g23;1 g23;2 1 m23;12
c23 ¼ r11 þ r22 þ r23 þ r12 ;
E1 E2 G23 G12
ð14:4000 Þ
g31;1 g31;2 1 m31;12
c31 ¼ r11 þ r22 þ r31 þ r12 ;
E1 E2 G31 G12
m12;23 m12;31 1
c12 ¼ r23 þ r31 þ r12 ; ð14:40000 Þ
G23 G31 G12
where the Ox3 -axis has been chosen as axis of symmetry. There are 24 elastic
constants of the material, which verify the relations
m21 m12 m32 m23 m13 m31
¼ ; ¼ ; ¼ ;
E1 E2 E2 E3 E3 E1
g23;1 g1;23 g31;1 g1;31 g23;2 g2;23
¼ ; ¼ ; ¼ ; ð14:41Þ
E1 G23 E1 G31 E2 G23
g31;2 g2;31 m23;12 m12;23 m31;12 m12;31
¼ ; ¼ ; ¼ ;
E2 G31 G12 G23 G12 G31
because the Ox3 -axis is an axis of symmetry of third order, there appear following
relations between the elastic constants
1 1 m12
E1 ¼ E2 ; G23 ¼ G31 ; ¼2 þ ; m13 ¼ m23 ;
G12 E1 E 2
ð14:410 Þ
g23;1 g23;2 m31;12 g31;1 g31;2 m23;12
¼ ¼ ; ¼ ¼ :
E1 E2 G12 E1 E2 G12
Hence, there remain 24 ð9 þ 8Þ ¼ 7 independent elastic constants.
The relations (14.41), (14.410 ) allow to write Hooke’s law in the form
1 m0 1
e11 ¼ ðr11 mr22 Þ 0 r33 þ 0 ðgr23 g0 r31 Þ;
E E G ð14:42Þ
1 m0 1
e22 ¼ ðr22 mr11 Þ 0 r33 0 ðgr23 g0 r31 Þ;
E E G
m00 1
e33 ¼ ðr11 þ r22 Þ þ 0 r33 ; ð14:420 Þ
E E
14.1 Anisotropic Elastic Bodies 629
g00 1 m
c23 ¼ ðr11 r22 Þ þ 0 r23 þ r12 ;
E G G ð14:4200 Þ
g000 1 m0
c31 ¼ ðr11 r22 Þ þ 0 r31 þ r12 ;
E G G
1 00 1
c12 ¼ ðm r23 þ m000 r31 Þ þ r12 ; ð14:42000 Þ
G0 G
using thus 15 elastic constants; between these constants there exist 8 relations
m0 m00 m0 m000 m m00 1 2ð1 þ mÞ
0
¼ ; ¼ 0; ¼ 0; ¼ ;
E E G G G G G E ð14:43Þ
g g00 g0 g000 g00 m0 g000 m
¼ ; ¼ ; ¼ ; ¼ :
G0 E G0 E E G E G
If we put the conditions r33 ¼ 0; r31 ¼ r32 ¼ 0 for a state of plane stress (the
elastic axis has been considered to be normal to the plane in which take place the
relations)
1 1
e11 ¼ ðr11 mr22 Þ; e22 ¼ ðr22 mr11 Þ;
E E ð14:44Þ
2ð1 þ mÞ 1
c12 ¼ r12 ¼ r12 ;
E G
where appears the elastic constants of the isotropic bodies. The linear strain along
the elastic axis is given by
m00
e33 ¼ ðr11 þ r22 Þ; ð14:440 Þ
E
to which one adds the angular strains (which, in the isotropic case, are equal to
zero)
1 00
c23 ¼ ½m ðr11 r22 Þ þ mr12 ;
G ð14:4400 Þ
1
c31 ¼ ½mðr22 r11 Þ þ m0 r12 ;
G
where the relations (14.43) have been used.
In the case of a plane state of strain (e33 ¼ 0; c31 ¼ c32 ¼ 0) one uses the same
relations (14.44), where the generalized elastic constants
E G
E0 ¼ ; G0 ¼ ;
1 m0 m00 00
ðgg þ g g Þ0 000 1 ðmm m0 m000 Þ
00
ð14:45Þ
m þ m0 m00 ðgg00 þ g0 g000 Þ
m0 ¼
1 m0 m00 ðgg00 þ g0 g000 Þ
are introduced. We mention that these constants are independent.
630 14 Anisotropic and Non-homogeneous Bodies
Because the deformation is stopped in the direction of the axis, there appears
the normal stress
r33 ¼ m0 ðr11 þ r22 Þ ð14:46Þ
r31 ¼ m00 ðr11 r22 Þ m000 r12 ; r32 ¼ m000 ðr22 r11 Þ m00 r12 : ð14:460 Þ
The problem of plane stress is the same as in the isotropic case; but in the
problem of plane strain, the equation of continuity in stresses takes the form
1 1 m0
ðr11;22 r22;11 Þ ¼ 2 r12;12 : ð14:47Þ
E0 G0 E0
One uses the elastic representation of Airy, where the potential function F ¼
Fðx1 ; x2 Þ verifies the equation
E0
F ;1111 þ 2m0 F;1122 þ F ;2222 ¼ 0; ð14:48Þ
G0
where
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 E0 1 E0
s1;2 ¼ þ 2ð1 m0 Þ 2ð1 m0 Þ: ð14:4800 Þ
2 G0 2 G0
In displacements, the solution may be represented in the form
E0 m 0 E0
u1 ¼ þ G0 F ; 12 ; u2 ¼ F ; 11 þ G0 F ; 22 ð14:49Þ
1 m20 1 m20
or in the form
E0 E 0 m0
u1 ¼ F ; 22 þ G0 F ; 11 ; u 2 ¼ þ G0 F ; 12 ; ð14:490 Þ
1 m20 1 m20
m31 m32 1
e33 ¼ r11 r22 þ r33 ; ð14:500 Þ
E1 E2 E3
1 1
c23 ¼ r23 ; c31 ¼ r31 ; ð14:5000 Þ
G23 G31
g12;1 g12;2 1
c12 ¼ r11 þ r22 þ r12 ; ð14:50000 Þ
E1 E2 G12
where the Ox3 -axis is that of elastic symmetry.
There appear 16 elastic constants of the material, between which take place the
relations
m21 m12 m32 m23 m13 m31
¼ ; ¼ ; ¼ ;
E1 E 2 E2 E 3 E3 E1
g12;1 g1;12 g12;2 g2;12 ð14:51Þ
¼ ; ¼ :
E1 G12 E2 G12
Because the elastic axis is of fourth order, there take place the relations
E1 ¼ E2 ; G23 ¼ G31 ;
ð14:510 Þ
m13 ¼ m23 ; g1;12 ¼ g2;12
1 g 1 g
e11 ¼ ðr11 mr22 Þ þ r12 ; e22 ¼ ðr22 mr11 Þ r12 ; ð14:54Þ
E G E G
g0 1
c12 ¼ ðr11 r22 Þ þ r12 ; ð14:540 Þ
E G
where appear 5 elastic constants, connected by the second relation (14.53). The
linear strain along the elastic axis is given by
m00
e33 ¼ ðr11 þ r22 Þ: ð14:5400 Þ
E
For a plane state of strain, we use the same relations (14.54), (14.540 ), the
generalized elastic constants taking the form
E m þ m0 m00
E0 ¼ 0 00
; m0 ¼ ;
1mm 1 m0 m00 ð14:55Þ
g0
g00 ¼ ;
1 m0 m00
because the linear strain along the axis of symmetry is stopped, there appears the
normal stress
Let us consider the case of an orthotropic body (see Sect. 14.1.1.3). In a plane state
of stress, the equations of equilibrium read
r11;1 þ r12;2 þ F1 ¼ 0; r12;1 þ r22;2 þ F2 ¼ 0; ð14:59Þ
14.1 Anisotropic Elastic Bodies 633
which may take a simpler form in important particular cases. By means of such
particular integrals, one may construct, e.g., expansions into Fourier series too.
The problems considered above have been studied also in the dynamic case; we
mention, e.g., the results of R. E. D. Bishop [8] concerning both plane problems:
state of stress and state of strain.
In the case of an orthotropic elastic body, the equations corresponding to a
plane state of stress are
r11;1 þ r12;2 ¼ q€
u1 ; r21;1 þ r22;2 ¼ q€u2 ; ð14:66Þ
1 m12 m21 1
u1;1 ¼ r11 r22 ; u2;2 ¼ r11 þ r22 ; ð14:67Þ
E1 E2 E1 E2
1
u1;2 þ u2;1 ¼ r12 ; ð14:670 Þ
G12
where the elastic constants are related by
m12 m21
¼ : ð14:68Þ
E2 E1
In the case of a plane state of strain, the same equations are used, the elastic
constants being replaced by
E1 E2
E10 ¼ ; E0 ¼ ;
1 m13 m31 2 1 m23 m32
ð14:69Þ
m12 þ m13 m32 0 m23 þ m23 m31
m012 ¼ ; m23 ¼ ;
1 m23 m32 1 m13 m31
where m13 6¼ m31 ; m12 6¼ m21 are coefficients of Poisson type.
Eliminating the stresses between these equations, one obtains the equations of
Lamé type
14.1 Anisotropic Elastic Bodies 635
E1
ðu1;11 þ m12 u2;12 Þ þ G12 ðu1;22 þ u2;12 Þ ¼ q€u1 ;
1 m12 m21
ð14:70Þ
E2
G12 ðu1;12 þ u2;11 Þ þ ðu2;22 þ m21 u1;12 Þ ¼ q€u2 :
1 m12 m21
Because the differential operators which act on the displacements u1 and u2 are
prime between them, it results that the displacements may be represented by a
single potential function F in the form
E1 m12 E1 € ð14:71Þ
u1 ¼ þ G12 F ;12 ; u2 ¼ F ;11 þ G12 F ;22 qF
1 m12 m21 1 m12 m21
or in the form
E2 € E1 m12
u1 ¼ F ;22 þ G12 F ;11 qF ; u2 ¼ þ G12 F ;12 :
1 m12 m21 1 m12 m21
ð14:710 Þ
If we put the condition that each of these representations, given by one of the
equation (14.70), verify the other equation too, then we conclude that the function
F ¼ F ðx1 ; x2 ; tÞ is given by the equation
1 1 m21 m12 1
F ;1111 þ F ;1122 þ F ;2222
E2 G12 E1 E2 E1
q 1 m12 m21 1 € q 1 m12 m21 1 €
þ F ;11 þ F ;22
E2 E1 G12 E1 E2 G12
q2 ð1 m12 m21 Þ €
€ ¼ 0:
þ F ð14:72Þ
E1 E2 G12
Such a solution of the problem is useful in the case of the first problem of the
theory of elasticity.
The discriminant of the form (14.72), considered as a quadratic form, is given
by
1 1 1
2 m21
q 1m12 m21
þ 1
G12
E2 2 G12 E1 2E2
E1
1 1 m21
2 G12 2 E1
1
2Eq 1 1mE122 m21 þ G112 : ð14:73Þ
E1
q 1m12 m21 1 q 1m12 m21 1 q2 ð1m12 m21 Þ
2E E þ G 2E E þ G E E G
2 1 12 1 2 12 1 2 12
One observes easily that this discriminant does not vanishes (if we consider it as a
polynomial in 1/G12 then we observe that the term of third degree of it has a non-
zero coefficient). Hence, unlike the isotropic case, the elastic waves can no more
be obtained by superposing two types of waves (irrotational and shearing waves,
for instance), but there exists only one wave front.
636 14 Anisotropic and Non-homogeneous Bodies
must be fulfilled; hence, taking into account (14.80), the displacements must verify
the equations
E1 u1;22 þ ½2G12 ð1 m12 m21 Þ þ E1 m12 u1;22 qð1 m12 m21 Þ€u1 ¼ 0;
ð14:74Þ
½2G12 ð1 m12 m21 Þ þ E2 m21 u2;11 u2;11 þ E2 u2;22 qð1 m12 m21 Þ€u2 ¼ 0:
On the other hand, these displacements are integrals of the equation (14.72). These
conditions may be fulfilled simultaneously only in particular cases.
For a solution in stresses of the problems, one eliminates u1 and u2 between the
equations of the problem; we get
1 m12
r11;11 þ r12;12 ¼ q €11
r r
€22 ;
E1 E2
ð14:75Þ
1 m21
r12;12 þ r22;22 ¼ q €22
r r
€11 :
E2 E1
Subtracting these equations one of the other, we can state that the normal
stresses may be expressed in the form
q €
r11 ¼ F;22 ð1 þ m12 Þ F;
E2
q € ð14:76Þ
r22 ¼ F;11 ð1 þ m21 Þ F;
E1
while the tangential stress will be given by
ZZ
o2 1 1 1 m12 m21 o2
r12 ¼ F;12 þ q 2 o11 þ o22 q 2 Fdx1 dx2
ot E2 E1 E1 E2 ot
2
o 1 1
þq 2 f1 ðx2 ; tÞ þ f2 ðx1 ; tÞ : ð14:760 Þ
ot E2 E2
The relation (14.670 ) allows to determine the functions f1 ðx2 ; tÞ and f2 ðx1 ; tÞ in
the form
Z Z
1 1 1 m21 m12
F;22 dx1 þ F;11 dx2 þ F;12
E1 E2 G12 E1 E2
Z Z
o2 1 1
ð1 m12 m21 Þq 2 F;2 dx1 þ F;1 dx2
ot E2 E1
2 Z Z ð14:79Þ
q o 1 1 1 m12 m21 o2
o 11 þ o 22 q Fdx dx
1 2
G12 ot2 E2 E1 E 1 E2 ot2
1 1 q o2 1 1
þ f1;22 þ f2;11 f1 þ f2 ¼ 0:
E1 E2 G12 ot2 E1 E2
If we introduce everywhere double integrals with respect to x1 and x2 , we obtain
under the integral sign the expression in (14.72). Hence, Eq. (14.79) is with
separate variables and can be decomposed in two equations of the form
1€
f1;22 f1 þ F1 ðx2 ; tÞ þ E1 x0 ðtÞ ¼ 0;
a2 ð14:790 Þ
1
f2;11 2 €f2 þ F2 ðx1 ; tÞ E2 x0 ðtÞ ¼ 0;
a
the function x0 ðtÞ being arbitrary, while a is a velocity of waves propagation,
given by
1 q
2
¼ : ð14:80Þ
a G12
By integration, one obtains
1 2 0
f1 ðx2 ; tÞ u1 ðx2 atÞ þ w1 ðx2 þ atÞ þ f1 ðx2 ; tÞ þ E x0 ðtÞx2 þ u1 ðtÞx2 ;
2
1 2 0
f2 ðx2 ; tÞ u2 ðx1 atÞ þ w1 ðx1 þ atÞ þ f2 ðx2 ; tÞ þ E x0 ðtÞx1 þ u1 ðtÞx1 ;
2
ð14:81Þ
where u1 ; u2 ; w1 ; w2 are arbitrary functions; f1 ðx2 ; tÞ; f2 ðx1 ; tÞ are particular inte-
grals of the Eq. (14.790 ), while u01 ðtÞ; u02 ðtÞ and x0 ðtÞ are the displacements and the
rotation of the body, respectively, considered as a rigid. This constitutes a gen-
eralization for the dynamic case of the results obtained in the static one.
Choosing a function F of the form
Fðx1 ; x2 ; tÞ ¼ Fðx1 ; x2 ÞTðtÞ;
one may assume that the stress function is obtained by a superposition of harmonic
vibrations
638 14 Anisotropic and Non-homogeneous Bodies
X
1
2np
Fðx1 ; x2 ; tÞ ¼ Fn ðx1 ; x2 Þ cosðhn þ cn tÞ; cn ¼ ; ð14:82Þ
n¼1
T
Crystallography is a branch of science which studies the external form and the
internal structure of crystals, their development and growth and their properties.
Initially, crystallography was a part of mineralogy, becoming an independent
discipline only at the end of nineteenth century, when it has been found that there
exist non-mineral substances with crystalline structure and that the number of such
substances is sufficiently great. The bodies with such geometrical characteristics
have many specific properties. Nowadays, cristallography is developing in close
relationship with mineralogy and metallography, which use the results of crysta-
lography, as well as with chemistry, physics and mathematics, the results of which
are applied in crystallography.
The first references concerning the regular structure of the spatial arrangement of
the particles that make up a crystalline substance can be found in the works of
J. Kepler (1619), R. Hooke (1665) and C. Huygens (1690). In 1669, N. Stensen
discovered the law of constancy of the angle between the crystalline planes and, a
century later, in 1774 R. J. Ha} uy formulated the law of integers (of rational
indices), relative to the characterization of the positions of crystalline planes.
In the first decade of the nineteenth century, the systematic measurement of the
angles formed by crystalline planes began and W. H. Miller introduced the use of
14.1 Anisotropic Elastic Bodies 639
The crystals are bounded by faces (planes), edges (straight lines) and vertices
(points); the exterior faces of perfect crystals have the following properties:
(i) the angles between the corresponding faces of the crystals of a given chemical
compound are constant;
(ii) if the positions of the faces are expressed in a definite co-ordinate system, then
the position of any face can always be specified by three integral numbers.
The study of physical properties of crystals shows that, in general, the prop-
erties are not the same in different directions; for instance, mechanical rigidity,
heat conductivity, dielectric permittivity vary relative to the direction in which
they are determined; for this reason, we say that the crystals are anisotropic.
However, there can exist several different directions along which some properties
640 14 Anisotropic and Non-homogeneous Bodies
of a given crystal are identical; the spatial arrangement of these directions reveals
the existence of a symmetry of the geometrical structure of crystalline bodies.
The observed characteristics can be partly expressed without taking into
account the nature and the form of the constituent particles of crystals, if we
suppose that the crystalline substances, more precisely the monocrystalline sub-
stances which form the real polycrystalline media, have a regular reticular struc-
ture; for this reason, the spatial arrangement of the particles which form a perfect
crystal can be represented mathematically through a rectilinear, three-dimensional,
regular net (lattice); formed by an enormous, but finite, number of identical par-
allelepipeds, called unit cells, without any gaps between them. The vertices of the
parallelepipeds constitute the space lattice; these points are called nodes, the lines
drawn through them are rows and the planes determined by two concurrent rows
are called plane nets. Using three concurrent non-coplanar edges of a unit cell, an
oblique co-ordinate system can be introduced, having as bases just the vectors
determined by the three edges and as origin their common vertex; in this system,
the co-ordinates of this nodes are represented by triplets of integers. We say that
the geometrical structure of a crystal is characterized by a simple Bravais net if
all the component particles of the crystal are mutually identical and their totality
forms the nodes of a regular net generated by a unit cell, having particles as its
nodes. By superposition of a finite number of simple Bravais nets, we obtain a
complex Bravais net. Real crystals are formed by particles of different types, hence
their geometrical structure is characterized by a complex Bravais net; every
component simple net is constituted by particles of a certain type. With these
specifications, we can attack one of the fundamental problems of crystallography:
the determination of all classes of symmetry of real crystals.
of the tensor Q depend obviously on the co-ordinate system through which the
point transformation (14.84) is specified; in what follows, we shall choose co-
ordinate transformations such that these components have the simplest possible
form.
Symmetry is a special regularity observable in the spatial arrangement of objects
or of parts of objects. Any geometrical body is formed of systems of points. We
say that two geometrical figures (bodies) are mutually symmetric or congruent if
there exists a motion through which one of the geometrical figures can coincide
point by point with the other one. If a figure is formed of two congruent parts, we
say that it is self-symmetric.
The motions that realize the coincidence of two mutually symmetric figures or
of two parts of a self-symmetric figure are called operations of symmetry, corre-
sponding to the two geometrical figures and to the self-symmetric figure,
respectively, and the geometrical elements (points, lines, planes a.s.o.) charac-
teristic for such operations are called elements of symmetry. These motions are
defined by the relations (14.84).
Further, we shall deal only with those operations and elements of symmetry
which can occur with finite figures. The terminology and the notations used are
specific for crystallography.
(i) Reflection relative to a plane P. We suppose that the plane P contains the Ox3 -
axis and its intersection with the Ox1 x2 -plane forms the angle a with the Ox1 -
axis. We denote by P the orthogonal tensor which characterizes a reflection
relative to P. In the chosen co-ordinate system, the motion corresponding to a
reflection relative to P is described by the relations
2 3
cos 2a sin 2a 0 x01 ¼ x1 cos 2a þ x2 sin 2a;
0 4 5
r ¼ Pr; P sin 2a cos 2a 0 ; x02 ¼ x1 sin 2a x2 cos 2a; ð14:85Þ
0 0 1 x03 ¼ x3 :
The element of symmetry in this motion is the plane P, called the plane of
symmetry.
(ii) Rotation about an axis L. Sometimes, by rotating a figure through a certain
angle about an axis L, it can come into a congruent figure (position). In this
case, L is called the axis of symmetry and it represents an element of sym-
metry in the considered motion. The angle of rotation must be an exact
divisor of 360 . Sometimes, the congruence can be obtained by rotating a
figure through different angles; the smallest of them, denoted by a, is called
the elementary angle of rotation. The number
360
n¼ ð14:86Þ
a
represents the order of the axis L, which we denoted further by Ln and will be
called the symmetry axis of nth order or n-fold axis of symmetry.
642 14 Anisotropic and Non-homogeneous Bodies
Let us suppose that Ln coincides with the Ox3 -axis and let us denote by Ln the
orthogonal tensor which characterizes the rotation about Ln . In this case, the
motion corresponding to the considered rotation is described by the relation
2 3
cos a sin a 0 x01 ¼ x1 cos a x2 sin a;
r ¼ L r; L 4 sin a cos a 0 5; x02 ¼ x1 sin a þ x2 cos a;
0 n n
ð14:87Þ
0 0 1 x03 ¼ x3 :
We stress the fact that, in general, a roto-reflection does not reduce to a rotation
with a separate reflection.
Applying a roto-reflection two fold, we obtain a motion which is described by
the relations 2 3
cos 2a sin 2a 0
r0 ¼ Ln2n Ln2n r; Ln2n Ln2n ¼ 4 sin 2a cos 2a 0 5; ð14:90Þ
0 0 1
14.1 Anisotropic Elastic Bodies 643
We leave to the reader the proof of the fact that L3i may be expressed through L3
and C; hence, among the inversion axes, only L4i and L6i have a proper meaning.
We shall state hereafter some theorems, necessary for deducing all possible
combinations of symmetry elements for crystals. Thus, we state
L5
N1 N3
N
N2
14.1 Anisotropic Elastic Bodies 645
Theorem 14.2 If a geometrical figure has two or more symmetry planes P, then
their lines of intersection are symmetry axes L. The elementary angle of rotation
corresponding to such an axis is twice the angle formed by the two planes which
determine this axis.
Let P1 and P2 be two planes of symmetry. The co-ordinate axes can be always
be chosen in such a way that the line of intersection of the two planes coincide
with the Ox3 -axis and P1 coincide with the Ox1 x3 -plane. We denote by a the angle
formed by the Ox1 -axis with the intersection of the Ox1 x2 -plane and P2 . In this
case
2 3 2 3
1 0 0 cos 2a sin 2a 0
P1 4 0 1 0 5; P2 4 sin 2a cos 2a 0 5;
0 0 12 0 30 1
ð14:93Þ
cos 2a sin 2a 0
P1 P2 ¼ 4 sin 2a cos 2a 0 5:
0 0 1
The comparison of the last equality (14.33) with the relation (14.27) shows that the
successive application of reflections relative to the planes P1 and P2 , respectively,
results in a rotation about the Ox3 -axis, through the angle 2a, which represents
twice the angle formed by the too planes. Since Ox3 is the line of intersection of
the two planes of symmetry, the theorem is proved.
There result two important consequences:
(i) The action of a symmetry axis can be replaced by the combined action of two
symmetry planes, intersecting along it.
(ii) If a figure has a symmetry plane P1 with a symmetry axis Ln contained in it,
then this figure must also have a second plane of symmetry P2 , intersecting P1
along the axis Ln , at an angle determined by the order n of the axis.
Let us state now
Theorem 14.3 (Euler) If a geometrical figure has two concurrent axes of sym-
metry Ln and Lm , with elementary angles of rotation a and b, respectively, then it
must also have a resultant axis of symmetry Lp .
According to the consequence (i), we can replace the axis Ln by two symmetry
planes P1 and P2 , which intersect along Ln , the angle between them being a=2. We
choose the plane P1 such that it passes through Lm . Analogically, we replace the
axis Lm by two planes of symmetry P3 and P4 , intersecting along Lm at an angle
b=2, such that P3 passes through Ln . The combined action of the planes P1 and P3
reduces to the identical transformation. Thus, the combined action of the axes Ln
and Lm is equivalent to two successive reflections through the symmetry planes P2
and P4 . But, according to the Theorem 14.2, the final result of these two motions
corresponding to a rotation through a known angle about the symmetry axis Lr ,
which coincides with the line of intersection of the two planes. The theorem is thus
proved.
646 14 Anisotropic and Non-homogeneous Bodies
We remark that, without changing the order of the axes Ln and Lm or the angle
between them, their replacement with couples of symmetry planes can be realized
in two different ways, which shows that there also exists a second resultant
0
symmetry axis Ln . Hence, Lm and Ln have two resultant axes of symmetry,
arranged simmetrically relative to the plane formed by Lm and Ln . We leave to the
reader the task of verifying this. We state now in Theorem 14.4.
Theorem 14.4 The point at which a symmetry axis of an even order pierces a
symmetry plane which is normal to it is a centre of symmetry.
We suppose that the symmetry axis L2n coincides with the Ox3 -axis and the
symmetry plane P coincides with the Ox1 x2 -plane. From (14.85)–(14.87) it results
that
2 3 2 3
cos 180 =n sin 180 =n 0 1 0 0
L2n 4 sin 180 =n cos 180 =n 0 5; P 4 0 1 0 5 ð14:94Þ
0 0 1 0 0 1
The last relation (14.95), compared with (14.88), shows that the origin of the co-
ordinates, hence the point of intersection of the axis L2n with the plane P, is indeed
a centre of symmetry. We state hereafter
Theorem 14.5 A centre of symmetry C is equivalent to an infinite number of
twofold roto-reflection axes, i.e., C ¼ 1L12 .
We suppose that the centre of symmetry C coincides with the origin of the
co-ordinate axes. From (14.86), (14.89) it follows that
2 3
1 0 0
L12 ¼ C ¼ 4 0 1 0 5: ð14:96Þ
0 0 1
Hence, the Ox3 -axis is a roto-reflection axis of second order. However, since the
relations (14.96) have the same structure in any orthogonal Cartesian co-ordinate
system, it follows that any line which passes through the centre of symmetry is
also a twofold roto-reflection axis. We state now
Theorem 14.6 The existence of a roto-reflection axis Ln2n , where n is an odd
number, always implies the existence of a centre of symmetry situated on this axis.
We suppose that the roto-reflection axis L2pþ1
2ð2pþ1Þ coincides with the Ox3 -axis.
From (14.86) and (14.89) it follows that
14.1 Anisotropic Elastic Bodies 647
2 3
cos 180 =ð2p þ 1Þ sin 180 =ð2p þ 1Þ 0
6 7
L2pþ1
2ð2pþ1Þ 4 sin 180 =ð2p þ 1Þ cos 180 =ð2p þ 1Þ 0 5;
0 0 1
2 3 ð14:97Þ
2pþ1 1 0 0
6 7
L2pþ1
2ð2pþ1Þ ¼ 4 0 1 0 5:
0 0 1
The last relation shows that the origin of the co-ordinate axes is a centre of
symmetry. We may also state
Theorem 14.7 If a roto-reflection axis Ln2n , where n is an even number, is
combined with a twofold symmetry axis L2 , normal to L2n , then we get n symmetry
planes passing through Ln2n .
We suppose that the axis L2n ¼ L4p coincides with the Ox3 -axis and L2 coin-
cides with the Ox1 -axis; from (14.86), (14.87) and (14.89) it follows that
2 3 2 3
cos 90 =p sin 90 =p 0 1 0 0
L2p 4 sin 90 =p cos 90 =p 0 5; L2 4 0 1 0 5;
4p ð14:98Þ
0 0 1 0 0 1
The last relation, compared with (14.85), shows that there exists a symmetry plane
which passes through the Ox3 -axis and the intersection of which with the Ox1 x2 -
plane forms an angle of 45 =p with the Ox1 -axis. Determining successively the
components of the orthogonal tensors ðL2p 2 2p 4p
4p Þ ; . . .; ðL4p Þ and combining the
2
results thus obtained with those relative to L , the reader can himself verify the fact
that there exist still n 1 symmetry planes which pass through the Ox3 -axis.
(a) We assume that the entire symmetry of the figure is characterized by a single
symmetry axis Ln ; in this case, the following symmetries are possible in
crystals:
L1 ; L2 ; L3 ; L4 ; L6 : ð14:100Þ
(b) It is easy to see that, when the figure possesses a symmetry axis Ln , with
n [ 2, then, in order to avoid the occurrence of a new axis of order greater
than 2, only an L2 -axis that is normal to Ln can be added to Ln ; the addition of
an axis of symmetry of second order, that is not normal to Ln , would result in
the displacement of the Ln -axis in a new position, which is equivalent to the
existence of a new axis of symmetry of the same order n.
At the same time, we can see, on the basis of the Theorem 14.2 and on
Euler’s theorem concerning the composition of rotations, that the addition of
the L2 -axis to the Ln -axis, when L2 and Ln are mutually perpendicular, will
necessarily result in the occurrence of other n 1 axes of symmetry of
second order, perpendicular to Ln . This new type of symmetry will be denoted
by Ln nL2 . We shall therefore have the following symmetries:
On the basis of similar reasoning, we can conclude that for the crystal classes of
subset A, a symmetry plane can be added to Ln either perpendicularly or parallel to
Ln , hence passing through this axis.
(c) In the first case, according to the Theorem 14.4, the point of intersection of a
symmetry axis of an even order with the symmetry plane normal to it will
necessarily be a centre of symmetry. The new types of possible symmetries
are denoted by Ln P and Ln PC; their complete list looks as follows:
another symmetry plane, which also contains the axis Ln and forms an angle
of a=2ða ¼ 360 =nÞ with the first plane. Repeating this reasoning, we can
conclude that n 1 such additional symmetry planes, passing through Ln , can
occur, with angles a=2 between each other. The new types of symmetry are
denoted by Ln nP; their complete list is:
L1 L2 2P; 3L2 3PC; L3 3L2 4P; L4 4L2 5PC; L6 6L2 7PC: ð14:104Þ
(f) We suppose, first, that the entire symmetry of the figure is characterized by a
single roto-reflection axis Ln2n . According to the Theorem 14.6, if n is odd,
there necessarily occurs also a centre of symmetry C, situated on the axis Ln2n .
Thus, the complete list of possible symmetries is the following:
(g) Combining a roto-reflection axis Ln2n with a symmetry axis of order 2L2 ,
normal to Ln2n (see the reasoning of (b)), we necessarily obtain, according to
the Theorem 14.7, when n is even, n additional symmetry planes which pass
through Ln2n . Taking into account the Theorem 14.6 as well, we obtain the
following complete list of the symmetries possible in this case:
the relations (14.107), the third vertex can possibly correspond to a symmetry axis
of second order, as a possible resultant axis. We give the list of all combinations
which are not excluded by the above considerations:
From (14.107) and (14.108) it follows that only the combinations ðL4 ; L3 ; L2 Þ
and ðL3 ; L3 ; L2 Þ are admissible. Knowing the angles of the two types of spherical
triangles corresponding to these combinations and using the formulae of spherical
trigonometry, we can compute the angles corresponding to the sides of the tri-
angles; we obtain:
If the surface of the sphere R can be covered continuously with such triangles,
then their number and mutual arrangement determine the possible symmetries in
the case B. Using also Gauss’s formula
S ¼ p þ A; ð14:110Þ
which relates the sum S of the angles of a spherical triangle to the area A, it can be
shown that this covering is realizable and, in the first case, only the combination
3L4 4L3 6L2 can exist, whereas the second case permits only the combination
3L4 4L3 , the mutual position of the axes of symmetry being determined by the
relations (14.109).
Applying repeatedly the consequence (ii) of the Theorem 14.2, we conclude
that the addition of a plane of symmetry to the combination 3L4 4L3 6L2 , such that
no new axes of symmetry occur, will result in the occurrence of 8 additional planes
of symmetry, the positions of which can be determined from the fact that the plane
652 14 Anisotropic and Non-homogeneous Bodies
added initially must be normal to one of the L4 -axes. At the same time, the
Theorem 14.5 shows that the common point of the 13 symmetry axes is, in this
case, a centre of symmetry; therefore, the resultant combination has the structure
3L4 4L3 6L2 9PC.
Since the occurrence of the axes of symmetry is not permitted, the addition of a
plane of symmetry cannot be done arbitrarily in the second case, either. We leave
to the reader to verify that a plane of symmetry can be added to the combinations
3L2 4L3 only in two ways:
(i) the plane passes through the axes L3 and L4 ;
(ii) the plane passes through the L2 -axis, symmetrically relative to the L3 -axis.
In (i) results a symmetry of the type 3L2 4L3 6P, whereas in (ii) the symmetry of
the type 3L2 4L3 3PC. Hence, in the subset B there exist only five crystal classes
3L4 4L3 6L2 ; 3L4 4L3 6L2 9PC; 3L2 4L3 ; 3L2 4L3 6P; 3L2 4L3 3PC: ð14:111Þ
In the first and the fifth classes, the threefold symmetry axes are also sixfold
roto-reflection axes, and in the fourth class, the twofold symmetry axes become
roto-reflection axes of the fourth order.
Consequently, the totality of crystal classes can contain only 32 distinct ele-
ments; the Table 14.1 together with (14.111), exhaust all possible types of geo-
metrical symmetries for the external form of real crystals.
We give first, in what follows, some general results concerning the static problem.
The dynamics case will be emphasized too.
Poisson’s ratio m has, in general, small variations, one may take m ¼ const,
remaining with only one elastic constant as function of point (see [34, 35]).
We remember that
1 i ¼ 1 Fi ; i; j ¼ 1; 2; 3;
ij ¼
r rij ; F ð14:112Þ
E E
are the reduced stresses and the reduced volume forces, respectively. Hooke’s law
becomes
eij ¼ ð1 þ mÞ
rij m
rkk dij ; i; j ¼ 1; 2; 3; ð14:113Þ
E ¼ E0 e f ; ð14:114Þ
ij;i þ f;i r
r j ¼ 0; j ¼ 1; 2; 3;
ij þ F ð14:116Þ
In a vector form, these equations become (in the absence of the volume forces and
taking into account (14.115))
1
ðD þ a rÞu þ ½rðr uÞ þ 2maðr uÞ þ rða uÞ ¼ 0: ð14:1170 Þ
1 2m
Using an analogue C ¼ CðC1 ; C2 ; C3 Þ, of Galerkin’s vector, which verifies the
equation
n m h 2 io
ðD þ a rÞ ðD þ a rÞ2 a D ða rÞ2 C ¼ 0; ð14:118Þ
1m
654 14 Anisotropic and Non-homogeneous Bodies
By means of Boggio’s theorem (see Sect. A.1.2.7), one can express the
potential vector C in the form
C ¼ C1 þ C2 ; ð14:120Þ
ij;i þ f;i r
r j ¼ q
ij þ F €uj ; j ¼ 1; 2; 3; ð14:121Þ
where
1
¼
q q ð14:1160 Þ
E
is the reduced unit mass (the reduced density). The equations of Lamé in the
dynamic case take the form (in the absence of the volume forces)
1
uj;ii þ ðui;ij þ 2mf;j ui;i Þ þ f;i ðui;j þ uj;i Þ
1 2m
þ 2ð1 þ mÞðF j quvj Þ ¼ 0: ð14:1170 Þ
Also in this case, one can give a representation of Somigliana-Iacovache type too.
n n
pi ¼ r
ij nj ; i ¼ 1; 2; 3: ð14:123Þ
We express the stresses in the form
X
n
ðkÞ
r 0ij þ
ij ¼ r r
ij : ð14:124Þ
k¼1
Neglecting the terms with variable coefficients (if a r 6¼ const), the equations in
stresses take a form analogue to that which corresponds to homogeneous bodies,
one obtains an approximation of zero order, solving the problem for an homo-
geneous body acted upon by given volume forces and with the boundary condi-
tions (14.123).
To obtain the approximation of k-th order, assuming that one knows the
approximation of (k 1)-th order, which leads to the conventional volume forces
jðkÞ ¼ f;i r
F
ðk1Þ
ij ; j ¼ 1; 2; 3; ð14:125Þ
one solves the problem for a homogeneous body, acted upon by the volume forces
jðkÞ , with zero loads on the frontier.
F
The solution in stresses of the static elastic problem of a non-homogeneous
body may be thus reduced to the solution in stresses of a succession of elastic
problems for the same body, considered as homogeneous, acted upon by con-
ventional volume forces and (except the approximation of zero-th order) with
vanishing boundary conditions.
The convergence of the iterative process may be practically appreciated by a
mechanical interpretation (the conventional volume forces F iðkÞ must give a state of
stress negligible with respect to the state of stress given by F iðk1Þ ).
In the case of the first fundamental problem (in displacements), one may use an
analogue procedure, using the equations of Lamé type (14.117).
The corresponding dynamic problem can be treated similarly. Obviously, one
must also put initial conditions for the time, i.e.:
n n
t t0 :
pi ¼ r
ji nj ; i ¼ 1; 2; 3; ð14:126Þ
or
n
t t0 : ui ¼ u i ; i ¼ 1; 2; 3; ð14:1260 Þ
as well as
One assumes that in the case of a plane state of strain the longitudinal modulus of
elasticity does not depend on x3 , i.e., E ¼ Eðx1 ; x2 Þ. In the case of a plane state of
stress one uses an analogous hypothesis, but with a mean elasticity moduli on the
thickness h of a plate, i.e.,
Z Z
1 h=2 1 h=2
Eðx1 ; x2 Þ ¼ Eðx3 Þdx3 ; Gðx1 ; x2 Þ ¼ Gðx3 Þdx3 ; ð14:128Þ
h h=2 h h=2
22 Þ ¼ ð1 þ mÞða21 r
r11 þ r
Dð 11 þ a22 r
22 þ 2a1 a2 r
12 Þ: ð14:1320 Þ
Hence, the partial differential equations of the problem are (14.1300 ) and (14.1320 ).
The Eqs. (14.1300 ) lead to
so that
11 ¼ ðo2 þ a2 Þ2 F; r
r 22 ¼ ðo1 þ a1 Þ2 F;
ð14:133Þ
12 ¼ ðo1 þ a1 Þðo2 þ a2 ÞF;
r
which must be satisfied by the stress function F; after elementary calculations, one
obtains
h i
D ðo1 þ a1 Þ2 þ ðo2 þ a2 Þ2 F ð1 þ mÞða1 o2 a2 o1 Þ2 F ¼ 0 ð14:1340 Þ
or
ðD þ a1 o1 þ a2 o2 Þ2 F mða1 o2 a2 o1 Þ2 F ¼ 0; ð14:13400 Þ
wherefrom
pffiffiffi pffiffiffi
pffiffiffi
D þ ða1 a2 mÞo1 þ ða2 þ a1 mÞo2 D þ ða1 þ a2 mÞo1
pffiffiffi
þða2 a1 mÞo2 F ¼ 0; ð14:134000 Þ
658 14 Anisotropic and Non-homogeneous Bodies
By a change of function, one may obtain a total differential in the right member
n
p 1 ea1 x1 þa2 x2 ds ¼ d½o2 ðFea1 x1 þa2 x2 Þ;
n
p 2 ea1 x1 þa2 x2 ds ¼ d½o1 ðFea1 x1 þa2 x2 Þ;
wherefrom, by integration,
Z sn
p 2 ds ¼ V0 V;
ðEFÞ;1 ¼ ðEFÞ;1
0 0
Z sn ð14:137Þ
p 1 ds ¼ H0 þ H;
ðEFÞ;2 ¼ ðEFÞ;2
0 0
where s is the curvilinear co-ordinate along the frontier of the plane domain.
In the case of a simply connected domain, one can take the constants H0 and V0
equal to zero, because a stress function of the form
F ¼ ðH0 x2 þ V0 x1 þ constÞea1 x1 a2 x2 ð14:138Þ
leads to no any stress, taking into account (14.133). The quantities H and V (which
play the rôle of the functions u and w) correspond to the components along the co-
ordinate axes of the external load acting on a bar along the frontier of the con-
sidered domain, starting from an arbitrary point. One observes that
o dx1 dx2
ðEFÞ ¼ ðEFÞ;1 þ ðEFÞ;2 ;
os ds ds ð14:139Þ
o dx1 dx2
ðEFÞ ¼ ðEFÞ;1 þ ðEFÞ;2 :
on dn dn
14.2 Non-homogeneous Elastic Bodies 659
integrating by parts
Z s Z s
ðHdx2 Vdx1 Þ ¼ ðHx1 Þs ðVx1 Þs ðx2 dH x1 dVÞ
0 0
Z s Z s
n n
¼ p 1 (x2 js x2 Þds þ p 2 (x1 x1 js Þds;
0 0
h i
1
Du1 þ 1m ð1 þ mÞðu1;1 þ u2;2 Þ;1 þ 2a1 ðu1;1 þ mu2;2 Þ
h þa2 ðu1;2 þ u2;1 Þ ¼ 0; i ð14:141Þ
1
Du2 þ 1m ð1 þ mÞðu1;1 þ u2;2 Þ;2 þ 2a2 ðu2;2 þ mu1;1 Þ
þa1 ðu1;2 þ u2;1 Þ ¼ 0;
by a change of function
X ¼ eðc1 x1 þc2 x2 Þ=2 X; ð14:144Þ
one obtains the equation
1 ðc2 þ c2 ÞX
DX ¼ 0; ð14:1440 Þ
4 1 2
14.2 Non-homogeneous Elastic Bodies 661
1 ð1 þ mÞða2 þ a2 ÞX
DX ¼ 0; ð14:146Þ
1 2
4
which is obtained by two different changes of function.
If the modulus of longitudinal elasticity is not taken in the form (14.132), but
has a more general form
one may continue on the same way as before, obtaining a representation of the
stresses in a form analogous to the Airy one, i.e.,
or the equation
h i
D ðo1 þ f;1 Þ2 þ ðo2 þ f;2 Þ2 F ð1 þ mÞðf;1 o2 f;2 o1 Þ2 F
¼ ð1 þ mÞ f;12 ½o1 ðo2 þ f;2 Þ þ o2 ðo1 þ f;1 Þ
ð14:1510 Þ
f;11 o2 ðo2 þ f;2 Þ f;22 o1 ðo1 þ f;1 Þ F:
662 14 Anisotropic and Non-homogeneous Bodies
Let us search now some particular integrals of the Eqs. (14.143) and (14.146).
First of all, let us search a particular integral in form of a polynomial of nth
degree (except a polynomial of first degree). But only one polynomial is not
sufficient because, by applying it, one obtains a polynomial of ðn 1Þth degree
and a polynomial of ðn 2Þth degree and each of them must identically vanish.
Hence, the polynomial of nth degree must verify each of the equations
c1 v;1 þ c2 v;2 ¼ 0; Dv ¼ 0: ð14:152Þ
We observe that the first Eq. (14.152) has an integral of the form
v ¼ vðc2 x1 c1 x2 Þ;
v ¼ ðc2 x1 c1 x2 Þn ;
but it is not a harmonic one. Hence, one introduces a polynomial of ðn 1Þth
degree of the form
11 ¼ b2 C; r
r 22 ¼ a2 C; r
12 ¼ abC; ð14:163Þ
where
np
cn ¼ ; n ¼ 1; 2; 3. . . ð14:1640 Þ
a
and 2a is the period; the quantities an ; bn ; dn ; en ; kn ; ln ; vn ; ,n may be expressed as
functions of cn by means of the relations (14.160). Under the trigonometric lines,
one takes the sign þ or as one has c1 [ 0 or c1 \0, hence after the sign of the
pffiffiffi
expression a b m. One obtains another stress function by interchanging x1 and
x2 .
Let be the elastic half-plane x2 0 acted upon on the separation line x2 ¼ 0 by the
reduced periodic load
X X
pðx1 Þ ¼
an sin cn x1 þ bn cos cn x1 ; ð14:165Þ
n n
where
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 1 2 1
kn ¼ b þ 2cn þ b þ 4c4n þ 2ð1 þ 2mÞb2 c2n þ b4 ;
2
2 2 2 4
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð14:168Þ
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 1
an ¼ 2c2n þ b2 þ 4c4n þ 2ð1 þ 2mÞb2 c2n þ b4 :
2 2 4
The first sum above satisfies the second Eq. (14.1350 ) and the second sum
verifies the first of these equations.
666 14 Anisotropic and Non-homogeneous Bodies
which does no more verify each of the equations (14.1350 ), but the Eq. (14.134000 ).
One also remarks that the functions (14.1670 ) are valid only for elasticity
moduli of the form (14.166).
By means of the formulae (14.133), one obtains the reduced stresses
X
22 ¼
r c2n ekn x2 ðA
n sin an x2 þ B
n cos an x2 Þ sin cn x1
n
X
0
c2n ekn x2 A 0n cos an x2 cos cn x1 ;
sin an x2 þ B
n
n
X
0
12 ¼
r cn ekn x2 ðb kn Þ A n
0n cos an x2
sin an x2 þ B
n ð14:168Þ
0
cos an x2 B
þan A n
0n sin an x2 sin cn x1
X
cn ekn x2 ½ðb kn ÞðA n sin an x2 þ B
n cos an x2 Þ
n
n cos an x2 B
þan ðA n sin an x2 Þ cos cn x1 :
n ¼ ðb kn Þan ; A
A 0 ¼ ðb kn Þbn ;
n
an c2n an c2n
ð14:170Þ
n ¼ an ; B
B 0 ¼ bn :
c2n n c2n
The stress function (14.168) becomes
X1
kn x2 b kn
Fðx1 ; x2 Þ ¼ e sin an x2 cos an x2 ðan sin cn x1 þ bn cos cn x1 Þ
n
c2n an
ð14:171Þ
X1 h ib k
kn x2 2 2 n
r
11 ¼ e an þ ðb kn Þ sin an x2 þ cos an x2 ðan sin cn x1 þ bn cos cn x1 Þ;
n
c2n an
X
b kn
r
22 ¼ ekn x2 sin an x2 : cos an x2 ðan sin cn x1 þ bn cos cn x1 Þ;
n
an
ð14:172Þ
X 1 h i
12 ¼
r ekn x2 a2n þ ðb kn Þ2 sin an x2 ðan cos cn x1 bn sin cn x1 Þ:
n
an c n
ð14:1720 Þ
One observes that the normal stresses on the separation line are given by
X a2 þ ðb kn Þ2
n
11 ðx1 ; 0Þ ¼
r ðan sin cn x1 þ bn cos cn x1 Þ: ð14:173Þ
n
c2n
In the case of local normal loads acting upon the separation line, one may study
the problem on the same way, replacing the Fourier series by Fourier integrals. The
final results are thus similar to (14.172), (14.173), where the parameters kn ; an ; cn
become functions and the sums integrals.
References
A. Books
1. Beju, I., Soós, E., Teodorescu, P.P.: Euclidean Tensor Calculus with Applications. Abacus
Press, Tunbridge Wells (1983). (Ed. Tehnică, Bucuresti)
2. Brilla, J.: Anizotropické steny (Anisotropic Plates). Slov. Akad. Vied., Bratislava (1958)
3. Cristescu, N.D., Crăciun, E.M., Soós, E.: Mechanics of Elastic Composites. Chapman and
Hall/CRC, Boca Raton (2004)
4. Hearmon, R.F.S.: An Introduction to Applied Anisotropic Elasticity. Clarendon Press, Oxford
(1961)
5. Lekhnitskiǚ, S.G.: Teoriya uprugosti anizotropnogo tela (Theory of Elasticity of Anisotropis
Bodies). Ogiz, Moskva-Leningrad (1947)
6. Love, A.-E.-H.: A Treatise on the Mathematical Theory of Elasticity, 4th edn. Cambridge
University Press, Cambridge (1934)
B. Papers
anisotropic bodies. VI. Bodies with an axis of elastic symmetry of fourth order). Bul. Acad.
Rom. VII, 753 (1957)
30. Teodorescu, P. P.: Asupra problemei plane a elasticitătii unor corpuri anizotrope. VII.
Corpuri actionate de forte masice oarecare (On the plane problem of elasticity of some
anisotropic bodies. VII. Bodies acted upon by arbitrary volume forces). Com. Acad. Rom.
VIII, 887 (1958)
31. Teodorescu, P.P.: Asupra problemei plane a elasticitătii unor corpuri anizotrope. VIII.
Influenta variatiei de temperatură (On the plane problem of elasticity of some anisotropic
bodies. VIII. Influence of the temperature variation). Com. Acad. Rom. VIII, 1119 (1958)
32. Teodorescu, P. P.: Asupra problemei plane a elasticitătii unor corpuri anizotrope. IX. Cazul
micilor miscări elastice (On the plane problem of elasticity of some anisotropic bodies. IX.
Case of small elastic motions). Com. Acad. Rom. VIII, 1243 (1958)
33. Teodorescu, P. P.: Ůber das Kinetische Problem nichthomogener elasticher Körper. Bull.
Acad. Pol. Sci., sér. Sci. Techn. XII, 595 (1964)
34. Teodorescu, P. P., Predeleanu, M.: Quelques considérations sur le problème des corps
élastiques hétérogènes (Non-homogeneity in Elasticity and Plasticity). In: IUTAM-
Symposium, pp. 31 (1958)
35. Teodorescu, P. P., Predeleanu, M.: Quelques considérations sur le problème des corps
élastiques hétérogènes. Bull. Acad. Pol. Sci., sér. Sci. Techn. VII, 81 (1959)
36. Teodorescu, P. P., Predeleanu, M.: Ůber das ebene Problem nichthomogeneuer elastischer
Körper. Acta Techn. Acad. Sci. Hung. XXVII, 349 (1959)
Chapter 15
Introduction to Thermoelasticity
Thermoelasticity contains the theory of heat conduction and the theory of strains
and stresses due to the flow of heat, when coupling of temperature and strain fields
occurs. An important case is that in which the coupling of temperature and strain is
neglected. In what follows some quasi-static and dynamic problems are dealt with.
Most of the monographs and treatises on the theory of elasticity contain smaller
of larger chapters on thermoelasticity. We mention some monographs dedicated to
heat conduction, i.e. those of H. C. Carlslaw [1], H. C. Carlslaw and J. C. Jaeger
[2] and Hamburger [3]; as to monographs dedicated to thermoelasticity, we
mention those of B. A. Boley and J. H. Weiner [4], E. Melan and H. Parkns [5],
W. Nowacki [6], H. Parkus [7] and O. Tedone [8].
First of all, using also ideas from [6], we deal with some problems concerning heat
conduction and the relations to the equations of equilibrium and motion of elastic
bodies. Stationary and dynamic quasi-static problems will be then considered, as
well as dynamic ones. Some general principles and theorems will be presented too.
if inside of the considered domain there are heat sources too, the quantity of heat is
completed by the quantity
ZZZ
DQ00 ¼ WdVDt;
V1
where W is the quantity of heat generated per unit volume and per unit time. The
sum DQ ¼ DQ0 þ DQ00 may be determined on the basis of the temperature dif-
ferences in the domain D1 in the time interval Dt; thus, we obtain
ZZZ
DQ ¼ _
cqTdVDT;
V1
where q is the unit mass and c is the specific heat, i.e. the quantity of heat required
for a unit increase of temperature of unit mass. It results the equation
ZZZ ZZ
oT
cqT_ W dV k dS ¼ 0:
V1 S1 on
The temperature field is completely determined by the Eq. (15.3), to which one
must add appropriate boundary and initial conditions.
The boundary conditions are put for t [ t0 on the surface S and correspond to
the interaction between the body and the surrounding medium. They may be of
three kinds:
(i) The temperature is prescribed at all points of the surface S:
(ii) The temperature gradient is prescribed at all points of the surface S:
(iii) The function oT=on þ aT ¼ b; a; b ¼ const, is prescribed at all points of the
surface S:
The boundary condition (ii) occurs if the intensity of the heat flow from outside
into inside of the body is known. If there is no heat exchange with the surrounding
medium across the surface S (the body is thermally insulated), then oT=on ¼ 0 on
S: The condition (iii) corresponds to a free heat exchange on the surface S:
The initial condition determines the distribution of temperature T ¼ TðrÞ at the
initial moment t ¼ t0 :
If a part of the frontier is thrown to infinite, the boundary conditions are
replaced by the requirement that the temperature is bounded or takes a given value
on this part. If the conduction of heat is stationary, then the initial condition is
irrelevant; one puts only boundary conditions for the Eq. (15.4).
In what follows, we admit that the heat equation is solved, so that the tem-
perature Tðr; tÞ is known.
674 15 Introduction to Thermoelasticity
which represents the Duhamel-Neumann law. Making i ¼ j and summing, one has
1 2m
h 3at ¼ H; ð15:7Þ
E
the law mentioned above takes also the form
rij ¼ ðkh bTÞdij þ 2leij ; ð15:8Þ
First of all, we deal with problems in which the temperature has not a decisive
importance in the variation of strains and stresses, i.e. with stationary and quasi-
static problems; we will consider thus both displacement and stress equations.
15.1 Basic Relations and Equations 675
Eliminating the strains and the stresses, as in the classical case, the equation of
Lamé type which must be verified by the displacement vector reads
lDu þ ðk þ lÞgrad divu þ F ¼ bgrad T ð15:11Þ
or
1 1
Du þ grad divu þ ðF bgrad TÞ ¼ 0; ð15:110 Þ
1 2m l
We observe that the volume force F is replaced by F bgrad T; obviously, in
the absence of volume forces one remains only with the quantity bgradT. As
usual, the boundary conditions are put in displacements (the first fundamental
problem).
J. N. Goodier has introduced the thermoelastic displacement potential Uðr; tÞ by
the relation
u ¼ grad U: ð15:12Þ
where
1 1
Ui ðr; nÞ ¼ ; i ¼ 1; 2; 3: ð15:160 Þ
4pðk þ 2lÞ R ;i
This function can be regarded as the displacement of the point r in the direction of
the Oxi -axis, due to the action of a centre of pressure situated at the point n of the
elastic space, i.e. the dilatation at the point n; due to the action of a concentrated
unit force situated at the point r and having the direction of the Oxi -axis.
Denoting
Hi ¼ ð3k þ 2lÞUi ; i ¼ 1; 2; 3;
we get
ZZZ
ui ðrÞ ¼ a TðnÞHi ðn; rÞdVðnÞ; i ¼ 1; 2; 3: ð15:17Þ
V
The formulae (15.15) and (15.17) are obtained for the stationary case. If one
replaces the function TðnÞ by the function Tðn; tÞ; it follows that the results remain
valid for quasi-static problems.
Taking into account the homogeneous equation of that conduction (corre-
sponding to (15.3))
1
\T ¼ DT T_ ¼ 0; ð15:300 Þ
,
_ it results
differentiating (15.13) with respect to time and eliminating T,
DU_ ¼ c,DT;
hence
Z t
U ¼ c, Tdt þ U0 þ U1 t; ð15:18Þ
0
where
lD
ui þ ðk þ lÞD
u;i ¼ 0; i ¼ 1; 2; 3: ð15:190 Þ
The vector of components
ui may be expressed in the form
1
ui ¼ ðk þ 2lÞDdij ðk þ lÞoij uj ; i ¼ 1; 2; 3; ð15:21Þ
l
where the displacement functions uj ; j ¼ 1; 2; 3, are biharmonic
DDuj ; j ¼ 1; 2; 3: ð15:210 Þ
one obtains thus the Beltrami type equations (in the absence of the volume forces).
The sum of the normal stresses is given by
2aE
DH þ DT ¼ 0: ð15:24Þ
1m
In the particular case of a stationary temperature field T_ ¼ 0 and no heat
sources acting inside the body ðQ ¼ 0Þ, the temperature function becomes a
harmonic one ðDT ¼ 0Þ; the sum of the normal stresses is a harmonic function too
ðDH ¼ 0Þ: Hence,
1
Drij þ ðH þ aET Þ;ij ¼ 0: ð15:25Þ
1þm
Starting from these equations, one has obtained various solutions for the elastic
space, for the elastic half-space, for axisymmetric problems a.s.o.
678 15 Introduction to Thermoelasticity
Eliminating the stresses and the strains between the equations of thermoelasticity
in the dynamic case, one obtains the equations of Lamé type in the form (see §4.3)
lh2 u þ ðk þ lÞgrad divu ¼ bgradT ð15:26Þ
or in the form
1 b
h2 u þ grad divu ¼ gradT; ð15:260 Þ
ð1 2mÞ l
in the absence of the volume forces.
Another form of these equations is
1 b
h1 u þ curlcurlu ¼ grad T: ð15:27Þ
2ð1 mÞ k þ 2l
By means of a scalar potential U ¼ Uðr; tÞ and of a vector potential W ¼
Wðr; tÞ; the displacement uðr; tÞ can be represented in the form
u ¼ gradU þ curlW; ð15:28Þ
ZZZ
c Tðn; t R=c1 Þ
Uðr; tÞ ¼ dVðnÞ; R c1 t; ð15:31Þ
4p V Rðr; nÞ
where Rðr; nÞ is the distance between the points r and n: The integration is carried
over a volume of the body situated inside the sphere of radius R ¼ c1 t and of
centre at the point r: The formula corresponds for a body initially at rest, in the
natural state.
The displacement u can be obtained also as a sum u ¼ u þu ; where u
satisfies
the non-homogeneous system (15.26). while u satisfies the homogeneous system
(of Lamé type) and is given by
1
¼ h2 u þ
u ðDu graddivuÞ; ð15:32Þ
1 2m
where the vector potential u satisfies the double wave equation
h1 h2 u ¼ 0; ð15:320 Þ
We deal hereafter with some general considerations concerning the elastic bodies
subjected to thermal actions too, i.e. d’Alembert’s and Hamilton’s principles,
reciprocal theorems and coupling problems.
where
1 Ea
W ¼ rij eij ekk T: ð15:410 Þ
2 1 2m
Hence, the principle takes the form
ZZZ
dWe ui dui dV ¼ dWi ;
q€ ð15:42Þ
V
if we assume that one has synchronous virtual displacements (the virtual dis-
placements vanish at t0 and t1 ), then
Z t1 Z t1 ZZZ
dTdt ¼ dt q€ui dui dV:
t0 t0 V
It follows that
Z t1 Z t1
d ðWi TÞdt ¼ dWe dt:
t0 t0
682 15 Introduction to Thermoelasticity
Assuming that the external loads do not vary, we introduce the potential energy
P ¼ Wi 2We ; ð15:45Þ
where We is the external work (6.12), corresponding to the static case; we obtain
thus Hamilton’s principle in the form
Z t1
d ðP TÞdt ¼ 0; ð15:46Þ
t0
where the integral is called action. Hamilton’s principle states that the action is
steady, while the motion of the elastic body corresponds to the extremals of this
functional in the synchronous case.
If the external load is static and the temperature field is steady, the relation
(15.46) yields the principle of minimum of the potential energy.
dP ¼ 0; ð15:47Þ
Let us consider two states of strain and stress, due to two cases of loading, marked
by (‘) and (’’), respectively; we may state the principle (theorem) of Betti in the
form (6.36).
n
Replacing the volume forces Fi00 by bT;i and the superficial loads p 00i by bTni
and neglecting the volume forces Fi ; one may express this principle as follows
ZZZ ZZ ZZZ ZZ
n
Fi0 u00i dV þ p 0i u00i dS ¼ b T;i u0i dV þ b Tni u0i dS: ð15:48Þ
V S V S
Let be the state of strain and state denoted by the dash, due to a concentrated
unit force acting at the point r in the direction of the Oxi -axis; the relation (15.19)
leads to
15.1 Basic Relations and Equations 683
ZZZ
ui ðrÞ ¼ a TðnÞHðiÞ ðn; rÞdVðnÞ
V
ZZZ
¼b TðnÞhðiÞ ðn; rÞdVðnÞ; i ¼ 1; 2; 3: ð15:50Þ
V
One obtains thus Maysel’s formulae, which specify the displacements in a body,
due to the action of a steady temperature field. Obviously, these formulae may be
used also in the case of quasi-static problems, in which the time t is a parameter,
ZZZ
ui ðr; tÞ ¼ a Tðn; tÞHðiÞ ðn; rÞdVðnÞ; i ¼ 1; 2; 3: ð15:500 Þ
V
If we construct now the equation of virtual work, we may write the relation of
reciprocity in the form
ZZZ ZZ ZZZ
n
Fi0 du00i dV þ p 0i du00i dS þ a T 0 dH00 dV
V S V
ZZZ ZZ ZZZ
00 0 n 00 0
¼ dFi ui dV þ d p i ui dS þ a dT 00 H0 dV: ð15:51Þ
V S V
The above results are valid for stationary and quasi-static thermoelastic prob-
lems. If we wish to include also dynamic effects, then we must introduce inertia
terms in the relation (15.51); we obtain thus
ZZZ ZZ ZZZ
n
ðFi0 q€u0i Þdu00i dV þ p 0i du00i dS þ a T 0 dH00 dV
V S V
ZZZ ZZ ZZZ
00 00 0 n 00 0
¼ ðdFi qd€ ui Þui dV þ d p i ui dS þ a dT 00 H0 dV: ð15:52Þ
V S V
If the thermal stresses are produced only by the temperature field T 0 ; i.e. if Fi0 ¼ 0;
n
p 0i ¼ 0; i ¼ 1; 2; 3; and dT 00 ¼ 0; then the formula (15.52) takes the form
ZZZ ZZ
00 00 0 n
ðdFi qd€ ui Þui dV þ d p 00i u0i dS
V
ZZZ ZZZ S
0 00
þ q€ui dui dV ¼ a T 0 dH00 dV: ð15:53Þ
V V
The coupling of the strain and temperature fields represents a more correct
approach of the problem, which becomes thus more intricate too. The equation of
heat propagation (15.3), (15.30 ) is completed as follows
1 1
\T ¼ DT T_ ¼ Q þ gdivu;
_ ð15:54Þ
, ,
where one takes use of the notation (15.30 ), as well as of
bT0
g¼ ; ð15:540 Þ
h
where h is the heat conductivity. Obviously, the coupling is due to this latter term.
We remember now Lamé’s type equation
lh2 u þ ðk þ lÞgraddivu þ F ¼ bgradT ð15:55Þ
1
h2 W ¼ w: ð15:5700 Þ
c22
15.2 Applications
Hereafter we will present some elementary examples and then some applications
concerning problems with axial symmetry using results due to S. Timoshenko and
J. N. Goodier [9]; as well, we deal with some plane problems of thermoelasticity,
using our results [10].
First of all, we deal with some simple problems concerning thin plates and then we
deal with a sphere of large radius.
Let us consider a thin rectangular plate referred to the Ox1 and Ox2 axes, situated
in the middle plane; the frontier is specified by the sides xi ¼ ai ; i ¼ 1; 2: We
suppose firstly that T ¼ Tðx2 Þ is an even function. The longitudinal thermal
expansion aT is suppressed by applying the longitudinal stress r011 ¼ aET; lat-
erally we assume that the expansion is free, but at the ends of the plate there must
be forces of compression which must suppress any expansion in the direction of
the Ox1 -axis. If the plate is free from external forces, then we must superpose
distributed at the ends x1 ¼ a1 of resultant
Z a2
aETdx2 ; ð15:59Þ
a2
at a certain distance from the ends will appear approximate uniformly distributed
stresses (principle of Saint-Venant), of magnitude
Z a2
1
aETdx2 :
2a2 a2
The thermal stresses in the plate with free ends, at a considerable distance from the
ends, becomes
Z a2
1
r11 ¼ aETdx2 aET: ð15:60Þ
2a2 a2
686 15 Introduction to Thermoelasticity
then we obtain
2 x2
r11 ¼ aET0 aET0 1 22 ; ð15:610 Þ
3 a2
hence, to the stresses r011 one must add the bending stresses r0011 ¼ rx2 =a2 ; which
have to be determined by equating to zero the total moment over the cross section.
It results
Z a2 Z a2
1 2
rx2 dx2 aETx2 dx2 ¼ 0:
a2 a2 a2
so that
Z a2 Z a2
1 3
r11 ¼ aET þ aETdx2 þ 3 x2 aETx2 dx2 : ð15:62Þ
2a2 a2 2a 2 a2
If the plate is not thin in the Ox3 -direction, i.e. if the plate is large in this
direction, we get a plate with the Ox1 x3 -plane as middle plane; we assume, as
before, that the temperature T ¼ Tðx2 Þ depends, only on one variable. Putting
e11 ¼ e33 ¼ aT; r22 ¼ 0; it results
aET
r11 ¼ r33 ¼ : ð15:63Þ
1m
By a similar reasoning, one obtains
Z a2
aET 1
r11 ¼ r33 ¼ þ aETdx2
1 m 2ð1 mÞa2 a2
Z a2
3x2
þ aETx2 dx2 ; ð15:64Þ
2ð1 mÞa32 a2
15.2 Applications 687
a result analogous to the previous one. One may thus easily calculate the state of
stress in the plate if the distribution of the temperature T over the thickness of the
plate is known.
If the faces x2 ¼ a2 are maintained at two different temperatures Tþ and T ;
then, after a certain time, the temperature is given by
1 1 x2
T ¼ ðTþ þ T Þ þ ðTþ T Þ : ð15:65Þ
2 2 a2
If the plate is not restrained, then the thermal stresses vanish, but if the edges
are perfectly restrained against expansion and rotation, then the stress thus induced
is given by the relations (15.63). For instance, if T2 ¼ T1 ; we get
x2
T ¼ T1 ; ð15:650 Þ
a2
so that
aET1 x2
r11 ¼ r33 ¼ ; ð15:66Þ
1 m a2
the maximum stress is
aET1
r11max ¼ r33max ¼ : ð15:660 Þ
1m
Let be now the case of a sphere of large radius; we assume that in a small spherical
element of radius q; situated at the centre of the sphere, there occurs a temperature
rise T: The expansion of this element being not possible, a pressure p appears on
its surface. Assuming that the radius of the sphere is much greater than q; we
obtain for r [ q the normal stress rrr and the stress rt (in the plane normal to r; the
centre of the sphere being taken as origin)
q3 p
q3
rrr ¼ p ; rt ¼ ; ð15:67Þ
r 2 r
for r ¼ q; we get
p
rrr ¼ p; rt ¼ : ð15:670 Þ
2
The increase of the radius r ¼ q; due to the pressure p; is given by
q pq pq
Dr ¼ ðqet Þr¼q ¼ ½rt mðrrr þ rt Þr¼q ¼ ð1 þ mÞ ¼ :
E 2E 4l
688 15 Introduction to Thermoelasticity
It must equate the increase of the radius of the heated spherical element produced
by the rise of temperature and pressure; it follows that
pq pq
aqT ð1 2mÞ ¼ ð1 þ mÞ ;
E 2E
wherefrom
2 aET
p¼ : ð15:68Þ
31 m
Introducing in the relation (15.67), we get the formulae
2 aET
q3 1 aET
q3
rrr ¼ ; rt ¼ : ð15:69Þ
31 m r 31m r
Hereafter we deal with two problems with axial symmetry, i.e.: the case of a
circular disk and the case of a circular cylinder.
Let be a thin circular disk for which the temperature T ¼ TðrÞ does not vary over
its thickness; we assume that the stresses and the displacements have the same
property. The temperature, as well as the displacements and the stresses are
symmetric about the centre and do not depend on h:
The Duhamel-Neumann law takes the form
1 1
err aT ¼ ðrrr mrhh Þ; ehh aT ¼ ðrhh mrrr Þ: ð15:70Þ
E E
and corresponds to a plane state of stress.
Solving with respect to the stresses, one obtains,
E
rrr ¼ ½err þ mehh ð1 þ mÞaT ;
1 m2 ð15:700 Þ
E
rhh ¼ ½ehh þ merr ð1 þ mÞaT :
1 m2
The equation of equilibrium
drrr 1
þ ðrrr rhh Þ ¼ 0 ð15:71Þ
dr r
15.2 Applications 689
d2 ur 1 dur ur dT
2
þ 2 ¼ ð1 þ mÞa ;
dr r dr r dr
wherefrom
d 1 dðrur Þ dT
¼ ð1 þ mÞa : ð15:73Þ
dr r dr dr
By integration, we get
Z r
1 C2
ur ¼ ð1 þ mÞa TðrÞrdr þ C1 r þ ; ð15:74Þ
r r0 r
so that
Z r1 Z
1 1 r
rrr ¼ aE 2 TðrÞrdr 2 TðrÞrdr ;
r 0 r 0
1 Z r1 Z
ð15:76Þ
1 1 r
rhh ¼ aE 2 TðrÞrdr þ 2 TðrÞrdr TðrÞ :
r1 0 r 0
At the centre, one has
Z r
1 1
lim TðrÞrdr ¼ T0 ; ð15:77Þ
r!0 r 2 0 2
where T0 is the temperature at the centre.
Let be a circular cylinder for which the Oz-axis is a geometrical axis of symmetry,
as well as for the temperature TðrÞ: We suppose that all the geometrical quantities
do not depend on z: We deal with a long cylinder (theoretically infinite, practically
finite) and suppose that the displacement uz vanishes.
The angular strains and the tangential stresses vanish and the Duhamel-Neumann
law takes the form
1
err aT ¼ ½rrr mðrhh þ rzz Þ;
E
1
ehh aT ¼ ½rrr mðrzz þ rrr Þ; ð15:78Þ
E
1
ezz aT ¼ ½rzz mðrrr þ rhh Þ
E
in cylindrical co-ordinates. Since ur ¼ 0; so that ezz ¼ 0; the third Eq. (15.78)
leads to
rzz ¼ mðrrr þ rhh Þ aET; ð15:79Þ
wherefrom
1 m2
m
err ð1 þ mÞaT ¼ rrr rhh ;
E 1m ð15:790 Þ
1 m2
m
ehh ð1 þ mÞaT ¼ rhh rrr :
E 1m
These equations correspond for the case of a plane state of strain and may be
obtained from the equations (15.70) (a plane state of stress) if one replaces E by
E=ð1 m2 Þ; m by m=ð1 mÞ and a by ð1 þ mÞa: Therefore, we may use the formulae
(15.74) and (15.75) obtained in the preceding subsection; it results
15.2 Applications 691
Z r
1 þ ma C2
ur ¼ TðrÞrdr þ C1 r þ
; ð15:80Þ
1mr r0 r
Z
aE 1 r E C1 C2
rrr ¼ TðrÞrdr þ ;
1 m r 2 r0 1 þ m 1 2m r 2
Z ð15:81Þ
aE 1 r aET E C1 C2
rhh ¼ TðrÞrdr þ þ
1 m r 2 r0 1 m 1 þ m 1 2m r 2
and the resultant of the uniform axial stress C3 is C3 pr12 ; the total axial force
must vanish, so that
Z
2 2paE r1 2mEC1
C3 2pr1 TðrÞrdr þ pr 2 :
1m 0 ð1 þ mÞð1 2mÞ 1
Finally, the state of stress reads
Z Z
aE 1 r1 1 r
rrr ¼ TðrÞrdr 2 TðrÞrdr ;
1 m r12 0 r 0
Z Z
aE 1 r1 1 r
rhh ¼ TðrÞrdr þ TðrÞrdr ; ð15:82Þ
1 m r12 0 r2 0
Z
aE 2m r1
rzz ¼ TðrÞrdr T
1 m r12 0
Z Z
ð1 þ mÞa ð1 2mÞr r1 1 r
ur ¼ TðrÞrdr þ TðrÞrdr : ð15:820 Þ
1m r12 0 r2 0
Let us assume now that the initial temperature is T0 and that, beginning from
t ¼ 0; the boundary is maintained at the temperature T1 : The distribution of
temperature is given by
X1
r pn t
T ¼ ðT0 T1 Þ An J0 bn e ; ð15:83Þ
n¼1
r1
where J0 ðbn r=r1 Þ is the Bessel function of zeroth order and where bn are the roots
of the equation J0 ðbÞ ¼ 0; as well, one has
2 k b2n
An ¼ ; pn ¼ ; ð15:830 Þ
bn J1 ðbn Þ cq r12
as well as
2aEðT0 T1 Þ X1
1 2 J0 ðbn r=r1 Þ pn t
rzz ¼ e : ð15:840 Þ
1m b bn
n¼1 n
J1 ðbn Þ
one obtains
15.2 Applications 693
Z r1
EC1 aE
¼ 2 TðrÞrdr;
ð1 þ mÞð1 2mÞ ð1 mÞ r1 r02 r0
Z r1
EC2 aEr02
¼ 2 TðrÞrdr:
1 þ m ð1 mÞ r1 r02 r0
In the case of a steady heat flow, we assume that T0 is the temperature on the
inner surface; if the temperature on the outer surface is T1 ¼ 0; then the temper-
ature is given by
T0 r1
TðrÞ ¼ ln : ð15:86Þ
lnðr1 =r0 Þ r
In this particular case, the state of stress becomes
aET0 r1 r2 r2 r1
rrr ¼ ln þ 2 0 2 1 12 ln ;
2ð1 mÞ lnðr1 =r0 Þ r r1 r0 r r0
ð15:87Þ
aET0 r1 r2 r2 r1
rhh ¼ 1 ln 2 0 2 1 þ 12 ln ;
2ð1 mÞ lnðr1 =r0 Þ r r1 r0 r r0
aET0 r1 2r 2 r1
rzz ¼ 1 2 ln 2 0 2 ln : ð15:870 Þ
2ð1 mÞ lnðr1 =r0 Þ r r1 r0 r0
The problems considered in the previous section are plane problems. Hereafter
we deal with a more general case in Cartesian co-ordinates; we will thus give
formulations in stresses and displacements.
694 15 Introduction to Thermoelasticity
1
e12 ¼ r12 ð15:910 Þ
2G12
to which we add
m31 m32
e33 a3 T ¼ r11 r22 : ð15:9100 Þ
E1 E2
Between the elastic constants takes place the first relation (15.89), so that one has
only 6 independent elastic constants.
In the case of a plane state of strain we assume that
e33 ¼ 0; e31 ¼ e32 ¼ 0; ð15:92Þ
we are thus led to the same relations (15.91), (15.910 ) in which the elastic constants
E1 ; E2 ; m12 6¼ m21 are replaced by the generalized elastic constants
15.2 Applications 695
E1 E2
E10 ¼ ; E20 ¼ ;
1 m13 m31 1 m23 m32
ð15:93Þ
m12 þ m13 m32 m21 þ m23 m31
m012 ¼ ; m021 ¼
1 m23 m32 1 m13 m31
and the coefficients of linear dilatation a1 ; a2 by the generalized coefficients of
linear dilatation
One can verify that between these generalized constants remains valid the first
relation (15.89). In this case, one obtains also
r33 ¼ m13 r11 þ m23 r22 E3 a3 T; ð15:95Þ
because the linear strain e33 vanishes and we took into consideration the relation
(15.89). In this case we have only 6 elastic constants too.
Excepting the determination of e33 for the state of plane stress and the deter-
mination of r33 for the state of plane strain, the problems are identical from the
mathematical point of view; there are involved only 4 distinct elastic constants (the
other two constants are used to determine the two quantities mentioned above).
Using the relations (15.91), (15.910 ) and the equations of equilibrium mentioned
above, the equations of continuity in stresses read
1 1 1 m21 m12
r11 þ a1 T þ r22 þ a2 T ¼ r12;12 : ð15:96Þ
E1 ;22 E2 ;11 l12 E1 E2
This leads to a solution in stresses of the problem. In the case of arbitrary volume
forces, one must add some particular integrals.
We introduce a representation of Airy type of the form
r11 ¼ F;22 ; r22 ¼ F;11 ; r12 ¼ F;12 ; ð15:97Þ
If the temperature variation is not known, then one must obtain it from the
equation of heat conduction in orthotropic media
where
k1 k2
,1 ¼ ; ,2 ¼ ; ð15:1010 Þ
cq cq
one supposes that Q ¼ Qðx1 ; x2 ; tÞ; T ¼ Tðx1 ; x2 ; tÞ.
Taking into account (15.990 ), we obtain
T ¼ DF ; ð15:102Þ
where we have introduced a new potential function F ¼ F ðx1 ; x2 ; tÞ; which ver-
ifies the equation
Dð,1 o11 þ ,2 o22 ot ÞF ¼ Q: ð15:103Þ
The connection between the two potential functions is made by
F ¼ ða1 o11 þ a2 o22 ÞF ; ð15:104Þ
0 0
,1 F;11 þ ,2 F;22 ¼ F_ 0 : ð15:107Þ
If the flow of heat is stationary, one has T_ ¼ 0 and if one has not heat sources one
has Q ¼ 0 too; in this case, the potential function is given by
Dð,1 o11 þ ,2 o22 Þ F ¼ 0: ð15:108Þ
It is interesting to remark that, in the case of an anisotropic (in particular
orthotropic) body, a variation of the temperature leads not only to a variation of the
strain tensor but also to a variation of the stress tensor.
Using Cauchy’s equations, one may easily calculate the components of the
displacement vector.
To determine the potential function F; one puts three conditions on the
boundary: two for the stresses and one for the temperature.
In conclusion, to solve the thermoelastic problem one must integrate a partial
differential equation of the form
Dða1 o11 þ a2 o22 ot Þ F ¼ 0; ð15:109Þ
on the boundary being given the function F; the normal derivative oF=on and DF:
One may also put an initial condition concerning the distribution of temperature
Tðx1 ; x2 ; tÞ for t ¼ t0 :
In the general case (the function T is not known), we eliminate the temperature
function between the Eq. (15.110), obtaining an equation verified by the dis-
placements u1 and u2 : Because the differential operators which act on these
functions are prime each other, it results that we can express the displacements by
means of only one function in the form
a1 a2 þ m21 a1 a1 þ m12 a2
u1 ¼ F ;22 þ F ;11 ;
G12 E1 E2 ;1
ð15:111Þ
a2 a1 þ m12 a2 a2 þ m21 a1
u2 ¼ F ;11 þ F ;22 :
G12 E2 E1 ;2
698 15 Introduction to Thermoelasticity
The results is no more valid in the particular case of the isotropic body, because
the above operators are no more prime each other.
References
A. Books
B. Papers
10. Teodorescu, P.P.: Asupra problemei plane a elasticitătßii unor corpuri anizotrope. VIII.
Influentßa variatßiei de temperatură (On the plane problem of elasticity of certain anisotropic
bodies. VIII. Influence of the temperature variation). Com. Acad. Rom., VIII (1958)
Chapter 16
Introduction to Linear Viscoelasticity
We have assumed in the classical theory of elasticity that the constitutive law (the
strain-stress relations) is linear and independent on time. As well, we have
assumed the hypothesis of small deformations with respect to unity, so that the
principle of superposition of effects may be applied. On the other hand, many
bodies do not respect the above hypotheses, appearing the influence of the time
too. A constitutive law has a more general form, i.e.
f ðr; e; tÞ ¼ 0;
where appears also the variable t.
To take into consideration the variation in time of the mechanical properties of
the body, we will assume that the medium is a viscoelastic one, i.e. it is formed of
two media: a perfectly elastic medium and a medium with properties of viscous
fluid, described by Hooke’s law and by Newton’s law, respectively.
In this order of ideas, we assume the superposition principle of Boltzmann as a
basis of the mathematical theory of linear viscoelastic bodies. In the case of a
Boltzmannian body, if the stress r1 ðtÞ leads to the strain e1 ðtÞ and the stress r2 ðtÞ
leads to the strain e2 ðtÞ, then the sum r1 ðtÞ þ r2 ðtÞ leads to the strain e1 ðtÞ þ e2 ðtÞ;
in particular, if r2 ¼ kr1 , then e2 ¼ ke1 (k = const) and if r2 ¼ r1 , then e2 ¼ e1 .
Thus, the above constitutive law is a linear differential or integral equation.
By means of the creep and relaxation functions we will construct one- and three-
dimensional laws, putting in evidence their differential and integro-differential
form too. Complex moduli of creep and relaxation will also be put in evidence.
w 0; w_ 0; w
€ 0: ð16:1Þ
Moreover, the experiments show that the curvature of the relaxation function w is a
decreasing function, hence
v 0:
w ð16:10 Þ
Indeed, because the curvature 1=q corresponding to the function rðtÞ is given
by
1 rv
¼ ;
q ð1 þ r_ 2 Þ3=2
the condition put to the relaxation function and the relation (16.1) lead to
_ r2
3r€
rv 0;
1 þ r_ 2
wherefrom the relation (16.10 ). A study of the properties of the linear viscoelastic
bodies by means of the properties (16.1), (16.10 ) is due to M. Alain [8].
The viscoelastic solids are bodies with memory, because the state of stress at the
moment t depends on the whole history of the deformations, i.e. on the state of
deformation on the interval ð1; t: Hereafter, we will characterize the linear
viscoelastic solid, deducing the corresponding constitutive law by means of the
methods used by M. Gurtin and E. Sternberg [10] and by Yu. N. Rabotnov [5].
16.1 Linear Viscoelastic Solids 701
Let be rðr; tÞ and eðr; tÞ; ðr; tÞ 2 X R R3 R; the tensor fields of stress
and strain, the components of which are functions of class C0;0 ðX RÞ; which
satisfy the conditions
which correspond to the natural state before the initial moment t = 0. We remark
that, for a fixed r in X; the component rij ðr; tÞ of the stress tensor at the moment
t represents a real functional of the components ekl ðr; sÞ; k; l ¼ 1; 2; 3; s 2
ð1; s of the strain tensor. Consequently, the stress rij ðr; tÞ may be represented
as a functional in the form
rij ðr; tÞ ¼ Lij ðekl Þ; i; j; k; l ¼ 1; 2; 3; ð16:3Þ
where Lij represents a real functional defined on the set of functions ekl 2
C0;0 ðX RÞ, equal to zero for t \ 0: For r 2 X, variable, Lij represents an oper-
ator. In a concentrated form, the relation (16.3) may be written in the form
r ¼ LðeÞ; ð16:4Þ
where keðr; tÞk ¼ max eij ðr; tÞ :
s2½0;t
The condition (i) expresses the linearity of the functional L and, from the
mechanical point of view, ensures the application of the principle of superposition
of effects of Boltzmann. The condition (ii) puts in evidence the fact that the
functional L is invariant with respect to the translation operation in time, hence
independent on the choice of the origin of time measure; on the other hand, this
condition shows that the constitution of the body does not depend on time and,
consequently, the correspondence between the tensors r and e defined by (16.4) is
independent on the choice of their origin, Finally, the condition (iii) expresses the
fact that the functional L is bounded and, together with (i), ensures its continuity.
Hence, it results that eðr; sÞ ¼ 0 for s 2 ½0; t implies rðr; tÞ ¼ 0 on ½0; t:
To write the law (16.4) in components, let us observe that the functional L is
equivalent to the lot of functionals Lijkl so that the constitutive law may be written
in the form
rij ðr; tÞ ¼ Lijkl ekl ðr; tÞ; i; j; k; l ¼ 1; 2; 3; ð16:6Þ
one may say that the functional L is a fourth order tensor functional of components
Lijkl ; i; j; k; l ¼ 1; 2; 3: Because the tensors r and e are symmetric, it follows the
symmetry with respect to the pairs of indices ði; jÞ, ðk; lÞ; i.e.:
Lijkl ¼ Ljikl ¼ Lijlk ; i; j; k; l ¼ 1; 2; 3: ð16:7Þ
It results that from the 34 ¼ 81 components only 36 are distinct; they satisfy the
conditions (i), (ii), (iii) and characterize the general anisotropic linear viscoelastic
solid. If the functional is invertible, then there exists the tensor functional p of
components pijkl ; i; j; k; l ¼ 1; 2; 3; so that
where wijkl ðr; tÞ are functions with bounded variation with respect to t 2 R on any
interval and equal to zero for t 0: These functions are called relaxation functions
and are the components of the relaxation tensor w, which is in complete con-
cordance with the function introduced in the one-dimensional case.
Denoting
8 8
< f ðaÞ; x\a; < gðaÞ; x\a;
f ðxÞ ¼ f ðxÞ; x 2 ½a; b; gðxÞ ¼ gðxÞ; x 2 ½a; b; ð16:11Þ
: :
f ðbÞ; x [ b; gðbÞ; x [ b;
because ðdðx xi Þ;
gðxÞÞ ¼ gðxi Þ ¼ gðxi Þ. Taking into account (16.11), one
obtains
Z b
~
f 0 ;
g ¼ gðxÞ~f 0 ðxÞdx;
a
so that
Z b X
n Z b
ðf 0 ;
gÞ ¼ gðxÞ~f 0 ðxÞdx þ si ðf Þgðxi Þ ¼ gðxÞdf ðxÞ: ð16:13Þ
a i¼0 a
If we assume that wijkl ðr; tÞ 2 C0;1 ðX ½0; 1ÞÞ and take into account that
wijkl ¼ 0 for t 0; then it results that wijkl ðr; tÞ 2 C0;1 ðX Rnf0gÞ; while t ¼ 0
represents a point of discontinuity of the first species, with the jump s0 ðwijkl Þ
¼ wijkl ðr; 0 þ 0Þ wijkl ðr; 0 0Þ ¼ wijkl ðr; 0þÞ:
Assuming that wijkl ðr; tÞ has, at the origin, a discontinuity of the first species
with respect to t 2 R and that rij ðr; tÞ 2 C0;0 ðX RÞ, while wijkl and rij vanish for
t 0; then, taking into account (16.13), the constitutive law (16.10) takes the form
Z t ~o
rij ðr; tÞ ¼ ekl ðr; tÞwijkl ðr; 0þÞ þ ekl ðr; t sÞ wijkl ðr; sÞds
0 os
Z t
¼ ekl ðr; t sÞdwijkl ðr; sÞ; i; j ¼ 1; 2; 3; ð16:14Þ
0
where ~o=os is the derivative in the usual sense; the derivative in the sense of the
theory of distributions is denoted by o=os. Using the convolution product in
the space of distributions with regard to t; we may write the constitutive law in the
form
704 16 Introduction to Linear Viscoelasticity
o o
rij ðr; tÞ ¼ ekl ðr; tÞ wijkl ðr; tÞ ¼ ekl ðr; tÞ wijkl ðr; tÞ; i; j ¼ 1; 2; 3: ð16:15Þ
ðtÞ ot ot ðtÞ
This relation may be considered in the distribution space K0þ with the supports
in ½0; 1Þ if we assume that rij ; eij ; wijkl ; with respect to t 2 R; are distributions in
K0þ , while r 2 X is a parameter.
Because of the relations
wijkl ¼ wjikl ¼ wijlk ; i; j; k; l ¼ 1; 2; 3; ð16:16Þ
in the case of an anisotropic linear viscoelastic solid, the relaxation tensor w has 36
distinct components.
In the same space K0þ ; the constitutive law (16.15) may be written in the
equivalent form
o o
eij ðr; tÞ ¼ rkl ðr; tÞ uijkl ðr; tÞ ¼ rkl ðr; tÞ uijkl ðr; tÞ; i; j ¼ 1; 2; 3: ð16:17Þ
ðtÞ ot ot ðtÞ
where u represents the creep tensor of components uijkl in the distribution space
K0þ ; hence, uijkl ¼ 0 for t 0. As in the case of the relaxation tensor, one has
r ~ ðr; pÞ; i; j ¼ 1; 2; 3;
~kl ðr; pÞ ¼ ~eij ðr; pÞpw ð16:18Þ
ijkl
~eij ðr; pÞ ¼ r
~kl ðr; pÞp~
uijkl ðr; pÞ; i; j ¼ 1; 2; 3; ð16:180 Þ
Thus, a dual of Hooke’s law is obtained for the linear anisotropic viscoelastic
bodies; hence, the results obtained for the classical linear elastic bodies may be
transposed to the linear viscoelastic bodies.
In the case of the isotropic elastic solid, Hooke’s tensor is given by (4.89);
taking into account (16.19), one may write
16.1 Linear Viscoelastic Solids 705
where kðr; tÞ and lðr; tÞ are distribution in K0þ . Applying the inverse Laplace
transform, one obtains
o
w ðr; tÞ ¼ kðr; tÞdij dkl þ lðr; tÞðdik djl þ dil djk Þ; i; j; k; l ¼ 1; 2; 3: ð16:21Þ
ot ijkl
Substituting in the constitutive law (16.15), one obtains the constitutive law for an
isotropic linear viscoelastic solid in the form
rij ðr; tÞ ¼ kðr; tÞ ekk ðr; tÞdij þ 2lðr; tÞ eij ðr; tÞ; i; j ¼ 1; 2; 3: ð16:22Þ
ðtÞ ðtÞ
~ij ðr; tÞ ¼ ~
r kðr; pÞ~ekk ðr; pÞdij þ 2~
lðr; pÞ~eij ðr; pÞ; i; j ¼ 1; 2; 3:
Introducing the deviators of stresses and strains r0ij ; e0ij and the spheric tensors r0 dij
and e0 dij ; respectively, one obtains
~0ij ðr; pÞ ¼ 2~
r lðr; pÞ~e0ij ðr; pÞ;
ð16:23Þ
~ pÞ~e0 ðr; pÞ;
~0 ðr; pÞ ¼ 3Kðr;
r
where
~ pÞ ¼ ~ 2
Kðr; kðr; pÞ þ l~ðr; pÞ; ð16:24Þ
3
the inverse Laplace transform leads to
where
2
Kðr; tÞ ¼ kðr; tÞ þ lðr; tÞ: ð16:240 Þ
3
The distribution l is the relaxation modulus to sliding, while the distribution K is
the relaxation modulus to volume dilatation.
The relaxation distributions wi ðr; tÞ 2 K0þ ; i ¼ 1; 2; are defined by the relations
o o
w1 ðr; tÞ ¼ 2lðr; tÞ; w2 ðr; tÞ ¼ 3Kðr; tÞ; ð16:25Þ
ot ot
wherefrom
w1 ðr; tÞ ¼ 2hðtÞ lðr; tÞ; w2 ðr; tÞ ¼ 3hðtÞ Kðr; tÞ; ð16:250 Þ
ðtÞ ðtÞ
706 16 Introduction to Linear Viscoelasticity
o o
r0ij ðr; tÞ ¼ w1 ðr; tÞ ~e0ij ðr; tÞ ¼ w1 ðr; tÞ ~e0ij ðr; tÞ; i; j ¼ 1; 2; 3; ð16:26Þ
ot ðtÞ ðtÞ ot
o o
r0 ðr; tÞ ¼ w ðr; tÞ e0 ðr; tÞ ¼ w2 ðr; tÞ e0 ðr; tÞ: ð16:260 Þ
ot 2 ðtÞ ðtÞ ot
From (16.21), (16.230 ) and (16.25), one obtains (by integrating with respect to t)
the components of the relaxation tensor
1
wijkl ðr; tÞ ¼ ½w ðr; tÞ w1 ðr; tÞdij dkl
3 2
1
þ w1 ðr; tÞðdik djl þ dil djk Þ; i; j; k; l ¼ 1; 2; 3: ð16:27Þ
2
Hence, the constitutive law (16.22) becomes
1 o
rij ðr; tÞ ¼ ½w2 ðr; tÞ w1 ðr; tÞ ekk ðr; tÞdij
3 ðtÞ ot
o
þ w1 ðr; tÞ eij ðr; tÞ; i; j ¼ 1; 2; 3: ð16:28Þ
ðtÞ ot
where 2 K0þ for an isotropic linear viscoelastic body. By means of the same
k; l
notations, one may write
e0ij ðr; tÞ ¼ 2
lðr; tÞ r0ij ðr; tÞ;
ðtÞ
ð16:31Þ
tÞ r0 ðr; tÞ;
e0 ðr; tÞ ¼ 3Kðr;
ðtÞ
where
tÞ ¼ 2 tÞ 2 K0þ ðRÞ:
Kðr; kðr; tÞ þ lðr; tÞ; Kðr; ð16:310 Þ
3
16.1 Linear Viscoelastic Solids 707
obtaining, finally,
1 o
eij ðr; tÞ ¼ ½u2 ðr; tÞ u1 ðr; tÞ rkk ðr; tÞdij
3 ðtÞ ot
o
þ u1 ðr; tÞ rij ðr; tÞ; i; j ¼ 1; 2; 3: ð16:33Þ
ðtÞ ot
with the notation u2 ðtÞ ¼ uðtÞ: For rðtÞ ¼ r0 hðtÞ one obtains uðtÞ ¼ e=r0 ; the
denomination of creep distribution being justified.
If
u1 ðtÞ ¼ 2 0 hðtÞ;
l0 hðtÞ; u2 ðtÞ ¼ 3K ð16:35Þ
then one obtains, as a particular case, the constitutive law of a linear elastic body.
Thus, the creep and relaxation distributions completely characterize the
mechanical properties with respect to time of the linear viscoelastic solids.
Because the constitutive laws (16.28), (16.33) are equivalent, it results that the
creep and relaxation distributions, as well as the distributions k; l; k; l are not
independent. To obtain the corresponding relations, we apply the Laplace trans-
form to the above mentioned laws; it follows
~0ij ðr; pÞ ¼ 2~
r lðr; pÞ~e0ij ðr; pÞ; i; j ¼ 1; 2; 3; r ~ pÞ~e0 ðr; pÞ;
~0 ðr; pÞ ¼ 3Kðr; ð16:36Þ
~e0ij ðr; pÞ ¼ 2l
~ ~
ðr; pÞr ~ pÞ~
0ij ðr; pÞ; i; j ¼ 1; 2; 3; ~e0 ðr; pÞ ¼ 3Kðr; r0 ðr; pÞ; ð16:360 Þ
wherefrom
~
lðr; pÞl
4~ ðr; pÞ ¼ 9Kðr; ~
pÞKðr;
pÞ ¼ 1: ð16:37Þ
Applying the inverse Laplace transform and taking into account (16.25) and
(16.32), one obtains
o o
lðr; tÞ 2lðr; tÞ ¼
2 u1 ðr; tÞ w1 ðr; tÞ ¼ dðtÞ;
ðtÞ ot ðtÞ ot
ð16:38Þ
tÞ 3Kðr; tÞ ¼ u2 ðr; tÞ o w2 ðr; tÞ ¼ dðtÞ;
3Kðr;
o
ðtÞ ot ðtÞ ot
708 16 Introduction to Linear Viscoelasticity
hence, the distributions oui =ot; i ¼ 1; 2 are inverse in the convolution algebra K0þ ;
i.e.
1 1
oui owi owi ou
¼ ; ¼ i ; i ¼ 1; 2: ð16:39Þ
ot ot ot ot
One may remark from (16.38) that
o2
ðu wi Þ ¼ dðtÞ; i ¼ 1; 2;
ot2 i
wherefrom
ui ðr; tÞ wi ðr; tÞ ¼ thðtÞ ¼ tþ ; i ¼ 1; 2: ð16:40Þ
ðtÞ
which corresponds to a spatial linear viscoelastic solid. What concerns the relation
between the spherical tensors, that one may be taken arbitrarily, i.e.
o
r0 ¼ 3KðtÞ e0 ¼ w2 ðtÞ e0 : ð16:410 Þ
ðtÞ ðtÞ ot
wherefrom
r0 ¼ 3K0 e0 :
In the case of a Kelvin-Voigt model, we replace the constitutive law (4.176) by
o 0
r0ij ðr; tÞ ¼ Ee0ij ðr; tÞ þ g e ðr; tÞ; i; j ¼ 1; 2; 3; ð16:42Þ
ot ij
r0 ðr; tÞ ¼ 3K0 e0 ðr; tÞ: ð16:43Þ
The first equation may be written in the form
r0 ¼ 3K0 e0 ; ð16:450 Þ
the creep and relaxation distributions are
1 1 1
u1 ðtÞ ¼ hðtÞ þ thðtÞ; u2 ðtÞ ¼ hðtÞ; ð16:46Þ
E g 3K0
where PðDÞ; QðDÞ are differential operators with constant coefficients in the space
of distributions. We suppose that the solid is subjected to a periodic stress with a
harmonic variation in time, of the form
r1 ðtÞ ¼ r0 hðtÞ cosðxt þ bÞ; ð16:48Þ
where r0 ¼ r0 ðxÞ is the amplitude, function of the pulsation x, while T ¼ 2p=x
is the period of the motion. One may associate a complex stress rðtÞ given by
r ¼ r0 eib : ð16:490 Þ
The corresponding strain is of the form
e1 ðtÞ ¼ e0 hðtÞ cosðxt þ aÞ; ð16:480 Þ
e ¼ e0 eia : ð16:500 Þ
Hence, r1 ¼ Re rðtÞ; e1 ¼ Re eðtÞ; so that the constitutive law (16.47) takes the
form PðDÞRe rðtÞ ¼ QðDÞRe eðtÞ; i.e.
Re½PðDÞrðtÞ ¼ Re½QðDÞeðtÞ: ð16:470 Þ
Hence, instead of the equivalent Eqs. (16.47), (16.470 ), one may consider the
equation
PðDÞr ¼ QðDÞe; ð16:4700 Þ
a complex constitutive equation of the unidimensional viscoelastic solid.
Because r and e are distributions of function type of class C1 ðRÞ; excepting the
origin, where they have discontinuities of the first species; one has
P ðDÞr
~ ¼ QðDÞe;
~ t [ 0; ð16:51Þ
where D ~ ¼~ d=dt represents the derivative in the usual sense. Hence, substituting
0
(16.49 ), (16.50) in (16.51), we get
16.1 Linear Viscoelastic Solids 711
r ðxÞPðDÞe
~ ixt ¼ e ðxÞQðDÞe
~ ixt ;
wherefrom
r ðxÞPðixÞ ¼ e ðxÞQðixÞ; ð16:52Þ
because
~ ixt ¼ eixt PðixÞ; QðDÞe
PðDÞe ~ ixt ¼ eixt QðixÞ:
It results
r r ðxÞ QðixÞ
¼ ¼ ; t [ 0: ð16:520 Þ
e e ðxÞ PðixÞ
We call complex modulus of relaxation corresponding to the viscoelastic solid
the complex number
QðixÞ
E ðixÞ ¼ : ð16:53Þ
PðixÞ
By means of the complex modulus E ; the relation (16.520 ) becomes
r ¼ E e ð16:54Þ
or
where
d
PðDÞ ¼ 1; QðDÞ ¼ gD þ E; D ¼ : ð16:570 Þ
dt
We have PðixÞ ¼ 1; QðixÞ ¼ E þ ixg; consequently,
QðixÞ
E ðixÞ ¼ ¼ E þ ixg ¼ E1 ðxÞ þ iE2 ðxÞ;
PðixÞ
hence
E1 ðxÞ ¼ E; E2 ðxÞ ¼ xg: ð16:58Þ
The complex creep modulus is
1 E ixg
JðixÞ ¼ ¼ ¼ J1 ðxÞ iJ2 ðxÞ;
E ðixÞ E2 þ x2 g2
wherefrom
E xg
J1 ðxÞ ¼ ; J2 ðxÞ ¼ 2 : ð16:580 Þ
E2 2
þx g 2 E þ x 2 g2
Hence, the constitutive law of the Kelvin-Voigt model in the complex form reads
ðP1 ðDÞdðtÞ þ f1 ðtÞÞ r0ij ðr; tÞ ¼ ðQ1 ðDÞdðtÞ þ g1 ðtÞÞ e0ij ðr; tÞ; ð16:62Þ
ðtÞ ðtÞ
ðP2 ðDÞdðtÞ þ f2 ðtÞÞ r0 ðr; tÞ ¼ ðQ2 ðDÞdðtÞ þ g2 ðtÞÞ e0 ðr; tÞ; ð16:620 Þ
ðtÞ ðtÞ
16.1 Linear Viscoelastic Solids 713
r0 r 0 Q2 ðixÞ þ g ~2 ðixÞ
¼ ¼ ; ð16:640 Þ
e0 e0 P2 ðixÞ þ ~f2 ðixÞ
Q2 ðixÞ þ ~
g2 ðixÞ
3K ðixÞ ¼ ; ð16:650 Þ
P2 ðixÞ þ ~f2 ðixÞ
thus, the constitutive law takes the form
where
where r
ij ; eij ; e0 are the amplitudes of the stresses, strains and volume strain,
respectively, when the loads have an harmonic variation. One obtains thus the
relations between the amplitudes
r
ij ðr; xÞ ¼ k ðixÞe0 ðr; xÞdij þ 2l ðixÞeij ðr; xÞ; i; j ¼ 1; 2; 3: ð16:70Þ
so that
r ij e ij r 0 e 0
~0ij ¼
r ; ~e0ij ¼ ~0 ¼
;r ; ~e0 ¼ : ð16:71Þ
p ix p ix p ix p ix
One obtains thus
r ij ðr; ixÞ
~ ðixÞ ¼ 2l ðixÞ;
¼ ixw1
e ij ðr; ixÞ
r 0 ðr; ixÞ ~ ðixÞ ¼ 3K ðixÞ:
¼ ixw2
e 0 ðr; ixÞ
Because
w0k u0k ¼ dðtÞ; k ¼ 1; 2;
ðtÞ
ðixÞ2 u ~ ðixÞ ¼ 1; k ¼ 1; 2;
~ k ðixÞwk
We deal now with the formulation of the problems of the linear theory of visco-
elasticity, presenting then some limit problems.
16.2 Limit Problems 715
ij ðr; tÞ ¼ hðtÞrij ðr; tÞ; eij ðr; tÞ ¼ hðtÞeij ðr; tÞ; i; j ¼ 1; 2; 3;
r
ð16:73Þ
i ðr; tÞ ¼ hðtÞui ðr; tÞ; F
u i ðr; tÞ ¼ hðtÞFi ðr; tÞ; i ¼ 1; 2; 3:
r i ¼ q€
ij;j þ F _
ui q½u_ 0i dðtÞ þ u0i dðtÞ; i ¼ 1; 2; 3: ð16:75Þ
The constitutive law reads
ij ¼ kðtÞ ekk ðr; tÞdij þ 2lðtÞ eij ðr; tÞ; i; j ¼ 1; 2; 3;
r ð16:76Þ
ðtÞ ðtÞ
where
dw1 ðtÞ dw ðtÞ
2lðtÞ ¼ ; 3KðtÞ ¼ 2 ; l; K; w1 ; w2 2 K0þ : ð16:760 Þ
dt dt
The equations of Cauchy are
1
eij ¼ ui;j þ
uj;i ; i; j ¼ 1; 2; 3; ð16:77Þ
2
716 16 Introduction to Linear Viscoelasticity
Eliminating the stresses and the strains, one obtains the equations in displacements
of Lamé’s type, i.e.
lðtÞ l
i;jj þ ½kðtÞ þ lðtÞ i
uj;ji þ F
ðtÞ ðtÞ
¼ q€
ui
q½u_ 0i dðtÞ þ _
u0i dðtÞ; i ¼ 1; 2; 3: ð16:79Þ
Applying Laplace’s transforms in distributions with respect to t 2 R; one
obtains the equations
l ui;jj ðr; pÞ þ ½~
~ðpÞ~
kðpÞ þ l ~ðpÞ~ ~
uj;ji ðr; pÞ þ F
i
¼ qp 2~ 0 0
ui ðr; pÞ qðu_ i þ pui Þ; i ¼ 1; 2; 3; ð16:790 Þ
where the initial conditions have been put too for the volume strain.
In particular, in the case of the quasistatic problem, the volume forces and the
initial conditions being equal to zero, one obtains
ðk þ 2lÞ Dekk ¼ 0: ð16:81Þ
ðtÞ
o2
h 1 ¼ D qðk þ lÞ1 ; ð16:83Þ
ðtÞ ot2
where we assumed that kðtÞ þ 2lðtÞ 2 K0þ is invertible in the convolution algebra
K0þ :
We have seen that in solving the limit problems of the linear viscoelasticity the
Laplace transform plays an important rôle. Comparing the formulation of the
problems of linear viscoelasticity to the similar ones of linear elasticity, one
observes that the difference consists in the constitutive law; but the Laplace image
of the constitutive law of linear viscoelastic bodies has the same mathematical
structure as the corresponding linear elastic law of Hooke. This remark led
T. Alfrey [9] and E. H. Lee [15] to the formulation of the principle of corre-
spondence as a method of solving the above mentioned problems. Vito Volterra
[18] dealt with this principle and Yu. N. Rabotnov and A. A. Ilyushin [6] called it
the principle of Volterra. This principle has been generalized and used by
W. T. Read [16], H. S. Tsien [17] and J. Mandel [4].
At it has been shown in [9], [16], [17], to a quasistatic problem of viscoelas-
ticity one may associate an elastostatic problem. W. T. Read [16] showed that,
using the Fourier transform, one can associate, in general, to a dynamic visco-
elastic problem an elastostatic problem too, but with a special character. Thus, to
the inertial forces in the viscoelastodynamics correspond volume forces direct
proportional to the components of the displacement in the elastostatic problem.
Let us thus remember the common equations (motion, Cauchy, continuity) of
both isotropic elastic and viscoelastic solids, i.e.:
r i ðr; tÞ ¼ q€
ij;j ðr; tÞ þ F _
ui q½u_ 0i dðtÞ þ u0i dðtÞ; i ¼ 1; 2; 3; ð16:85Þ
eij ¼ ui;j þ
uj;i ; i; j ¼ 1; 2; 3; ð16:850 Þ
2
eij;kl þ ekl;ij ¼ eil;jk þ ejk;il ; i; j; k; l ¼ 1; 2; 3; ð16:8500 Þ
718 16 Introduction to Linear Viscoelasticity
o
0ij ðr; tÞ ¼ 2l ðr; tÞ e0ij ðr; tÞ ¼
r w ðr; tÞ e0ij ðr; tÞ; i; j ¼ 1; 2; 3; ð16:88Þ
ðtÞ ot 1 ðtÞ
o
0 ðr; tÞ ¼ 3K ðr; tÞ e0 ðr; tÞ ¼
r w ðr; tÞ e0 ðr; tÞ ð16:880 Þ
ðtÞ ot 2 ðtÞ
too, then one has the complete system of equations of the mixed fundamental
problem of the linear elastoviscodynamics.
Obviously, for both solids we assume that the geometric form is the same, as
well as the volume forces, the boundary and the initial conditions. We mention
also that l, K and l*, K* are the same constants but with different notations, to
may distinguish the two solids.
Applying the Laplace transform in distributions to the constitutive laws (16.87),
(16.870 ), (16.88), (16.880 ) one obtains
0 ðr; pÞ ¼ 3KðrÞ~
~
r e0 ðr; pÞ; ð16:890 Þ
Comparing the Eqs. (16.89) and (16.890 ) with (16.90) and (16.900 ), respectively,
we see that they have the same structure, which leads to the bijection
~ ðr; pÞ;
lðrÞ $ l ð16:91Þ
~ ðr; pÞ
KðrÞ $ K ð16:910 Þ
between the quantities which characterize the two media; the complex variable
p plays the rôle of a parameter.
The other relations are the same for both media. It results that, knowing the
solution of a problem in elastodynamics, one may obtain the solution for the
analogue problem in elastoviscodynamics.
Hence, in conformity with the principle of correspondence, one must solve
firstly the correspondent problem of elastodynamics, considering then the Laplace
image of the solution thus obtained. Then, the elastic constants l and K are
replaced by Laplace’s images l ~ and K ~ ; respectively; one obtains thus the
Laplace image of the solution of the mixed problem of viscoelastodynamics.
Applying now the inverse Laplace transform, one gets the solution of the searched
problem.
A class of problems to which the principle of correspondence may be suc-
cessfully applied is that of quasi-static problems.
r i ðr; tÞ ¼ 0; i ¼ 1; 2; 3;
ij;j ðr; tÞ þ F ð16:92Þ
1
eij ðr; tÞ ¼ ui;j ðr; tÞ þ
uj;i ðr; tÞ ; i; j ¼ 1; 2; 3; ð16:920 Þ
2
eij;kl ðr; tÞ þ ekl;ij ðr; tÞ ¼ eil;jk ðr; tÞ þ ejk;il ðr; tÞ; i; j; k; l ¼ 1; 2; 3; ð16:9200 Þ
n
ui ðr; tÞ on Su ; i ¼ 1; 2; 3; t 2 R;
ui ðr; tÞ ¼ ð16:94Þ
720 16 Introduction to Linear Viscoelasticity
n
ij ðr; tÞnj ðrÞ ¼
r pi ðr; tÞ on Sr ; i; j ¼ 1; 2; 3: ð16:940 Þ
Applying the Laplace transform in distributions, one obtains
~ ~
i ðr; pÞ ¼ 0; i ¼ 1; 2; 3;
ij;j ðr; pÞ þ F
r ð16:95Þ
1
~eij ðr; pÞ ¼ ~
ui;j ðr; pÞ þ ~
uj;i ðr; pÞ ; i; j ¼ 1; 2; 3;
ð16:950 Þ
2
eij;kl ðr; pÞ þ ~
~ ekl;ij ðr; pÞ ¼ ~
eil;jk ðr; pÞ þ ~
ejk;il ðr; pÞ; i; j; k; l ¼ 1; 2; 3; ð16:9500 Þ
0ij ðr; pÞ ¼ 2~
~
r l ðr; pÞ~
e0ij ðr; pÞ; i; j ¼ 1; 2; 3; ð16:96Þ
~ ~ ðr; pÞ~
0 ðr; pÞ ¼ 3K
r e0 ðr; pÞ; ð16:960 Þ
n
~ u~i ðr; pÞ on Su ; i ¼ 1; 2; 3;
ui ðr; pÞ ¼
ð16:97Þ
n
~
ij ðr; pÞnj ðrÞ ¼ ~
r pi ðr; pÞ on Sr ; i; j ¼ 1; 2; 3:
ð16:970 Þ
Let us consider now following bijections between the quantities in elastostatics
and the Laplace images of the quantities in viscoelasticity
~
rij ðrÞ $ r ~
i ðr; pÞ; i; j ¼ 1; 2; 3;
ij ðr; pÞ; Fi ðrÞ $ F ð16:98Þ
16.2.2 Applications
We shall apply now the above results in some particular cases, i.e.: the viscoelastic
disk in uniform circular motion, the viscoelastic tube with thick walls and the
viscoelastic cylindrical bar subjected to its own weight and to an axial force.
16.2 Limit Problems 721
If the disk is a linear elastic one, then it is subjected to a plane state of stress; using
Airy’s formulae, one obtains
" #
2
1 2 k þ 2l ðk þ 2lÞ
ur ¼ qx r ð2re2 r 2 Þ þ re2 : ð16:101Þ
8 4lðk þ lÞ 4lðk þ lÞð3k þ 2lÞ
where
2~ ~ ðpÞ; 3~
l ¼ pw k þ 2~
l ¼ 3K ~ ðpÞ;
~ ¼ pw ð16:102Þ
1 2
Fr
P
r θ
O
re x1
722 16 Introduction to Linear Viscoelasticity
ð~ lÞ2
k þ 2~ ~ l ~k þ 2~
k þ 2~ l
¼
4~ ~
l ðk þ l ~
~Þð3k þ 2~ ~
lðk þ l
lÞ 4~ ~
~Þ ð3k þ 2~
lÞ
~ þw
ð2w ~ Þ2
¼ 1 2 ~
¼ GðpÞ; ð16:103Þ
~ w
3pw ~ ðw~ þ 2w ~ Þ
1 2 1 2
~
k þ 2~
l 2w~ þw ~
1 2 ~
¼ ¼ HðpÞ: ð16:1030 Þ
lð~
4~ kþl ~ ðw
~ Þ pw 1
~ þ 2w
1
~ Þ
2
~
Denoting GðtÞ ¼ L1 ½GðpÞ; ~
HðtÞ ¼ L1 ½HðpÞ; it results
1
~ ~
ur ¼ L½ur ¼ qx2 r ð2re2 r 2 ÞHðpÞ ~
þ re2 GðpÞ ;
8
wherefrom
1
ur ðr; tÞ ¼ qx2 r ð2re2 r 2 ÞHðtÞ þ re2 GðtÞ ; ð16:104Þ
8
i.e. the state of displacement corresponding to the viscoelastic solid.
Let us suppose now that the viscoelastic body is of Kelvin-Voigt type, which
behaves as an elastic one at simple compression or traction; one has thus w2 ¼
3K0 hðtÞ; K0 [ 0, hence w ~ ðpÞ ¼ 3K0 =p: The relaxation distribution w ¼ gdðtÞ þ
2 1
~
EhðtÞ; leads to w1 ðpÞ ¼ ðgp þ EÞ=p:
Substituting in (16.103), (16.1030 ), one obtains
but Dirac’s distribution dðtÞ puts in evidence only that the circular motion of the
disk is due at t ¼ 0 to a sudden cause. So that, the radial displacement of the linear
viscoelastic disk of Kelvin-Voigt type is given by
qx2 r hg i
ur ðr; tÞ ¼ hðtÞ re2 r 2 eðE=gÞt þ 3ðre2 r 2 Þe½ðEþ6K0 Þ=gt ð16:106Þ
16g 3
for t [ 0:
Let us consider now the Lamé problem for a homogeneous and isotropic visco-
elastic tube subjected to an uniformly distributed internal and external compres-
sion of variable intensities pi ðtÞ and pe ðtÞ on the faces of radii ri and rc ;
respectively (Fig. 16.2). We neglect the volume forces, as well the forces of
inertia. The intensities pi ðtÞ and pe ðtÞ are distributions in K0þ :
We assume that the tube is subjected to a plane state of strain; because of the
axial symmetry, one has rrr ¼ rrr ðr; tÞ; rhh ¼ rhh ðr; tÞ; rrh ¼¼ 0: The equation
of motion becomes
o 1
rrr ðr; tÞ þ ½rrr ðr; tÞ rhh ðr; tÞ ¼ 0 ð16:107Þ
or r
and the boundary conditions are of the form
rrr ðr; tÞjr¼ri ¼ pi ðtÞ; rrr ðr; tÞjr¼re ¼ pe ðtÞ: ð16:108Þ
P re
ri
O
pi (t)
724 16 Introduction to Linear Viscoelasticity
1 oF o2 F
rrr ¼ ; rhh ¼ 2 :
r or or
One obtains the stresses
one has
3K ~ ðpÞ ¼ 2~
~ ¼ pw l þ 3~ ~ ðpÞ:
l ¼ pw
k; 2~ ð16:110Þ
2 1
We observe that
~
k ~ w
w ~
~ 1 ðpÞ ¼
G ¼ 2 1
;
~þl
lðk
4~ ~ ðw
~ Þ pw ~ þ 2w ~ Þ
1 1 2
as well, one uses the Laplace image HðpÞ~ given by (16.105). Applying the inverse
Laplace transform, one obtains
ur ðr; tÞ ¼ r rrr ðtÞ G1 ðtÞ rhh ðtÞ HðtÞ : ð16:111Þ
ðtÞ ðtÞ
11 ¼ r
r 22 ¼ r
23 ¼ r
31 ¼ r
12 ¼ 0;
PðtÞ ð16:113Þ
33 ðr; tÞ ¼ qghðtÞðl x3 Þ þ
r ;
A
wherefrom one obtains the state of displacement.
ρg
P(t)
x3
726 16 Introduction to Linear Viscoelasticity
~
kðpÞx1 qg ~
PðtÞ
~
u1 ðr; pÞ ¼
ðl x3 Þ þ ;
lð3~
2~ lÞ p
k þ 2~ A
ð16:114Þ
~
kðpÞx2 qg ~
PðtÞ
~
u2 ðr; pÞ ¼
ðl x3 Þ þ ;
lð3~
2~ lÞ p
k þ 2~ A
~
kðpÞqg x21 þ x22
~
u3 ðr; pÞ ¼
lð3~
2~ lÞ p
k þ 2~
~
kþl ~ qg 2 ~
PðtÞ
þ ðl x23 Þ þ 2x3 ; ð16:1140 Þ
lð3~
2~ lÞ p
k þ 2~ A
where
~ ; 3K
l ¼ pw
2~ ~ :
~ ¼ pw ð16:115Þ
1 2
Because
dui ðtÞ dwi ðtÞ
¼ dðtÞ; i ¼ 1; 2; ð16:116Þ
dt ðtÞ dt
one has
~ ¼ 1
u
~ iwi : ð16:1160 Þ
p2
Taking into account (16.116), (16.1160 ), one obtains
~
k 2
¼ p½u
~ 1 ðpÞ u
~ 2 ðpÞ;
~
~ð3k þ 2~
l lÞ 3
~
kþl ~ p
¼ ½2~
u1 ðpÞ þ u~ 2 ðpÞ;
~
~ð3k þ 2~
l lÞ 3
wherefrom
" #
~
k 2 du1 ðtÞ du2 ðtÞ
1
L ¼ ;
~ð3~
l k þ 2~
lÞ 3 dt dt
" #
~
kþl ~ 1 du1 ðtÞ du2 ðtÞ
1
L ¼ 2 þ :
~ð3~
l k þ 2~
lÞ 3 dt dt
Applying the inverse Laplace transform, the relations (16.73), (16.1160 ) lead to
16.2 Limit Problems 727
1 du1 ðtÞ du2 ðtÞ PðtÞ
u1 ðr; tÞ ¼
qghðtÞðl x3 Þ þ x1
3 dt dt ðtÞ A
qg x1 du1 ðtÞ du2 ðtÞ PðtÞ
¼ x1 ðl x3 Þðu1 u2 Þ ;
3 3 dt dt ðtÞ A
qg x2 du1 ðtÞ du2 ðtÞ PðtÞ ð16:117Þ
u2 ðr; tÞ ¼ x2 ðl x3 Þðu1 u2 Þ
;
3 3 dt dt ðtÞ A
qg 2
u3 ðr; tÞ ¼
x1 þ x22 ðu1 u2 Þ
3
x3 du1 ðtÞ du2 ðtÞ PðtÞ
þ 2 þ qghðtÞðl x3 Þ þ 2 :
6 dt dt ðtÞ A
In particular, if PðtÞ is a force constant in time, then PðtÞ ¼ P0 hðtÞ and the
formulae (16.117) become
1 P0
u1 ðr; tÞ ¼ ðu1 u2 Þ qgx1 ðl x3 Þ þ
;
3 A
1 P0
u2 ðr; tÞ ¼ ðu1 u2 Þ qgx2 ðl x3 Þ þ
;
3 A ð16:118Þ
qg 2 2
u3 ðr; tÞ ¼
x þ x2 ðu1 u2 Þ
3 1
x3 P0
þ ð2u1 þ u2 Þ qgðl x3 Þ þ 2 :
6 A
References
A. Books
1. Bland, D.R.: The Theory of Linear Viscoelasticity. Pergamon Press, London (1960)
2. Kecs, W.:Elasticitate si vâscoelasticitate (Elasticity and Viscoelasticity). Ed. Tehnică,
Bucuresti (1986)
3. Kolski, H.: Volny napryazheniya v tverdykh telakl (Stress Waves in Solid Bodies). Inostr.
Lit., Moskva (1955)
4. Mandel, J.: Cours de mécanique des milieux continus. I, II. Gauthier–Vilars, Paris (1966)
5. Rabotnov, Yu. N.: Elements of Hereditary Solid Mechanics. Mir Publ, Moecow (1980)
6. Rabotnov, Yu. N., A, Ilyushin: Methoden der Viskoelastizitätstheorie. VEB Fachbuchverlag,
Leipzig (1970)
7. Riesz, F., Sz Nagy, B.: Leçons d’analyse fonctionnelle. Budapest (1972)
728 16 Introduction to Linear Viscoelasticity
B. Papers
In what follows, we shall deal with elements of tensor calculus, with the introduction
of curvilinear co-ordinates and with some notions of the theory of distributions;
as well, we shall introduce certain notations and integrals.
We shall make some considerations concerning the algebra and the analysis
of tensors, dealing successively with scalars, tensors of first order and tensors of
nth order. Various particular tensors will be put into evidence. As well, we will
express in a tensor form the algebraic and analytic operations effected with
vectors. We consider only orthogonal affine tensors in R3 , the corresponding
notions being sufficient for the mathematical representation of the mechanical
phenomena with which we deal.
P
i3 r i'2
x'2
i'3
i1 x2
O i2
i1'
x1
x'1
while ‘‘’’ indicates the scalar (dot) product of two vectors, represented by bold
letters; analogically, we may write
ij ¼ akj i0k ; j ¼ 1; 2; 3: ðA:10 Þ
A scalar product of the relation (A.1) by i0l and, analogically, a scalar product of
the relation (A.10 ) by il ; l ¼ 1; 2; 3, leads to the relations between the cosines
introduced above
aij aik ¼ aji aki ¼ djk ; ðA:3Þ
where xj ; x0k ; j; k ¼ 1; 2; 3, are the co-ordinates of the point P in the two systems of
co-ordinates, respectively. Taking into account (A.1) and (A.10 ), respectively, we
get the relations which allow to pass from a system of co-ordinates to another one,
in the form
x0k ¼ xj akj ; xj ¼ x0k akj ; ðA:6Þ
these linear transformations are orthogonal, their coefficients verifying the relations
(A.3).
Hence, it results easily
ox0k oxj
¼ ¼ akj : ðA:7Þ
oxj ox0k
where aj ¼ aj ðx1 ; x2 ; x3 Þ; a0k ¼ a0k ðx01 ; x02 ; x03 Þ; j; k ¼ 1; 2; 3. Taking into account
(A.1) and (A.10 ), respectively, we get the relations which allow passing from the
components of the vector a, with respect to the basis B, to the components of the
same vector with respect to the basis B0 and inversely, in the form
a0k ¼ aj akj ; aj ¼ a0k akj ; k; j ¼ 1; 2; 3: ðA:10Þ
In general, the sets of three functions, in each basis, which are transformed
according to the formulae (A.10), by a change of orthogonal Cartesian
co-ordinates, are the components of a tensor of first order (or the components of
a vector). A tensor of first order may be represented by its components with regard
to a certain basis B in the form of a column matrix or by means of its transpose
(a row matrix)
732 Appendix
32
a1
a ½ai 4 a2 5 ½ a1 a2 a3 T : ðA:11Þ
a3
Taking into account (A.7), we may express the transformation relations in the
form
ox0k oxj
a0k ¼ aj ; aj ¼ a0k 0 ; j; k ¼ 1; 2; 3: ðA:100 Þ
oxj oxk
Extending the result obtained in the previous section, we call tensor of second
order the totality of the sets of 32 ¼ 9 magnitudes aij ; i; j ¼ 1; 2; 3, which, by a
change of orthogonal Cartesian co-ordinates, become
a0kl ¼ aij aki alj ; aij ¼ a0kl aki alj ; i; j; k ¼ 1; 2; 3; ðA:12Þ
in a basis B, then we say that the tensor is symmetric with respect, to the indices ij
and ik ; if the property (A.16) holds for any indices ij and ik , then the tensor a is
totally symmetric.
If the components of the tensor a verify the relation
ai1 i2 ...ij ...ik ...in ¼ ai1 i2 ...ik ...ij ...in ; ðA:17Þ
If we use the upper index ‘‘ 0 ’’ for the quantities which are obtained from the
relation (A.2), by a change of co-ordinate axes, then it follows that
a0lm ¼ i0l im ; l; m ¼ 1; 2; 3:
the relation of definition of these cosines remaining the same; indeed, the scalar
products (A.2) are invariant with respect to any frame of reference, in particular
with respect to the frames B and B0 . We see that a relation of the form (A.12)
holds, because
a0lm ¼ akj alk amj ¼ alk dkm ¼ alm ; l; m ¼ 1; 2; 3;
which makes clear the position of the orthonormed basis B0 with respect to the
basis B and inversely.
734 Appendix
We can define Kronecker’s symbol with respect to the basis B0 in the form
0 0; k 6¼ l;
dkl ¼ ; k:l ¼ 1; 2; 3;
1; k ¼ l:
which is just the unit tensor (d ¼ 1); this tensor is symmetric and we may write
dji ¼ dij ; i; j ¼ 1; 2; 3: ðA:200 Þ
Let us consider also Ricci’s permutation symbol, defined by the formula
8
< 1; ði; j; kÞ ¼ ð1; 2; 3Þ;
ijk ¼ 1; ði; j; kÞ ¼ ð2; 1; 3Þ; ðA:21Þ
:
0; i ¼ j or j ¼ k or k ¼ j;
where by ði; j; kÞ ¼ ð1; 2; 3Þ we understand that the indices i; j; k take the distinct
values 1; 2; 3 or a cyclic permutation of them. We notice that
2 3
al1 al2 al3
cijk ali amj ank 4 am1 am2 am3 5; l; m; n ¼ 1; 2; 3; ðA:22Þ
an1 an2 an3
taking into account (A.1) and introducing the scalar triple product of three vectors
ðii ; ij ; ik Þ ¼ cijk ; ði0l ; i0m ; i0n Þ ¼ c 0lmn ; i; j; k; l; m; n ¼ 1; 2; 3; ðA:23Þ
it results
c0lmn ¼ cijk ali amj ank ; l; m; n ¼ 1; 2; 3:
Hence, the permutation symbol is a tensor of third order; this tensor is totally
antisymmetric and we may write
cijk ¼ cjki ¼ ckij ¼ cjik ¼ cikj ¼ ckji ; i; j; k ¼ 1; 2; 3: ðA:24Þ
Let be now the determinant
2 3
a11 a12 a13
det apq 4 a21 a22 a23 5; ðA:25Þ
a31 a32 a33
A.1 Elements of Tensor Calculus 735
too or
1
det½apq ¼ cijk clmn ail ajm akn ; ðA:25000 Þ
6
which constitutes an extension of the above results.
In what follows we shall deal with algebraic or analytic operations effected with
tensors, as well as with the tensor expression of such operations effected with
vectors. By the way, we shall give some results of the field theory.
Two tensors a and b of the same order n are equal (a ¼ b) if they have the same
components
ai1 i2 ...in ¼ bi1 i2 ...in ; ij ¼ 1; 2; 3; j ¼ 1; 2; . . .; n; ðA:26Þ
in an arbitrary frame B; this relation has the well known properties of reflexivity,
symmetry and transitivity.
The sum of two tensors a and b of the same order n is a tensor c = a ? b of the
same order; hence, in a frame B, we have
ai1 i2 ...in þ bi1 i2 ...in ¼ ci1 i2 ...in : ðA:27Þ
one antisymmetric with respect to the same indices; we can thus write
1
ai1 i2 ...ij ...ik ...in ¼ ai1 i2 ...ij ...ik ...in þ ai1 i2 ...ik ...ij n
2
1
þ ai1 i2 ...ij ...ik ...in ai1 i2 ...ik ...ij ...in : ðA:28Þ
2
The tensor product (external product) of two tensors a and b of nth and mth
order, respectively, is a tensor c ¼ a b of (n þ m)th order; in a frame B we can
write
ai1 i2 ...in bj1 j2 ...jm ¼ ck1 k2 ...knþm ; ðA:29Þ
this operation is called the contraction of the tensor. Hence, by the contraction of
two indices of a tensor of nth order one obtains a tensor of (n 2)th order.
For instance, by the contraction of a tensor of second order, of components aij
in a basis B one obtains a scalar aii , called the trace of the tensor a and denoted
tr a ¼ aii : ðA:310 Þ
In particular, by the contraction of Kronecker’s tensor one obtains tr1 ¼ dii ¼ 3.
The internal product (the contracted tensor product) of two tensors a and b of
nth and mth order, respectively, is a tensor ab of (n þ m 2p)th order, where p is
the number of effected contractions.
For instance, the scalar product of two vectors a and b, of components ai and
bi , respectively, will be given by the contracted product (Fig. A.2a)
a b ¼ ab cosða; bÞ ¼ ai bi ; ðA:32Þ
where a; b are the moduli of the respective vectors. If this product vanishes, then
the two vectors are orthogonal one to the other (or one of them is null).
If c ¼ a b is the vector product of the two vectors, then we may write
(Fig. A.2b)
ci ¼ cijk aj bk ; i ¼ 1; 2; 3; ðA:33Þ
A.1 Elements of Tensor Calculus 737
b c c b
(a,b) b
a a a
(a) (b) (c)
we notice that the vector c is normal to the plane formed by the vectors a and b,
while its modulus is equal to the area of the parallelogram formed by these two
vectors. If this product vanishes, then the two vectors are collinear.
The triple scalar product of three vectors a; b and c can be expressed by means
of a contracted product too, in the form
ða; b; cÞ ¼ cijk ai bj ck ; ðA:34Þ
obtaining thus the components of a tensor of nth order in a basis B if b and c are
tensors.
With two permutation tensors one may effect the external product
dil dim din
ijk lmn ¼ djl djm djn ; i; j; k; l; m; n ¼ 1; 2; 3: ðA:37Þ
dkl dkm dkn
To prove this identity, one considers all the possible values which may be given to
the indices i; j; k and l; m; n, respectively; if two indices are equal, then one obtains
zero in both members, while if the indices are different, then one considers
successively the cases in which they form a permutation of the numbers 1; 2; 3 or a
permutation of the numbers 2; 1; 3.
738 Appendix
while the relations (A.33) allows to write the triple vector product of three vectors
a, b and c (Fig. A.3) in the form
a ðb cÞ ¼ ða cÞb ða bÞc: ðA:41Þ
In the case of a continuous system of vectors W ¼ WðrÞ, definite on the
domain D of volume V, one obtains the resultant and the resultant moment
ZZZ
R¼ WðrÞ dV;
V
ZZZ ðA:42Þ
MO ¼ r WðrÞ dV;
V
of components
ZZZ
Ri ¼ Wi ðrÞ dV;
V
ZZZ i ¼ 1; 2; 3: ðA:420 Þ
MOi ¼ cijk xj Wk ðrÞ dV;
V
What concerns the symmetric tensor of second order, its canonic decomposition
in the spheric tensor and the stress deviator, as well as the study of the variation of
its components along various directions, the determination of the principal
directions, the extreme values, the corresponding graphic representations etc.,
these are the problems which have been considered in the Chaps. 2 and 3, in
connection with the particular tensors Te and Tr .
The antisymmetric tensor a of second order will be represented in the form
0 a12 a31
½aij ¼ a12 0 a23 ; ðA:44Þ
a31 a23 0
ones, hence one cannot add a vector product to an axial vector. Instead, the scalar
product of a polar vector by an axial one (which leads to a scalar, i.e. the mixed
product of three vectors), as well as the vector product of two such vectors (which
leads to a polar vector, i.e. the double vector product of three vectors) are operations,
which have sense.
where we have taken into account the relations (A.7) and the relation (A.8), which
defines a scalar field. Therefore, the derivative of a scalar with respect to an
independent variable leads to a tensor of first order; we denote
oU
¼ oi U ¼ U;i ; i ¼ 1; 2; 3; ðA:47Þ
oxi
where the indices at the right of the comma indicate the differentiation with respect
to the corresponding variable.
Similarly, starting from the vector field Vi ¼ Vi ðx1 ; x2 ; x3 Þ with Vi 2 C1 ðDÞ,
i ¼ 1; 2; 3, and making a change of co-ordinates of the form (A.6), we may write
oVl0 oðVk alk Þ oxi oVk
¼ ¼ alk aji ; l; j ¼ 1; 2; 3; ðA:48Þ
ox0j oxi ox0j oxi
hence, the derivatives of first order of a tensor of first order are the components of
a tensor of second order. We may write
oVi
¼ oj Vi ¼ Vi;j ; i; j ¼ 1; 2; 3: ðA:49Þ
oxj
In general, the derivatives of first order of tensor of nth order are the
components of a tensor of (n þ 1)th order.
Similarly, one can define derivatives of higher order of a tensor.
If ai 2 C2 ðDÞ, then we may write
ai;jk ¼ ai;kj ; i; j; k ¼ 1; 2; 3; ðA:50Þ
hence, the mixed derivatives of second order do not depend on the order of
differentiation (Schwarz’s theorem). We notice thus that the tensor ai;jk is
symmetric with respect to the indices j and k.
A.1 Elements of Tensor Calculus 741
Let us introduce the vector differential operator nabla (del) of Hamilton, denoted
by r, in the form
o
r ¼ ij ¼ ij oj : ðA:51Þ
oxj
In this case, the conservative vector field (A.47) will define the gradient of the
scalar field U by the relation
gradU ¼ rU ¼ oj Uij ¼ U;j ij : ðA:52Þ
As well, we assume that Vi;j are the components of the tensor of second order,
denoted by GradV; in general, the gradient of a tensor of nth order will be a tensor
of (n þ 1)th order.
In this case, the differential of the scalar field U ¼ Uðx1 ; x2 ; x3 Þ may be written
in the form
dU ¼ gradU dr: ðA:53Þ
Let us consider the field of vectors V ¼ Vðx1 ; x2 ; x3 Þ defined by the relations
Vi ¼ U;i ; ðA:54Þ
O
(a) (b)
We denote by
dW ¼ VðrÞ dr ¼ Vi ðrÞdxi ðA:57Þ
the elementary work of the vector V ¼ VðrÞ; this one is, in general, not an exact
differential. We notice that the work of the sum of n vectors applied at the same
point is equal to the sum of the works of each vector; this result is obvious, taking
into account the property of distributivity of the scalar product with respect to the
addition of vectors.
The work of a vector V along a curve C, between the points P0 and P1
(Fig. A.4a), is given by
ðA:58Þ
the travelling sense along this curve being from P0 to P1 . As well, we notice that
the work of a vector is a scalar quantity.
In the case of a closed curve C (Fig. A.4b), we consider the curvilinear vector
integral
I I
WC ðVÞ ¼ VðrÞ dr ¼ Vi dxi ; ðA:59Þ
C C
the travelling sense being the counterclockwise one; this work is called the
circulation of the vector V along the closed curve C.
We mention that the curvilinear vector integrals along a closed curve do not
depend on the initial point on the curve.
In the case of a conservative vector field, the elementary work of a conservative
vector is given by
dW ¼ gradU dr ¼ dU; ðA:60Þ
so that it is an exact differential; it results that
A.1 Elements of Tensor Calculus 743
ðA:600 Þ
where we used the formula (A.58). Hence, in the case of a conservative vector, the
work between two points does not depend on the path, bat only on the values of
the potential at the ends; similarly, starting from the formula (A.59), we notice that
the work of a conservative vector along a closed curve vanishes (if the curve
belongs to a simply connected domain).
Let be
UðrÞ ¼ C; C ¼ const; ðA:61Þ
the equation of a surface having the property that at each of its points the scalar
potential is constant; we assume that U 2 C1 . This surface is called an
equipotential surface; in the case in which U ¼ Uðr; tÞ, we have an equiquasi-
potential surface
Uðr; tÞ ¼ C; C ¼ const; ðA:610 Þ
obtaining thus the derivative on the direction of unit vector s of the tangent to this
curve.
We mention the properties:
gradðU1 U2 Þ ¼ U1 gradU2 þ U2 gradU1 ; ðA:64Þ
as well, by applying this operator to the vector field V, one gets the vector
ðA gradÞV ¼ Aj Vk;j ik : ðA:70Þ
A.1 Elements of Tensor Calculus 745
then it results
dV ¼ GradV dr ¼ V;i dxi : ðA:730 Þ
The curves for which the tangents to each point of them are directed along the
vectors V ¼ VðrÞ of the field are called vector lines; these lines form a congruence
of curves. Because the differential dr is tangent to these lines, their vector equation
will be of the form
VðrÞ dr ¼ 0; ðA:74Þ
equivalent to
cijk Vj dxk ¼ 0; i ¼ 1; 2; 3: ðA:740 Þ
As well, we introduce the vector differential operator
o
A r ¼ cjkl Aj il ¼ cjkl Aj il ok : ðA:75Þ
oxk
The differential operator ‘‘r’’ if applied to a vector V, then it defines its
divergence in the form
oVi
divV ¼ r V ¼ ¼ oi Vi ¼ Vi;i ; ðA:76Þ
oxi
this scalar quantity is invariant to a change of co-ordinate axes.
The differential operator ‘‘r’’ if applied to a vector V, then it leads to the curl
of this vector; we may write
oVk
curlV ¼ r V ¼ cjkl il ¼ cjkl oj Vk il ¼ cjkl Vk;j il : ðA:77Þ
oxj
746 Appendix
Similarly, one can introduce the operators div and curl for tensors of higher
order.
A field of vectors for which
curlV ¼ r V ¼ 0 ðA:78Þ
is called irrotational. One easily observer that a field of gradients
V ¼ gradU ¼ rU ðA:780 Þ
is irrotational (curl gradU ¼ 0); hence, the fields of conservative vectors are
irrotational, these fields being the only ones which have this property. The property
is maintained in the case of a quasi-conservative field too; as we observe, this is
also the condition that the integral in (A.58) be an exact differential.
A vector field for which
divV ¼ r V ¼ 0 ðA:79Þ
o2 U o 2 U o2 U o2 U
DU ¼ ¼ 2 þ 2 þ 2 ðA:81Þ
oxi oxi ox1 ox2 ox3
D2 U ¼ 0; ðA:84Þ
A.1 Elements of Tensor Calculus 747
in the domain D, is called biharmonic in this domain. On the same way one may
introduce polyharmonic functions.
Laplace’s operator is of elliptic type. Analogically, we introduce d’Alembert’s
operator of hyperbolic type, in the form
1 o2
hc ¼ D ; c ¼ const; ðA:85Þ
c2 ot2
assuming that U ¼ Uðr; tÞ, the equation
hc U ¼ 0; ðA:86Þ
where U 2 C2 ðDÞ, is called the waves equation, while c is the propagation velocity
of the waves.
If U 2 C4 ðDÞ, then one may introduce the double waves equation
h1 h2 U ¼ 0; ðA:87Þ
too, where
1 o2
hi ¼ D ; ci ¼ const, i ¼ 1; 2; ðA:870 Þ
c2i ot2
where U 2 C2 ðDÞ will be called caloric equation; similarly, one may introduce
polycaloric functions.
because the left member of the formula depends only on the curve C, we may
replace the surface S by any other surface SO D, which satisfies analogous
conditions.
P S Σ
r C
O x2
x1
A.1 Elements of Tensor Calculus 749
The above curvilinear integral is called circulation (it represents the work of a
field of vectors), while the surface integral represents the flux of a field of curls; in
the above mentioned conditions, it results that the circulation of a field of vectors
along a closed curve is equal to the flux of the curl of the very same field of vectors
through a sufficiently smooth arbitrary surface, bounded by a given curve. We also
mention that the circulation of a field of irrotational vectors, hence of a field of
conservative vectors, vanishes.
With respect to an orthonormed frame of reference Ox1 x2 x3 , we may write
I ZZ
Wi dxi ¼ ijk ni Wk;i dS: ðA:940 Þ
C S
P
D
r x2
O
x1
750 Appendix
where n is the unit vector of the external normal to the surface; the surface integral
represents the flux of the field of vectors through the surface S, so that the formula
is called the flux-divergent formula too.
In components, we may write
ZZ ZZZ
Wi ni dS ¼ Wi;i ds: ðA:950 Þ
S D
so that
ZZ ZZZ
n V dS ¼ curlV ds: ðA:950000 Þ
S D
for a field of conservative vectors W ¼ gradU, where we took into account the
formula (A.550 ) and the definition of Laplace’s operator. If W ¼ wU; w scalar,
then we may write
ZZ ZZZ
wU n dS ¼ ðwdivU þ U gradwÞds;
S D
A.1 Elements of Tensor Calculus 751
where we took into account the formula (A.90). If, after this, we take U ¼ gradu,
u scalar, then we get
ZZ ZZZ
ou
w dS ¼ ðwDu þ gradu gradwÞds; ðA:97Þ
S on D
inverting u and w and subtracting the relation thus obtained from (A.97). it results
ZZ
ZZZ
ou ow
w u dS ¼ ðwDu uDwÞds: ðA:98Þ
S on on D
Often, one can reduce the study of certain partial differential equations to the study
of several differential equations of the same type, but of a lower order.
Thus, let be the biharmonic equation (A.84) and two functions U1 , and U2 ,
harmonic in the domain D and verifying the Eq. (A.83); Almansi showed that the
biharmonic function U may be expressed univocally in the form
U ¼ U 1 þ x1 U 2 ; ðA:99Þ
hence, a biharmonic function is, in a certain way, equivalent to two harmonic
functions. We notice that we can replace the variable x1 by anyone of the other
variables; as well, we may write
U ¼ U1 þ r 2 U2 : ðA:990 Þ
To verify the above formulae, we mention the relations
Dðxi UÞ ¼ 2U;i þ xi DU; i ¼ 1; 2; 3; ðA:100Þ
Dn U ¼ 0; ðA:103Þ
where U 2 Cmn ðDÞ, may be written in the form
U ¼ U0 þ q1 U1 þ q21 U2 þ þ qn1
1
Un1 ; ðA:1030 Þ
analogue to the formula (A.102); if the vector V satisfies a simple waves equation,
then it becomes
hi ðr VÞ ¼ 2divV; i ¼ 1; 2; ðA:1070 Þ
A.1 Elements of Tensor Calculus 753
If we take
V ¼ curl W; ðA:108Þ
First of all, we deal with some general results concerning orthogonal curvilinear
co-ordinates, particularizing then the results thus obtained to cylindrical and
spherical co-ordinates.
where xi are functions at least of class C1 , put in evidence three curvilinear lines
passing through the point M, as we take q2 ; q3 ¼ const or q3 ; q1 ¼ const or
r M ( x1, x 2, x3)
i3
i1 O x2
i2
x1
754 Appendix
In the Sect. A.2, the sign (!) indicates ‘‘without summation with respect to
Greek indices’’.
In case of orthogonal curvilinear co-ordinates, one has
isb isc ¼ dbc ; b; c ¼ 1; 2; 3; ðA:114Þ
wherefrom, taking into account (A.112) and that ij has an analogous property, we
get
oqb oqc
¼ hb hc dbc ð!Þ; b; c ¼ 1; 2; 3; ðA:115Þ
oxi oxi
in particular,
oqb oqc
¼ 0; b 6¼ c; b; c ¼ 1; 2; 3: ðA:1150 Þ
oxi oxi
These are necessary and sufficient conditions for the transformation (A.1100 ) to be
orthogonal.
The direction cosines of the directions qa will be given by the scalar products
cosðqa ; xi Þ ¼ isa ii ;
A.2 Curvilinear Co-ordinates 755
which leads to
1 oqa
cosðqa ; xi Þ ¼ ; i; a ¼ 1; 2; 3; ðA:116Þ
ha oxi
where we took into account (A.112).
We notice that
oqb oqb oxi
¼ ¼ dbc ð!Þ; b; c ¼ 1; 2; 3:
oqc oxi oqc
To solve the system of nine linear equations thus obtained, we assume that
oxi oqc
f ðcÞ ð!Þ;
oqc oxi
wherefrom, for b ¼ c;
1
f ðcÞ ¼ ;
h2c
hence,
oxi 1 oqa
¼ 2 ð!Þ; i; a ¼ 1; 2; 3: ðA:117Þ
oqa ha oxi
in particular,
oxi oxi
¼ 0; b 6¼ c; ðA:1180 Þ
oqb oqc
obtaining thus the necessary and sufficient conditions for the transformation
(A.110) to be orthogonal.
Starting from (A.116), we get the direction cosines in the form
oxi
cosðqa ; xi Þ ¼ ha ð!Þ; i; a ¼ 1; 2; 3; ðA:1160 Þ
oqa
756 Appendix
Taking into account (A.117), we may express the differential parameters of first
order also in the form
In the case of cylindrical co-ordinates (Fig. A.8a), we may write the relations
(orthonormed basis ir ; ih ; iz )
x1 ¼ r cos h; x2 ¼ r sin h; x3 ¼ z; r 0; 0
h\2p; z 2 R; ðA:119Þ
wherefrom qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
x2
r¼ x21 þ x22 ; h ¼ arg tan ; z ¼ x3 : ðA:1190 Þ
x1
The differential parameters of first order are
1
hr ¼ hz ¼ 1; hh ¼ : ðA:120Þ
r
In the case of spherical co-ordinates (space polar co-ordinates) (Fig. A.8b), we
have (orthonormed frame of unit vectors iR ; iu ; ih )
One may easily pass from cylindrical co-ordinates to spherical ones, observing
that
r ¼ R cos u: ðA:122Þ
The differential parameters of first order are given by
1 1
hR ¼ 1; hu ¼ ; hh ¼ : ðA:123Þ
R R sin u
Starting from
oxi
dxi ¼ dqa ; i ¼ 1; 2; 3; ðA:124Þ
oqa
the arc element in orthogonal curvilinear co-ordinates is given by (Fig. A.9)
X3
1 2
ds2 ¼ dxi dxi ¼ dq ; ðA:125Þ
h2 a
a¼1 a
i s3
ds3
i s2 i s1
ds1
ds2
M ( x1, x 2, x3)
758 Appendix
1
dsa ¼ dqa ð!Þ; a ¼ 1; 2; 3; ðA:126Þ
ha
and
o2 ohb o o2
¼ þ hc hb ð!Þ;
osc osb osc oqb oqc oqb
similarly
o2 ohc o o2
¼ þ hb hc ð!Þ:
osb osc osb oqc oqb oqc
Subtracting one relation of the other, taking into account the symmetry of the
mixed derivative operators with respect to qb and qc and using the formula
(A.128), we get the relation between the mixed derivative operators with respect to
sb and sc in the form
o2 o2 1 ohc o 1 ohb o
¼ ð!Þ; b; c ¼ 1; 2; 3: ðA:133Þ
osb osc osc osb hc osb osc hb osc osb
Taking into account the relation (A.132), the condition (A.114) leads to
oxi oxi
¼ dbc ; b; c ¼ 1; 2; 3; ðA:134Þ
osb osc
one can obtain this result also starting from the relation (A.118) and from the
operator (A.128).
Differentiating the relation (A.134) with respect to sa , we get
o2 xi oxi o2 xi oxi
þ ¼ 0; a; b; c ¼ 1; 2; 3; ðA:135Þ
osa osb osc osa osc osb
o2 xi oxi
¼ 0; a ¼ 1; 2; 3; ðA:136Þ
osa osb osb
relation which is true if one sums with respect to b, as well as if one does not sum
with respect to this index.
Taking into account the relations (A.133, A.134), the relation (A.135) also
allows to write
1 ohb 1 ohc
dbc þ ð!Þ; a; b; c ¼ 1; 2; 3: ðA:1350 Þ
hb osa hc osa
Making b ¼ c, we get, in particular,
1 oðhb hc Þ o 1
¼ ¼ hb hc ð!Þ;
hb hc osa osa hb hc
a 6¼ b 6¼ c 6¼ a; with summation with respect to d: ðA:13600 Þ
o o xi o 2 xi 1 oha oha
þ ¼ dab þ dac
osa osb osc osc osb ha osc osb
1 ohb 1 ohc
þdbc þ ð!Þ; a; b; c ¼ 1; 2; 3: ðA:136000 Þ
hb osa hc osa
The divergence of a vector V is given by
o oVsb ois
divV ¼ r V ¼ isa Vsb isb ¼ isa isb þ Vsb isa b ;
osa osa osa
where we used the operator (A.131); taking into account (A.132), we may write
oisb oxi o2 xi
isa ¼ ; ðA:137Þ
osa osa osa osb
oisb o 1
isa ¼ ha hc ð!Þ: ðA:1370 Þ
osa osb ha hc
Finally, we get
where from
A.2 Curvilinear Co-ordinates 761
Then
oisb oxi o2 xj oxk
isa ¼ cijk is ;
osa osa osa osb osc c
but
where we took into account (A.129) and (A.133). The relation (A.141) becomes
oisb 1 ohb
isa ¼ cabc is ð!Þ;
osa hb osa c
with summation with respect to a and c: ðA:1410 Þ
Finally, we get
X
3
oVsb 1 ohb
curlV ¼ cabc Vs isc
b¼1
osa hb osa b
X3
o oVsb
¼ cabc hb isc : ðA:142Þ
b¼1
osa hb
Taking into account the definition formula (A.80) and the differential operators
(A.131) and (A.138), we may write Laplace’s operator in the form
762 Appendix
o2 1 ohb 1 ohc o o2
D¼ þ ¼
osa osa hb osa hc osa osa osa osa
ðA:143Þ
1 oðhb hc Þ o o 1 o
¼ hb hc ð!Þ:
hb hc osa osa osa hb hc osa
1o o 1 o2 o2
D¼ r þ 2 2þ 2: ðA:149Þ
r or or r oh oz
A.2 Curvilinear Co-ordinates 763
1 o 2 o 1 o o 1 o2
D¼ 2 R þ 2 sin u þ 2 2 : ðA:155Þ
R oR oR R sin u ou ou R sin u oh2
exists.
If the function f ðxÞ is absolutely integrable in any finite interval of R, then we
say that f ðxÞ is a locally integrable function. We mention that an absolutely
integrable function is integrable too, i.e. the integral
Zb
f ðxÞdx ðA:1560 Þ
a
exists.
A.3 Elements of the Theory of Distributions 765
where f ðxÞ is a locally integrable function and ½a; b is the support of uðxÞ. It is
easy to see that the functional thus defined is linear and continuous. The functional
defined on the space K by means of the locally integrable function f ðxÞ represents
a distribution on this space which will be denoted by f ðxÞ too, like the generating
function. Such distributions are called regular distributions (distributions of
functional type). Similarly, one can define temperate regular distributions on the
space S.
The distributions which are not regular distributions are called singular
distributions.
If to any function uðxÞ 2 K we attach its value at the origin (the value uð0Þ),
then we see that the respective functional is linear and continuous, hence it is a
distribution which is not regular; this is the Dirac distribution, which will be
denoted by the symbol dðxÞ. We may write
ðdðxÞ; uðxÞÞ ¼ uð0Þ: ðA:158Þ
We can define the Dirac distribution on the space K 0 too; it will be called, in
this case, the Dirac measure.
If the fundamental functions u 2 K are defined in Rn , then we have
ðdðx1 ; x2 ; . . .; xn Þ; uðx1 ; x2 ; . . .; xn ÞÞ ¼ uð0; 0; . . .; 0Þ: ðA:1580 Þ
The relation
ðf ðx x0 Þ; uðxÞÞ ¼ ðf ðxÞ; uðx þ x0 ÞÞ; x; x0 2 Rn ; ðA:159Þ
defines the translated distribution f ðx x0 Þ. In particular, for the Dirac
distribution dðx x0 Þ we may write
ðdðx x0 Þ; uðxÞÞ ¼ uðx0 Þ; x; x0 2 Rn ; ðA:1590 Þ
For the distributions subjected to a homothetic transformation with respect to
the independent variable, we shall use—by definition—the formula
ðf ðax; uðxÞÞ ¼ jajn ðf ðxÞ; uðx=aÞÞ; x 2 Rn : ðA:160Þ
The equality of two distributions f ðxÞ and gðxÞ is defined by the relation
ðf ; uÞ ¼ ðg; uÞ; 8u 2 K; ðA:162Þ
hence, we may write
f ¼ g: ðA:163Þ
If the distributions f and g are generated by continuous functions f ðxÞ and gðxÞ,
then the equality (A.163) occurs in the usual sense, i.e. punctual, because—in this
case—the distributions f and g coincide everywhere with the functions f and g. If
the functions f and g are locally integrable and coincide almost everywhere, then
the distributions generated by them will be equal and the relation (A.163) holds.
A distribution f ðxÞ is equal to zero (f ¼ 0) if, for any fundamental function
uðxÞ, we have ðf ðxÞ; uðxÞÞ ¼ 0.
We define the Heaviside function (the unit function) on the real axis in the form
(Fig. A.10)
1; x 0;
hðxÞ ¼ ðA:164Þ
0; x \ 0;
O x
O x O x
(a) (b)
A.3 Elements of the Theory of Distributions 767
O x
In particular, for the function f ðxÞ ¼ x, we can introduce the positive part
(Fig. A.11b)
x; x 0;
xþ ¼ xhðxÞ ¼ ðA:1650 Þ
0; x \ 0;
being thus led to a distribution of function type; this distribution may be expressed
also by means of the Heaviside distribution
hðxÞ ¼ hðx þ aÞ hðx aÞ ¼ hða jxjÞ; x 2 R: ðA:1660 Þ
If wðxÞ is a function of class C1 , then we can write the equality
wðxÞdðxÞ wð0ÞdðxÞ; x 2 R; ðA:167Þ
in particular, for wðxÞ ¼ xn , we obtain
xn dðxÞ ¼ 0; n ¼ 1; 2; . . . ðA:168Þ
f2
f1
O x
From the very beginning, we mention that the distributions admit derivatives of
any order, which is a great advantage with respect to usual functions.
Let f ðxÞ be a function of class C1 and uðxÞ a fundamental function belonging to
the fundamental space K; considering the corresponding distribution of function
type, we obtain the rule of differentiation in the form
ðf 0 ; uÞ ¼ ðf ; u0 Þ: ðA:170Þ
In particular, we have
h0 ðxÞ ¼ dðxÞ: ðA:171Þ
In case of a distribution of several variables f ðx1 ; x2 ; . . .; xn Þ we can write
o
f ðx1 ; x2 ; :::; xn Þ; uðx1 ; x2 ; :::; xn Þ
oxi
o ðA:172Þ
¼ f ðx1 ; x2 ; :::; xn Þ; uðx1 ; x2 ; :::; xn Þ ; i ¼ 1; 2; :::; n;
oxi
one obtains also the property
o2 f o2 f
¼ ; i; j ¼ 1; 2; . . .; n; ðA:173Þ
oxi oxj oxj oxi
which shows that, in case of distributions, the derivatives do not depend on the
order of differentiation.
A.3 Elements of the Theory of Distributions 769
In general, the product of two distributions has no meaning; we have seen that the
products of a distribution by a function of class C1 has sense. That is why we will
define products of a special type (composition of distributions). We introduce thus
the direct (or tensor) product and the convolution product.
The direct product f ðxÞ gðyÞ of two distributions f ðxÞ and gðyÞ; defined on the
basic spaces Kx ðx 2 X n Þ and Ky ðy 2 Y m Þ; respectively, is given by the relation
ðf ðxÞ gðyÞ; uðx; yÞÞ ¼ ðf ðxÞ; ðgðyÞ; uðx; yÞÞÞ; ðA:178Þ
where uðx; yÞ is a fundamental function defined on X n Y m ; this product is a
distribution defined on the fundamental space Kx Ky . In the case of usual
functions, this product coincides with their usual product.
The direct product is commutative
f ðxÞ gðyÞ ¼ gðyÞ f ðxÞ ðA:179Þ
and associative
½f ðxÞ gðyÞ hðzÞ ¼ f ðxÞ ½gðyÞ hðzÞ ¼ f ðxÞ gðyÞ hðzÞ: ðA:180Þ
The first of these properties allows to write the definition relation (A.178) also in
the form
ðf ðxÞ gðyÞ; uðx; yÞÞ ¼ ðgðyÞ; ðf ðxÞ; uðx; yÞÞÞ: ðA:1780 Þ
The second property shows that the direct product may be defined for an arbitrary
finite number of distributions.
Let Dx and Dy be two differential operators with respect to the variables x and y,
respectively; we may write the relation
Dx Dy ½f ðxÞ gðyÞ ¼ Dx f ðxÞ Dy gðyÞ: ðA:181Þ
In particular, we get
o2 ohðxÞ ohðyÞ
½hðxÞ hðyÞ ¼ ¼ dðxÞ dðyÞ ¼ dðx; yÞ: ðA:182Þ
oxoy ox oy
Let f ðxÞ and gðxÞ be locally integrable functions of x; their convolution product is
the function defined by
Z1
f ðxÞ gðxÞ ¼ f ðnÞgðx nÞdn; ðA:183Þ
1
sufficient condition in this respect is that the support of the two functions f ðxÞ and
gðxÞ be compact.
If f ðxÞ and gðxÞ are two distributions on Rn , then their convolution product
f ðxÞ gðxÞ represents a new distribution on Rn , defined by the formula
ðf ðxÞ gðxÞ; uðxÞÞ ¼ ðf ðxÞ gðyÞ; uðx þ yÞÞ
¼ ðf ðxÞ; ðgðyÞ; uðx þ yÞÞÞ ¼ ðgðyÞ; ðf ðxÞ; uðx þ yÞÞÞ; ðA:184Þ
this definition is reduced to the first one in the case of usual functions. We may
show that the convolution product has a meaning if one of the following conditions
is satisfied:
1. one of the distributions f ðxÞ; gðxÞ has a compact support;
2. the distributions f ðxÞ and gðxÞ have the support bounded on the same side;
thus, if f ðxÞ ¼ 0 for x\a and gðxÞ ¼ 0 for x\b, then the supports of the two
distributions are bounded on the same side.
We remark that the convolution product may be defined for an arbitrary finite
number of distributions. Under the conditions required for the existence of the
convolution product, one may prove the property of commutativity
f ðxÞ gðxÞ ¼ gðxÞ f ðxÞ ðA:185Þ
and the property of associativity
½f ðxÞ gðxÞ hðxÞ ¼ f ðxÞ ½gðxÞ hðxÞ ¼ f ðxÞ gðxÞ hðxÞ: ðA:186Þ
We notice that
dðxÞ f ðxÞ ¼ f ðxÞ dðxÞ ¼ f ðxÞ; ðA:187Þ
hence, Dirac’s distribution is a unit element for the convolution product.
If D is an arbitrary differential operator, then we may write
D½f ðxÞ gðxÞ ¼ Df ðxÞ gðxÞ ¼ f ðxÞ DgðxÞ: ðA:188Þ
A strong tool for the integration of differential equations is the method of integral
transforms. We give, in the following, some general results concerning Fourier and
Laplace transforms.
772 Appendix
exists and is called the Fourier transform of the function f ðxÞ; we shall write
noting that the variable u is real. In general, the image function FðuÞ is complex,
although the function f ðxÞ may be a real function.
Assuming that the function FðuÞ is given, the equality (A.189) may be
considered as an integral equation with respect to the unknown function f ðxÞ under
the integral symbol; the solution of this integral equation is written in the form
Z1
1
f ðxÞ ¼ FðuÞ eiux du: ðA:190Þ
2p
1
The function f ðxÞ is called the inverse Fourier transform of the function FðuÞ; we
have
The function wðuÞ may be definite for complex values s ¼ u þ iv too, namely
Z1 Z1
wðsÞ ¼ isx
uðxÞ e dx ¼ uðxÞ evx eiux dx: ðA:1910 Þ
1 1
The set of functions wðsÞ ¼ F½uðxÞ, where the support of the fundamental
functions is included in the segment ½a; a, forms the vector space ZðaÞ. We
denote by
Z ¼ [ ZðaÞ; K ¼ [ KðaÞ ðA:192Þ
a a
A.3 Elements of the Theory of Distributions 773
0
the new complex linear space; then, Z is the set of linear and continuous
functional defined on Z (ultradistributions).
If FðsÞ is a distribution defined on Z and f ðxÞ is a distribution defined on K, then
0
the functional FðsÞ 2 Z specified by the equality of Parseval type
ðFðsÞ; wðsÞÞ ¼ 2pðf ðxÞ; uðxÞÞ; ðA:193Þ
is called the Fourier transform of the distribution f ðxÞ and is denoted by
d d
P FðsÞ ¼ P F½f ðxÞ ¼ F½PðixÞf ðxÞ; P polynomial; ðA:195Þ
ds ds
d
F P f ðxÞ ¼ PðisÞF½f ðxÞ ¼ PðisÞFðsÞ; P polynomial; ðA:196Þ
dx
1
F½hðxÞ ¼ pdðuÞ þ ; ðA:202Þ
u
1 0
F½xþ ¼ ipd ðuÞ; ðA:203Þ
u2
F½1ðxÞ ¼ 2pdðuÞ; ðA:204Þ
where 1ðxÞ is the function which takes the value 1 for any x.
Let f ðxÞ be a complex function of a real variable, which satisfies the conditions:
1. f ðxÞ ¼ 0 for x\0,
2. f ðxÞ is piecewise differentiable,
3. jf ðxÞj
Meax , where M is a positive constant, while the non-negative constant a
represents the incremental ratio of the function.The function LðpÞ of the
complex variable p ¼ u þ iv, defined by the expression
Z1
LðpÞ ¼ f ðxÞepx dx; ðA:205Þ
0
If f ðxÞ is a distribution having its support on the half-line x 0 and is such that
the distribution f ðxÞepx is a temperate distribution, then
represents the Laplace transform of that distribution. It is obvious that the relation
(A.207) generalizes the relation (A.205)
One verifies easily the delay theorem
L½f ðx aÞ ¼ epa L½f ðxÞ; ðA:208Þ
1 p
L½f ðkxÞ ¼ L ; k [ 0; ðA:209Þ
k k
and the theorem of translation (the damping theorem)
L½f ðxÞeqx ¼ Lðp qÞ: ðA:210Þ
In the case of a derivative of a distribution one may write
L½f 0 ðxÞ ¼ pL½f ðxÞ: ðA:211Þ
One may give analogous results for the distributions of several variables.
For a convolution product it results
L½f ðxÞ gðxÞ ¼ L½f ðxÞL½gðxÞ: ðA:212Þ
We may write the Laplace transforms
L½dðxÞ ¼ 1: ðA:213Þ
L½dðx1 ; x2 ; . . .; xn Þ ¼ 1: ðA:2130 Þ
DYðxÞ ¼ 0; ðA:218Þ
while Eþ ðxÞ is a particular fundamental solution {corresponding to the non-
homogeneous equations (A.216)}. We shall give a simple method for determining
this solution; to this end, we determine first the solution YðxÞ of the homogeneous
equation, which verifies the initial conditions
00
Yð0Þ ¼ Y 0 ð0Þ ¼ Y ð0Þ ¼ ¼ Y ðn2Þ ð0Þ ¼ 0; Y ðn1Þ ð0Þ ¼ 1; ðA:219Þ
one can prove that a fundamental particular solution is, in this case, given by
Eþ ðxÞ ¼ hðxÞYðxÞ ðA:220Þ
The fundamental solution of a differential equation is useful for the determination
of the general solution; thus, the general solution of the Eq. (A.215) is given by
yðxÞ ¼ EðxÞ f ðxÞ: ðA:221Þ
Let now be again the Eq. (A.215) with x 0, f ðxÞ being a continuous function
with the support in ½0; 1Þ; in the case of initial conditions of Cauchy type
xn1 1 n1
Eþ ðxÞ ¼ hðxÞ ¼ x ; x 2 R; ðA:228Þ
ðn 1Þ! ðn 1Þ! þ
the solution of the boundary value problem is given by
x2 xn1
yðxÞ ¼ y0 þ y1 x þ y2 þ þ yn1
2 ðn 1Þ!
Zx
1
þ ðx nÞn1 f ðnÞ dn; ðA:229Þ
ðn 1Þ!
0
where the latter integral, which represents the convolution product, is known as the
Cauchy formula; for n 2, we get
Eþ ðxÞ ¼ xþ : ðA:230Þ
The above ideas may be extended to systems of ordinary differential equations
with constant coefficients.
Problems similar to those in the preceding subsection may be put in the case of
partial differential equations. Let thus be
o o o o
P ; ; . . .; ; uðx1 ; x2 ; . . .; xm ; tÞ ¼ 0 ðA:231Þ
ox1 ox2 oxm ot
a homogeneous linear partial differential equation of nth order with respect to the
variable t, with constant coefficients. For instance, the Cauchy problem for this
equation consists in the determination of the function uðx1 ; x2 ; . . .; xm ; tÞ which
satisfies the Eq. (A.231) and the initial conditions
uðx1 ; x2 ; . . .; xm ; t0 Þ ¼ u0 ðx1 ; x2 ; . . .; xm Þ;
o
uðx1 ; x2 ; . . .; xm ; t0 Þ ¼ u1 ðx1 ; x2 ; . . .; xm Þ;
ot
. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . . ðA:232Þ
on1
uðx1 ; x2 ; . . .; xm ; t0 Þ ¼ un1 ðx1 ; x2 ; . . .; xm Þ:
otn1
To solve this problem, let us consider the function
uðx1 ; x2 ; . . .; xm ; tÞ ¼ uðx1 ; x2 ; . . .; xm ; tÞhðt t0 Þ;
ðA:233Þ
as well as the corresponding regular distribution; taking into account the formula
which links the derivative in the sense of the theory of distributions to the
derivative in the usual sense and using the initial conditions (A.232), it results
778 Appendix
o ~
o
uðx1 ; x2 ; . . .; xm ; tÞ ¼ uðx1 ; x2 ; . . .; xm ; tÞ
ot ot
þu0 ðx1 ; x2 ; . . .; xm Þdðt t0 Þ;
o 2 ~
o2
u ðx 1 ; x 2 ; . . .; x m ; tÞ ¼ uðx1 ; x2 ; . . .; xm ; tÞ
ot2 ot2
þu1 ðx1 ; x2 ; . . .; xm Þdðt t0 Þ þ u0 ðx1 ; x2 ; . . .; xm Þdðt _ t0 Þ;
ðA:234Þ
. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .
on ~
on
u ðx 1 ; x 2 ; . . .; xm ; tÞ ¼ uðx1 ; x2 ; . . .; xm ; tÞ
otn otn
þun1 ðx1 ; x2 ; . . .; xm Þdðt t0 Þ þ un2 ðx1 ; x2 ; . . .; xm Þdðt _ t0 Þ
þ. . . þ u0 ðx1 ; x2 ; . . .; xm Þdðn1Þ ðt t0 Þ:
Noting that the derivatives in the usual sense with respect to the temporal variable
of the function uðx1 ; x2 ; . . .; xm ; tÞ are equal to the corresponding ones of the
function uðx1 ; x2 ; . . .; xm ; tÞ for t t0 , the Eq. (A.231) takes the form
o o o o
P ; ; . . .; ; uðx1 ; x2 ; . . .; xm ; tÞ ¼ f ðx1 ; x2 ; . . .; xm ; tÞ;
ðA:2310 Þ
ox1 ox2 oxm ot
in distributions, where f ðx1 ; x2 ; . . .; xm ; tÞ is a given distribution, which contains the
initial conditions considered above.
Thus, we call fundamental solution of the Eq. (A.2310 ) the distribution
Eðx1 ; x2 ; . . .; xm ; tÞ which satisfies the equation
o o o o
P ; ; . . .; ; Eðx1 ; x2 ; . . .; xm ; tÞ ¼ dðx1 ; x2 ; . . .; xm ; tÞ: ðA:235Þ
ox1 ox2 oxm ot
The solution of the above Cauchy problem is given by (A.233) where
uðx1 ; x2 ; . . .; xm ; tÞ ¼ Eðx1 ; x2 ; . . .; xm ; tÞ f ðx1 ; x2 ; . . .; xm ; tÞ;
ðA:236Þ
the convolution product corresponding to all m þ 1 variables.
It should be noted that some equations of mathematical physics cannot be
always deduced directly in the space of distributions, owing to the difficulties
encountered in modelling physical phenomena. In general, the equations which
describe such phenomena are obtained first by classical method. Next, an
extension is effected, where the unknown functions take zero values, so that they
be defined on the whole space; the derivatives, considered in the usual sense, are
expressed by relations which connect derivatives in the sense of the theory of
distributions to the derivatives in the usual sense of a distribution corresponding to
an almost everywhere continuous function, having a finite number of
discontinuities of the first species. In this way, the unknowns of the problem
will be regular distributions; then it will be assumed that these unknowns may be
arbitrary distributions. Another possibility, which is frequently used, is to suppose
A.3 Elements of the Theory of Distributions 779
from the very beginning, that the unknowns of the problem are arbitrary
distributions, assuming the same form in distributions for the differential equation
obtained by classical methods (obviously, these ones are no longer valid for the
whole space). However, there are not general methods for passing to differential
equations in distributions.
1 R
Ei ðx1 ; x2 ; x3 ; tÞ ¼ d t ; i ¼ 1; 2: ðA:2450 Þ
4pR ci
1 R
uðx1 ; x2 ; x3 ; tÞ ¼ , t ; i ¼ 1; 2: ðA:2460 Þ
R ci
The solution of the equation
h1 h2 uðx1 ; x2 ; x3 ; tÞ þ 4p,ðtÞ dðx1 ; x2 ; x3 Þ ¼ 0 ðA:247Þ
A.3 Elements of the Theory of Distributions 781
reads
c2 c2 R R
uðx1 ; x2 ; x3 ; tÞ ¼ 2 1 2 2 t t ,ðtÞ; ðA:2470 Þ
c1 c2 c1 þ c2 þ ðtÞ
where K0 is the modified Bessel function of order zero, we obtain the fundamental
solutions
1 r
Eðx1 ; x2 ; tÞ ¼ f0 t; ; i ¼ 1; 2; ðA:2490 Þ
2p ci
for the Eq. (A.249). As well, if we introduce the distribution defined by the
function
r 1 1 r
f2 t; ¼L K0 p
ci p2 ci
" sffiffiffiffiffiffiffiffiffiffiffiffiffiffi! sffiffiffiffiffiffiffiffiffiffiffiffiffiffi#
r r2 r2
¼h t t ln t þ t2 2 t2 2 ; t [ 0; i ¼ 1; 2;
ci ci ci
ðA:252Þ
c2 c2 r r
Eðx1 ; x2 ; tÞ ¼ 21 2 2 f2 t; f2 t; ; ðA:2500 Þ
2p c1 c2 c1 c2
We give, in the following, some functions used in the study of the problems of the
elastic parallelepiped to simplify the representations of the results. As well, we
give final forms for some integrals which appear in the formulae thus obtained.
A.4 Notations and Integrals 783
A.4.1 Notations
We present the functions U; W and W0 which appear in the study of the problems
specified above and also put in evidence the differential relations which exist
between them.
where 0
m
1=2 is Poisson’s ratio.
hs
vðhsÞ ¼ W2 ðhs; hsÞ ¼ coth hs þ hs: ðA:261Þ
sinh2 hs
hs
v0 ðhsÞ ¼ W02 ðhs; hsÞ ¼ tanh hs hs: ðA:265Þ
cosh2 hs
A.4.2 Integrals
We shall give some results concerning integrals of rational functions and of
rational functions and radicals; as well, we give results for double Fourier
integrals.
Z1
dx p
¼ ; ðA:2660 Þ
ðp2 þ x2 Þ2 4p3
0
Z1
x2 dx p
¼ ; ðA:267Þ
ð p2 þ x 2 Þ2 4p
0
Z1
xdx p
¼ ; ðA:2670 Þ
ðp2 þ x2 Þ2 2p2
0
Z1
dx p
¼ ; ðA:268Þ
ðp2 þ x2 Þðq2 þ x2 Þ 2pqðp þ qÞ
0
Z1
x2 dx p
¼ ; ðA:2680 Þ
ðp2 þ x2 Þðq2 þ x2 Þ 2pðp þ qÞ
0
786 Appendix
Z1
dx pð2p þ qÞ
¼ ; ðA:269Þ
ð p2 þ x2 Þ2 ðq2 þ x2 Þ 4p3 qðp þ qÞ2
0
Z1
x2 dx p
¼ ; ðA:2690 Þ
ð p2 þ x 2 Þ 2 ð q2 þ x2 Þ 4pðp þ qÞ2
0
Z1
x4 dx pðp þ 2qÞ
¼ ; ðA:26900 Þ
ð p2 þ x 2 Þ 2 ð q2 þ x2 Þ 2ðp þ qÞ2
0
Z1
x2 dx p
¼ ; ðA:270Þ
ðp2 þ x2 Þ2 ðq2 þ x2 Þ2 4pqðp þ qÞ3
0
Z1
x4 dx p
¼ : ðA:2700 Þ
ð p2 þ x2 Þ2 ðq2 þ x2 Þ2 4ðp þ qÞ3
0
Z1
dx 2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 4 ; ðA:2710 Þ
ð p2 þ x2 Þ2 2
p þx 2 3p
0
Z1
dx 1 pþq
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ ln ; ðA:272Þ
ð p2 þ x2 Þ 2
q þx 2 2pq p q
0
Z1
dx 1 1 pþq
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2 2 þ ln ðp 6¼ qÞ; ðA:2720 Þ
ð p2 þ x 2 Þ 2 q2 þ x2 2p ðp q2 Þ 4p3 q p q
0
Z1
x2 dx 1 1 pþq
2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2 q2 Þ
þ ln ðp 6¼ qÞ; ðA:27200 Þ
2 2 2
ðp þ x Þ q þ x 2 2ðp 4pq pq
0
A.4 Notations and Integrals 787
Z1
dx
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð p2 þ x 2 Þ ð q2 þ x 2 Þ r 2 þ x 2
0
1 1 pþr 1 qþr
¼ ln ln ; p 6¼ r; q 6¼ r; ðA:273Þ
2rðp2 q2 Þ p p r q q r
Z1
dx 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2 2
ð p2 þ x 2 Þ 2 ð q2 þ x2 Þ 2
r þx 2 2p ðp q 2 Þðp2 r 2 Þ
0
3p2 q2 pþr 1 qþr
ln þ ln ; p 6¼ r; q 6¼ r; ðA:274Þ
4p3 ðp2 q2 Þ2 r p r 2qrðp2 q2 Þ2 q r
Z1
x2 dx 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼
ð p2 þ x2 Þ2 ðq2 þ x2 Þ r 2 þ x2 2ðp2 q2 Þðp2 r 2 Þ
0
p2 þ q2 pþr q qþr
þ ln ln ; p 6¼ r; q 6¼ r; ðA:2740 Þ
4prðp2 q2 Þ 2 p r 2rðp q Þ
2 2 2 qr
Z1
x2 dx p2 þ q2 2r 2
p ffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼
ðp2 þ x2 Þ2 ðq2 þ x2 Þ2 r 2 þ x2 2ðp2 q2 Þðp2 r 2 Þðq2 r 2 Þ
0
2
1 3p þ q2 p þ r 3q2 þ p2 q þ r
ln ln ; p 6¼ r; q 6¼ r;
4rðp2 q2 Þ3 p pr q qr
ðA:275Þ
Z1
x4 dx p4 ðp2 þ q2 Þr 2 þ q4
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼
ð p2 þ x2 Þ2 ðq2 þ x2 Þ2 r 2 þ x2 2ðp2 q2 Þ2 ðp2 r 2 Þðq2 r 2 Þ
0
1 pþr
þ 3
pðp2 þ 3q2 Þ ln
2 2
4rðp q Þ pr
qþr
qðq2 þ 3p2 Þ ln ; p 6¼ r; q 6¼ r; ðA:2750 Þ
qr
Z1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dx 1 p2 þ q2 þ q
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ln pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; p 6¼ 0; ðA:276Þ
2 2 2 2
ðp þ q þ x Þ p þ x 2 2
2q p þ q 2 p2 þ q2 q
0
Z1
x2 dx 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2 ; ðA:277Þ
ð p2 þ x 2 Þ ð q2 þ x 2 Þ q2 þ x 2 p q2
0
788 Appendix
Z1
x2 dx 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ ; ðA:2770 Þ
ð p2 þ x2 Þ2 ðq2 þ x2 Þ 2
q þx 2 ð p q2 Þ 2
2
0
as well as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r ¼ x1 þ x2 þ x3 ; r3 ¼ x21 þ x22
2 2 2 ðA:279Þ
for the radii, one obtains, for some double Fourier integrals, the formulae
Z1 Z1
2 x3
ecx3 cos ax1 cos bx2 dadb ¼ ; ðA:280Þ
p r3
0 0
Z1 Z1 x 2
2 1 3
cecx3 cos ax1 cos bx2 dadb ¼ 1 3 ; ðA:281Þ
p r3 r
0 0
Z1 Z1
2 1
c cos ax1 cos bx2 dadb ¼ ; ðA:2810 Þ
p r33
0 0
Z1 Z1
2 1 cx3 1
e cos ax1 cos bx2 dadb ¼ ; ðA:282Þ
p c r
0 0
Z1 Z1
2 1 1
cos ax1 cos bx2 dadb ¼ ; ðA:2820 Þ
p c r3
0 0
Z1 Z1
2 x3 x1
aecx3 sin ax1 cos bx2 dadb ¼ 3 ; ðA:283Þ
p r5
0 0
Z1 Z1
2 x2 x3
becx3 cos ax1 sin bx2 dadb ¼ 3 ; ðA:2830 Þ
p r5
0 0
A.4 Notations and Integrals 789
Z1 Z1
2 a cx3 x1
e sin ax1 cos bx2 dadb ¼ 3 ; ðA:284Þ
p c r
0 0
Z1 Z1
2 b cx3 x2
e cos ax1 sin bx2 dadb ¼ 3 ; ðA:2840 Þ
p c r
0 0
Z1 Z1
2 a x1
sin ax1 cos bx2 dadb ¼ 3 ; ðA:285Þ
p c r3
0 0
Z1 Z1
2 b x2
cos ax1 sin bx2 dadb ¼ 3 ; ðA:2850 Þ
p c r3
0 0
Z1 Z1 x 2
2 cx3 x1 3
ace sin ax1 cos bx2 dadb ¼ 3 5 1 5 ; ðA:286Þ
p r r
0 0
Z1 Z1 x 2
2 cx3 x2 3
bce cos ax1 sin bx2 dadb ¼ 3 5 1 5 ; ðA:2860 Þ
p r r
0 0
Z1 Z1
2 x1
ac sin ax1 cos bx2 dadb ¼ 3 ; ðA:287Þ
p rx53
0 0
Z1 Z1
2 x2
bc cos ax1 sin bx2 dadb ¼ 3 ; ðA:2870 Þ
p rx53
0 0
Z1 Z1 x 2
2 2 cx3 x3 1
ae cos ax1 cos bx2 dadb ¼ 3 5 1 5 ; ðA:288Þ
p r r
0 0
Z1 Z1 x 2
2 x3 2
b2 ecx3 cos ax1 cos bx2 dadb ¼ 3 1 5 ; ðA:2880 Þ
p r5 r
0 0
Z1 Z1
2 x1 x2 x3
abecx3 sin ax1 sin bx2 dadb ¼ 15 ; ðA:28800 Þ
p r7
0 0
790 Appendix
Z1 Z1
2 a cx3 x1
e sin ax1 cos bx2 dadb ¼ ; ðA:289Þ
p c2 rðr þ x3 Þ
0 0
Z1 Z1
2 b cx3 x2
e cos ax1 sin bx2 dadb ¼ ; ðA:2890 Þ
p c2 rðr þ x3 Þ
0 0
Z1 Z1
2 a x1
2
sin ax1 cos bx2 dadb ¼ 2 ; ðA:290Þ
p c rx 3
0 0
Z1 Z1
2 b x2
cos ax1 sin bx2 dadb ¼ 2 ; ðA:2900 Þ
p c2 rx 3
0 0
Z1 Z1
2
2 a ðr 2 x21 Þðr þ x3 Þ rx21
ecx3 cos ax1 cos bx2 dadb ¼ ; ðA:291Þ
p c r 3 ðr þ x3 Þ2
0 0
Z1 Z1
2
2 b ðr 2 x22 Þðr þ x3 Þ rx22
ecx3 cos ax1 cos bx2 dadb ¼ ; ðA:2910 Þ
p c r 3 ðr þ x3 Þ2
0 0
Z1 Z1
2 ab cx3 2ðr þ x3 Þx1 x2
e sin ax1 sin bx2 dadb ¼ ; ðA:292Þ
p c2 r 3 ðr þ x3 Þ2
0 0
Z1 Z1
2 ab cx3 3x1 x2
e sin ax1 sin bx2 dadb ¼ 5 ; ðA:2920 Þ
p c r
0 0
Z1 Z1
2
2 a x2 x2
cos ax1 cos bx2 dadb ¼ 1 4 2 ; ðA:293Þ
p c r3
0 0
Z1 Z1
2
2 b x21 x22
cos ax1 cos bx2 dadb ¼ ; ðA:2930 Þ
p c r34
0 0
Z1 Z1
2 ab 2x1 x2
sin ax1 sin bx2 dadb ¼ 4 ; ðA:294Þ
p c2 r3
0 0
A.4 Notations and Integrals 791
Z1 Z1
2 ab 3x1 x2
sin ax1 sin bx2 dadb ¼ 5 ; ðA:2940 Þ
p c r3
0 0
Z1 Z1 x 2
2 a2 cx3 1 1
e cos ax1 cos bx2 dadb ¼ 3 1 3 ; ðA:295Þ
p c r r
0 0
Z1 Z1 x 2
2 b2 cx3 1 2
e cos ax1 cos bx2 dadb ¼ 3 1 3 ; ðA:2950 Þ
p c r r
0 0
Z1 Z1
2 a2 x2 x2
cos ax1 cos bx2 dadb ¼ 1 5 2 ; ðA:296Þ
p c r3
0 0
Z1 Z1
2 b2 x2 x2
cos ax1 cos bx2 dadb ¼ 1 5 2 ; ðA:2960 Þ
p c r3
0 0
Z1 Z1 x 2
2 2 cx3 x3 3
c e cos ax1 cos bx2 dadb ¼ 3 5 3 5 : ðA:297Þ
p r r
0 0
Z1 Z1 x 2
2 x3 3
ð1 cx3 Þecx3 cos ax1 cos bx2 dadb ¼ 2 3 ; ðA:2980 Þ
p r3 r
0 0
Z1 Z1
2
cð1 þ cx3 Þecx3 cos ax1 cos bx2 dadb
p
0 0
x 2 x 4
1 3 3
¼ 3 1þ6 15 ; ðA:299Þ
r r r
Z1 Z1
2
cð1 cx3 Þecx3 cos ax1 cos bx2 dadb
p
0 0
x 2 x 4
1 3 3
¼ 1 12 þ15 ; ðA:2990 Þ
r3 r r
792 Appendix
Z1 Z1
2
2 a
ð1 cx3 Þecx3 cos ax1 cos bx2 dadb
p c
0 0
" #
1 x 2 ðr 2 x2 Þðr þ x Þ rx2
1 2 3 2
¼ 3 3x3 ; ðA:300Þ
r r ðr þ x3 Þ2
Z1 Z1
2
2 b
ð1 cx3 Þecx3 cos ax1 cos bx2 dadb
p c
0 0
" #
1 x 2 ðr 2 x2 Þðr þ x Þ rx2
2 3
¼ 3 3x3 1 1
; ðA:3000 Þ
r r ðr þ x3 Þ2
Z1 Z1
2 a2
ð1 cx3 Þecx3 cos ax1 cos bx2 dadb
p c
0 0
x 2
1 2 x23 x21
¼ 3 23 15 4 ; ðA:301Þ
r r r
Z1 Z1
2 b2
ð1 cx3 Þecx3 cos ax1 cos bx2 dadb
p c
0 0
x 2
1 1 x2 x2
¼ 3 23 15 2 4 3 : ðA:3010 Þ
r r r
Subject Index
A C
Anisotropy, 138, 624 Computation methods, 266
axis of symmetry, 627 direct, 280
dynamics problems, 634 fundamental solutions, 288
general case, 138 Fourier representations, 272
elastic constants, 619 in elastodynamics, 295
orthotropy, 142, 622 in elastostaties, 292
plane of symmetry, 141, 625 particular integrals, 266
transverse isotropy, 619 polynomials, 269
volume forces, 632 wave equations, 276
Antiplane problems, 604 point matching, 302
state of strain, 608 successive approximations, 301
state of stress, 567 variational, 281
Bubnov-Galerkin, 287
least squares, 284
B Ritz, 286
Basic problems, 192 Concentrated loads, 357
first, 192 Cosserat type solution, 383
limit conditions, 192 centre of dilatation, 386
boundary, 192 directed, 385
initial, 197 force, 384
mixed, 194 moments, 385
second, 193 rotational, 386
Body, 8 elastic solutions, 377
bar, 8 dynamic, 380
block, 10 static, 376
connection, 262 force, 358
multiply, 263 system of, 360
simple, 262 moment, 361
distorsion, 264 dipole type, 367
incompressible, 150 directed, 361
plate, 9 linear, 367
P
F Particular states, 547
Force, 73 of strain, 592
lost, 78 a zero linear strain, 592
superficial, 73 two zero angular strains, 602
volume, 76 of stress, 548
a zero normal stress, 548
two zero tangential
H stresses, 577
History, 15 straight cylinder, 587
deformable solids, 14 Plane problems, 582
elastic solids, 18 state of strain, 597
dynamic, 605
static, 604
I state of stress, 573, 582, 604
Inelastic bodies, 155 Principles, 248
hardening material, 162 Clapeyron, 247
plastic models, 156 d’Alembert-Lagrange, 249
theories, 158 Hamilton, 251
viscoelastic models, 184 minimum internal work, 247
Burgers, 182 minimum potential energy, 252
creep, 170 reciprocity, 254
generalized; Kelvin, Maxwell, 177, 178 Betti, 254
grouping; Kelvin–Voigt, Maxwell, Maxwell, 255
relaxation, 171, 174, 175, 180 Saint–Venant, 260
Integrals, 785 virtual variations, 251
796 Subject Index
S classification, 764
Strain, 39 composition, 769
angular, 40 differential equations, 775
curvilinear co-ordinates, 65 elliptic, 779
cylindrical, 66 hyperbolic, 780
spherical, 67 ordinary, 775
initial, 155 parabolic, 782
linear, 39 partial, 777
octahedral, 51 differentiation, 768
principal, 48 integral transforms, 771
tensor, 45 Fourier, 772
volume, 42, 56 Laplace, 774
Strength of materials, 2 product, 769
Stress, 73 convolution, 770
Cauchy’s theorem, 79 direct, 769
curvilinear co-ordinates, 108 Thermoelasticity, 671
cylindrical, 110 applications, 685
spherical, 111 axial symmetry, 688
deviator, 92 elementary, 685
elementary states, 98 plane problems, 693
ellipsoid, 90 coupled problems, 684
equilibrium, 95 general equations, 672
finite, 108 in displacements, 674, 678
initial, 154 in stresses, 677, 679
Mohr’s circles, 90 heat conduction, 671
motion, 97 Maysel’s formulae, 683
octahedral, 92 reciprocity principle, 682
principal, 85 Thick plate, 505
symmetry, 82
tensor, 84
Cauchy, 105 V
Piola–Kirchhoff, 103 Viscoelasticity, 699
variation around a point, 78 applications, 720
vector, 73 complex modulus, 707
normal, 75 one-dimensional case, 710
tangential, 75 three-dimensional case, 712
constitutive laws, 700
creep, 700
T formulation, 715
Tensor calculus, 729 in displacements, 715
operators, 735 principle of correspondence, 717
algebraic, 735 quasi-static problems, 719
analytic, 740 relaxation, 700
integral formulae, 748
orthogonal affine, 729
differential operators of first W
and second order, 741, 746 Work, 124, 243
Theorems, 257 elementary, 124
existence, 257 external, 244
uniqueness, 257 formulae of Castigliano, 247
Kirchhoff, 257 formulae of Green, 246
Neumann, 265 of deformation, 245
Theory of distributions, 763
Author Index
Burger, 182 E
Burzynski, W., 308 Eason, G., 529
Butty, H., 29 Engesser, Friedrich, 19
Ericksen, J. L., 309
Eringen, Cemal A., 309, 310
C Essmann, U., 311
Cagniard, L., 531 Eubanks, R. A., 195, 198, 224, 225, 344, 345
Carathéodory, 130 Euler, Leonhard, 16, 18, 35, 38, 101, 102, 124,
Carlslaw, H. C., 671 283, 284, 646
Carlslaw, M. S., 188, 671, 698 Ewing, W. M., 529
Carlson, D. E., 348
Castigliano, Alberto, 19, 132, 135, 246,
247, 287 F
Cauchy, Augustin-Louis, 17, 38, 46, 51, 79, Fedorov, E. S., 525, 637
82, 104, 120, 198, 308 Filipescu, Gheorghe Emanoil, 23
Cerruti, V., 399 Filon, Louis-Napoleon George, 22
Cesàro, Ernesto, 59, 64 Filonenko-Borodich, Mikhail Mitrofanovich,
Chadwick, P. D., 198 22, 479
Chao, C. C., 529 Finzi, Bruno, 195, 205, 208, 209, 233, 235
Chentsov, N. G., 18, 616 Flamant, Alfred Aimé, 20, 21
Chladni, 17 Flint, 647
Clapeyron, Benoît-Paul Émile, 17, 197, Flitman, L. M., 531
248, 250 Flügge, S., 30
Clausis, 129 Fontana, 16
Clebsch, R. F. A., 226, 676 Fourier, François-Marie-Charles, 18, 269, 272,
Clintock, F. A. Mc, 309 278, 672
Cosserat, Eugène-Maurice-Pierre, 22, 25, 310, Föppl, August, 19, 481
346 Föppl, Ludwig, 19, 20, 481
Cosserat, François, 22, 25, 310, 348 Franciosi, V., 30
Coulomb, Charles-Augustin de, 16 Fredholm, Erik Ivar, 18
Cove, Y. Le, 310 Fresnel, 526
Crăciun, Marius E., 613 Fridman, M. M., 633
Cristea, M., 618 Friedrichs, K. O., 259
Cristescu, Dan Nicolae, 613 Fung, V. C., 30
D G
Davidescu Moisil, Ana, 545 Gadolin, Aksel Wilhelmovich, 20, 639
Davidoglu, Anton, 23 Galerkin, Boris Grigorievich, 21, 194, 198,
Davies, R. M., 529 202, 203, 287, 333
Derevianko, N. I., 311 Galileo, Galilei, 15
Dhaliwal, R. S., 390 Galletto, Dionigi, 312
Dinnik, Aleksandr Nicolaevich, 21 Gauss, Karl Friedrich, 17, 64
Dirac, Paul Adrien Maurice, 771 Gazis, D. C., 311
Dixon, R. C., 310 Gegelya, T. G., 257
Djurić, S., 309, 311 Germain, Paul, 30
Dougall, 503 Germain, Sophie, 17
Drăgan, Mircea, 24 Gibbs, Josiah Dixon Willard, 137
Drucker, 160 Golitsyn, Boris Borisovich, 21
Duhamel, Jean-Marie-Constant, 18 Golovin, Harlampii Sergeevich, 20
Duhem, Paul, 197 Goodier, J. N., 675, 685
Author Index 799