Beruflich Dokumente
Kultur Dokumente
DRAFT
Lecture Notes
Introduction to
CONTINUUM MECHANICS
and Elements of
Elasticity/Structural Mechanics
c VICTOR
E. SAOUMA
Draft
02
Victor Saouma
Draft
03
PREFACE
Une des questions fondamentales que lingnieur des Matriaux se pose est de conna le comportee
e
itre
ment dun materiel sous leet de contraintes et la cause de sa rupture. En dnitive, cest prcisment la
e
e e
rponse a c/mat es deux questions qui vont guider le dveloppement de nouveaux matriaux, et dterminer
e
`
e
e
e
leur survie sous direntes conditions physiques et environnementales.
e
Lingnieur en Matriaux devra donc possder une connaissance fondamentale de la Mcanique sur le
e
e
e
e
plan qualitatif, et tre capable deectuer des simulations numriques (le plus souvent avec les Elments
e
e
e
Finis) et den extraire les rsultats quantitatifs pour un probl`me bien pos.
e
e
e
Selon lhumble opinion de lauteur, ces nobles buts sont idalement atteints en trois tapes. Pour
e
e
commencer, ll`ve devra tre confront aux principes de base de la Mcanique des Milieux Continus.
ee
e
e
e
Une prsentation dtaille des contraintes, dformations, et principes fondamentaux est essentiel. Par
e
e
e
e
la suite une briefe introduction a lElasticit (ainsi qu` la thorie des poutres) convaincra ll`ve quun
`
e
a
e
ee
probl`me gnral bien pos peut avoir une solution analytique. Par contre, ceci nest vrai (` quelques
e
e e
e
a
exceptions prts) que pour des cas avec de nombreuses hypoth`ses qui simplient le probl`me (lasticit
e
e
e
e
e
linaire, petites dformations, contraintes/dformations planes, ou axisymmetrie). Ainsi, la troisi`me
e
e
e
e
et derni`re tape consiste en une briefe introduction a la Mcanique des Solides, et plus prcisment
e e
`
e
e e
au Calcul Variationel. A travers la mthode des Puissances Virtuelles, et celle de Rayleigh-Ritz, ll`ve
e
ee
sera enn prt ` un autre cours dlments nis. Enn, un sujet dintrt particulier aux tudiants en
e a
ee
e e
e
Matriaux a t ajout, ` savoir la Rsistance Thorique des Matriaux cristallins. Ce sujet est capital
e
ee
e a
e
e
e
pour une bonne comprhension de la rupture et servira de lien a un ventuel cours sur la Mcanique de
e
`
e
e
la Rupture.
Ce polycopi a t enti`rement prpar par lauteur durant son anne sabbatique a lEcole Polye ee
e
e e
e
`
technique Fdrale de Lausanne, Dpartement des Matriaux. Le cours tait donn aux tudiants en
e e
e
e
e
e
e
deuxi`me anne en Franais.
e
e
c
Ce polycopi a t crit avec les objectifs suivants. Avant tout il doit tre complet et rigoureux. A
e ee e
e
tout moment, ll`ve doit tre ` mme de retrouver toutes les tapes suivies dans la drivation dune
ee
e
a e
e
e
quation. Ensuite, en allant a travers toutes les drivations, ll`ve sera a mme de bien conna les
e
`
e
ee
` e
itre
limitations et hypoth`ses derri`re chaque model. Enn, la rigueur scientique adopte, pourra servir
e
e
e
dexemple a la solution dautres probl`mes scientiques que ltudiant pourrait tre emmen ` rsoudre
`
e
e
e
ea e
dans le futur. Ce dernier point est souvent nglig.
e e
Le polycopi est subdivis de faon tr`s hirarchique. Chaque concept est dvelopp dans un parae
e
c
e
e
e
e
graphe spar. Ceci devrait faciliter non seulement la comprhension, mais aussi le dialogue entres levs
e e
e
e e
eux-mmes ainsi quavec le Professeur.
e
Quand il a t jug ncessaire, un bref rappel mathmatique est introduit. De nombreux exemples
ee
e e
e
sont prsents, et enn des exercices solutionns avec Mathematica sont prsents dans lannexe.
e
e
e
e
e
Lauteur ne se fait point dillusions quand au complet et a lexactitude de tout le polycopi. Il a t
`
e
ee
enti`rement dvelopp durant une seule anne acadmique, et pourrait donc bncier dune rvision
e
e
e
e
e
e e
e
extensive. A ce titre, corrections et critiques seront les bienvenues.
Enn, lauteur voudrait remercier ses levs qui ont diligemment suivis son cours sur la Mcanique
e e
e
de Milieux Continus durant lanne acadmique 1997-1998, ainsi que le Professeur Huet qui a t son
e
e
ee
hte au Laboratoire des Matriaux de Construction de lEPFL durant son sjour a Lausanne.
o
e
e
`
Victor Saouma
Ecublens, Juin 1998
Victor Saouma
Draft
04
PREFACE
One of the most fundamental question that a Material Scientist has to ask him/herself is how a
material behaves under stress, and when does it break. Ultimately, it its the answer to those two
questions which would steer the development of new materials, and determine their survival in various
environmental and physical conditions.
The Material Scientist should then have a thorough understanding of the fundamentals of Mechanics
on the qualitative level, and be able to perform numerical simulation (most often by Finite Element
Method) and extract quantitative information for a specic problem.
In the humble opinion of the author, this is best achieved in three stages. First, the student should
be exposed to the basic principles of Continuum Mechanics. Detailed coverage of Stress, Strain, General
Principles, and Constitutive Relations is essential. Then, a brief exposure to Elasticity (along with Beam
Theory) would convince the student that a well posed problem can indeed have an analytical solution.
However, this is only true for problems problems with numerous simplifying assumptions (such as linear
elasticity, small deformation, plane stress/strain or axisymmetry, and resultants of stresses). Hence, the
last stage consists in a brief exposure to solid mechanics, and more precisely to Variational Methods.
Through an exposure to the Principle of Virtual Work, and the Rayleigh-Ritz Method the student will
then be ready for Finite Elements. Finally, one topic of special interest to Material Science students
was added, and that is the Theoretical Strength of Solids. This is essential to properly understand the
failure of solids, and would later on lead to a Fracture Mechanics course.
These lecture notes were prepared by the author during his sabbatical year at the Swiss Federal
Institute of Technology (Lausanne) in the Material Science Department. The course was oered to
second year undergraduate students in French, whereas the lecture notes are in English. The notes were
developed with the following objectives in mind. First they must be complete and rigorous. At any time,
a student should be able to trace back the development of an equation. Furthermore, by going through
all the derivations, the student would understand the limitations and assumptions behind every model.
Finally, the rigor adopted in the coverage of the subject should serve as an example to the students of
the rigor expected from them in solving other scientic or engineering problems. This last aspect is often
forgotten.
The notes are broken down into a very hierarchical format. Each concept is broken down into a small
section (a byte). This should not only facilitate comprehension, but also dialogue among the students
or with the instructor.
Whenever necessary, Mathematical preliminaries are introduced to make sure that the student is
equipped with the appropriate tools. Illustrative problems are introduced whenever possible, and last
but not least problem set using Mathematica is given in the Appendix.
The author has no illusion as to the completeness or exactness of all these set of notes. They were
entirely developed during a single academic year, and hence could greatly benet from a thorough review.
As such, corrections, criticisms and comments are welcome.
Finally, the author would like to thank his students who bravely put up with him and Continuum
Mechanics in the AY 1997-1998, and Prof. Huet who was his host at the EPFL.
Victor E. Saouma
Ecublens, June 1998
Victor Saouma
Draft
Contents
I
CONTINUUM MECHANICS
07
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
11
11
12
14
14
15
16
16
18
18
110
110
110
110
111
111
111
111
111
113
113
113
114
114
114
115
2 KINETICS
2.1 Force, Traction and Stress Vectors . . . . . . . . . . . .
2.2 Traction on an Arbitrary Plane; Cauchys Stress Tensor
E 2-1 Stress Vectors . . . . . . . . . . . . . . . . . . . .
2.3 Symmetry of Stress Tensor . . . . . . . . . . . . . . . .
2.3.1 Cauchys Reciprocal Theorem . . . . . . . . . . .
2.4 Principal Stresses . . . . . . . . . . . . . . . . . . . . . .
2.4.1 Invariants . . . . . . . . . . . . . . . . . . . . . .
2.4.2 Spherical and Deviatoric Stress Tensors . . . . .
2.5 Stress Transformation . . . . . . . . . . . . . . . . . . .
E 2-2 Principal Stresses . . . . . . . . . . . . . . . . . .
E 2-3 Stress Transformation . . . . . . . . . . . . . . .
2.5.1 Plane Stress . . . . . . . . . . . . . . . . . . . . .
2.5.2 Mohrs Circle for Plane Stress Conditions . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
21
21
23
24
25
25
26
28
28
29
29
210
210
210
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
02
2.6
CONTENTS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
212
214
214
214
217
DIFFERENTIATION
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
31
31
31
33
34
34
36
36
36
36
38
38
39
310
311
311
311
313
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
41
41
41
42
42
44
44
47
48
48
48
49
410
410
411
411
411
412
412
414
414
415
415
415
416
416
417
417
418
4 KINEMATIC
4.1 Elementary Denition of Strain . . . . . . . . . . . . . . . . . . . . . . .
4.1.1 Small and Finite Strains in 1D . . . . . . . . . . . . . . . . . . .
4.1.2 Small Strains in 2D . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Strain Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.1 Position and Displacement Vectors; (x, X) . . . . . . . . . . . . .
E 4-1 Displacement Vectors in Material and Spatial Forms . . . . . . .
4.2.1.1 Lagrangian and Eulerian Descriptions; x(X, t), X(x, t) .
E 4-2 Lagrangian and Eulerian Descriptions . . . . . . . . . . . . . . .
4.2.2 Gradients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.2.1 Deformation; (x X , X x ) . . . . . . . . . . . . . . . .
4.2.2.1.1 Change of Area Due to Deformation . . . . .
4.2.2.1.2 Change of Volume Due to Deformation . . .
E 4-3 Change of Volume and Area . . . . . . . . . . . . . . . . . . . . .
4.2.2.2 Displacements; (u X , u x ) . . . . . . . . . . . . . . .
4.2.2.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . .
E 4-4 Material Deformation and Displacement Gradients . . . . . . . .
4.2.3 Deformation Tensors . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.3.1 Cauchys Deformation Tensor; (dX)2 . . . . . . . . . .
4.2.3.2 Greens Deformation Tensor; (dx)2 . . . . . . . . . . . .
E 4-5 Greens Deformation Tensor . . . . . . . . . . . . . . . . . . . . .
4.2.4 Strains; (dx)2 (dX)2 . . . . . . . . . . . . . . . . . . . . . . . .
4.2.4.1 Finite Strain Tensors . . . . . . . . . . . . . . . . . . .
4.2.4.1.1 Lagrangian/Greens Strain Tensor . . . . . . .
E 4-6 Lagrangian Tensor . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.4.1.2 Eulerian/Almansis Tensor . . . . . . . . . . .
4.2.4.2 Innitesimal Strain Tensors; Small Deformation Theory
4.2.4.2.1 Lagrangian Innitesimal Strain Tensor . . . .
4.2.4.2.2 Eulerian Innitesimal Strain Tensor . . . . . .
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
CONTENTS
03
4.2.4.3 Examples . . . . . . . . . . . . . . . . . . . . . . .
E 4-7 Lagrangian and Eulerian Linear Strain Tensors . . . . . . .
4.2.5 Physical Interpretation of the Strain Tensor . . . . . . . .
4.2.5.1 Small Strain . . . . . . . . . . . . . . . . . . . . .
4.2.5.2 Finite Strain; Stretch Ratio . . . . . . . . . . . . .
4.3 Strain Decomposition . . . . . . . . . . . . . . . . . . . . . . . . .
4.3.1 Linear Strain and Rotation Tensors . . . . . . . . . . . . .
4.3.1.1 Small Strains . . . . . . . . . . . . . . . . . . . . .
4.3.1.1.1 Lagrangian Formulation . . . . . . . . . .
4.3.1.1.2 Eulerian Formulation . . . . . . . . . . .
4.3.1.2 Examples . . . . . . . . . . . . . . . . . . . . . . .
E 4-8 Relative Displacement along a specied direction . . . . . .
E 4-9 Linear strain tensor, linear rotation tensor, rotation vector .
4.3.2 Finite Strain; Polar Decomposition . . . . . . . . . . . . . .
E 4-10 Polar Decomposition I . . . . . . . . . . . . . . . . . . . . .
E 4-11 Polar Decomposition II . . . . . . . . . . . . . . . . . . . .
E 4-12 Polar Decomposition III . . . . . . . . . . . . . . . . . . . .
4.4 Summary and Discussion . . . . . . . . . . . . . . . . . . . . . . .
4.5 Compatibility Equation . . . . . . . . . . . . . . . . . . . . . . . .
E 4-13 Strain Compatibility . . . . . . . . . . . . . . . . . . . . . .
4.6 Lagrangian Stresses; Piola Kircho Stress Tensors . . . . . . . . .
4.6.1 First . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.6.2 Second . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
E 4-14 Piola-Kircho Stress Tensors . . . . . . . . . . . . . . . . .
4.7 Hydrostatic and Deviatoric Strain . . . . . . . . . . . . . . . . . .
4.8 Principal Strains, Strain Invariants, Mohr Circle . . . . . . . . . .
E 4-15 Strain Invariants & Principal Strains . . . . . . . . . . . . .
E 4-16 Mohrs Circle . . . . . . . . . . . . . . . . . . . . . . . . . .
4.9 Initial or Thermal Strains . . . . . . . . . . . . . . . . . . . . . . .
4.10 Experimental Measurement of Strain . . . . . . . . . . . . . . . .
4.10.1 Wheatstone Bridge Circuits . . . . . . . . . . . . . . . . . .
4.10.2 Quarter Bridge Circuits . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
418
418
419
419
421
422
422
422
422
424
425
425
425
426
426
427
428
430
430
432
432
433
433
434
434
436
436
438
439
439
440
441
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
51
51
51
52
52
52
52
52
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
61
61
61
62
63
63
64
64
64
65
66
66
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
04
6.4
CONTENTS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
71
. 71
. 71
. 72
. 73
. 73
. 74
. 75
. 75
. 75
. 76
. 76
. 77
. 77
. 78
. 79
. 710
. 710
. 710
. 711
. 712
. 712
. 713
. 713
. 713
. 714
. 714
. 715
. 716
6.5
6.6
6.7
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
8 INTERMEZZO
II
81
ELASTICITY/SOLID MECHANICS
Victor Saouma
67
67
67
69
69
610
610
611
611
612
613
614
83
ELASTICITY
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
91
91
91
93
93
93
Draft
CONTENTS
05
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
a
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
Circular
9.5
9.6
9.7
9.8
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
95
95
95
95
96
96
97
98
99
910
910
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
Hole
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
in a Plate
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
101
. 101
. 101
. 103
. 103
. 105
. 107
. 107
. 107
. 108
. 1010
. 1010
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
111
. 111
. 113
. 113
. 115
. 116
12 BEAM THEORY
12.1 Introduction . . . . . . . . . . . . . . . . . . . .
12.2 Statics . . . . . . . . . . . . . . . . . . . . . . .
12.2.1 Equilibrium . . . . . . . . . . . . . . . .
12.2.2 Reactions . . . . . . . . . . . . . . . . .
12.2.3 Equations of Conditions . . . . . . . . .
12.2.4 Static Determinacy . . . . . . . . . . . .
12.2.5 Geometric Instability . . . . . . . . . . .
12.2.6 Examples . . . . . . . . . . . . . . . . .
E 12-1 Simply Supported Beam . . . . . . . . .
12.3 Shear & Moment Diagrams . . . . . . . . . . .
12.3.1 Design Sign Conventions . . . . . . . . .
12.3.2 Load, Shear, Moment Relations . . . . .
12.3.3 Examples . . . . . . . . . . . . . . . . .
E 12-2 Simple Shear and Moment Diagram . .
12.4 Beam Theory . . . . . . . . . . . . . . . . . . .
12.4.1 Basic Kinematic Assumption; Curvature
12.4.2 Stress-Strain Relations . . . . . . . . . .
12.4.3 Internal Equilibrium; Section Properties
12.4.3.1 Fx = 0; Neutral Axis . . . .
12.4.3.2 M = 0; Moment of Inertia .
12.4.4 Beam Formula . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
121
. 121
. 121
. 121
. 123
. 123
. 124
. 124
. 125
. 125
. 126
. 126
. 126
. 128
. 128
. 1210
. 1210
. 1211
. 1211
. 1211
. 1212
. 1212
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
06
CONTENTS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
131
. 131
. 132
. 133
. 134
. 134
. 135
. 135
. 136
. 136
. 136
. 137
. 139
. 1310
. 1311
. 1311
. 1313
. 1314
. 1316
1
A1
B SECTION PROPERTIES
B1
C MATHEMATICAL PRELIMINARIES;
C.1 Euler Equation . . . . . . . . . . . . . .
E C-1 Extension of a Bar . . . . . . . .
E C-2 Flexure of a Beam . . . . . . . .
D1
E1
Victor Saouma
Draft
List of Figures
1.1
1.2
1.3
1.4
1.5
1.6
1.7
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
System
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
12
12
13
14
15
17
18
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11
2.12
2.13
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
22
22
23
26
27
211
213
214
215
215
216
216
217
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
3.10
3.11
3.12
.
.
.
.
.
.
.
.
.
.
a
.
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
Vector .
. . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
32
32
33
34
34
35
35
37
38
310
311
312
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
41
43
45
46
413
420
423
436
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Draft
02
LIST OF FIGURES
4.9
4.10
4.11
4.12
5.1
6.1
6.2
6.3
6.4
6.5
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
62
66
613
614
614
9.1
9.2
9.3
9.4
9.5
9.6
9.7
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
92
93
94
96
97
97
99
10.1
10.2
10.3
10.4
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
102
109
1010
1010
11.1
11.2
11.3
11.4
11.5
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
111
112
113
114
115
12.1
12.2
12.3
12.4
12.5
12.6
12.7
Types of Supports . . . . . . . . . . . . . . . . . . . . . . . . .
Inclined Roller Support . . . . . . . . . . . . . . . . . . . . . .
Examples of Static Determinate and Indeterminate Structures .
Geometric Instability Caused by Concurrent Reactions . . . . .
Shear and Moment Sign Conventions for Design . . . . . . . . .
Free Body Diagram of an Innitesimal Beam Segment . . . . .
Deformation of a Beam under Pure Bending . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
123
124
124
125
126
127
1210
13.1
13.2
13.3
13.4
13.5
13.6
13.7
13.8
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
132
138
1310
1312
1313
1315
1317
1318
14.1
14.2
14.3
14.4
14.5
14.6
test . .
mod1 .
v-kv .
vis .
vis .
comp .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
2
2
3
3
3
.
.
.
.
.
.
Victor Saouma
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
439
440
441
442
Draft
LIST OF FIGURES
03
14.7 epp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
14.8 ehs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
C.1 Variational and Dierential Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C2
Victor Saouma
Draft
LIST OF FIGURES
Victor Saouma
04
Draft
LIST OF FIGURES
Symbol
05
NOTATION
Denition
SCALARS
A
Area
c
Specic heat
e
Volumetric strain
E
Elastic Modulus
g
Specicif free enthalpy
h
Film coecient for convection heat transfer
h
Specic enthalpy
I
Moment of inertia
J
Jacobian
K
Bulk modulus
K
Kinetic Energy
L
Length
p
Pressure
Q
Rate of internal heat generation
r
Radiant heat constant per unit mass per unit time
s
Specic entropy
S
Entropy
t
Time
T
Absolute temperature
u
Specic internal energy
U
Energy
Complementary strain energy
U
W
Work
W
Potential of External Work
Potential energy
Shear modulus
Poissons ratio
mass density
ij
Shear strains
1
ij
Engineering shear strain
2
Lames coecient
Stretch ratio
G
Lames coecient
Lames coecient
Temperature
Dimension
SI Unit
L2
m2
N.D.
L1 M T 2
L2 T 2
Pa
JKg 1
L2 T 2
L4
JKg 1
m4
L1 M T 2
L2 M T 2
L
L1 M T 2
L2 M T 3
M T 3 L4
L2 T 2 1
M L2 T 2 1
T
L2 T 2
L2 M T 2
L2 M T 2
L2 M T 2
L2 M T 2
L2 M T 2
1
L1 M T 2
N.D.
M L3
N.D.
N.D.
L1 M T 2
N.D.
L1 M T 2
L1 M T 2
Pa
J
m
Pa
W
W m6
JKg 1 K 1
JK 1
s
K
JKg 1
J
J
J
J
J
T 1
Pa
Kgm3
Pa
Pa
Pa
L2 M T 2
TENSORS order 1
b
b
q
t
t
u
Victor Saouma
N Kg 1
M T 3
L1 M T 2
L1 M T 2
L
W m2
Pa
Pa
m
Draft
06
u(x)
u
x
X
0
(i)
LIST OF FIGURES
L
L
L
L
L1 M T 2
L1 M T 2
m
m
m
m
Pa
Pa
TENSORS order 2
B1
C
D
E
E
E
F
H
I
J
k
K
L
R
T0
T
U
V
W
0
k
, T
T
W m1 K 1
-
Pa
Pa
Pa
Pa
TENSORS order 4
D
L1 M T 2
Constitutive matrix
Pa
L2
L2
L2
L2
L2
L2
L2
L3
m2
m2
m2
m2
m2
m2
m2
m3
t
u
T
c
q
, V
Contour line
Surface of a body
Surface
Boundary along which
Boundary along which
Boundary along which
Boundary along which
Boundary along which
Volume of body
FUNCTIONS, OPERATORS
Victor Saouma
Draft
LIST OF FIGURES
u
2
07
Victor Saouma
uz
z
Draft
LIST OF FIGURES
Victor Saouma
08
Draft
Part I
CONTINUUM MECHANICS
Draft
Draft
Chapter 1
MATHEMATICAL
PRELIMINARIES; Part I Vectors
and Tensors
Physical laws should be independent of the position and orientation of the observer. For this reason,
physical laws are vector equations or tensor equations, since both vectors and tensors transform
from one coordinate system to another in such a way that if the law holds in one coordinate system, it
holds in any other coordinate system.
1.1
Vectors
A vector is a directed line segment which can denote a variety of quantities, such as position of point
with respect to another (position vector), a force, or a traction.
A vector may be dened with respect to a particular coordinate system by specifying the components
of the vector in that system. The choice of the coordinate system is arbitrary, but some are more suitable
than others (axes corresponding to the major direction of the object being analyzed).
The rectangular Cartesian coordinate system is the most often used one (others are the cylindrical, spherical or curvilinear systems). The rectangular system is often represented by three mutually
perpendicular axes Oxyz, with corresponding unit vector triad i, j, k (or e1 , e2 , e3 ) such that:
ij = k;
jk = i;
ki = j;
(1.1-a)
ii = jj = kk = 1
(1.1-b)
ij = jk = ki = 0
(1.1-c)
(1.2)
where
vx
vy
=
=
vi = v cos
vj = v cos
(1.3-a)
(1.3-b)
vz
vk = v cos
(1.3-c)
Draft
12
v
= cos i + cos j + cos k
v
(1.4)
Since v is arbitrary, it follows that any unit vector will have direction cosines of that vector as its
Cartesian components.
7
v =
2
2
2
v1 + v2 + v3 .
k
8 We will denote the contravariant components of a vector by superscripts v , and its covariant
components by subscripts vk (the signicance of those terms will be claried in Sect. 1.1.2.1.
1.1.1
Operations
Addition: of two vectors a + b is geometrically achieved by connecting the tail of the vector b with the
head of a, Fig. 1.2. Analytically the sum vector will have components a1 + b1 a2 + b2 a3 + b3 .
u+v
a1
a2
a3 .
Victor Saouma
Draft
1.1 Vectors
13
Dot Product (or scalar product) is a scalar quantity which relates not only to the lengths of the
vector, but also to the angle between them.
3
ab a
cos (a, b) =
ai bi
(1.5)
i=1
where cos (a, b) is the cosine of the angle between the vectors a and b. The dot product
measures the relative orientation between two vectors.
The dot product is both commutative
ab = ba
(1.6)
(1.7)
and distributive
The dot product of a with a unit vector n gives the projection of a in the direction of n.
The dot product of base vectors gives rise to the denition of the Kronecker delta dened
as
ei ej = ij
(1.8)
where
ij =
1
0
if
if
i=j
i=j
(1.9)
Cross Product (or vector product) c of two vectors a and b is dened as the vector
c = ab = (a2 b3 a3 b2 )e1 + (a3 b1 a1 b3 )e2 + (a1 b2 a2 b1 )e3
(1.10)
e2
a2
b2
e1
a1
b1
e3
a3
b3
(1.11)
and is equal to the area of the parallelogram described by a and b, Fig. 1.3.
axb
A(a,b)=||a x b||
b
A(a, b) = ab
Victor Saouma
(1.12)
Draft
14
The cross product is not commutative, but satises the condition of skew symmetry
ab = ba
(1.13)
(1.14)
Triple Scalar Product: of three vectors a, b, and c is desgnated by (ab)c and it corresponds
to the (scalar) volume dened by the three vectors, Fig. 1.4.
n=a x b
||a x b||
c
c.n
=
=
(1.17)
(1.18)
Vector Triple Product is a cross product of two vectors, one of which is itself a cross product.
a(bc) = (ac)b (ab)c = d
(1.19)
1.1.2
Coordinate Transformation
1.1.2.1
General Tensors
9 Let us consider two bases bj (x1 , x2 , x3 ) and bj (x1 , x2 x3 ), Fig. 1.5. Each unit vector in one basis must
be a linear combination of the vectors of the other basis
bj = ap bp and bk = bk bq
q
j
Victor Saouma
(1.20)
Draft
1.1 Vectors
15
(summed on p and q respectively) where ap (subscript new, superscript old) and bk are the coecients
q
j
for the forward and backward changes respectively from b to b respectively. Explicitly
1 1 1
b1 b2 b3 e1
a1 a2 a3 e1
e1
e1
1
1
e2
e2
e2
e2
= b2 b2 b2
and
= a1 a2 a3
(1.21)
1
2
3
2
2
2
3
3
3
e3
b1 b2 b3
e3
e3
a1 a2 a3
e3
3
3
3
X2
X2
X1
-1
2
cos a1
X1
X3
X3
Figure 1.5: Coordinate Transformation
10
x1
1
x2
x
1
x3
x
x1
x1
2
x2
x
2
x3
x
x2
x1
3
x2
x
3
x3
x
x3
=0
(1.22)
1.1.2.1.1
12
Contravariant Transformation
(1.23)
since the base vectors bq are linearly independent, the coecients of bq must all be zero hence
v q = bq v k and inversely v p = ap v j
j
k
(1.24)
showing that the forward change from components v k to v q used the coecients bq of the backward
k
change from base bq to the original bk . This is why these components are called contravariant.
13 Generalizing, a Contravariant Tensor of order one (recognized by the use of the superscript)
transforms a set of quantities rk associated with point P in xk through a coordinate transformation into
Victor Saouma
Draft
16
xq k
r
xk
(1.25)
bq
k
14 By extension, the Contravariant tensors of order two requires the tensor components to obey
the following transformation law
rij =
1.1.2.1.2
xi xj rs
r
xr xs
(1.26)
Covariant Transformation
v j = ap vp and inversely vk = bk v q
q
j
(1.27)
We note that contrarily to the contravariant transformation, the covariant transformation uses the same
transformation coecients as the ones for the base vectors.
16
1.1.2.2
=
=
xk
rk
xq
r
x xs
rrs
xi xj
(1.28)
(1.29)
17 If we consider two dierent sets of cartesian orthonormal coordinate systems {e1 , e2 , e3 } and {e1 , e2 , e3 },
any vector v can be expressed in one system or the other
v = vj ej = v j ej
(1.30)
18 To determine the relationship between the two sets of components, we consider the dot product of v
with one (any) of the base vectors
ei v = v i = vj (ei ej )
(1.31)
(since v j (ej ei ) = v j ij = v i )
19
(1.32)
which arise from the dot products of base vectors as the direction cosines. (Since we have an orthonormal system, those values are nothing else than the cosines of the angles between the nine pairing
of base vectors.)
20 Thus, one set of vector components can be expressed in terms of the other through a covariant
transformation similar to the one of Eq. 1.27.
Victor Saouma
Draft
1.1 Vectors
17
vj
vk
ap vp
j
(1.33)
bk v q
q
(1.34)
we note that the free index in the rst and second equations appear on the upper and lower index
respectively.
21
As a further illustration of the above derivation, let us consider the transformation of a vector V from
(X, Y, Z) coordinate system to (x, y, z), Fig. 1.6:
22
23
ax
Vx
Vy
= aX
y
Vz
aX
z
or
aY
x
aY
y
aY
z
aZ VX
x
aZ
VY
y
VZ
aZ
z
(1.35)
(1.36)
Finally, for the 2D case and from Fig. 1.7, the transformation matrix is written as
T =
a1
1
a1
2
cos
cos
cos
cos
(1.37)
Victor Saouma
a2
1
a2
2
cos
sin
sin
cos
(1.38)
Draft
18
X2
X1
X1
1.2
Tensors
25 We now seek to generalize the concept of a vector by introducing the tensor (T), which essentially
exists to operate on vectors v to produce other vectors (or on tensors to produce other tensors!). We
designate this operation by Tv or simply Tv.
26
We hereby adopt the dyadic notation for tensors as linear vector operators
u
u
= Tv or ui = Tij vj
= vS where S = TT
(1.39-a)
(1.39-b)
ui
ui
=
=
Tij v j ; ui
Ti.j vj ; ui
=
=
T ij vj
i
T.j v j
(1.40)
involving the covariant components Tij , the contravariant components T ij and the mixed comi
ponents T.j or Ti.j .
28 Whereas a tensor is essentially an operator on vectors (or other tensors), it is also a physical quantity,
independent of any particular coordinate system yet specied most conveniently by referring to an
appropriate system of coordinates.
29 Tensors frequently arise as physical entities whose components are the coecients of a linear relationship between vectors.
30 A tensor is classied by the rank or order. A Tensor of order zero is specied in any coordinate system
by one coordinate and is a scalar. A tensor of order one has three coordinate components in space, hence
it is a vector. In general 3-D space the number of components of a tensor is 3n where n is the order of
the tensor.
31
1.2.1
Indicial Notation
32 Whereas the Engineering notation may be the simplest and most intuitive one, it often leads to long
and repetitive equations. Alternatively, the tensor and the dyadic form will lead to shorter and more
compact forms.
Victor Saouma
Draft
1.2 Tensors
19
33 While working on general relativity, Einstein got tired of writing the summation symbol with its range
n=3
of summation below and above (such as i=1 aij bi ) and noted that most of the time the upper range
(n) was equal to the dimension of space (3 for us, 4 for him), and that when the summation involved a
product of two terms, the summation was over a repeated index (i in our example). Hence, he decided
that there is no need to include the summation sign
if there was repeated indices (i), and thus any
repeated index is a dummy index and is summed over the range 1 to 3. An index that is not repeated
is called free index and assumed to take a value from 1 to 3.
34
35
a1
a2
ai = ai = a1 a2 a3 =
i = 1, 3
(1.41)
a3
assuming that n = 3.
2. A repeated index will take on all the values of its range, and the resulting tensors summed. For
instance:
(1.42)
a1i xi = a11 x1 + a12 x2 + a13 x3
3. Tensors order:
First order tensor (such as force) has only one free index:
ai = ai =
a1
a2
a3
(1.43)
(1.44)
xi
36
= ,i
vi
xi
= vi,i
vi
xj
= vi,j
Ti,j
xk
= Ti,j,k
(1.45)
Usefulness of the indicial notation is in presenting systems of equations in compact form. For instance:
xi = cij zj
(1.46)
=
=
x3
(1.47-a)
Similarly:
Aij = Bip Cjq Dpq
Victor Saouma
(1.48)
Draft
110
A11
A12
A21
A22
37
= B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22
= B11 C11 D11 + B11 C12 D12 + B12 C11 D21 + B12 C12 D22
= B21 C11 D11 + B21 C12 D12 + B22 C11 D21 + B22 C12 D22
= B21 C21 D11 + B21 C22 D12 + B22 C21 D21 + B22 C22 D22
(1.49-a)
Using indicial notation, we may rewrite the denition of the dot product
ab = ai bi
(1.50)
ab = pqr aq br ep
(1.51)
we note that in the second equation, there is one free index p thus there are three equations, there are
two repeated (dummy) indices q and r, thus each equation has nine terms.
1.2.2
Tensor Operations
1.2.2.1
Sum
38
(1.52)
1.2.2.2
39
Multiplication by a Scalar
(1.53)
Sij = Tij
1.2.2.3
Contraction
40 In a contraction, we make two of the indeces equal (or in a mixed tensor, we make a ubscript equal to
the superscript), thus producing a tensor of order two less than that to which it is applied. For example:
Tij
u i vj
Amr
..sn
Eij ak
Ampr
qs
Victor Saouma
Tii ;
u i vi ;
r
Amr = B.s ;
..sm
Eij ai = cj ;
mp
Ampr = Bq ;
qr
2
2
4
3
5
0
0
2
1
3
(1.54)
Draft
1.2 Tensors
1.2.2.4
1.2.2.4.1
111
Products
Outer Product
41 The outer product of two tensors (not necessarily of the same type or order) is a set of tensor
components obtained simply by writing the components of the two tensors beside each other with no
repeated indices (that is by multiplying each component of one of the tensors by every component of
the other). For example
ai bj
.k
A Bj
=
=
Tij
C i.k .j
(1.55-a)
(1.55-b)
vi Tjk
Sijk
(1.55-c)
1.2.2.4.2
Inner Product
42 The inner product is obtained from an outer product by contraction involving one index from each
tensor. For example
ai bj
ai Ejk
Eij Fkm
.k
Ai Bi
1.2.2.4.3
43
ai bi
ai Eik = fk
(1.56-a)
(1.56-b)
(1.56-c)
(1.56-d)
Scalar Product
T : U = Tij Uij
in any rectangular system.
44
1.2.2.4.4
(1.58-a)
(1.58-b)
(1.58-c)
(1.58-d)
Tensor Product
45 Since a tensor primary objective is to operate on vectors, the tensor product of two vectors provides
a fundamental building block of second-order tensors and will be examined next.
Victor Saouma
Draft
112
The Tensor Product of two vectors u and v is a second order tensor u v which in turn operates
on an arbitrary vector w as follows:
46
[u v]w (vw)u
(1.59)
In other words when the tensor product u v operates on w (left hand side), the result (right hand
side) is a vector that points along the direction of u, and has length equal to (vw)||u||, or the original
length of u times the dot (scalar) product of v and w.
47 Of particular interest is the tensor product of the base vectors ei ej . With three base vectors, we
have a set of nine second order tensors which provide a suitable basis for expressing the components of a
tensor. Again, we started with base vectors which themselves provide a basis for expressing any vector,
and now the tensor product of base vectors in turn provides a formalism to express the components of
a tensor.
48 The second order tensor T can be expressed in terms of its components Tij relative to the base
tensors ei ej as follows:
Tij [ei ej ]
(1.60-a)
Tij [ei ej ] ek
(1.60-b)
i=1 j=1
3
Tek
=
i=1 j=1
[ei ej ] ek
(ej ek )ei = jk ei
(1.60-c)
Tek
Tik ei
(1.60-d)
i=1
Thus Tik is the ith component of Tek . We can thus dene the tensor component as follows
Tij = ei Tej
(1.61)
Now we can see how the second order tensor T operates on any vector v by examining the components
of the resulting vector Tv:
49
Tv =
Tij [ei ej ]
i=1 j=1
vk ek
(1.62)
k=1
Tv =
Tij vj ei
(1.63)
i=1 j=1
(Tv)i =
Tij vj
(1.64)
i=1
50
Victor Saouma
(1.65)
Introduction to Continuum Mechanics
Draft
1.2 Tensors
113
51 A simple example of a tensor and its operation on vectors is the projection tensor P which generates
the projection of a vector v on the plane characterized by a normal n:
PInn
(1.66)
the action of P on v gives Pv = v (vn)n. To convince ourselves that the vector Pv lies on the plane,
(1.67)
(1.68-a)
(1.68-b)
(R + U)T = RT + UT
(TU) = (T)U = T(U)
(1.68-c)
(1.68-d)
1T =
T1 = T
(1.68-e)
1.2.3
Dyads
54 The indeterminate vector product of a and b dened by writing the two vectors in juxtaposition as
ab is called a dyad. A dyadic D corresponds to a tensor of order two and is a linear combination of
dyads:
(1.69)
D = a1 b1 + a2 b2 an bn
1.2.4
55
(1.70)
Rotation of Axes
The rule for changing second order tensor components under rotation of axes goes as follow:
ui
= aj uj
i
= aj Tjq vq
i
= aj Tjq aq v p
p
i
(1.71)
But we also have ui = T ip v p (again from Eq. 1.39-a) in the barred system, equating these two expressions
we obtain
T ip (aj aq Tjq )v p = 0
(1.72)
i p
hence
T ip
Tjq
Victor Saouma
(1.73)
aj aq T ip
i p
(1.74)
Draft
114
By extension, higher order tensors can be similarly transformed from one coordinate system to another.
56
cos sin 0
A = sin cos 0
0
0
1
Txx Txy 0
T = Txy Tyy 0
0
0
0
T xx T xy 0
T = AT T A = T xy T yy 0
0
0
0
2
2
cos Txx + sin Tyy + sin 2Txy
= 1 ( sin 2Txx + sin 2Tyy + 2 cos 2Txy
2
0
(1.75-a)
(1.75-b)
(1.75-c)
1
2 ( sin 2Txx + sin 2Tyy + 2 cos 2Txy
sin2 Txx + cos (cos Tyy 2 sin Txy
0
0
0
(1.75-d)
alternatively, using sin 2 = 2 sin cos and cos 2 = cos2 sin2 , this last equation can be rewritten
as
sin2
2 sin cos
cos2
Txx
T xx
(1.76)
Tyy
=
sin2
cos2
2 sin cos
T
yy
Txy
sin cos cos sin cos2 sin2
T xy
1.2.5
Trace
The trace of a second-order tensor, denoted tr T is a scalar invariant function of the tensor and is
dened as
57
tr T Tii
(1.77)
1.2.6
58
Inverse Tensor
(1.78)
1
1
alternatively T1 T = TT1 = I, or Tik Tkj = ij and Tik Tkj = ij
1.2.7
59 Since the two fundamental tensors in continuum mechanics are of the second order and symmetric
(stress and strain), we examine some important properties of these tensors.
60 For every symmetric tensor Tij dened at some point in space, there is associated with each direction
(specied by unit normal nj ) at that point, a vector given by the inner product
vi = Tij nj
Victor Saouma
(1.79)
Draft
1.2 Tensors
115
If the direction is one for which vi is parallel to ni , the inner product may be expressed as
Tij nj = ni
(1.80)
and the direction ni is called principal direction of Tij . Since ni = ij nj , this can be rewritten as
(Tij ij )nj = 0
(1.81)
which represents a system of three equations for the four unknowns ni and .
(T11 )n1 + T12 n2 + T13 n3
T21 n1 + (T22 )n2 + T23 n3
=
=
0
0
(1.82-a)
To have a non-trivial slution (ni = 0) the determinant of the coecients must be zero,
|Tij ij | = 0
61
(1.83)
(1.84)
IIT
IIIT
Tij = tr Tij
1
(Tii Tjj Tij Tij )
2
|Tij | = det Tij
(1.85)
(1.86)
(1.87)
are called the rst, second and third invariants respectively of Tij .
62
63 For a symmetric tensor with real components, the principal values are also real. If those values are
distinct, the three principal directions are mutually orthogonal.
1.2.8
64
When expressed in term of the principal axes, the tensor array can be written in matrix form as
(1)
0
0
0
(2)
T = 0
(1.88)
0
0
(3)
65 By direct matrix multiplication, the quare of the tensor Tij is given by the inner product Tik Tkj , the
cube as Tik Tkm Tmn . Therefore the nth power of Tij can be written as
0
0
n
(1)
0
n
Tn= 0
(1.89)
(2)
0
0
n
(3)
Victor Saouma
Draft
116
Since each of the principal values satises Eq. 1.84 and because the diagonal matrix form of T given
above, then the tensor itself will satisfy Eq. 1.84.
T 3 IT T 2 + IIT T IIIT I = 0
(1.90)
where I is the identity matrix. This equation is called the Hamilton-Cayley equation.
Victor Saouma
Draft
Chapter 2
KINETICS
Or How Forces are Transmitted
2.1
1
body forces: act on the elements of volume or mass inside the body, e.g. gravity, electromagnetic
elds. dF = bdV ol.
surface forces: are contact forces acting on the free body at its bounding surface. Those will be dened
in terms of force per unit area.
The surface force per unit area acting on an element dS is called traction or more accurately stress
vector.
tdS = i
tx dS + j
ty dS + k
tz dS
(2.1)
Most authors limit the term traction to an actual bounding surface of a body, and use the term stress
vector for an imaginary interior surface (even though the state of stress is a tensor and not a vector).
The traction vectors on planes perpendicular to the coordinate axes are particularly useful. When the
vectors acting at a point on three such mutually perpendicular planes is given, the stress vector at
that point on any other arbitrarily inclined plane can be expressed in terms of the rst set of tractions.
4 A stress, Fig 2.1 is a second order cartesian tensor, ij where the 1st subscript (i) refers to the
direction of outward facing normal, and the second one (j) to the direction of component force.
11 12 13
t1
t2
= ij = 21 22 23 =
(2.2)
31 32 33
t3
5 In fact the nine rectangular components ij of turn out to be the three sets of three vector components
(11 , 12 , 13 ), (21 , 22 , 23 ), (31 , 32 , 33 ) which correspond to the three tractions t1 , t2 and t3 which
are acting on the x1 , x2 and x3 faces (It should be noted that those tractions are not necesarily normal
to the faces, and they can be decomposed into a normal and shear traction if need be). In other words,
stresses are nothing else than the components of tractions (stress vector), Fig. 2.2.
The state of stress at a point cannot be specied entirely by a single vector with three components; it
requires the second-order tensor with all nine components.
Draft
22
KINETICS
X3
33
32
31
23
13
21
X3
22
X2
12
X1
11
X2
X1
X3
X3
V3
33
t3
t2
13
21
X1
t1
32
23
31
11
V2
X2
22
V1
X1
X2
12
Victor Saouma
Draft
2.2
23
Let us now consider the problem of determining the traction acting on the surface of an oblique plane
(characterized by its normal n) in terms of the known tractions normal to the three principal axis, t1 , t2
and t3 . This will be done through the so-called Cauchys tetrahedron shown in Fig. 2.3.
7
X2
-t
1
-t
S3
*
tn S
h N
n
A
X1
-t 2 S2
*
X3
*
The components of the unit vector n are the direction cosines of its direction:
n1 = cos( AON );
n2 = cos( BON );
n3 = cos( CON );
(2.3)
The altitude ON , of length h is a leg of the three right triangles AN O, BN O and CN O with hypothenuses OA, OB and OC. Hence
h = OAn1 = OBn2 = OCn3
9
(2.4)
The volume of the tetrahedron is one third the base times the altitude
V =
1
1
1
1
hS = OAS1 = OBS2 = OCS3
3
3
3
3
(2.5)
S2 = Sn2 ;
S3 = Sn3 ;
(2.6)
or Si = Sni .
10 In Fig. 2.3 are also shown the average values of the body force and of the surface tractions (thus the
asterix). The negative sign appears because t denotes the average traction on a surface whose outward
i
normal points in the negative xi direction. We seek to determine t .
n
11 We invoke the momentum principle of a collection of particles (more about it later on) which
is postulated to apply to our idealized continuous medium. This principle states that the vector sum
Victor Saouma
Draft
24
KINETICS
of all external forces acting on the free body is equal to the rate of change of the total momentum1 .
vdm. By the mean-value theorem of the integral calculus, this is equal
to v m where v is average value of the velocity. Since we are considering the momentum of a given
collection of particles, m does not change with time and m dv = V dv where is the average
dt
dt
density. Hence, the momentum principle yields
t S + b V t S1 t S2 t S3 = V
n
1
2
3
dv
dt
(2.7)
(2.8)
(2.9)
We observe that we dropped the asterix as the length of the vectors approached zero.
12 It is important to note that this result was obtained without any assumption of equilibrium and that
it applies as well in uid dynamics as in solid mechanics.
13 This equation is a vector equation, and the corresponding algebraic equations for the components of
tn are
tn1 = 11 n1 + 21 n2 + 31 n3
tn2 = 12 n1 + 22 n2 + 32 n3
tn3 = 13 n1 + 23 n2 + 33 n3
(2.10)
Indicial notation tni = ji nj
dyadic notation
tn = n = T n
14 We have thus established that the nine components ij are components of the second order tensor,
Cauchys stress tensor.
15 Note that this stress tensor is really dened in the deformed space (Eulerian), and this issue will be
revisited in Sect. 4.6.
7 5
= 5 3
0
1
0
t1
1 =
t2
t3
2
(2.11)
We seek to determine the traction (or stress vector) t passing through P and parallel to the plane ABC
where A(4, 0, 0), B(0, 2, 0) and C(0, 0, 6). Solution:
The vector normal to the plane can be found by taking the cross products of vectors AB and AC:
N
= ABAC =
=
1 This
Victor Saouma
e2
2
0
e1
4
4
e3
0
6
(2.12-a)
(2.12-b)
m dv
dt
Draft
25
3
6
2
e1 + e2 + e3
7
7
7
3
7
and thus t = 9 e1 + 5 e2 +
7
7
2.3
16
6
7
2
7
7 5
5 3
0
1
0
1 =
2
(2.13)
9
7
5
7
10
7
(2.14)
10
7 e3
From Fig. 2.1 the resultant force exerted on the positive X1 face is
11 X2 X3
12 X2 X3
13 X2 X3
(2.15)
22 X3 X1
23 X3 X1
(2.16)
We now consider moment equilibrium (M = Fd). The stress is homogeneous, and the normal
force on the opposite side is equal opposite and colinear. The moment (X2 /2)31 X1 X2 is likewise
balanced by the moment of an equal component in the opposite face. Finally similar argument holds for
32 .
17
18
(2.17)
(2.18)
ij = ji
20
A more rigorous proof of the symmetry of the stress tensor will be given in Sect. 6.3.2.1.
2.3.1
21 If we consider t1 as the traction vector on a plane with normal n1 , and t2 the stress vector at the
same point on a plane with normal n2 , then
t1 = n1 and t2 = n2
(2.19)
(2.20)
or in matrix form as
If we postmultiply the rst equation by n2 and the second one by n1 , by virtue of the symmetry of []
we have
(2.21)
[n1 ]n2 = [n2 ]n1
Victor Saouma
Draft
26
KINETICS
n
tn
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
11111111111111111111111
00000000000000000000000
111
000
11
00
11111111111111111111111
00000000000000000000000
111
000
11
00
11111111111111111111111
00000000000000000000000
11
00
11
00
11111111111111111111111
00000000000000000000000
11
00
11111111111111111111111
00000000000000000000000
11
00
11
00
11111111111111111111111
00000000000000000000000
11
00
1111
0000
11
00
1111
0000
11
00
1111
0000
11
00
1111
0000
11
00
11
00
11
00
11
00
11
00
t -n
-n
Figure 2.4: Cauchys Reciprocal Theorem
or
t1 n2 = t2 n1
22
(2.22)
(2.23)
23 We should note that this theorem is analogous to Newtons famous third law of motion To every
action there is an equal and opposite reaction.
2.4
Principal Stresses
24 Regardless of the state of stress (as long as the stress tensor is symmetric), at a given point, it is
always possible to choose a special set of axis through the point so that the shear stress components
vanish when the stress components are referred to this system of axis. these special axes are called
principal axes of the principal stresses.
To determine the principal directions at any point, we consider n to be a unit vector in one of the
unknown directions. It has components ni . Let represent the principal-stress component on the plane
whose normal is n (note both n and are yet unknown). Since we know that there is no shear stress
component on the plane perpendicular to n,
the stress vector on this plane must be parallel to n and
25
tn = n
26
(2.24)
(2.25)
nr rs = ns
(2.26)
Victor Saouma
Draft
27
tn = t
12
n2
12
11
11= t
n1
tn
s
s =0
n2
=tn
n
t n2
n1
t n2
n1
tn1
Arbitrary Plane
Principal Plane
(2.27)
n ([] [I]) = 0
(2.28)
(2.29)
they can not all be zero. hence Eq.2.28 has solutions which are not zero if and only if the determinant
of the coecients is equal to zero, i.e
11
21
31
12
22
32
13
23
33
(2.30)
|rs rs | =
(2.31)
| I| =
(2.32)
28 For a given set of the nine stress components, the preceding equation constitutes a cubic equation for
the three unknown magnitudes of .
29 Cauchy was rst to show that since the matrix is symmetric and has real elements, the roots are all
real numbers.
30 The three lambdas correspond to the three principal stresses (1) > (2) > (3) .
When any one
of them is substituted for in the three equations in Eq. 2.28 those equations reduce to only two
independent linear equations, which must be solved together with the quadratic Eq. 2.29 to determine
the direction cosines ni of the normal ni to the plane on which i acts.
r
31
Victor Saouma
(2.33)
Draft
28
32
KINETICS
In 2D, it can be shown that the principal stresses are given by:
1,2 =
2.4.1
x + y
x y
2
2
2
+ xy
(2.34)
Invariants
33 The principal stresses are physical quantities, whose values do not depend on the coordinate system
in which the components of the stress were initially given. They are therefore invariants of the stress
state.
34
When the determinant in the characteristic Eq. 2.32 is expanded, the cubic equation takes the form
3 I 2 II III = 0
(2.35)
where the symbols I , II and III denote the following scalar expressions in the stress components:
I
II
=
=
=
=
III
35
11 + 22 + 33 = ii = tr
2
2
2
(11 22 + 22 33 + 33 11 ) + 23 + 31 + 12
1
1
1 2
(ij ij ii jj ) = ij ij I
2
2
2
1
2
( : I )
2
1
det = eijk epqr ip jq kr
6
(2.38)
(2.39)
(2.40)
2.4.2
=
=
(2.41)
(2.42)
III
36
(2.36)
(2.37)
(2.43)
1
1
1
(11 + 22 + 33 ) = ii = tr
3
3
3
(2.44)
then the stress tensor can be written as the sum of two tensors:
Hydrostatic stress in which each normal stress is equal to p and the shear stresses are zero. The
hydrostatic stress produces volume change without change in shape in an isotropic medium.
p 0
0
p 0
(2.45)
hyd = pI = 0
0
0
p
Deviatoric Stress: which causes the change in shape.
11 12
22
dev = 21
31
32
Victor Saouma
13
23
33
(2.46)
Draft
2.5
37
29
Stress Transformation
From Eq. 1.73 and 1.74, the stress transformation for the second order stress tensor is given by
ip
(2.47)
jq
38
=
=
(2.48)
sin2
cos2
xx
2
yy
=
sin
cos2
xy
sin cos cos sin
2 sin cos
xx
yy
2 sin cos
xy
cos2 sin2
(2.49)
3
= 1
1
1
0
2
1
2
0
(2.50)
(2.51)
Or upon expansion (and simplication) ( + 2)( 4)( 1) = 0, thus the roots are (1) = 4, (2) = 1
and (3) = 2. We also note that those are the three eigenvalues of the stress tensor.
If we let x1 axis be the one corresponding to the direction of (3) and n3 be the direction cosines of
i
this axis, then from Eq. 2.28 we have
(3 + 2)n3 + n3 + n3 = 0
1
2
3
1
1
n3 + 2n3 + 2n3 = 0 n3 = 0;
n3 =
(2.52)
n3 = ;
1
2
3
1
2
3
2
2
n3 + 2n3 + 2n3 = 0
1
2
3
Similarly If we let x2 axis be the one corresponding to the direction of (2) and n2 be the direction
i
cosines of this axis,
2n2 + n2 + n2 = 0
1
2
3
1
1
1
n2 n2 + 2n2 = 0 n2 = ;
n2 = ;
n2 =
(2.53)
1
2
3
1
2
3
2
3
3
3
n1 + 2n2 n2 = 0
2
3
Finally, if we let x3 axis be the one corresponding to the direction of (1) and n1 be the direction cosines
i
of this axis,
n1 + n1 + n1 = 0
1
2
3
2
1
1
n1 4n1 + 2n1 = 0 n1 = ;
n1 = ;
n1 =
(2.54)
1
2
3
1
2
3
1
6
6
6
1
1
n1 + 2n2 4n3 = 0
Finally, we can convince ourselves that the two stress tensors have the same invariants I , II and III .
Victor Saouma
Draft
210
KINETICS
0
1
3
2
6
2
= 0
0
2.5.1
39
2
1
3
1
6
0 0
1 0
0 4
1
2
3
1
3 1
1
1
6
1
0
2
0
1
1
2 2
1
2
2
3
1
3
1
3
2
6
1
6
1
6
(2.55-a)
(2.55-b)
Plane Stress
Plane stress conditions prevail when 3i = 0, and thus we have a biaxial stress eld.
40 Plane stress condition prevail in (relatively) thin plates, i.e when one of the dimensions is much smaller
than the other two.
2.5.2
41 The Mohr circle will provide a graphical mean to contain the transformed state of stress ( xx , yy , xy )
at an arbitrary plane (inclined by ) in terms of the original one (xx , yy , xy ).
42
Substituting
cos2
cos 2
=
=
1+cos 2
2
2
cos sin
2
sin2
sin 2
= 1cos 2
2
= 2 sin cos
(2.56)
1
1
(xx + yy ) + (xx yy ) cos 2 + xy sin 2
2
2
1
= xy cos 2 (xx yy ) sin 2
2
=
(2.57-a)
(2.57-b)
Points (xx , xy ), (xx , 0), (yy , 0) and [(xx +yy )/2, 0] are plotted in the stress representation of Fig.
2.6. Then we observe that
1
(xx yy ) = R cos 2
(2.58-a)
2
(2.58-b)
xy = R sin 2
43
where
R
tan 2
Victor Saouma
=
=
1
2
(xx yy )2 + xy
4
2xy
xx yy
(2.59-a)
(2.59-b)
Draft
211
yy
yx
y
yy
yx
xy
xx
xy
xx
xy
xx
xx
yx
yx
xy
yy
yy
(b)
(a)
n
xy
xx
X( xx xy )
,
xy
xx
x
O
yy
xy
X( xx xy )
,
2 2
2
xx
D
yx
yx
1 ( + )
xx yy
2
yy
1 ( + )
1
2
2
1 ( - )
xx yy
2
1( - )
1
2
2
(c)
(d)
Victor Saouma
Draft
212
KINETICS
then after substitution and simpliation, Eq. 2.57-a and 2.57-b would result in
xx
xy
1
(xx + yy ) + R cos(2 2)
2
R sin(2 2)
(2.60)
(2.61)
We observe that the form of these equations, indicates that xx and xy are on a circle centered at
1
2 (xx + yy ) and of radius R. Furthermore, since xx , yy , R and are denite numbers for a given
state of stress, the previous equations provide a graphical solution for the evaluation of the rotated
stress xx and xy for various angles .
44
By eliminating the trigonometric terms, the Cartesian equation of the circle is given by
1
[ xx (xx + yy )]2 + 2 = R2
xy
2
45
(2.62)
Finally, the graphical solution for the state of stresses at an inclined plane is summarized as follows
1. Plot the points (xx , 0), (yy , 0), C : [ 1 (xx + yy ), 0], and X : (xx , xy ).
2
2. Draw the line CX, this will be the reference line corresponding to a plane in the physical body
whose normal is the positive x direction.
3. Draw a circle with center C and radius R = CX.
4. To determine the point that represents any plane in the physical body with normal making a
counterclockwise angle with the x direction, lay o angle 2 clockwise from CX. The terminal
side CX of this angle intersects the circle in point X whose coordinates are ( xx , xy ).
5. To determine yy , consider the plane whose normal makes an angle + 1 with the positive x axis
2
in the physical plane. The corresponding angle on the circle is 2 + measured clockwise from
the reference line CX. This locates point D which is at the opposite end of the diameter through
X. The coordinates of D are ( yy , xy )
(2.63)
Victor Saouma
=
=
1
(15 5)2 + 42 = 6.403
4
2(4)
= 0.8 2 = 38.66o ;
15 5
(2.64-a)
= 19.33o
(2.64-b)
Draft
213
n
=25.7
6.4
X(15,4) =0
15
=109.3
38.66
=19.3
15
15
80
41.34
4
4
=40
=90
=64.3
10
10.00
3.6
5.19
14.81
o
40
16.4
o
19.3
6.40
4.23
25.7
10.00
(2.65-a)
(2.65-b)
(2.66-a)
(2.66-b)
10 + 6.4 = 16.4
(2.67-a)
(2)
10 6.4 = 3.6
(2.67-b)
(1) acts on a plane dened by the angle of +19.3o clockwise from the x axis, and (2) acts at an
o
o
angle of 38.66 2+180 = 109.3o with respect to the x axis.
7. The maximum and minimum shear stresses are equal to the radius of the circle, i.e 6.4 at an angle
of
90o 38.66o
= 25.70
(2.68)
2
Victor Saouma
Draft
214
2.5.3
KINETICS
46 There can be an innite number of planes passing through a point O, each characterized by their own
normal vector along ON , Fig. 2.8. To each plane will correspond a set of n and n .
Y II
B
E
J
C
Z III
Figure 2.8: Unit Sphere in Physical Body around O
47 It can be shown that all possible sets of n and n which can act on the point O are within the shaded
area of Fig. 2.9.
2.6
48 For many applications of continuum mechanics the problem of determining the three-dimensional
stress distribution is too dicult to solve. However, in many (civil/mechanical)applications, one or more
dimensions is/are small compared to the others and possess certain symmetries of geometrical shape and
load distribution.
49 In those cases, we may apply engineering theories for shells, plates or beams. In those problems,
instead of solving for the stress components throughout the body, we solve for certain stress resultants
(normal, shear forces, and Moments and torsions) resulting from an integration over the body. We
consider separately two of those three cases.
50 Alternatively, if a continuum solution is desired, and engineering theories prove to be either too
restrictive or inapplicable, we can use numerical techniques (such as the Finite Element Method) to
solve the problem.
2.6.1
Arch
51 Fig. 2.10 illustrates the stresses acting on a dierential element of a shell structure. The resulting
forces in turn are shown in Fig. 2.11 and for simplication those acting per unit length of the middle
surface are shown in Fig. 2.12. The net resultant forces are given by:
Victor Saouma
Draft
215
1 ( - )
2
1 ( - )
2
III
O
C
CII
1 ( - )
2
II
C III
1 ( + )
2
1 ( + )
2
1
Victor Saouma
Draft
216
KINETICS
Victor Saouma
Draft
+h
2
z
dz
1
h
r
2
Bending Moments
+h
2
M =
h
2
z 1
z
dz
r
2.6.2
52
z
dz
1
r
h
2
217
xx
Nyy
Nxy
N
yx
Mxx
Myy
Mxy
M
yx
+h
2
=
h
2
+h
2
=
h
2
+h
2
=
h
2
+h
2
=
h
2
+h
2
yy 1
z
rx
dz
xy 1
z
ry
dz
xy 1
z
rx
dz
z
yy z 1
rx
h
2
dz
xx z 1
+h
2
h
2
+h
2
=
h
2
Qy
dz
dz
h
2
+h
2
Qx
z
ry
z
ry
xx 1
+h
2
h
2
+h
2
h
2
xy z 1
xy z 1
z
ry
z
rx
xz 1
z
ry
z
rx
dz
dz
dz
yz 1
(2.69)
dz
Plates
Considering an arbitrary plate, the stresses and resulting forces are shown in Fig. 2.13, and resultants
Victor Saouma
Draft
218
KINETICS
t
2
Membrane Force
N =
t
2
t
2
Bending Moments
M =
t
2
t
2
=
t
2
dz
zdz
dz
Nxx =
=
N
yy
xy =
xx =
Myy =
M
xy =
Vx =
V
y
t
2
t
2
xx dz
t
2
t
2
yy dz
t
2
t
2
xy dz
t
2
t
2
xx zdz
(2.70-a)
t
2
t
2
yy zdz
t
2
t
2
t
2
t
2
xy zdz
xz dz
t
2
=
t
2
yz dz
53 Note that in plate theory, we ignore the eect of the membrane forces, those in turn will be accounted
for in shells.
Victor Saouma
Draft
Chapter 3
MATHEMATICAL
PRELIMINARIES; Part II
VECTOR DIFFERENTIATION
3.1
Introduction
A eld is a function dened over a continuous region. This includes, Scalar Field g(x), Vector
Field v(x), Fig. 3.1 or Tensor Field T(x).
i+
j+
k
x
y
z
(3.1)
We also note that there are as many ways to dierentiate a vector eld as there are ways of multiplying
vectors, the analogy being given by Table 3.1.
Multiplication
uv
dot
uv
cross
u v tensor
Dierentiation
v
divergence
v curl
v
gradient
Tensor Order
?
6
3.2
4
The derivative of a vector p(u) with respect to a scalar u, Fig. 3.2 is dened by
dp
p(u + u) p(u)
lim
du u0
u
(3.2)
(3.3)
Draft
32
-1
-2
-2
-1
ContourGraphics
Plot3D@Exp@Hx ^ 2 + y ^ 2LD, 8x, 2, 2<, 8y, 2, 2<, FaceGrids > AllD
1
0.75
0.5
0.25
0
-2
2
2
1
0
-1
-1
0
1
2 -2
SurfaceGraphics
)
u
(u+
p
p (u)
Draft
33
dp
dt
is the velocity
In dierential geometry, if we consider a curve C dened by the function p(u) then dp is a vector
du
tangent ot C, and if u is the curvilinear coordinate s measured from any point along the curve, then dp
ds
is a unit tangent vector to C T, Fig. 3.3. and we have the following relations
7
dp
ds
(3.4)
(3.5)
(3.6)
curvature
(3.7)
1
Radius of Curvature (3.8)
=
= 0.
dp
dt
dp
dt
=
=
T =
=
d
(t2 + 1)i + (4t 3)j + (2t2 6t)k = 2ti + 4j + (4t 6)k
dt
(2t)2 + (4)2 + (4t 6)2
(3.9-a)
(3.9-b)
(3.9-c)
(3.9-d)
Victor Saouma
Draft
34
Parametric Plot in 3D
ParametricPlot3D@8t ^ 2 + 1, 4 t 3, 2 t ^ 2 6 t<, 8t, 0, 4<D
10
5
0
0
5
10
15
Graphics3D
3.3
3.3.1
Divergence
Vector
The divergence of a vector eld of a body B with boundary , Fig. 3.5 is dened by considering
that each point of the surface has a normal n, and that the body is surrounded by a vector eld v(x).
The volume of the body is v(B).
8
v(x)
v(B)0
1
v(B)
vndA
(3.10)
where v.n is often referred as the ux and represents the total volume of uid that passes through
dA in unit time, Fig. 3.6 This volume is then equal to the base of the cylinder dA times the height of
Victor Saouma
Draft
3.3 Divergence
35
n
dA
v
v.n
11 The denition is clearly independent of the shape of the solid region, however we can gain an insight
into the divergence by considering a rectangular parallelepiped with sides x1 , x2 , and x3 , and with
normal vectors pointing in the directions of the coordinate axies, Fig. 3.7. If we also consider the corner
x3
-e
e3
-e
x3
e2
e1
x2
-e
x2
x1
x1
Figure 3.7: Innitesimal Element for the Evaluation of the Divergence
closest to the origin as located at x, then the contribution (from Eq. 3.10) of the two surfaces with
normal vectors e1 and e1 is
1
x1 ,x2 ,x3 0 x1 x2 x3
lim
(3.11)
x2 x3
or
lim
x1 ,x2 ,x3 0
1
x2 x3
v(x + x1 e1 ) v(x)
e1 dx2 dx3
x1
x2 x3
=
=
lim
x1 0
v
e1 (3.12-a)
x1
v
e1
x1
(3.12-b)
Victor Saouma
v(x)
ei
xi
(3.13)
Draft
36
12
or alternatively
div v
=
=
e1 +
e2 +
e3 )(v1 e1 + v2 e2 + v3 e3 ) (3.14)
x1
x2
x3
v1
v2
v3
vi
+
+
=
= i vi = vi,i
(3.15)
x1
x2
x3
xi
v = (
13
14
(3.16)
F = F,ii
i+
j+
k (x2 zi 2y 3 z 2 j + xy 2 zk)
x
y
z
2y 3 z 2
xy 2 z
x2 z
+
+
=
x
y
z
2 2
2
= 2xz 6y z + xy
= 2(1)(1) 6(1)2 (1)2 + (1)(1)2 = 3 at (1, 1, 1)
=
(3.17-a)
(3.17-b)
(3.17-c)
(3.17-d)
3.3.2
15
Second-Order Tensor
v(B)0
1
v(B)
TndA
(3.18)
3.4
3.4.1
Tpq
eq
xp
(3.19)
Gradient
Scalar
dg
=
ds
Victor Saouma
gv
(3.20)
Introduction to Continuum Mechanics
Draft
3.4 Gradient
37
mdiver.nb
Divergence of a Vector
<< CalculusVectorAnalysis
V = 8x ^ 2 z, 2 y ^ 3 z ^ 2, x y ^ 2 z<;
Div@V, Cartesian@x, y, zDD
-6 z2 y2 + x y2 + 2 x z
<< GraphicsPlotField3D
PlotVectorField3D@8x ^ 2 z, 2 y ^ 3 z ^ 2, x y ^ 2 z<, 8x, 10, 10<, 8y, 10, 10<, 8z, 10, 10<,
Axes > Automatic, AxesLabel > 8"X", "Y", "Z"<D
10
5
0
-5
-10
10
10
5
Z
0
-5
-10
-10
-5
0
X
5
10
Graphics3D
Div@Curl@V, Cartesian@x, y, zDD, Cartesian@x, y, zDD
0
where v = dp We note that the denition made no reference to any coordinate system. The gradient is
ds
thus a vector invariant.
17
dg
ds
dxi
dp
=
ei
ds
ds
g dxi
xi ds
(3.21-a)
(3.21-b)
g
ei
xi
(3.22)
or
=
i+
j+
k
x
y
z
i+
j+
k
x
y
z
(3.23-a)
(3.23-b)
Victor Saouma
Draft
38
19
g(x)n gives the rate of change of the scalar eld in the direction of n.
(8i j 10k)
1
2
2
i j k
3
3
3
16 1 20
37
+ +
=
3
3
3
3
(3.24-a)
(3.24-b)
(3.24-c)
(3.24-d)
3x1 x2 5x2
0
2
0
2x2
= 5x2
2
3
0
2x3
0
(3.25)
Determine the stress vector (or traction) at the point P (2, 1, 3) of the plane that is tangent to the
cylindrical surface x2 + x2 = 4 at P , Fig. 3.9.
2
3
x3
n
x2
P
3
1
x1
Victor Saouma
Draft
3.4 Gradient
39
Solution:
At point P , the stress tensor is given by
6
5
0
= 5
0 2 3
2 3
0
(3.26)
(3.27)
(3.28)
At point P ,
1
3
e3
n = e1 +
2
2
Thus the traction vector will be determined from
6
5
0
0 5/2
3
1/2
0
=
= 5
2 3
3/2
3
0
0 2 3
and thus the unit normal at P is
or tn = 5 e1 + 3e2 +
2
3.4.2
(3.29)
(3.30)
3e3
Vector
We can also dene the gradient of a vector eld. If we consider a solid domain B with boundary ,
Fig. 3.5, then the gradient of the vector eld v(x) is a second order tensor dened by
20
x v(x)
lim
v(B)0
1
v(B)
v ndA
(3.31)
and with a construction similar to the one used for the divergence, it can be shown that
x v(x)
vi (x)
[ei ej ]
xj
(3.32)
x v]
[v
Victor Saouma
x]
vx
x
vx
y
vx
z
vx
x
vy
x
vz
x
x
vy
y
vy
z
vx
y
vy
y
vz
y
vz
x
vz
y
vz
z
vx
z
vy
z
vz
z
(3.33)
(3.34)
Draft
310
that is [ v]ij gives the rate of change of the ith component of v with respect to the jth coordinate axis.
22 Note the diference between v
other.
23
and
x v.
24 We can interpret the gradient of a vector geometrically, Fig. 3.10. If we consider two points a and b
that are near to each other (i.e s is very small), and let the unit vector m points in the direction from a
to b. The value of the vector eld at a is v(x) and the value of the vector eld at b is v(x + sm). Since
the vector eld changes with position in the domain, those two vectors are dierent both in length and
orientation. If we now transport a copy of v(x) and place it at b, then we compare the dierences between
those two vectors. The vector connecting the heads of v(x) and v(x + sm) is v(x + sm) v(x), the
change in vector. Thus, if we divide this change by s, then we get the rate of change as we move in
the specied direction. Finally, taking the limit as s goes to zero, we obtain
lim
s0
(3.35)
v(x+ s m ) -v(x)
v(x+ s m )
x3
v(x)
a sm b
x2
x1
Figure 3.10: Gradient of a Vector
The quantity Dv(x)m is called the directional derivative because it gives the rate of change of
the vector eld as we move in the direction m.
x v(x)
e1 ] +
e2 ] + 2x1 x2 x3 [e3 e3 ]
2
x1 /x2 x1 /x3
2
x2 /x3
= x1 x2 x3 x2 /x1
x3 /x1 x3 /x2
2
+x2 x2 [e3
3
Victor Saouma
(3.36-a)
x1 x2 [e3
3
(3.36-b)
Draft
3.5 Curl
3.4.3
25
311
Mathematica Solution
2
1
PlotVectorField3D@vecfield, 8x1, -10, 10<, 8x2, -10, 10<, 8x3, -10, 10<, Axes -> Automatic,
AxesLabel -> 8"x1", "x2", "x3"<D
Gradient
x2
10
Scalar
-10
f = x ^ 2 y z + 4 x z ^ 2;
10
Gradf = Grad@f, Cartesian@x, y, zDD
8 x3+ 2 x y z, x2 z, y x2 + 8 z x<
4 z2
<< GraphicsPlotField3D
PlotGradientField3D@f, 8x, 0, 2<, 8y, -3, -1<, 8z, -2, 0<D
-10
-10
0
x1
10
Graphics3D
MatrixForm@Grad@vecfield, Cartesian@x1, x2, x3DDD
i2 x1 x2 x3 x12 x3
x12 x2 y
j
z
j
z
j
j
z
j x22 x3 2 x1 x2 x3 x1 x22 z
z
j
z
j
z
j
z
j
z
j
z
j
z
2
2
x1 x3
2 x1 x2 x3 {
k x2 x3
Graphics3D
x = 1; y = -2; z = -1;
vect = 82, -1, -2< Sqrt@4 + 1 + 4D
2
1
2
9 ,- ,- =
3
3
3
Gradf . vect
37
3
Gradient of a Vector
vecfield = x1 x2 x3 8x1, x2, x3<
Figure 3.11: Mathematica Solution for the Gradients of a Scalar and of a Vector
8 2 x2 x3, x1 x22 x3, x1 x2 x32 <
x1
3.5
Curl
26
curl v
v =
e1
x1
v1
=
=
Victor Saouma
v2
v3
x2
x3
eijk j vk
e2
x2
v2
e1 +
e3
x3
(3.37)
v3
v3
v1
x3
x1
e2 +
v1
v2
x1
x2
e3
(3.38)
(3.39)
Draft
312
i+
j+
k (xz 3 i 2x2 yzj + 2yz 4 k)
x
y
z
A =
i
=
x
3
y
2
2x yz
xz
(3.40-b)
2yz 4
2x2 yz
2yz 4
y
z
(3.40-a)
i+
2yz 4
xz 3
z
x
xz 3
2x2 yz
x
y
k (3.40-c)
j+
3j + 4k at (1, 1, 1)
(3.40-d)
(3.40-e)
Curl
<< CalculusVectorAnalysis
A = 8x z ^ 3, 2 x ^ 2 y z, 2 y z ^ 4<;
CurlOfA = Curl@A, Cartesian@x, y, zDD
82 z4 + 2 x2 y, 3 x z2 , -4 x y z<
<< GraphicsPlotField3D
PlotVectorField3D@CurlOfA, 8x, 0, 2<, 8y, 2, 0<, 8z, 0, 2<, Axes > Automatic, AxesLabel > 8"x", "y", "z"<D
0
y
-0.5
-1
-1.5
-2
2
2
1.5
z
1
0.5
0
0
0.5
1
1.5
x
Graphics3D
Div@CurlOfA, Cartesian@x, y, zDD
0
x = 1; y = 1; z = 1;
CurlOfA
80, 3, 4<
Victor Saouma
Draft
3.6
27
313
= AdB + dAB
(3.41-a)
= AdB + dAB
(3.41-b)
( + )
(A + B)
A +
v
(A)
(3.41-c)
B
(3.41-d)
= v
= ( )A + ( A)
(3.41-e)
(3.41-f)
(AB) = B( A) A( B)
(AB) = (B )A + (A )B + B( A) + A( B)
2 2 2
2
+ 2 + 2 Laplacian Operator
( )
x2
y
z
( A) = 0
( )
Victor Saouma
= 0
(3.41-g)
(3.41-h)
(3.41-i)
(3.41-j)
(3.41-k)
Draft
314
Victor Saouma
Draft
Chapter 4
KINEMATIC
Or on How Bodies Deform
4.1
We begin our detailed coverage of strain by a simplied and elementary set of denitions for the 1D
and 2D cases. Following this a mathematically rigorous derivation of the various expressions for strain
will follow.
4.1.1
2 We begin by considering an elementary case, an axial rod with initial lenght l0 , and subjected to a
deformation l into a nal deformed length of l, Fig. 4.1.
l0
l
Figure 4.1: Elongation of an Axial Rod
3 We seek to quantify the deformation of the rod and even though we only have 2 variables (l0 and l),
there are dierent possibilities to introduce the notion of strain. We rst dene the stretch of the rod
as
l
(4.1)
l0
This stretch is one in the undeformed case, and greater than one when the rod is elongated.
4
Using l0 , l and we next introduce four possible denitions of the strain in 1D:
Draft
42
KINEMATIC
Engineering Strain
Natural Strain
Lagrangian Strain
Eulerian Strain E
l l0
=1
l0
1
l l0
=1
2
1 l 2 l0
=
2
2
l0
2
1 l 2 l0
=
2
l2
(4.2)
(4.3)
1 2
( 1)
2
1
1
1 2
2
(4.4)
(4.5)
we note the strong analogy between the Lagrangian and the engineering strain on the one hand, and the
Eulerian and the natural strain on the other.
The choice of which strain denition to use is related to the stress-strain relation (or constitutive law)
that we will later adopt.
4.1.2
6
Small Strains in 2D
The elementary denition of strains in 2D is illustrated by Fig. 4.2 and are given by
xx
yy
xy
xy
ux
X
uy
Y
uy
ux
= 2 + 1 =
+
2
Y
X
uy
1
1 ux
xy
+
2
2 Y
X
(4.6-a)
(4.6-b)
(4.6-c)
(4.6-d)
ux
;
X
yy =
uy
;
Y
xy =
1
1
xy =
2
2
uy
ux
+
Y
X
(4.7)
We note that in the expression of the shear strain, we used tan which is applicable as long as is
small compared to one radian.
We have used capital letters to represent the coordinates in the initial state, and lower case letters
for the nal or current position coordinates (x = X + ux ). This corresponds to the Lagrangian strain
representation.
7
4.2
Strain Tensor
Following the simplied (and restrictive) introduction to strain, we now turn our attention to a rigorous
presentation of this important deformation tensor.
Victor Saouma
Draft
Victor Saouma
1 u y
43
ux
Draft
44
KINEMATIC
5. Introduce the notion of a material displacement gradient and spatial displacement gradient.
6. Dene Cauchys and Greens deformation tensors (in terms of (dX)2 and (dx)2 respectively.
7. Introduce the notion of strain tensor in terms of (dx)2 (dX)2 as a measure of deformation
in terms of either spatial coordinates or in terms of displacements.
4.2.1
We consider in Fig. 4.3 the undeformed conguration of a material continuum at time t = 0 together
with the deformed conguration at coordinates for each conguration.
10
11
(4.8)
13
14 The relative orientation of the material axes (OX1 X2 X3 ) and the spatial axes (ox1 x2 x3 ) is specied
through the direction cosines aX .
x
15 The displacement vector u connecting P0 and P is the displacement vector which can be expressed
in both the material or spatial coordinates
U = UK IK
u = u k ik
(4.10-a)
(4.10-b)
again Uk and uk are interrelated through the direction cosines ik = aK IK . Substituting above we
k
obtain
(4.11)
u = uk (aK IK ) = UK IK = U UK = aK uk
k
k
16 The vector b relates the origin o with respect to O. From geometry X+u = b+x, thus u = b+xX
or if the origins are the same (superimposed axis), Fig. 4.4:
u = xX
uk = xk aK XK
k
(4.12-a)
(4.12-b)
kK
uk
xk Xk
(4.12-c)
Victor Saouma
Draft
45
Spatial
X1
I3
O
Material
I2
X3
t=0
P0
x3
X2
x1
i3
o
i1
t=t
i2
x2
Victor Saouma
Draft
46
KINEMATIC
t=t
X
,x
t=0
U
P0
I 3 , 3i
Oo
I
I 2 , 2i
X2
,i
,x
2
Material/Spatial
X1
u
x
b=0
,x
1
= x1 X1 = 0
= x2 X2 = AX3
= x3 X3 = AX2
1 0 0 X1
x1
x2
X2
= 0 1 A
x3
X3
0 A 1
or upon inversion
1 A2
X1
1
X2
0
=
1 A2
X3
0
0
0
x1
x2
1
A
x3
A
1
(4.13-a)
(4.13-b)
(4.13-c)
(4.14)
(4.15)
u3
Victor Saouma
x1 X1 = 0
A(x3 Ax2 )
x2 X2 =
1 A2
A(x2 Ax3 )
x3 X3 =
a A2
(4.16-a)
(4.16-b)
(4.16-c)
Draft
47
4. The displacements for the edge of the cube are determined as follows:
(a) L1 : Edge X1 = X1 , X2 = X3 = 0, u1 = u2 = u3 = 0
(b) L2 :X1 = X3 = 0, X2 = X2 , u1 = u2 = 0, u3 = AX2 .
(c) L3 : Edge X1 = X2 = 0, X3 = X3 , u1 = u3 = 0, u2 = AX3 , thus points along this edge are
displaced in the X2 direction proportionally to their distance from the origin.
5. For the circular surface, and by direct substitution of X2 = (x2 Ax3 )/(1 A2 ), and X3 =
2
2
(x3 Ax2 )/(1 A2 ) in X2 + X3 = 1/(1 A2 ), the circular surface becomes the elliptical surface
(1 + A2 )x2 4Ax2 x3 + (1 + A2 )x2 = (1 A2 ) or for A = 1/2, 5x2 8x2 x3 + 5x2 = 3 . When
2
3
2
3
expressed in its principal axes, Xi (at /4), it has the equation x2 + 9x2 = 3
2
3
X3
*
X
2
dX
1/ 3
L
2/ 3
dX
dX
2
X +X =4/3
4.2.1.1
17 When the continuum undergoes deformation (or ow), the particles in the continuum move along
various paths which can be expressed in either the material coordinates or in the spatial coordinates
system giving rise to two dierent formulations:
Lagrangian Formulation: gives the present location xi of the particle that occupied the point (X1 X2 X3 )
at time t = 0, and is a mapping of the initial conguration into the current one.
xi = xi (X1 , X2 , X3 , t) or x = x(X, t)
(4.17)
Eulerian Formulation: provides a tracing of its original position of the particle that now occupies the
location (x1 , x2 , x3 ) at time t, and is a mapping of the current conguration into the initial one.
Xi = Xi (x1 , x2 , x3 , t) or X = X(x, t)
(4.18)
(X, t) and (x, t) are the Lagrangian and Eulerian variables respectivly.
19
If X(x, t) is linear, then the deformation is said to be homogeneous and plane sections remain plane.
Victor Saouma
Draft
48
KINEMATIC
20 For both formulation to constitute a one-to-one mapping, with continuous partial derivatives, they
must be the unique inverses of one another. A necessary and unique condition for the inverse functions
to exist is that the determinant of the Jacobian should not vanish
|J| =
xi
=0
Xi
(4.19)
(4.20)
x1 + x2 (et 1)
x1 (et 1) x2
; X2 =
; X3 = x3
t et
1e
1 et et
(4.21)
(e2 1)
(e2 e2 )
e2
= e2 = 0
(4.22)
1 0
0 1
0 0
1
(e2 1)
1
(e2 e2 ) = 0
e2
0
4.2.2
Deformation; (x
(e2 1)
X1
X2
(e4 1)
X3
e2
=
=
=
x1 + (e2 1)x3
x2 + (e4 1)x3
e2 x3
(4.23)
Gradients
4.2.2.1
0
1
0
X, X
x)
Fx
The matrix form of F is
x1
x2
F=
x3
Victor Saouma
x
x
x
xi
e1 +
e2 +
e3 =
X1
X2
X3
Xj
X1
X2
X3
x1
X1
x2
X1
x3
X1
x1
X2
x2
X2
x3
X2
x1
X3
x2
X3
x3
X3
(4.24)
xi
Xj
(4.25)
Draft
22
49
Similarly, dierentiation of Eq. 4.18 with respect to xj produces the spatial deformation gradient
H=X
X1
X2
H=
X3
23
X
X
X
Xi
e1 +
e2 +
e3 =
x1
x2
x3
xj
x1
x2
x3
X1
x1
X2
x1
X3
x1
X1
x2
X2
x2
X3
x2
X1
x3
X2
x3
X3
x3
(4.26)
Xi
=
xj
(4.27)
The material and spatial deformation tensors are interrelated through the chain rule
Xi xj
xi Xj
=
= ik
Xj xk
xj Xk
(4.28)
H = F1
(4.29)
and thus F 1 = H or
24 The deformation gradient characterizes the rate of change of deformation with respect to coordinates.
It reects the stretching and rotation of the domain in the innitesimal neighborhood at point x (or X).
25 The deformation gradient is often called a two point tensor because the basis ei Ej has one leg
in the spatial (deformed), and the other in the material (undeformed) conguration.
F is a tensor of order two which when operating on a unit tangent vector in the undeformed conguration will produce a tangent vector in the deformed conguration. Similarly H is a tensor of order two
which when operating on a unit tangent vector in the deformed conguration will produce a tangent
vector in the undeformed conguration.
26
4.2.2.1.1 Change of Area Due to Deformation 27 In order to facilitate the derivation of the
Piola-Kircho stress tensor later on, we need to derive an expression for the change in area due to
deformation.
If we consider two material element dX(1) = dX1 e1 and dX(2) = dX2 e2 emanating from X, the
rectangular area formed by them at the reference time t0 is
28
(4.30)
At time t, dX(1) deforms into dx(1) = FdX(1) and dX(2) into dx(2) = FdX(2) , and the new area is
dA
(4.31-a)
(4.31-b)
where the orientation of the deformed area is normal to Fe1 and Fe2 which is denoted by the unit
vector n. Thus,
(4.32)
Fe1 dAn = Fe2 dAn = 0
and recalling that abc is equal to the determinant whose rows are components of a, b, and c,
Fe3 dA = dA0 (Fe3 Fe1 Fe2 )
(4.33)
det(F)
Victor Saouma
Draft
410
KINEMATIC
or
e3 FT n =
dA0
det(F)
dA
(4.34)
dA0
det Fe3 dAn = dA0 det(F)(F1 )T e3
dA
(4.35)
which implies that the deformed area has a normal in the direction of (F1 )T e3 . A generalization of the
preceding equation would yield
dAn = dA0 det(F)(F1 )T n0
30
(4.36)
(4.37)
(4.38)
in spatial cordiantes:
If we dene
Fi =
xi
ei
Xj
(4.39)
(4.40)
d = det Fd0
(4.41)
or
and J is called the Jacobian and is the determinant of the deformation gradient F
J=
x1
X1
x2
X1
x3
X1
x1
X2
x2
X2
x3
X2
x1
X3
x2
X3
x3
X3
(4.42)
[F]
det F
Victor Saouma
1
0
0
1 2 3
0
0
2
0
3
0
(4.43-a)
(4.43-b)
Introduction to Continuum Mechanics
Draft
411
V
1 2 3
(4.43-c)
A0
n0
=
=
1
e3
(4.43-d)
(4.43-e)
An
(1)(det F)(F1 )T
1
0
1
0
0
= 1 2 3
1
0 2
=
0
0 0
1
3
0
1 2
=
0
1
0
An = 1 2 e2
4.2.2.2
(4.43-f)
(4.43-g)
(4.43-h)
Displacements; (u
X, u
x)
32 We now turn our attention to the displacement vector ui as given by Eq. 4.12-c. Partial dierentiation
of Eq. 4.12-c with respect to Xj produces the material displacement gradient
xi
ui
=
ij or J u
Xj
Xj
The matrix form of J is
u1
u2
J =
u3
X1
X2
X3
u1
X1
u2
X1
u3
X1
=FI
u1
X2
u2
X2
u3
X2
u1
X3
u2
X3
u3
X3
(4.44)
ui
Xj
(4.45)
33 Similarly, dierentiation of Eq. 4.12-c with respect to xj produces the spatial displacement gradient
Xi
ui
= ij
or K u
xj
xj
The matrix form of K is
u1
u2
K=
u3
4.2.2.3
x1
x2
x3
u1
x1
u2
x1
u3
x1
=IH
u1
x2
u2
x2
u3
x2
u1
x3
u2
x3
u3
x3
(4.46)
ui
=
xj
(4.47)
Examples
Victor Saouma
x2
(4.48)
x3
Draft
412
KINEMATIC
2. Thus
F
x1
X1
x2
X1
x3
X1
x
x
x
xi
e1 +
e2 +
e3 =
X1
X2
X3
Xj
x1
X2
x2
X2
x3
X2
2
1 + X3
2X1 X2
0
0
2
1 + X1
2X2 X3
x1
X3
x2
X3
x3
X3
(4.49-b)
2X1 X3
0
2
1 + X2
uX1
X1
uX2
X1
uX3
X1
2
X3
(4.49-a)
uX1
X2
uX2
X2
uX3
X2
2X1 X2
0
(4.49-c)
uX1
X3
uX2
X3
uX3
X3
0
2
X1
2X2 X3
2X1 X3
0
2
X2
(4.50-a)
(4.50-b)
4.2.3
Deformation Tensors
34 The deofrmation gradients, previously presented, can not be used to determine strains as embedded
in them is rigid body motion.
x
Having derived expressions for Xij and Xji we now seek to determine dx2 and dX 2 where dX and dx
x
correspond to the distance between points P and Q in the undeformed and deformed cases respectively.
35
36 We consider next the initial (undeformed) and nal (deformed) conguration of a continuum in which
the material OX1 , X2 , X3 and spatial coordinates ox1 x2 x3 are superimposed. Neighboring particles P0
and Q0 in the initial congurations moved to P and Q respectively in the nal one, Fig. 4.5.
4.2.3.1
38
This tensor is the inverse of the tensor B which will not be introduced until Sect. 4.3.2.
39
(4.51)
Xi
dxj or dX = Hdx
xj
(4.52)
Victor Saouma
Xk Xk
1
dxi dxj = Bij dxi dxj
xi xj
(4.53-a)
(dX)2
dxB1 dx
(4.53-b)
Introduction to Continuum Mechanics
Draft
413
P
0
dX
u +du
X 1, x 1
t=0
X 3 , x3
X+
d
t=t
Q
dx
X 2, x 2
Victor Saouma
Draft
414
KINEMATIC
Xk Xk
or B1 =
xi xj
x XX
(4.54)
Hc H
41 The square of the dierential element connecting Po and Q0 is now evaluated in terms of the spatial
coordinates
(4.55)
(dx)2 = dxdx = dxi dxi
dxi =
(4.56)
=
=
xk xk
dXi dXj = Cij dXi dXj
Xi Xj
dXCdX
(4.57-a)
(4.57-b)
xk xk
or C =
Xi Xj
X xx
(4.58)
Fc F
is Greens deformation tensor also known as metric tensor, or deformation tensor or right
Cauchy-Green deformation tensor. It relates (dx)2 to (dX)2 .
42
(4.59)
Victor Saouma
xi
Xj
1 0
= 0 1
0 A
=
(4.60-a)
0
A
(4.60-b)
1
Introduction to Continuum Mechanics
Draft
415
and thus
C =
=
4.2.4
Fc F
T
1 0 0
1
0 1 A 0
0 A 1
0
0 0
1
1 A = 0
0
A 1
(4.61-a)
0
1 + A2
2A
0
2A
1 + A2
(4.61-b)
With (dx)2 and (dX)2 dened we can now nally introduce the concept of strain through (dx)2
(dX)2 .
43
4.2.4.1
44 We start with the most general case of nite strains where no constraints are imposed on the deformation (small).
4.2.4.1.1
The dierence (dx)2 (dX)2 for two neighboring particles in a continuum is used as the measure
of deformation. Using Eqs. 4.57-a and 4.51 this dierence is expressed as
45
(dx)2 (dX)2
=
=
xk xk
ij dXi dXj = 2Eij dXi dXj
Xi Xj
dX(Fc F I)dX = 2dXEdX
(4.62-a)
(4.62-b)
1
2
xk xk
ij
Xi Xj
or E =
1
(
2
X xx
I)
(4.63)
Fc F=C
is called the Lagrangian (or Greens) nite strain tensor which was introduced by Green in 1841
and St-Venant in 1844.
46
The Lagrangian stress tensor is one half the dierence between the Green deformation tensor and I.
47 Note similarity with Eq. 4.4 where the Lagrangian strain (in 1D) was dened as the dierence between
the square of the deformed length and the square of the original length divided by twice the square of
2
l2 l0
). Eq. 4.62-a can be rewritten as
the original length (E 1
2
l2
0
(dx)2 (dX)2
2dXi dXj
(4.64)
Eij =
Victor Saouma
1
2
uj
uk uk
ui
+
+
Xj
Xi
Xi Xj
or E =
1
(u
2
Xu +
X uu
X)
(4.65)
J+Jc
Jc J
Introduction to Continuum Mechanics
Draft
416
KINEMATIC
or:
E11
E12
u1
1
+
X1
2
u1
X1
1 u1
u2
+
2 X2
X1
+
+
u2
X1
u3
X1
(4.66-a)
1 u1 u1
u2 u2
u3 u3
+
+
2 X1 X2
X1 X2
X1 X2
(4.66-b)
(4.66-c)
1
0
0
2A
C = 0 1 + A2
0
2A
1 + A2
1
(C I)
E =
2
0 0
0
1
0 A2 2A
=
2
0 2A A2
(4.67-a)
(4.67-b)
(4.67-c)
4.2.4.1.2
Eulerian/Almansis Tensor
Alternatively, the dierence (dx)2 (dX)2 for the two neighboring particles in the continuum can be
expressed in terms of Eqs. 4.55 and 4.53-b this same dierence is now equal to
49
(dx)2 (dX)2
Xk Xk
ij
(4.68-a)
(4.68-b)
Eij =
1
2
ij
Xk Xk
xi xj
or E =
1
(I
2
x XX
Hc
x)
(4.69)
H=B1
The Eulerian strain tensor is one half the dierence between I and the Cauchy deformation tensor.
51 Note similarity with Eq. 4.5 where the Eulerian strain (in 1D) was dened as the dierence between
the square of the deformed length and the square of the original length divided by twice the square of
2
l2 l0
). Eq. 4.68-a can be rewritten as
the deformed length (E 1
2
l2
Victor Saouma
(dx)2 (dX)2
2dxi dxj
(4.70)
Draft
417
For innitesimal strain it was introduced by Cauchy in 1827, and for nite strain by Almansi in 1911.
53 To express the Eulerian tensor in terms of the displacements, we substitute 4.46 in the preceding
equation, and after some simple algebraic manipulations, the Eulerian nite strain tensor can be rewritten
as
Eij =
54
1
2
uj
uk uk
ui
+
xj
xi
xi xj
or E =
1
(u
2
xu
x uu
x)
(4.71)
Kc K
K+Kc
Expanding
E11
E12
4.2.4.2
u1
1
x1
2
u1
x1
1 u1
u2
+
2 x2
x1
u2
x1
u3
x1
(4.72-a)
1 u1 u1
u2 u2
u3 u3
+
+
2 x1 x2
x1 x2
x1 x2
(4.72-b)
(4.72-c)
55 The small deformation theory of continuum mechanics has as basic condition the requirement that
the displacement gradients be small compared to unity. The fundamental measure of deformation is the
dierence (dx)2 (dX)2 , which may be expressed in terms of the displacement gradients by inserting Eq.
4.65 and 4.71 into 4.62-b and 4.68-b respectively. If the displacement gradients are small, the nite strain
tensors in Eq. 4.62-b and 4.68-b reduce to innitesimal strain tensors and the resulting equations
represent small deformations.
For instance, if we were to evaluate + 2 , for = 103 and 101 , then we would obtain 0.001001
0.001 and 0.11 respectively. In the rst case 2 is negligible compared to , in the other it is not.
56
4.2.4.2.1
u
In Eq. 4.65 if the displacement gradient components Xi are each small compared to unity, then the
j
third term are negligible and may be dropped. The resulting tensor is the Lagrangian innitesimal
strain tensor denoted by
57
Eij =
1
2
uj
ui
+
Xj
Xi
or E =
1
(u
2
X u)
(4.73)
J+Jc
or:
E11
E12
u1
X1
1 u1
u2
+
2 X2
X1
(4.74-a)
(4.74-b)
(4.74-c)
Victor Saouma
Draft
418
4.2.4.2.2
KINEMATIC
ui
Similarly, inn Eq. 4.71 if the displacement gradient components xj are each small compared to
unity, then the third term are negligible and may be dropped. The resulting tensor is the Eulerian
innitesimal strain tensor denoted by
58
Eij =
59
1
2
uj
ui
+
xj
xi
or E =
1
(u
2
x u)
(4.75)
K+Kc
Expanding
E11
E12
4.2.4.3
u1
x1
1 u1
u2
+
2 x2
x1
(4.76-a)
(4.76-b)
(4.76-c)
Examples
x1 X1 = X1 + AX2 X1 = AX2
(4.77-a)
u2
u3
=
=
x2 X2 = X2 + AX3 X2 = AX3
x3 X3 = X3 + AX1 X3 = AX1
(4.77-b)
(4.77-c)
Ju
From Eq. 4.73:
0
= 0
A
0
2E = (J + Jc ) = 0
A
0 A A
= A 0 A
A A 0
To determine the Eulerian tensor,
displacement eld given above:
1 A 0 X1
x1
x2
X2
= 0 1 A
x3
X3
A 0 1
Victor Saouma
A
0
0
0
A
0
A 0
0
0 A + A
0 0
0
(4.78)
0
0
A
A
0
0
(4.79-a)
(4.79-b)
1
A A2
x1
X1
1 2
X2
A
1
A
x2
1 + A3
X3
1
x3
A A2
Introduction to Continuum
(4.80)
Mechanics
Draft
419
u2
u3
1
A(A2 x1 + x2 Ax3 )
(x1 Ax2 + A2 x3 ) =
1 + A3
1 + A3
1
A(Ax1 + A2 x2 + x3 )
x2 X2 = x2
(A2 x1 + x2 Ax3 ) =
3
1+A
1 + A3
1
A(x1 Ax2 + A2 x3 )
x3 X3 = x3
(Ax1 + A2 x2 + x3 ) =
1 + A3
1 + A3
x1 X1 = x1
1
A
A
A
1
A A2
=
1 + A3
1
A A2
(4.81-a)
(4.81-b)
(4.81-c)
(4.82)
=
=
K + Kc
A
1 + A3
A
1 + A3
A2
A
1
A
A
1
A
1 +
A A2
1
A2 A
1 + A3
1
A A2
A
1
A2
2A2 1 A 1 A
1 A 2A2 1 A
1 A 1 A 2A2
(4.83-a)
(4.83-b)
(4.83-c)
as A is very small, A2 and higher power may be neglected with the results, then E E.
4.2.5
4.2.5.1
Small Strain
60 We nally show that the linear lagrangian tensor in small deformation Eij is nothing else than the
strain as was dened earlier in Eq.4.7.
61
=
or
=
(4.84-a)
(4.84-b)
but since dx dX under current assumption of small deformation, then the previous equation can be
rewritten as
du
dXi dXj
dx dX
= Eij
= Eij i j = E
dX
dX dX
(4.85)
62 We recognize that the left hand side is nothing else than the change in length per unit original length,
and is called the normal strain for the line element having direction cosines dXi .
dX
63 With reference to Fig. 4.6 we consider two cases: normal and shear strain.
Normal Strain: When Eq. 4.85 is applied to the dierential element P0 Q0 which lies along the X2
axis, the result will be the normal strain because since dX1 = dX3 = 0 and dX2 = 1. Therefore,
dX
dX
dX
Eq. 4.85 becomes (with ui = xi Xi ):
u2
dx dX
= E22 =
dX
X2
Victor Saouma
(4.86)
Draft
420
KINEMATIC
X3
x3
Q0
X2
dX
Normal
e3
n3
n2
X1
e2
X3
e1
M0
dX
Shear
P0
Q0
dX
X1
n3
P0
u1
u3
e1 + e2 +
e3
X2
X2
u1
u2
e1 +
e2 + e3
X3
X3
(4.87-a)
(4.87-b)
u1 u1
u2
u3
+
+
X2 X3
X3
X2
(4.88)
u2
u3
+
= 2E23
X3
X2
(4.89)
64 Finally taking the change in right angle between the elements as 23 = /2 , and recalling
that for small strain theory 23 is very small it follows that
(4.90)
Draft
421
Therefore the o diagonal terms of the linear strain tensor represent one half of the angle change
between two line elements originally at right angles to one another. These components are called
the shear strains.
64 The Engineering shear strain is dened as one half the tensorial shear strain, and the resulting
tensor is written as
1
1
11
2 12
2 13
1
1
(4.91)
Eij = 2 12 22
2 23
1
1
33
2 13
2 23
65 We note that a similar development paralleling the one just presented can be made for the linear
Eulerian strain tensor (where the straight lines and right angle will be in the deformed state).
4.2.5.2
66 The simplest and most useful measure of the extensional strain of an innitesimal element is the
stretch or stretch ratio as dx which may be dened at point P0 in the undeformed conguration or
dX
at P in the deformed one (Refer to the original denition given by Eq, 4.1).
Hence, from Eq. 4.57-a, and Eq. 4.63 the squared stretch at P0 for the line element along the unit
vector m = dX is given by
dX
67
2
m
dx
dX
= Cij
P0
dXi dXj
or 2 = mCm
m
dX dX
(4.92)
Thus for an element originally along X2 , Fig. 4.6, m = e2 and therefore dX1 /dX = dX3 /dX = 0 and
dX2 /dX = 1, thus Eq. 4.92 (with Eq. ??) yields
22 = C22 = 1 + 2E22
e
(4.93)
68
2
n
dX
dx
2
P
1
= Bij
1
dxi dxj
or 2 = nB1 n
dx dx
n
(4.94)
= 1 2E22
2 2
e
(4.95)
69 we note that in general e = e since the element originally along the X2 axis will not be along the
2
2
x2 after deformation. Furthermore Eq. 4.92 and 4.94 show that in the matrices of rectangular cartesian
components the diagonal elements of both C and B1 must be positive, while the elements of E must
be greater than 1 and those of E must be greater than + 1 .
2
2
70
(4.96)
and for the element P0 Q0 along the X2 axis, the unit extension is
dx dX
= E(2) = e2 1 =
dX
Victor Saouma
1 + 2E22 1
(4.97)
Draft
422
KINEMATIC
1
1 + 2E22 1
2
E22
(4.98)
sin 23 =
=
e2 e3
1 + 2E22 1 + 2E33
71
is given in terms of
(4.99)
Again, when deformations are small, this equation reduces to Eq. 4.90.
4.3
Strain Decomposition
72 In this section we rst seek to express the relative displacement vector as the sum of the linear
(Lagrangian or Eulerian) strain tensor and the linear (Lagrangian or Eulerian) rotation tensor. This is
restricted to small strains.
73 For nite strains, the former additive decomposition is no longer valid, instead we shall consider the
strain tensor as a product of a rotation tensor and a stretch tensor.
4.3.1
74 Strain components are quantitative measures of certain type of relative displacement between neighboring parts of the material. A solid material will resist such relative displacement giving rise to internal
stresses.
75 Not all kinds of relative motion give rise to strain (and stresses). If a body moves as a rigid body,
the rotational part of its motion produces relative displacement. Thus the general problem is to express
the strain in terms of the displacements by separating o that part of the displacement distribution
which does not contribute to the strain.
4.3.1.1
Small Strains
From Fig. 4.7 the displacements of two neighboring particles are represented by the vectors uP0 and
u and the vector
(4.100)
dui = uQ0 uP0 or du = uQ0 uP0
i
i
76
Q0
is called the relative displacement vector of the particle originally at Q0 with respect to the one
originally at P0 .
4.3.1.1.1
77
Lagrangian Formulation
ui
Xj
dXj or du = (u
P0
X )P 0
dX
(4.101)
78 We also dene a unit relative displacement vector dui /dX where dX is the magnitude of the
dierential distance dXi , or dXi = i dX, then
ui dXj
ui
du
dui
=
=
=u
j or
dX
Xj dX
Xj
dX
Victor Saouma
= J
(4.102)
Draft
423
Q
Q
du
dx
Q0
dX
P0
P0
Figure 4.7: Relative Displacement du of Q relative to P
u
The material displacement gradient Xi can be decomposed uniquely into a symmetric and an antij
symetric part, we rewrite the previous equation as
79
dui
uj
ui
+
Xj
Xi
1
2
Eij
or
1
= (u
2
X u) +
1
(u
2
or
E=
u1
X1
1
2
1
2
u1
X2
u1
X3
(4.103-a)
Wij
du
dXj
uj
ui
Xj
Xi
+
+
X u) dX
(4.103-b)
W
u1
X2 +
u2
X2
u2
X3 +
1
2
u2
X1
u3
X1
1
2
u2
X1
u1
X3 +
u2
X3 +
u3
X3
1
2
1
2
u3
X2
u3
X1
u3
X2
X u)
u1
X3
u2
X3
(4.104)
1
2
uj
ui
Xj
Xi
W = 1
2
1
2
1
2
0
u1
X2
u1
X3
or W =
u2
X1
u3
X1
u1
X2
1
(u
2
u2
X1
u2
X3
1
2
1
2
0
1
2
u3
X2
u3
X1
u3
X2
(4.105)
(4.106)
80 In a displacement for which Eij is zero in the vicinity of a point P0 , the relative displacement at that
point will be an innitesimal rigid body rotation. It can be shown that this rotation is given by the
Victor Saouma
Draft
424
KINEMATIC
wi =
ijk Wkj
1
2
or w =
X u
(4.107)
or
w = W23 e1 W31 e2 W12 e3
4.3.1.1.2
81
Eulerian Formulation
The derivation in an Eulerian formulation parallels the one for Lagrangian formulation. Hence,
dui =
82
ui
dxj or du = Kdx
xj
ui
ui dxj
du
=
=u
j or
xj dx
xj
dx
uj
ui
+
xj
xi
1
2
u1
x1
1
2
1
2
u1
x2
u1
x3
(4.111-a)
ij
x u) +
1
(u
2
E=
dxj
1
= (u
2
(4.110)
results in
du
or
= K
uj
ui
xj
xi
Eij
or
ui
xj
dui
84
(4.109)
83
(4.108)
+
+
x u) dx
(4.111-b)
u1
x2 +
u2
x2
u2
x3 +
1
2
u2
x1
u3
x1
1
2
u2
x1
u1
x3 +
u2
x3 +
u3
x3
1
2
1
2
u3
x2
u3
x1
u3
x2
(4.112)
in matrix form:
1
2
uj
ui
xj
xi
1
W = 2
1
2
u2
x1
u3
x1
u1
x2
1
2
0
u1
x2
u1
x3
or =
1
(u
2
u2
x1
u2
x3
1
2
1
2
0
1
2
u3
x2
u1
x3
u2
x3
(4.113)
x u)
u3
x1
u3
x2
(4.114)
Victor Saouma
1
2
ijk kj
or =
1
2
x u
(4.115)
Draft
425
Examples
2
2X1 X2
X1
0
ui
0
1
2X3
=
(4.116)
Xj
2
0
2X2 X3
X2
X )P
dX in the direction
4 1 0
{du} = 0 1 2
0 4 4
of X2 or
0 1
1
1
=
0
4
(4.117)
2e1 + e2 4e3
(4.118-a)
e1 e3
(4.118-b)
thus
uQ1 uP
uQ2 uP
uQ3 uP
uQ4 uP
e1 e2 + 3e3
1
(e1 e2 + 3.5e3 )
2
1
(e1 e2 + 3.75e3 )
4
1
(e1 e2 + 3.875e3 )
8
(4.119-a)
(4.119-b)
(4.119-c)
(4.119-d)
and it is clear that as Qi approaches P , the direction of the relative displacements of the two particles
approaches the limiting direction of du.
Example 4-9: Linear strain tensor, linear rotation tensor, rotation vector
Under the restriction of small deformation theory E = E , a displacement eld is given by u =
(x1 x3 )2 e1 + (x2 + x3 )2 e2 x1 x2 e3 . Determine the linear strain tensor, the linear rotation tensor and
the rotation vector at point P (0, 2, 1).
Solution:
the matrix form of the displacement gradient is
0
2(x1 x3 )
2(x1 x3 )
ui
0
2(x2 + x3 ) 2(x2 + x3 )
(4.120-a)
] =
[
xj
x1
0
x2
2 0 2
ui
= 0 2 2
(4.120-b)
xj P
2 0 0
Victor Saouma
Introduction to Continuum Mechanics
Draft
426
KINEMATIC
2 0 2
0 0 0
[Eij ] + [wij ] = 0 2 1 + 0 0 1
0 1 0
2 1 0
(4.121)
4.3.2
(4.122)
u
When the displacement gradients are nite, then we no longer can decompose Xi (Eq. 4.101) or
j
(Eq. 4.109) into a unique sum of symmetric and skew parts (pure strain and pure rotation).
85
ui
xj
86 Thus in this case, rather than having an additive decomposition, we will have a multiplicative
decomposition.
87 we call this a polar decomposition and it should decompose the deformation gradient in the product
of two tensors, one of which represents a rigid-body rotation, while the other is a symmetric positivedenite tensor.
88
xi
= Rik Ukj = Vik Rkj or F = RU = VR
Xj
(4.123)
where R is the orthogonal rotation tensor, and U and V are positive symmetric tensors known as
the right stretch tensor and the left stretch tensor respectively.
89
The interpretation of the above equation is obtained by inserting the above equation into dxi =
xi
Xj dXj
(4.124)
and we observe that in the rst form the deformation consists of a sequential stretching (by U) and
rotation (R) to be followed by a rigid body displacement to x. In the second case, the orders are
reversed, we have rst a rigid body translation to x, followed by a rotation (R) and nally a stretching
(by V).
90
(4.125)
Recalling that R is an orthonormal matrix, and thus RT = R1 then we can compute the various
tensors from
U =
FT F (4.126)
R = FU1
V = FRT
91
(4.127)
(4.128)
(4.129)
Draft
427
Given x1 = X1 , x2 = 3X3 , x3 = 2X2 , nd the deformation gradient F, the right stretch tensor U,
the rotation tensor R, and the left stretch tensor V.
Solution:
From Eq. 4.25
x1
X1
x2
X1
x3
X1
F=
From Eq. 4.126
1
U2 = FT F = 0
0
x1
X2
x2
X2
x3
X2
0
0
3
1
U= 0
0
From Eq. 4.127
R = FU1
1
= 0
0
0
0
2
1
= 0
0
1
0
2 0
0
0
thus
x1
X3
x2
X3
x3
X3
0
2
0
1
0
3 0
0
0
0 0
0 3
2 0
1
0
3 = 0
0
0
0
0
2
(4.130)
0
4
0
0
0
9
(4.131)
0
0
3
(4.132)
0
1
0 = 0
1
0
3
0
1
2
0 0
0 1
1 0
(4.133)
1
V = FRT = 0
0
0 0
1
0 3 0
0
2 0
0
0
1
1
0
1 = 0
0
0
0
3
0
0
0
2
(4.134)
[F]
2
1
= 0
0
0
0
2
0
3
0
(4.135)
[F]T [F]
0
0
0
1 0
1
0
2 0
0 3 =
= 0
0 3 0
0 2
0
0
1 0
[U] = 0 2 0
0
0 3
1
0
1 0
0
1
1
0 3 0 2
[R] = [F][U]1 = 0
0 2
0
0
0
[U]
2
1
0
0
0
2
2
0
(4.136)
0
0
2
3
(4.137)
(4.138)
0
1
0 = 0
1
0
3
0
0
1
0
1
0
(4.139)
Draft
428
KINEMATIC
Victor Saouma
Draft
429
mpolar.nb
2
In[4]:=
z
j
z
j
z
j
In[10]:=
. Ueigen the principal values of
Given x1 =XU_e = ,N@vnormalizedObtain C, b). vnormalized, 3DC and the corresponding directions, c) the
1 0.4142 1.
3
k +2X2 x2 =X2 ,0x{ =X3 , a)
-1 with respect to the principal directions, d) Obtain the matrix U and U -1 with respect to the e bas
matrix U and U
i
obtain the matrix R with respect to the ei basis.
i 0.707 0.707 0. y
z
j LinearAlgebraOrthogonalization
z
j
In[5]:=
<<
j 0.707 2.12 0. z
z
j
Out[10]=
z
j
z
j
z
j
0.
0. 1. {
k
In[6]:=
vnormalized = GramSchmidt@8v3, v2, v1<D
Determine the F matrix
In[11]:=
In[1]:=
Out[6]=
Out[11]=
Out[1]=
In[7]:=
U_einverse = N@Inverse@%D, 3D
F = 881, 2, 0<, 80, 1, 0<, 80, 0, 1<<
0.92388 0 y
i 0.382683
j
z
j
z
j 0.92388 -0.382683 0 z
j 2.12 -0.707 0.
z
j
z
j
i
y
j
z
j
z 1. z
j1 2 0 0
z
0
j
k -0.707 y 0.707 0. z
{
j
z
i
z
j
z
z
j
z
j0 1 0z
z
j
z
0. z
0. 1. {
j
k
j
z
j
z
k0 0 1{
CSTeigen = Chop@N@vnormalized . CST . vnormalized, 4DD
i 5.828
0y
j
z
j
Determine R 0 0.1716 0 z to the ei basis
with respect
z
j
z
j
Out[7]=
z
j
z
j
Solve for C
z
j
In[12]:=
In[2]:=
0
0 1. {
k
R = N@F . %, 3D
CST = Transpose@FD . F
i 0.707 0.707 0. y
j
z
j
z
j 1 2 0 y 0.707 0. z
j
z
Out[12]=
i -0.707 z
j
z
z
j
Determine U with respect
z
z
j2 5 0z
j
Out[2]=
0. z
0. 1. {
z
j
k
In[8]:=
z
j
z
j
k0 0 1{
Ueigen = N@Sqrt@CSTeigenD, 4D
0 0y
i 2.414
z
j
z
j
j
z
j
0 0.4142 0 z
z
j
z
j Eigenvalues and
Determine
z
j
0
0 1. {
k
Out[8]=
Eigenvectors
In[3]:=
N@Eigenvalues@CSTDD
In[9]:=
Ueigenminus1 = Inverse@UeigenD
Out[3]=
Out[9]=
Victor Saouma
Draft
430
4.4
92
KINEMATIC
thus the Eulerian and the Lagrangian innitesimal strain tensors may be taken as equal Eij = Eij .
2. If the displacement gradients are small, but the displacements are large, we should use the Eulerian
innitesimal representation.
3. If the displacements gradients are large, but the displacements are small, use the Lagrangian nite
strain representation.
4. If both the displacement gradients and the displacements are large, use the Eulerian nite strain
representation.
4.5
Compatibility Equation
If ij = 1 (ui,j + uj,i ) then we have six dierential equations (in 3D the strain tensor has a total
2
of 9 terms, but due to symmetry, there are 6 independent ones) for determining (upon integration)
three unknowns displacements ui . Hence the system is overdetermined, and there must be some linear
relations between the strains.
93
94 It can be shown (through appropriate successive dierentiation of the strain expression) that the
compatibility relation for strain reduces to:
2 jj
2 jk
2 ij
2 ik
+
= 0. or
xj xj
xi xk
xi xj
xj xk
x L
=0
(4.140)
+
+
x1
x1
x2
x3
31
12
23
+
x2 x1
x2
x3
31
12
23
+
x3 x1
x2
x3
=
=
=
=
=
=
2 12
x1 x2
2 23
2
x2 x3
2 31
2
x3 x1
2 11
x2 x3
2 22
x3 x1
2 33
x1 x2
2
(4.141-a)
(4.141-b)
(4.141-c)
(4.141-d)
(4.141-e)
(4.141-f)
(4.142)
Draft
431
P
x
u +d u
X
Oo
I2 ,i
X2
I 1 , 1i
b=0
Material/Spatial
X1
,x
2
,x
1
LAGRANGIAN
Material
Position Vector
x = x(X, t)
Deformation
F=x
Displacement
ui
Xj
Deformation
=
J=u
xi
Xj
ij or
=FI
xi
Xj
X
dX 2 = dxB1 dx
Cauchy
1
k
k
Bij = Xi Xj or
x x
B1 = x XX x = Hc H
Lagrangian
dx2 dX 2 = dX2EdX
Finite Strain
Eij = 1
2
E= 1(
2
Eij = 1
2
E = 1 (u
2
Small
Deformation
Victor Saouma
Small
deformation
EULERIAN
Spatial
X = X(x, t)
GRADIENTS
H = X x Xji
x
H = F1
ui
Xi
xj = ij xj or
Ku x =IH
TENSOR
dx2 = dXCdX
Green
x x
Cij = Xk Xk or
i
j
C = X xx X = Fc F
C1 = B1
STRAINS
Eulerian/Almansi
dx2 dX 2 = dx2E dx
ij or
X xx X I)
xk xk
Xi Xj
Fc F
uj
ui
Xj + Xi
X
X 1, x 1
,i
X 3 , x3
P0
U
I
t=0
t=0
dX
,x
t=t
X
X+
d
t=t
Q
dx
X 2, x 2
u u
+ Xk Xk or
i
j
Xu +
X uu X )
k
k
Eij = 1 ij Xi Xj or
2
x x
E = 1 (I x XX x )
2
Hc H
u
ui
Eij = 1 xj + xj uk uk or
2
xi xj
i
E = 1 (u x + x u x uu x )
2
K+Kc Kc K
u
ui
Eij = 1 xj + xj
2
i
+ Jc )
E = 1 (u x + x u) = 1 (K + Kc )
2
2
ROTATION TENSORS
Introduction touj
Continuum Mechanics
u
u
uj
u
u
1 ui
1 ui
dxj
[ 1 Xi + Xji + 1 Xi Xji ]dXj
2
2
2 xj + xi + 2 xj xi
j
j
1
1
1
1
[ (u X + X u) + (u X X u)]dX [ (u x + x u) + (u x x u)]dx
2
2
2
2
Eij = 1
2
E = 1 (u
2
ui
Xj
+
X+
J+Jc +Jc J
uj
Xi
1
X u) = 2 (J
Draft
432
KINEMATIC
= 2 (1 + )
x2
x2
x2
x2
x1 x2
2
2
1
1
(4.143)
X1
2
2
2(X1 +X2 )
X 2X2 2
+X
0
0
X1
2
2
2(X1 +X2 )
0
0
(4.144)
2 E11
2
X2
E11
X2
E12
2
X1
E22
2
X1
2 E22
+
2
X1
=
=
2
2
2
2
(X1 + X2 ) X2 (2X2 )
X2 X 1
=
2 + X 2 )2
2 + X 2 )2
(X1
(X1
2
2
2
2
2
2
(X1 + X2 ) X1 (2X1 )
X2 X 1
=
2
2
2
2
(X1 + X2 )2
(X1 + X2 )2
(4.145-a)
(4.145-b)
(4.145-c)
2 E12
X1 X2
(4.145-d)
Actually, it can be easily veried that the unique displacement eld is given by
u1 = arctan
X2
;
X1
u2 = 0;
u3 = 0
(4.146)
to which we could add the rigid body displacement eld (if any).
4.6
In Sect. 2.2 the discussion of stress applied to the deformed conguration dA (using spatial coordiantes
x), that is the one where equilibrium must hold. The deformed conguration being the natural one in
which to characterize stress. Hence we had
96
df
= tdA
t = Tn
(4.147-a)
(4.147-b)
(note the use of T instead of ). Hence the Cauchy stress tensor was really dened in the Eulerian
space.
97 However, there are certain advantages in referring all quantities back to the undeformed conguration
(Lagrangian) of the body because often that conguration has geometric features and symmetries that
are lost through the deformation.
98 Hence, if we were to dene the strain in material coordinates (in terms of X), we need also to express
the stress as a function of the material point X in material coordinates.
Victor Saouma
Draft
4.6.1
433
First
99 The rst Piola-Kircho stress tensor T0 is dened in the undeformed geometry in such a way that it
results in the same total force as the traction in the deformed conguration (where Cauchys stress
tensor was dened). Thus, we dene
(4.148)
df t0 dA0
where t0 is a pseudo-stress vector in that being based on the undeformed area, it does not describe
the actual intensity of the force, however it has the same direction as Cauchys stress vector t.
100 The rst Piola-Kircho stress tensor (also known as Lagrangian Stress Tensor) is thus the linear
transformation T0 such that
(4.149)
t0 = T0 n0
dA
t
ddA0
(4.150)
T0 n0 =
and using Eq. 4.36 dAn = dA0 (det F) F1
(4.151)
n0 we obtain
T0 n0 = T(det F) F1
n0
(4.152)
(det F)T F1
1
1
T0 FT or Tij =
(T0 )im Fjm
(det F)
(det F)
(4.153)
(4.154)
The rst Piola-Kircho stress tensor is not symmetric in general, and is not energitically correct.
That is multiplying this stress tensor with the Green-Lagrange tensor will not be equal to the product
of the Cauchy stress tensor multiplied by the deformation strain tensor.
101
102 To determine the corresponding stress vector, we solve for T0 rst, then for dA0 and n0 from
1
dA0 n0 = det F FT n (assuming unit area dA), and nally t0 = T0 n0 .
4.6.2
Second
The second Piola-Kircho stress tensor, T is formulated dierently. Instead of the actual force df
related to the force df in the same way that a material vector dX at X is
on dA, it gives the force df
related by the deformation to the corresponding spatial vector dx at x. Thus, if we let
103
d
f
df
= 0
tdA
and
= Fd
f
(4.155-a)
(4.155-b)
where d is the pseudo dierential force which transforms, under the deformation gradient F, the
f
(actual) dierential force df at the deformed position (note similarity with dx = FdX). Thus, the
pseudo vector t is in general in a diernt direction than that of the Cauchy stress vector t.
104
Victor Saouma
(4.156)
Introduction to Continuum Mechanics
Draft
434
KINEMATIC
df = FTn0 dA0
(4.157)
df = t0 dA0 = T0 n0 ddA0
(4.158)
T = F1 T0
(4.159)
which gives the relationship between the rst Piola-Kircho stress tensor T0 and the second Piola
Kircho stress tensor T.
Finally the relation between the second Piola-Kircho stress tensor and the Cauchy stress tensor can
be obtained from the preceding equation and Eq. 4.153
105
T = (det F) F1 T F1
(4.160)
and we note that this second Piola-Kircho stress tensor is always symmetric (if the Cauchy stress tensor
is symmetric). It can also be shown that it is energitically correct.
106
To determine the corresponding stress vector, we solve for T rst, then for dA0 and n0 from dA0 n0 =
= Tn0 .
(assuming unit area dA), and nally t
1
T
det F F n
4.7
85 The lagrangian and Eulerian linear strain tensors can each be split into spherical and deviator
tensor as was the case for the stresses. Hence, if we dene
1
1
e = tr E
3
3
(4.161)
(4.162)
We note that E measures the change in shape of an element, while the spherical or hydrostatic strain
1
3 e1 represents the volume change.
Victor Saouma
Draft
435
mpiola.nb
2
mpiola.nb
3
MatrixForm@Transpose@FD . n detFD
i0y
j z
The deformed4configuration of a body is described by x1 =X 1 2, x2 =X2 /2, x3 =4X3 ; If the Cauchy stress tensor is
j z
j z
j z
100 0 0 y = Inverse@FD . Tfirst
Tsecond
i j z
j k0{ z
j
z
j 0 0 0 z MPa; What are the corresponding first and second PiolaKirchoff stress tensors, and calculate the
z
j
given byj
z
j
z
z
j
k 0 0 0{
25
980, 0, 0 on n0 we obtain
Thus n0 =e2 and using t0<,0the e3 plane in the 80, 0, 0<=
respective stress tensors =T 90, , 0=, deformed state.
4
F tensor
MatrixForm@%D
i 0 y
z
j
j 0 z
j
z
j 0 z0 0
j
i 25 z
CST = 880, y 0<, 80, 0, 0<, 80, 0, 100<<
0,
j
z
{
k
j
z
25
j
z
j
z
j 0 0 z
4
j
z
j
z
k 0 0 0<, 80,
880, 0, 0 {
We note that this vector is in the0, 0<, 80, 0, 100<<
same direction as the Cauchy stress vector, its magnitude is one fourth of that of the
Cauchy stress vector, because the undeformed area is 4 times that of the deformed area
F = 881 2, 0,
Cuchy stress vector 0<, 80, 0, 1 2<, 80, 4, 0<<
PseudoStress vector associated with the Second PiolaKirchoff stress
tensor
Can be obtained from t=CST n
1
1
99 , 0, 0=, 90, 0, - =, 80, 4, 0<=
2
2
MatrixForm@%D
n = 80, 0, 1<
i0 0 0y
j
j 0 0, 1< z
z
j
80, 0 0 z
z
j
z
j
k 0 25 0 {
Victor Saouma
Draft
436
KINEMATIC
III
II
4.8
86 Determination of the principal strains (E(3) < E(2) < E(1) , strain invariants and the Mohr circle for
strain parallel the one for stresses (Sect. 2.4) and will not be repeated here.
3 IE 2 IIE IIIE = 0
(4.163)
where the symbols IE , IIE and IIIE denote the following scalar expressions in the strain components:
IE
IIE
=
=
=
=
IIIE
87
(4.166)
(4.167)
(4.168)
(4.169)
IIE
IIIE
88
(4.164)
(4.165)
=
=
(4.170)
(4.171)
The Mohr circle uses the Engineering shear strain denition of Eq. 4.91, Fig. 4.8
Example 4-15: Strain Invariants & Principal Strains
Victor Saouma
Draft
437
Determine the planes of principal strains for the following strain tensor
1
3 0
3 0 0
0
0 1
(4.172)
Solution:
The strain invariants are given by
IE
IIE
IIIE
= Eii = 2
1
(Eij Eij Eii Ejj ) = 1 + 3 = +2
=
2
= |Eij | = 3
(4.173-a)
(4.173-b)
(4.173-c)
3
0
0
0
0
1
1 + 13
(1 )
2
1 + 13
= 2.3
(1) =
2
(2) = 1
1 13
= 1.3
(3) =
2
Eij ij
=
E(1)
E(2)
E(3)
1+2 13
3
0
1+2 13
0
(4.174-a)
13
(4.174-b)
(4.174-c)
(4.174-d)
(4.174-e)
1+ 13
2
(1)
(1)
(1)
1 1+2 13 n1 + 3n2
0
n1
(1)
(1)
(1)
3n1 1+2 13 n2
=
0
n2
(1)
(1)
n3
1 1+2 13
1 1+ 13 n
3
0
0
=
(4.175)
which gives
(1)
n1
(1)
n3
1 + 13 (1)
n2
2 3
= 0
n(1) n(1)
n(1)
(4.176-a)
0.8
0.6
1 1
3
0
3
1
0
(1)
n2
= 1 n1 = 0.8;
2
(4.176-b)
1 + 2 13 + 13
+1
12
(2)
n1
0
(2)
0
n
2
(2)
11
n3
(4.176-c)
(4.176-d)
(2)
3n
(2) 2 (2)
=
3n1 n2
0
0
0
=
(4.177)
(4.178)
Draft
438
KINEMATIC
Finally, the third eigenvector can be obrained by the same manner, but more easily from
e1
n(3) = n(1) n(2) = det 0.8
0
Therefore
e2
0.6
0
(1)
0.8
n
(2)
aj =
= 0
n
i
(3)
0.6
n
0.8 0.6 0
1
0
1 3
[a][E][a]T = 0
0.6 0.8 0
0
0.8
3 0
0 0 0.6
0
0 1
e3
0
1
= 0.6e1 0.8e2
(4.179)
0.6 0
0
1
0.8 0
(4.180)
0.6
2.3
0.8 = 0
0
0
0
0
1
0
1
0
0
0
1.3
(4.181)
0
0
0
0
0
3
5
3 3
(4.182)
Solution:
1
2
60o
1
We note that since E(1) = 0 is a principal value for plane strain, ttwo of the circles are drawn as
shown.
Victor Saouma
Draft
4.9
89
439
ij =
0
T
= (1 + )
Plane Stress
T
0
0
T
(4.183)
Plane Strain
4.10
90 Typically, the transducer to measure strains in a material is the strain gage. The most common type
of strain gage used today for stress analysis is the bonded resistance strain gage shown in Figure 4.9.
a
b
c
=
=
=
cos2 a +
2
cos b +
cos2 c +
x
x
(4.184)
(4.185)
(4.186)
Victor Saouma
Draft
440
KINEMATIC
4.10.1
99 Due to their outstanding sensitivity, Wheatstone bridge circuits are very advantageous for the measurement of resistance, inductance, and capacitance. Wheatstone bridges are widely used for strain
measurements. A Wheatstone bridge is shown in Figure 4.11. It consists of 4 resistors arranged in a
diamond orientation. An input DC voltage, or excitation voltage, is applied between the top and bottom
of the diamond and the output voltage is measured across the middle. When the output voltage is zero,
the bridge is said to be balanced. One or more of the legs of the bridge may be a resistive transducer,
such as a strain gage. The other legs of the bridge are simply completion resistors with resistance equal
to that of the strain gage(s). As the resistance of one of the legs changes, by a change in strain from
a resistive strain gage for example, the previously balanced bridge is now unbalanced. This unbalance
causes a voltage to appear across the middle of the bridge. This induced voltage may be measured with
a voltmeter or the resistor in the opposite leg may be adjusted to re-balance the bridge. In either case
the change in resistance that caused the induced voltage may be measured and converted to obtain the
Victor Saouma
Draft
441
4.10.2
If a strain gage is oriented in one leg of the circuit and the other legs contain xed resistors as shown
Rgage
in Figure 4.11, the circuit is known as a quarter bridge circuit. The circuit is balanced when R1 = R3 .
R2
100
Rgage
Rgage +R3 ).
Wheatstone bridges may also be formed with two or four legs of the bridge being composed of
resistive transducers and are called a half bridge and full bridge respectively. Depending upon the type
of application and desired results, the equations for these circuits will vary as shown in Figure 4.12. Here
E0 is the output voltage in mVolts, E is the excitation voltage in Volts, is strain and is Poissons
ratio.
101
In order to illustrate how to compute a calibration factor for a particular experiment, suppose a
single active gage in uniaxial compression is used. This will correspond to the upper Wheatstone bridge
conguration of Figure 4.12. The formula then is
102
Victor Saouma
Draft
442
KINEMATIC
Victor Saouma
Draft
443
F (103 )
E0
=
E
4 + 2F (106 )
(4.187)
The extra term in the denominator 2F (106 ) is a correction factor for non-linearity. Because
this term is quite small compared to the other term in the denominator it will be ignored. For most
measurements a gain is necessary to increase the output voltage from the Wheatstone bridge. The
gain relation for the output voltage may be written as V = GE0 (103 ), where V is now in Volts. so
Equation 4.187 becomes
103
V
EG(103 )
V
=
=
F (103 )
4
4
F EG
(4.188)
Here, Equation 4.188 is the calibration factor in units of strain per volt. For common values where
4
F = 2.07, G = 1000, E = 5, the calibration factor is simply (2.07)(1000)(5) or 386.47 microstrain per volt.
104
Victor Saouma
Draft
KINEMATIC
Victor Saouma
444
Draft
Chapter 5
MATHEMATICAL
PRELIMINARIES; Part III
VECTOR INTEGRALS
5.1
1
Integral of a Vector
R1 (u)du + e2
d
du
R3 (u)du
(5.1)
(S(u)), then
d
(S(u)) du = S(u) + c
du
R(u)du =
5.2
R2 (u)du + e3
(5.2)
Line Integral
2 Given r(u) = x(u)e1 + y(u)e2 + z(u)e3 where r(u) is a position vector dening a curve C connecting
point P1 to P2 where u = u1 and u = u2 respectively, anf given A(x, y, z) = A1 e1 + A2 e2 + A3 e3 being
a vectorial function dened and continuous along C, then the integral of the tangential component of A
along C from P1 to P2 is given by
P2
Adr =
Adr =
C
P1
A1 dx + A2 dy + A3 dz
(5.3)
If A were a force, then this integral would represent the corresponding work.
3
Adr
A1 dx + A2 dy + A3 dz
(5.4)
then
is independent of the path C connecting P1 to P2
(5.5-a)
(5.5-b)
P1
Adr = 0
C
Draft
52
5.3
5
Integration by Parts
5.4
v(x)u (x)dx
(5.6)
The divergence theorem (also known as Ostrogradskis Theorem) comes repeatedly in solid mechanics
and can be stated as follows:
vd =
v.nd or
vi,i d =
vi ni d
(5.7)
That is the integral of the outer normal component of a vector over a closed surface (which is the
volume ux) is equal to the integral of the divergence of the vector over the volume bounded by the
closed surface.
7
(5.8)
5.5
9
qT nds
s
Stokes Theorem
Adr =
C
( A)dS
(5.9)
5.6
10
(Rdx + Sdy) =
Example 5-1:
Victor Saouma
R
S
x
y
dxdy
(5.10)
Draft
53
Z
E
Z
C
V
P(X,Y,Z) H
X
B
a)
S
dV=dxdydz
n
Vt
dS
dS
b)
c)
vx
vx
xx/2,y,z
vx
vx
x+x/2,y,z
(5.11-a)
1 vx
x AFED
2 x
1 vx
x GHCB
vx +
2 x
(5.11-b)
(5.11-c)
=
=
vx +
1 vx
1 vx
x yz vx
x yz
2 x
2 x
vx
xyz
x
(5.12-a)
(5.12-b)
Similarly
Vy
Vz
vy
xyz
y
vz
xyz
z
(5.13-a)
(5.13-b)
Hence, the total increase per unit volume and unit time will be given by
vx
x
vy
y
vz
z
xyz
xyz
= div v =
(5.14)
Furthermore, if we consider the total of uid crossing dS during t, Fig. 5.1-b, it will be given by
(vt)ndS = vndSt or the volume of uid crossing dS per unit time is vndS.
Victor Saouma
Draft
54
Thus for an arbitrary volume, Fig. 5.1-c, the total amount of uid crossing a closed surface S per
vdV (Eq. 5.14), thus
unit time is
S
vdV
vndS =
S
(5.15)
Victor Saouma
Draft
Chapter 6
FUNDAMENTAL LAWS of
CONTINUUM MECHANICS
6.1
Introduction
We have thus far studied the stress tensors (Cauchy, Piola Kircho), and several other tensors which
describe strain at a point. In general, those tensors will vary from point to point and represent a tensor
eld.
We have also obtained only one dierential equation, that was the compatibility equation.
In this chapter, we will derive additional dierential equations governing the way stress and deformation
vary at a point and with time. They will apply to any continuous medium, and yet we will not have
enough equations to determine unknown tensor eld. For that we need to wait for the next chapter
where constitututive laws relating stress and strain will be introduced. Only with constitutive equations
and boundary and initial conditions would we be able to obtain a well dened mathematical problem to
solve for the stress and deformation distribution or the displacement or velocity elds.
In this chapter we shall derive dierential equations expressing locally the conservation of mass, momentum and energy. These dierential equations of balance will be derived from integral forms of the
equation of balance expressing the fundamental postulates of continuum mechanics.
6.1.1
Conservation Laws
Conservation laws constitute a fundamental component of classical physics. A conservation law establishes a balance of a scalar or tensorial quantity in voulme V bounded by a surface S. In its most
general form, such a law may be expressed as
d
dt
AdV +
dS
Rate of variation
Exchange by Diffusion
AdV
(6.1)
V
Source
where A is the volumetric density of the quantity of interest (mass, linear momentum, energy, ...) a,
A is the rate of volumetric density of what is provided from the outside, and is the rate of surface
density of what is lost through the surface S of V and will be a function of the normal to the surface n.
Hence, we read the previous equation as: The input quantity (provided by the right hand side) is equal
to what is lost across the boundary, and to modify A which is the quantity of interest. The dimensions
of various quantities are given by
6
dim(a)
dim(AL3 )
(6.2-a)
Draft
62
=
=
dim(AL2 t1 )
3 1
dim(AL
(6.2-b)
(6.2-c)
Hence this chapter will apply the previous conservation law to mass, momentum, and energy. the
resulting dierential equations will provide additional interesting relation with regard to the imcompressibiltiy of solids (important in classical hydrodynamics and plasticity theories), equilibrium and
symmetry of the stress tensor, and the rst law of thermodynamics.
The enunciation of the preceding three conservation laws plus the second law of thermodynamics,
constitute what is commonly known as the fundamental laws of continuum mechanics.
6.1.2
Fluxes
Prior to the enunciation of the rst conservation law, we need to dene the concept of ux across a
bounding surface.
10 The ux across a surface can be graphically dened through the consideration of an imaginary surface
xed in space with continuous medium owing through it. If we assign a positive side to the surface,
and take n in the positive sense, then the volume of material owing through the innitesimal surface
area dS in time dt is equal to the volume of the cylinder with base dS and slant height vdt parallel to
the velocity vector v, Fig. 6.1 (If vn is negative, then the ow is in the negative direction). Hence, we
vdt
vn dt
n
dS
vndS =
S
vj nj dS
(6.3)
We can generalize this denition and dene the following uxes per unit area through dS:
Victor Saouma
Draft
Mass Flux
63
vndS =
vj nj dS
Momentum Flux
v(vn)dS =
=
S
qndS =
S
Electric ux
6.2.1
(6.5)
1
vi vi vj nj dS
S2
(6.6)
qj nj dS
(6.7)
Jj nj dS
(6.8)
JndS =
S
6.2
vk vj nj dS
S
1 2
v (vn)dS =
S2
(6.4)
(x, t)dV
M=
(6.9)
where (x, t) is a continuous function called the mass density. We note that this spatial form in terms
of x is most common in uid mechanics.
13
dV
t
(6.10)
14 The Law of conservation of mass requires that the mass of a specic portion of the continuum
remains constant. Hence, if no mass is created or destroyed inside V , then the preceding equation must
eqaul the inow of mass (of ux) through the surface. The outow is equal to vn, thus the inow
will be equal to vn.
(vn )dS =
S
must be equal to
M
t .
vndS =
(v)dV
+
t
(v) dV = 0
(6.11)
Thus
V
(6.12)
since the integral must hold for any arbitrary choice of dV , then we obtain
+
t
15
(v) or
(vi )
+
=0
t
xi
(6.13)
=
+ vi
xi
xi
xi
(6.14)
16 It can be shown that the rate of change of the density in the neighborhood of a particle instantaneously
at x by
d
=
+ v =
+ vi
(6.15)
dt
t
t
xi
where the rst term gives the local rate of change of the density in the neighborhood of the place of x,
while the second term gives the convective rate of change of the density in the neighborhood of a
Victor Saouma
Draft
64
particle as it moves to a place having a dierent density. The rst term vanishes in a steady ow, while
the second term vanishes in a uniform ow.
17
Upon substitution in the last three equations, we obtain the continuity equation
vi
d
d
+
+ v = 0
= 0 or
dt
xi
dt
(6.16)
The vector form is independent of any choice of coordinates. This equation shows that the divergence of
the velocity vector eld equals (1/)(d/dt) and measures the rate of ow of material away from the
particle and is equal to the unit rate of decrease of density in the neighborhood of the particle.
18 If the material is incompressible, so that the density in the neighborhood of each material particle
remains constant as it moves, then the continuity equation takes the simpler form
vi
= 0 or
xi
v = 0
(6.17)
6.2.2
Material Form
19 If material coordinates X are used, the conservation of mass, and using Eq. 4.41 (dV = |J|dV0 ),
implies
(X, t0 )dV0 =
V0
(x, t)dV =
(x, t)|J|dV0
(6.18)
V0
or
[0 |J|]dV0 = 0
(6.19)
V0
and for an arbitrary volume dV0 , the integrand must vanish. If we also suppose that the initial density
0 is everywhere positive in V0 (no empty spaces), and at time t = t0 , J = 1, then we can write
J = 0
(6.20)
d
(J) = 0
dt
(6.21)
or
which is the continuity equation due to Euler, or the Lagrangian dierential form of the continuity equation.
20 We note that this is the same equation as Eq. 6.16 which was expressed in spatial form. Those two
equations can be derived one from the other.
21
The more commonly used form if the continuity equation is Eq. 6.16.
6.3
6.3.1
The momentum principle states that the time rate of change of the total momentum of a given set of
particles equals the vector sum of all external forces acting on the particles of the set, provided Newtons
Third Law applies. The continuum form of this principle is a basic postulate of continuum mechanics.
22
tdS +
S
Victor Saouma
bdV =
V
d
dt
vdV
(6.22)
Draft
65
Tij
+ bi dV
xj
V
dvi
Tij
dV
+ bi
xj
dt
dvi
dV
dt
(6.23-a)
(6.23-b)
T + b =
dv
dt
(6.24)
which is Cauchys (rst) equation of motion, or the linear momentum principle, or more simply
equilibrium equation.
23
= 0
= 0
(6.25-a)
= 0
24 We note that these equations could also have been derived from the free body diagram shown in Fig.
6.2 with the assumption of equilibrium (via Newtons second law) considering an innitesimal element
of dimensions dx1 dx2 dx3 . Writing the summation of forces, will yield
(6.26)
Tij,j + bi = 0
where is the density, bi is the body force (including inertia).
2x1 x2
0
x2 + (x2 x2 )
x
1
x2 + (x2 x2 )
0
2x1 x2
1
2
1
2
2
0
0
(x1 + x2 )
(6.27)
T1j
xj
T2j
xj
T3j
xj
=
=
=
T11
T12
T13
+
+
= 2x1 2x1 = 0
x1
x2
x3
T21
T22
T23
+
+
= 2x2 + 2x2 = 0
x1
x2
x3
T31
T32
T33
+
+
=0
x1
x2
x3
(6.28-a)
(6.28-b)
(6.28-c)
Victor Saouma
Draft
66
yy
+
dy
yy d y
y
+
yx
yx y
d
y
xx
+
xx
xy
xx d x
x
xy x
d
x
+
xy
yx
yy
dx
6.3.2
25 The moment of momentum principle states that the time rate of change of the total moment of
momentum of a given set of particles equals the vector sum of the moments of all external forces acting
on the particles of the set.
Thus, in the absence of distributed couples (this theory of Cosserat will not be covered in this
course) we postulate the same principle for a continuum as
26
(rt)dS +
(rb)dV =
6.3.2.1
d
dt
(6.29)
(rv)dV
V
27 We observe that the preceding equation does not furnish any new dierential equation of motion. If
we substitute tn = Tn and the symmetry of the tensor is assumed, then the linear momentum principle
(Eq. 6.24) is satised.
28 Alternatively, we may start by using Eq. 1.18 (ci = ijk aj bk ) to express the cross product in indicial
form and substitute above:
(rmn xm tn )dS +
S
d
dt
(rmn xm bn )dV =
V
(rmn xm vn )dV
(6.30)
xm Tjn
+ xm bn dV =
xj
rmn
V
d
(xm vn )dV
dt
(6.31)
Victor Saouma
Tjn
+ bn
xj
+ mj Tjn dV =
rmn vm vn + xm
V
dvn
dt
dV
(6.32)
Draft
67
but rmn vm vn = 0 since vm vn is symmetric in the indeces mn while rmn is antisymmetric, and the last
term on the right cancels with the rst term on the left, and nally with mj Tjn = Tmn we are left with
rmn Tmn dV = 0
(6.33)
rmn Tmn = 0
(6.34)
= 1
= 2
= 3
T23 T32
T31 T13
T12 T21
=
=
=
0
0
0
(6.35)
establishing the symmetry of the stress matrix without any assumption of equilibrium or of uniformity
of stress distribution as was done in Sect. 2.3.
29
The symmetry of the stress matrix is Cauchys second law of motion (1827).
6.4
30 The rst principle of thermodynamics relates the work done on a (closed) system and the heat transfer
into the system to the change in energy of the system. We shall assume that the only energy transfers
to the system are by mechanical work done on the system by surface traction and body forces, by heat
transfer through the boundary.
6.4.1
31 We dene L as the spatial gradient of the velocity and in turn this gradient can be decomposed
into a symmetric rate of deformation tensor D (or stretching tensor) and a skew-symmeteric
tensor W called the spin tensor or vorticity tensor1 .
Lij
vi,j or L = v
D+W
1
(v x +
=
2
(6.36)
(6.37)
1
(v
x v) and W =
2
x v)
(6.38)
6.4.2
First Principle
If mechanical quantities only are considered, the principle of conservation of energy for the
continuum may be derived directly from the equation of motion given by Eq. 6.24. This is accomplished
by taking the integral over the volume V of the scalar product between Eq. 6.24 and the velocity vi .
32
vi Tji,j dV +
1 Note
Victor Saouma
bi vi dV =
V
vi
V
dvi
dV
dt
(6.39)
Draft
68
d
dvi
dV =
dt
dt
d
1
vi vi dV =
2
dt
dK
1 2
v dV =
2
dt
(6.40)
which represents the time rate of change of the kinetic energy K in the continuum.
33 Also we have vi Tji,j = (vi Tji ),j vi,j Tji and from Eq. 6.37 we have vi,j = Lij + Wij . It can be shown
that since Wij is skew-symmetric, and T is symmetric, that Tij Wij = 0, and thus Tij Lij = Tij Dij . TD
is called the stress power.
34
If we consider thermal processes, the rate of increase of total heat into the continuum is given by
Q=
qi ni dS +
rdV
(6.41)
Q has the dimension of power, that is M L2 T 3 , and the SI unit is the Watt (W). q is the heat ux
per unit area by conduction, its dimension is M T 3 and the corresponding SI unit is W m2 . Finally,
r is the radiant heat constant per unit mass, its dimension is M T 3 L4 and the corresponding SI
unit is W m6 .
35
36
We thus have
dK
+
dt
Dij Tij dV =
V
vi bi dV + Q
(6.42)
We next convert the rst integral on the right hand side to a surface integral by the divergence theorem
vdV = S v.ndS) and since ti = Tij nj we obtain
V
dK
+
dt
Dij Tij dV
vi ti dS +
S
dU
dK
+
dt
dt
vi bi dV + Q (6.43)
V
dW
+Q
dt
(6.44)
this equation relates the time rate of change of total mechanical energy of the continuum on the left side
to the rate of work done by the surface and body forces on the right hand side.
37 If both mechanical and non mechanical energies are to be considered, the rst principle states that the
time rate of change of the kinetic plus the internal energy is equal to the sum of the rate of work plus all
other energies supplied to, or removed from the continuum per unit time (heat, chemical, electromagnetic,
etc.).
38 For a thermomechanical continuum, it is customary to express the time rate of change of internal
energy by the integral expression
d
dU
=
udV
(6.45)
dt
dt V
where u is the internal energy per unit mass or specic internal energy. We note that U appears
only as a dierential in the rst principle, hence if we really need to evaluate this quantity, we need to
have a reference value for which U will be null. The dimension of U is one of energy dim U = M L2 T 2 ,
and the SI unit is the Joule, similarly dim u = L2 T 2 with the SI unit of Joule/Kg.
39
Rate of increae
d
dt
d
1
vi vi dV +
2
dt
dK
dt
udV =
V
dU
dt
Source
ti vi dS +
S
dW
dt
rdV
vi bi dV +
V
Exchange
Source
V
qi ni dS
(6.46)
S
Q
we apply Gauss theorem to convert the surface integral, collect terms and use the fact that dV is arbitrary
to obtain
Victor Saouma
Draft
du
dt
69
T:D + r
(6.47)
Tij Dij + r
qj
xj
(6.48)
or
du
dt
This equation expresses the rate of change of internal energy as the sum of the stress power plus
the heat added to the continuum.
40
41 In ideal elasticity, heat transfer is considered insignicant, and all of the input work is assumed
converted into internal energy in the form of recoverable stored elastic strain energy, which can be
recovered as work when the body is unloaded.
42 In general, however, the major part of the input work into a deforming material is not recoverably
stored, but dissipated by the deformation process causing an increase in the bodys temperature and
eventually being conducted away as heat.
6.5
The complete characterization of a thermodynamic system is said to describe the state of a system
(here a continuum). This description is specied, in general, by several thermodynamic and kinematic
state variables. A change in time of those state variables constitutes a thermodynamic process.
Usually state variables are not all independent, and functional relationships exist among them through
equations of state. Any state variable which may be expressed as a single valued function of a set of
other state variables is known as a state function.
43
The rst principle leaves unanswered the question of the extent to which conversion process is reversible or irreversible. If thermal processes are involved (friction) dissipative processes are irreversible
processes, and it will be up to the second principle of thermodynamics to put limits on the direction of
such processes.
45
6.5.1
Entropy
46 The basic criterion for irreversibility is given by the second principle of thermodynamics through
the statement on the limitation of entropy production. This law postulates the existence of two
distinct state functions: the absolute temperature and S the entropy with the following properties:
1. is a positive quantity.
2. Entropy is an extensive property, i.e. the total entropy is in a system is the sum of the entropies
of its parts.
47
where ds
(e)
(6.49)
(i)
ds(e)
Victor Saouma
>
0 irreversible process
(6.50-a)
ds(i)
0 reversible process
(6.50-b)
Draft
610
48
6.5.1.1
Statistical Mechanics
49 In statistical mechanics, entropy is related to the probability of the occurrence of that state among
all the possible states that could occur. It is found that changes of states are more likely to occur in the
direction of greater disorder when a system is left to itself. Thus increased entropy means increased
disorder.
50 Hence Boltzmans principle postulates that entropy of a state is proportional to the logarithm of its
probability, and for a gas this would give
3
S = kN [ln V + ln] + C
2
(6.51)
where S is the total entropy, V is volume, is absolute temperature, k is Boltzmans constant, and C
is a constant and N is the number of molecules.
6.5.1.2
51
Classical Thermodynamics
In a reversible process (more about that later), the change in specic entropy s is given by
ds =
52
dq
(6.52)
rev
(6.53)
where R is the gas constant, and assuming that the specic energy u is only a function of temperature
, then the rst principle takes the form
du = dq pdv
(6.54)
du = dq = cv d
(6.55)
dv
v
(6.57)
or division by yields
s s0 =
p,v
dq
=
p0 ,v0
cv ()
0
v
d
+ R ln
v0
(6.58)
which gives the change in entropy for any reversible process in an ideal gas. In this case, entropy is a
state function which returns to its initial value whenever the temperature returns to its initial value that
is p and v return to their initial values.
Victor Saouma
Draft
6.5.2
54
611
Clausius-Duhem Inequality
We restate the denition of entropy as heat divided by temperature, and write the second principle
d
dt
r
dV
dS
dt
q
ndS +
S
Sources
Exchange
Q
+ ;
(6.59)
Internal production
(6.60)
sdV is one
v
ds
dt
57
Sources
(6.61)
Exchange
We next seek to express the Clausius-Duhem inequality in terms of the stress tensor,
1
q
=
thus
q q
1
ds
dt
1
1
=
q +
1
q
2
1
r
q +
2
(6.62)
(6.63)
where q + r is the heat input into V and appeared in the rst principle Eq. 6.47
du
= T:D + r
dt
(6.64)
(6.65)
1
q 0
(6.66)
6.6
ds
du
dt
dt
58 In the preceding sections several equations and unknowns were introduced. Let us count them. for
both the coupled and uncoupled cases.
Victor Saouma
Draft
612
d
vi
dt + xi = 0
Tij
dvi
xj + bi = dt
du = Tij Dij + r
dt
Coupled
1
3
Uncoupled
1
3
1
5
Continuity Equation
Equation of motion
xj Energy equation
j
Total number of equations
59 Assuming that the body forces bi and distributed heat sources r are prescribed, then we have the
following unknowns:
Density
ds
dt
Coupled
1
3
6
3
1
1
1
16
1
div
Uncoupled
1
3
6
10
We thus need an additional 16 5 = 11 additional equations to make the system determinate. These
will be later on supplied by:
60
6
3
2
11
constitutive equations
temperature heat conduction
thermodynamic equations of state
Total number of additional equations
61 The next chapter will thus discuss constitutive relations, and a subsequent one will separately discuss
thermodynamic equations of state.
62
6.7
63 One of the relations which we will need is the one which relates temperature to heat ux.
This
constitutive realtion will be discussed in the next chapter under Fourriers law.
64 However to place the reader in the right frame of reference to understand Fourriers law, this section
will provide some elementary concepts of heat transfer.
65
Conduction: takes place when a temperature gradient exists within a material and is governed by
Fouriers Law, Fig. 6.3 on q :
Victor Saouma
Draft
613
qx
qy
T
x
T
ky
y
kx
(6.67)
(6.68)
where T = T (x, y) is the temperature eld in the medium, qx and qy are the componenets of the
heat ux (W/m2 or Btu/h-ft2 ), k is the thermal conductivity (W/m.o C or Btu/h-ft-o F) and T ,
x
T
are the temperature gradients along the x and y respectively. Note that heat ows from hot
y
to cool zones, hence the negative sign.
Convection: heat transfer takes place when a material is exposed to a moving uid which is at dierent
temperature. It is governed by the Newtons Law of Cooling
q = h(T T ) on c
(6.69)
where q is the convective heat ux, h is the convection heat transfer coecient or lm coecient
(W/m2 .o C or Btu/h-ft2 .o F). It depends on various factors, such as whether convection is natural or
forced, laminar or turbulent ow, type of uid, and geometry of the body; T and T are the surface
and uid temperature, respectively. This mode is considered as part of the boundary condition.
Radiation: is the energy transferred between two separated bodies at dierent temperatures by means
of electromagnetic waves. The fundamental law is the Stefan-Boltmans Law of Thermal Radiation
for black bodies in which the ux is proportional to the fourth power of the absolute temperature.,
which causes the problem to be nonlinear. This mode will not be covered.
6.7.1
66
Simple 2D Derivation
If we consider a unit thickness, 2D dierential body of dimensions dx by dy, Fig. 6.4 then
1. Rate of heat generation/sink is
I2 = Qdxdy
(6.70)
2. Heat ux across the boundary of the element is shown in Fig. ?? (note similarity with equilibrium
equation)
I1 =
qx +
Victor Saouma
qx
dx qx dx dy +
x
qy +
qy
qy
qx
dy qy dy dx =
dxdy +
dydx
y
x
y
(6.71)
Draft
614
6y +
q
qx
qy
y dy
qx +
qx
x dx
dy
?
6
qy
dx
67 From the rst law of thermodaynamics, energy produced I2 plus the net energy across the boundary
I1 must be equal to the energy absorbed I3 , thus
I1 + I2 I3
qy
d
qx
dxdy +
dydx + Qdxdy c dxdy
x
y
dt
I2
I1
6.7.2
68
(6.73-a)
(6.73-b)
I3
Generalized Derivation
The amount of ow per unit time into an element of volume and surface is
I1 =
q(n)d =
D .nd
(6.74)
div vd
(6.75)
div (D )d
(6.76)
Victor Saouma
Draft
615
I2 =
Q(x, y, z, t)d
(6.77)
Finally, we dene the specic heat of a solid c as the amount of heat required to raise a unit mass
by one degree. Thus if is a temperature change which occurs in a mass m over a time t, then the
corresponding amount of heat that was added must have been cm, or
71
I3 =
cd
(6.78)
I1 + I2 I3
d
div (D ) + Q c
t
(6.79-a)
(6.79-b)
=0
t
(6.80)
(6.81)
qy
qx
+
+ Q = c
x
y
t
(6.82)
div (D ) + Q c
or
div (D ) + Q = c
This equation can be rewritten as
1. Note the similarity between this last equation, and the equation of equilibrium
xy
xx
+
+ bx
x
y
xy
yy
+
+ by
y
x
=
=
2 ux
t2
2 uy
m 2
t
m
(6.83-a)
(6.83-b)
2. For steady state problems, the previous equation does not depend on t, and for 2D problems, it
reduces to
kx
+
ky
+Q=0
(6.84)
x
x
y
y
3. For steady state isotropic problems,
2 2 2 Q
=0
+ 2 + 2 +
x2
y
z
k
(6.85)
Victor Saouma
Draft
616
(6.86)
which is an Elliptic (or Laplace) equation. Solutions of Laplace equations are termed harmonic
functions (right hand side is zero) which is why Eq. 6.84 is refered to as the quasi-harmonic
equation.
5. If the function depends only on x and t, then we obtain
c
=
t
x
kx
+Q
(6.87)
Victor Saouma
Draft
Chapter 7
CONSTITUTIVE EQUATIONS;
Part I LINEAR
ceiinosssttuu
Hooke, 1676
Ut tensio sic vis
Hooke, 1678
7.1
7.1.1
Thermodynamic Approach
State Variables
The method of local state postulates that the thermodynamic state of a continuum at a given
point and instant is completely dened by several state variables (also known as thermodynamic
or independent variables). A change in time of those state variables constitutes a thermodynamic
process. Usually state variables are not all independent, and functional relationships exist among them
through equations of state. Any state variable which may be expressed as a single valued function of
a set of other state variables is known as a state function.
The time derivatives of these variables are not involved in the denition of the state, this postulate
implies that any evolution can be considered as a succession of equilibrium states (therefore ultra rapid
phenomena are excluded).
The basic assumption of thermodynamics is that in addition to the n substate variables, just one
additional dimensionally independent scalar paramer suces to determine the specic internal energy
u. This assumes that there exists a caloric equation of state
u = u(s, , X)
In general the internal energy u can not be experimentally measured but rather its derivative.
(7.1)
Draft
72
6 For instance we can dene the thermodynamic temperature and the thermodynamic tension j as
u
s
u
j
j = 1, 2, , n
(7.2)
s,i(i=j)
where the subscript outside the parenthesis indicates that the variables are held constant.
7 By extension Ai = i would be the thermodynamic force and its dimension depends on the
one of i .
7.1.2
8
Gibbs Relation
du
=
dt
9
ds
dp
+ p
dt
dt
(7.3)
dt
dt
T:D
we obtain
T:D +
ds
dt
u
s
1
q 0
+ Ap
(7.4)
1
dp
q 0
dt
(7.5)
but the second principle must be satised for all possible evolution and in particular the one for which
d
D = 0, dtp = 0 and = 0 for any value of ds thus the coecient of ds is zero or
dt
dt
=
u
s
(7.6)
thus
T:D + Ap
1
dp
q 0
dt
(7.7)
(7.8)
(7.9)
du = ds + p dp
10
u
s
;
v
u
v
(7.10)
s
where p is the thermodynamic pressure; and the thermodynamic tension conjugate to the specic volume
v is p, just as is conjugate to s.
Victor Saouma
Draft
7.1.3
73
11 From the caloric equation of state, Eq. 7.1, and the the denitions of Eq. 7.2 it follows that the
temperature and the thermodynamic tensions are functions of the thermodynamic state:
= (s, );
j = j (s, )
(7.11)
(7.12)
and substitute this into Eq. 7.1 to obtain an alternative form of the caloric equation of state with
corresponding thermal equations of state (obtained by simple substitution).
u = u(, , bX)
i = i (, , X)
i = i (, , X)
(7.13)
(7.14)
(7.15)
12 The thermal equations of state resemble stress-strain relations, but some caution is necessary in
interpreting the tesnisons as stresses and the j as strains.
7.1.4
Thermodynamic Potentials
13 Based on the assumed existence of a caloric equation of state, four thermodynamic potentials are
introduced, Table 7.1. Those potentials are derived through the Legendre-Fenchel transformation
Potential
Internal energy
Helmholtz free energy
Enthalpy
Free enthalpy
h
g
Relation to u
u
= u s
h = u j j
g = u s j j
Independent Variables
s, j
, j
s, j
, j
15
= ds + j dj
(7.16-a)
d = sd + j dj
dh = ds j dj
dg
(7.16-b)
(7.16-c)
= sd j dj
(7.16-d)
and from these dierentials we obtain the following partial derivative expressions
=
s=
u
;
s
h
;
s
g
;
=
=
Victor Saouma
j =
j =
j =
j =
u
j
s,i(i=j)
(7.17-a)
h
j
s,i(i=j)
g
j
(7.17-b)
(7.17-c)
(7.17-d)
Draft
74
where the free energy is the portion of the internal energy available for doing work at constant
temperature, the enthalpy h (as dened here) is the portion of the internal energy that can be released
as heat when the thermodynamic tensions are held constant.
7.1.5
16 Green dened an elastic material as one for which a strain-energy function exists. Such a material
is called Green-elastic or hyperelastic if there exists an elastic potential function W or strain
energy function, a scalar function of one of the strain or deformation tensors, whose derivative with
respect to a strain component determines the corresponding stress component.
17
For the fully recoverable case of isothermal deformation with reversible heat conduction we have
TIJ = 0
EIJ
(7.18)
hence W = 0 is an elastic potential function for this case, while W = 0 u is the potential for adiabatic
isentropic case (s = constant).
18 Hyperelasticity ignores thermal eects and assumes that the elastic potential function always exists,
it is a function of the strains alone and is purely mechanical
W (E)
TIJ =
EIJ
(7.19)
and W (E) is the strain energy per unit undeformed volume. If the displacement gradients are
small compared to unity, then we obtain
Tij =
W
Eij
(7.20)
which is written in terms of Cauchy stress Tij and small strain Eij .
19 We assume that the elastic potential is represented by a power series expansion in the small-strain
components.
1
1
(7.21)
W = c0 + cij Eij + cijkm Eij Ekm + cijkmnp Eij Ekm Enp +
2
3
where c0 is a constant and cij , cijkm , cijkmnp denote tensorial properties required to maintain the invariant
property of W . Physically, the second term represents the energy due to residual stresses, the third one
refers to the strain energy which corresponds to linear elastic deformation, and the fourth one indicates
nonlinear behavior.
Neglecting terms higher than the second degree in the series expansion, then W is quadratic in terms
of the strains
20
21
We next apply Eq. 7.20 to the quadratic expression of W and obtain for instance
T12 =
W
= 2c6 + c1112 E11 + c2212 E22 + c3312 E33 + c1212 E12 + c1223 E23 + c1231 E31
E12
Victor Saouma
(7.23)
Draft
75
if the stress must also be zero in the unstrained state, then c6 = 0, and similarly all the coecients in
the rst row of the quadratic expansion of W . Thus the elastic potential function is a homogeneous
quadratic function of the strains and we obtain Hookes law
7.2
Experimental Observations
We shall discuss two experiments which will yield the elastic Youngs modulus, and then the bulk
modulus. In the former, the simplicity of the experiment is surrounded by the intriguing character of
Hooke, and in the later, the bulk modulus is mathematically related to the Green deformation tensor
C, the deformation gradient F and the Lagrangian strain tensor E.
22
7.2.1
Hookes Law
23 Hookes Law is determined on the basis of a very simple experiment in which a uniaxial force is
applied on a specimen which has one dimension much greater than the other two (such as a rod). The
elongation is measured, and then the stress is plotted in terms of the strain (elongation/length). The
slope of the line is called Youngs modulus.
24 Hooke anticipated some of the most important discoveries and inventions of his time but failed to carry
many of them through to completion. He formulated the theory of planetary motion as a problem in
mechanics, and grasped, but did not develop mathematically, the fundamental theory on which Newton
formulated the law of gravitation.
His most important contribution was published in 1678 in the paper De Potentia Restitutiva. It
contained results of his experiments with elastic bodies, and was the rst paper in which the elastic
properties of material was discussed.
Take a wire string of 20, or 30, or 40 ft long, and fasten the upper part thereof to a nail,
and to the other end fasten a Scale to receive the weights: Then with a pair of compasses take
the distance of the bottom of the scale from the ground or oor underneath, and set down the
said distance, then put inweights into the said scale and measure the several stretchings of
the said string, and set them down. Then compare the several stretchings of the said string,
and you will nd that they will always bear the same proportions one to the other that the
weights do that made them.
This became Hookes Law
(7.24)
= E
25 Because he was concerned about patent rights to his invention, he did not publish his law when rst
discovered it in 1660. Instead he published it in the form of an anagram ceiinosssttuu in 1676 and
the solution was given in 1678. Ut tensio sic vis (at the time the two symbols u and v were employed
interchangeably to denote either the vowel u or the consonant v), i.e. extension varies directly with force.
7.2.2
Bulk Modulus
26 If, instead of subjecting a material to a uniaxial state of stress, we now subject it to a hydrostatic
pressure p and measure the change in volume V .
27
Victor Saouma
(det F)V0
=
det C =
(7.25-a)
det[I + 2E]
(7.25-b)
Draft
76
therefore,
V + V
= det[I + 2E]
V
we can expand the determinant of the tensor det[I + 2E] to nd
(7.26)
(7.27)
IIE
IIIE since the rst term is linear in E, the second is quadratic, and
but for small strains, IE
the third is cubic. Therefore, we can approximate det[I+2E] 1+2IE , hence we dene the volumetric
dilatation as
V
e IE = tr E
V
(7.28)
7.3
7.3.1
28
From Eq. 7.22 and 7.23 we obtain the stress-strain relation for homogeneous anisotropic material
T22
T33
c3333 c3312 c3323 c3331
E33
=
c1212 c1223 c1231 2E12 (12 )
T12
(7.29)
T23
SYM.
c2323 c2331 2E23 (23 )
T31
c3131
2E31 (31 )
cijkm
Tij
Ekm
1
Tij Eij
2
(7.30)
30 In general the elastic moduli cij relating the cartesian components of stress and strain depend on the
orientation of the coordinate system with respect to the body. If the form of elastic potential function
W and the values cij are independent of the orientation, the material is said to be isotropic, if not it
is anisotropic.
31
c3,2,1,1
c3,1,1,1 c3,1,1,2 c3,1,1,3
34 = 81 terms.
c1,2,1,2
c1,2,2,2
c1,2,3,2
c2,2,1,2
c2,2,2,2
c2,2,3,2
c3,2,1,2
c3,2,2,2
c3,2,3,2
c1,2,1,3
c1,2,2,3
c1,2,3,3
c2,2,1,3
c2,2,2,3
c2,2,3,3
c3,2,1,3
c3,2,2,3
c3,2,3,3
c1,3,1,1
c1,3,2,1
c1,3,3,1
c2,3,1,1
c2,3,2,1
c2,3,3,1
c3,3,1,1
c3,3,2,1
c3,3,3,1
c1,3,1,2
c1,3,2,2
c1,3,3,2
c2,3,1,2
c2,3,2,2
c2,3,3,2
c3,3,1,2
c3,3,2,2
c3,3,3,2
c1,3,1,3
c1,3,2,3
c1,3,3,3
c2,3,1,3
c2,3,2,3
c2,3,3,3
c3,3,1,3
c3,3,2,3
c3,3,3,3
(7.31)
Victor Saouma
Draft
77
But the matrix must be symmetric thanks to Cauchys second law of motion (i.e symmetry of both the
stress and the strain), and thus for anisotropic material we will have a symmetric 6 by 6 matrix with
(6)(6+1)
= 21 independent coecients.
2
32 By means of coordinate transformation we can relate the material properties in one coordinate system
(old) xi , to a new one xi , thus from Eq. 1.27 (v j = ap vp ) we can rewrite
j
W =
1
1
1
crstu Ers Etu = crstu ar as at au E ij E km = cijkm E ij E km
i j k m
2
2
2
(7.32)
thus we deduce
cijkm = ar as at au crstu
i j k m
(7.33)
that is the fourth order tensor of material constants in old coordinates may be transformed into a new
coordinate system through an eighth-order tensor ar as at au
i j k m
7.3.2
Monotropic Material
A plane of elastic symmetry exists at a point where the elastic constants have the same values
for every pair of coordinate systems which are the reected images of one another with respect to the
plane. The axes of such coordinate systems are referred to as equivalent elastic directions.
33
34
1 0 0
aj = 0 1 0
(7.34)
i
0 0 1
where the negative sign reects the symmetry of the mirror image with respect to the x3 plane.
We next substitute in Eq.7.33, and as an example we consider c1123 = ar as at au crstu = a1 a1 a2 a3 c1123 =
1 1 2 3
1 1 2 3
(1)(1)(1)(1)c1123 = c1123 , obviously, this is not possible, and the only way the relation can remanin
valid is if c1123 = 0. We note that all terms in cijkl with the index 3 occurring an odd number of times
will be equal to zero. Upon substitution, we obtain
0
0
c2222 c2233 c2212
0
0
c3333 c3312
(7.35)
cijkm =
0
0
c1212
SYM.
c2323 c2331
c3131
35
7.3.3
Orthotropic Material
36 If the material possesses three mutually perpendicular planes of elastic symmetry, (that is symmetric
with respect to two planes x2 and x3 ), then the transformation xi = aj xj is dened through
i
1 0
0
aj = 0 1 0
(7.36)
i
0 0 1
Victor Saouma
Draft
78
c1111 c1122
c2222
cijkm =
SYM.
0
0
0
0
0
0
0
c1212
0
0
0
0
0
c2323
(7.37)
c3131
We note that in here all terms of cijkl with the indices 3 and 2 occuring an odd number of times are
again set to zero.
37
Wood is usually considered an orthotropic material and will have 9 nonzero coecients.
7.3.4
38 A material is transversely isotropic if there is a preferential direction normal to all but one of the
three axes. If this axis is x3 , then rotation about it will require that
cos
sin 0
(7.38)
aj = sin cos 0
i
0
0
1
substituting Eq. 7.33 into Eq. 7.41, using the above transformation matrix, we obtain
(cos4 )c1111 + (cos2 sin2 )(2c1122 + 4c1212 ) + (sin4 )c2222
c1111
c1122
(cos sin )c1111 + (cos )c1122 4(cos sin )c1212 + (sin )c2211
(7.39-b)
(7.39-c)
+(sin2 cos2 )c2222
2
2
= (cos )c1133 + (sin )c2233
(7.39-d)
4
2
2
4
= (sin )c1111 + (cos sin )(2c1122 + 4c1212 ) + (cos )c2222
(7.39-e)
2
2
2
2
2
2
4
= (cos sin )c1111 2(cos sin )c1122 2(cos sin )c1212 + (cos )c1212 (7.39-f)
c1133
c2222
c1212
(7.39-a)
(7.39-g)
.
.
.
But in order to respect our initial assumption about symmetry, these results require that
c1111
c1133
cijkm
c1111
(7.40-a)
(7.40-b)
=
=
c3131
1
(c1111 c1122 )
2
(7.40-c)
c1212
c2222
c2233
c2323
yielding
=
=
c1122
c2222
SYM.
c1133
c2233
c3333
0
0
0
1
(c1111 c1122 )
2
(7.40-d)
0
0
0
0
c2323
0
0
0
0
0
(7.41)
c3131
we now have 5 nonzero coecients.
39 It should be noted that very few natural or man-made materials are truly orthotropic (certain crystals
as topaz are), but a number are transversely isotropic (laminates, shist, quartz, roller compacted concrete,
etc...).
Victor Saouma
Draft
7.3.5
79
Isotropic Material
40 An isotropic material is symmetric with respect to every plane and every axis, that is the elastic
properties are identical in all directions.
41 To mathematically characterize an isotropic material, we require coordinate transformation with
rotation about x2 and x1 axes in addition to all previous coordinate transformations. This process will
enforce symmetry about all planes and all axes.
42
cos 0 sin
j
1
0
ai = 0
sin 0 cos
(7.42)
43
=
=
c3333
1
(c1111 c1133 )
2
1
0
aj = 0 cos
i
0 sin
(7.43-a)
(7.43-b)
0
sin
cos
(7.44)
it follows that
c1122
c3131
c2323
which will nally give
cijkm
c1111
c1133
1
(c3333 c1133 )
2
1
(c2222 c2233 )
2
c1122
c2222
c1133
c2233
c3333
SYM.
0 0
0 0
0 0
a 0
b
(7.45-a)
(7.45-b)
(7.45-c)
0
0
0
0
0
c
(7.46)
+ 2
0 0 0
+ 2
0 0 0
+ 2 0 0 0
(7.47)
cijkm =
0 0
SYM.
0
Victor Saouma
ij km + (ik jm + im kj )
(7.48)
Draft
710
and we are thus left with only two independent non zero coecients and which are called Lames
constants.
45
(7.49)
Tij
ij Tkk
Eij =
2
3 + 2
or
or
T = IE + 2E
1
IT +
T
E=
2(3 + 2)
2
(7.50)
(7.51)
It should be emphasized that Eq. 7.47 is written in terms of the Engineering strains (Eq. 7.29)
that is ij = 2Eij for i = j. On the other hand the preceding equations are written in terms of the
tensorial strains Eij
46
7.3.5.1
Engineering Constants
47 The stress-strain relations were expressed in terms of Lames parameters which can not be readily
measured experimentally. As such, in the following sections we will reformulate those relations in terms
of engineering constants (Youngs and the bulks modulus). This will be done for both the isotropic
and transversely isotropic cases.
7.3.5.1.1
Isotropic Case
7.3.5.1.1.1
Youngs Modulus
In order to avoid certain confusion between the strain E and the elastic constant E, we adopt the
usual engineering notation Tij ij and Eij ij
48
49
If we consider a simple uniaxial state of stress in the x1 direction, then from Eq. 7.51
+
(3 + 2)
= 33 =
2(3 + 2)
0 = 12 = 23 = 13
11
22
(7.52-a)
(7.52-b)
(7.52-c)
Yet we have the elementary relations in terms engineering constants E Youngs modulus and
Poissons ratio
(7.53-a)
11 =
E
22
33
=
=
(7.53-b)
11
11
50
Victor Saouma
=
=
; =
(3 + 2)
2( + )
E
E
; = G =
(1 + )(1 2)
2(1 + )
(7.54)
(7.55)
Draft
51
711
212
all other ij = 0
(7.56-a)
(7.56-b)
53
54
=
=
1
xx
yy
1
zz
=
xy (2xy ) E 0
yz (2yz )
zx (2zx )
0
0
0
0
xx
1
0
0
0 yy
zz
1
0
0
0
0
0 1+
0
0 xy
0
0
0
1+
0 yz
zx
0
0
0
0
1+
xx
1
yy (1+)(12)
zz
=
xy
yz
zx
7.3.5.1.1.2
55
E
E
ij +
ij kk
+
I
or =
1+
1 2
1+
1 2
1+
1+
ij ij kk or =
I
E
E
E
E
1
G 0
0
0
1
0
(7.57)
(7.58)
(7.59)
xx
yy
zz
xy (2xy )
0
0 yz (2yz )
1
zx (2zx )
(7.60)
We can express the trace of the stress I in terms of the volumetric strain I From Eq. 7.50
ii = ii kk + 2ii = (3 + 2)ii 3Kii
(7.61)
2
K =+
3
(7.62)
or
56 We can provide a complement to the volumetric part of the constitutive equations by substracting
the trace of the stress from the stress tensor, hence we dene the deviatoric stress and strains as as
1
(tr )I (7.63)
3
1
(tr )I (7.64)
3
KeI + 2
1
p
I+
3K
2
(7.65)
(7.66)
Draft
712
7.3.5.1.1.3
57
(7.67)
but since dW is a scalar invariant (energy), it can be expressed in terms of volumetric (hydrostatic) and
deviatoric components as
(7.68)
dW = pde + ij dEij
substituting p = Ke and ij = 2GEij , and integrating, we obtain the following expression for the
isotropic strain energy
1
(7.69)
W = Ke2 + GEij Eij
2
and since positive work is required to cause any deformation W > 0 thus
2
+ GK
3
G
>
(7.70-a)
>
(7.70-b)
1
2
(7.71)
58
59
+ 2
3
(3+2)
+
2(+)
1
2
1 2
e + GEij Eij
2
implies G =
E
3,
and
(7.72)
1
K
= 0 or elastic incompressibility.
E,
E,
K,
E
(1+)(12)
E
2(1+)
E
3(12)
2
12
(E2)
3E
2(1+)
3(12)
3K
1+
3K(12)
2(1+)
E
3(3E)
2(1 + )
E
2
E
1
K
3K(1 2)
7.3.5.1.2
61
Victor Saouma
xx
yy
zz
xy
yz
xz
=
=
=
=
=
=
Draft
713
E (MPa)
196,000
68,000
61,000
2,900
2
60,000
60,000
0.3
0.33
0.34
0.4
0.5
0.2
0.27
1
;
E
a12 =
;
E
a13 =
;
E
a33 =
1
;
E
a44 =
(7.74)
where E is the Youngs modulus in the plane of isotropy and E the one in the plane normal to it.
corresponds to the transverse contraction in the plane of isotropy when tension is applied in the plane;
corresponding to the transverse contraction in the plane of isotropy when tension is applied normal
to the plane; corresponding to the shear moduli for the plane of isotropy and any plane normal to it,
and is shear moduli for the plane of isotropy.
7.3.5.2
62
Special 2D Cases
Often times one can make simplifying assumptions to reduce a 3D problem into a 2D one.
7.3.5.2.1
Plane Strain
For problems involving a long body in the z direction with no variation in load or geometry, then
zz = yz = xz = xz = yz = 0. Thus, replacing into Eq. 5.2 we obtain
(1 )
0
xx
xx
(1 )
0
yy
(7.75)
=
0 yy
zz (1 + )(1 2)
xy
12
xy
0
0
2
63
7.3.5.2.2
64
Axisymmetry
zz
rz
Victor Saouma
u
r
u
r
w
z
u w
+
z
r
(7.76-a)
(7.76-b)
(7.76-c)
(7.76-d)
Draft
714
65
rr
E
zz
1
=
(1 + )(1 2)
rz
0
0
0
7.3.5.2.3
0
0
0
0
12
2
rr
zz
rz
(7.77)
Plane Stress
If the longitudinal dimension in z direction is much smaller than in the x and y directions, then
yz = xz = zz = xz = yz = 0 throughout the thickness. Again, substituting into Eq. 5.2 we obtain:
1
0
xx
xx
1
1
0
yy
yy
=
(7.78-a)
1 2
xy
xy
0 0 1
2
1
(xx + yy )
(7.78-b)
zz =
1
66
7.4
Linear Thermoelasticity
67 If thermal eects are accounted for, the components of the linear strain tensor Eij may be considered
as the sum of
(T )
()
(7.79)
Eij = Eij + Eij
(T )
where Eij
eld.
()
68 When a body is subjected to a temperature change 0 with respect to the reference state
temperature, the strain componenet of an elementary volume of an unconstrained isotropic body are
given by
()
(7.80)
Eij = ( 0 )ij
Inserting the preceding two equation into Hookes law (Eq. 7.51) yields
Eij =
1
2
Tij
ij Tkk
3 + 2
+ ( 0 )ij
(7.81)
(7.82)
71 Alternatively, if we were to consider the derivation of the Green-elastic hyperelastic equations, (Sect.
7.1.5), we required the constants c1 to c6 in Eq. 7.22 to be zero in order that the stress vanish in the
unstrained state. If we accounted for the temperature change 0 with respect to the reference state
Victor Saouma
Draft
715
temperature, we would have ck = k ( 0 ) for k = 1 to 6 and would have to add like terms to Eq.
7.22, leading to
(7.83)
Tij = ij ( 0 ) + cijrs Ers
for linear theory, we suppose that ij is independent from the strain and cijrs independent of temperature
change with respect to the natural state. Finally, for isotropic cases we obtain
Tij = Ekk ij + 2Eij ij ( 0 )ij
which is identical to Eq. 7.82 with =
E
12 .
Hence
Tij =
72
(7.84)
E
1 2
(7.85)
Tij
2
(7.86)
7.5
p
+ 3( 0 )
K
(7.87)
Fourrier Law
73 Consider a solid through which there is a ow q of heat (or some other quantity such as mass, chemical,
etc...)
74
75 The direction of ow is in the direction of maximum potential (temperature in this case, but could
be, piezometric head, or ion concentration) decreases (Fourrier, Darcy, Fick...).
qx
x
qy
q=
= D
= D
(7.88)
y
qz
1
D = k 0
0
Anisotropic
kxx
D = kyx
kzx
Orthotropic
kxx
D= 0
0
0
1
0
0
0
1
(7.89)
kxy
kyy
kzy
kxz
kyz
kzz
(7.90)
0
kyy
0
0
0
kzz
(7.91)
Note that for ow through porous media, Darcys equation is only valid for laminar ow.
Victor Saouma
Draft
716
7.6
76 In light of the new equations introduced in this chapter, it would be appropriate to revisit our balance
of equations and unknowns.
d
vi
dt + xi = 0
Tij
dvi
xj + bi = dt
du = Tij Dij + r
dt
Coupled
1
3
Continuity Equation
Equation of motion
qj
xj
Energy equation
T = IE + 2E
Hookes Law
q = D
Heat Equation (Fourrier)
= (s, ); j = j (s, ) Equations of state
Total number of equations
Uncoupled
1
3
1
6
3
2
16
10
ds
dt
Coupled
1
3
6
3
1
1
1
16
1
div
Uncoupled
1
3
6
10
which governs entropy production
77 Hence we now have as many equations as unknowns and are (almost) ready to pose and solve problems
in continuum mechanics.
Victor Saouma
Draft
Chapter 8
INTERMEZZO
In light of the lengthy and rigorous derivation of the fundamental equations of Continuum Mechanics in
the preceding chapter, the reader may be at a loss as to what are the most important ones to remember.
Hence, since the complexity of some of the derivation may have eclipsed the nal results, this handout
seeks to summarize the most fundamental relations which you should always remember.
X3
X3
V3
33
t3
t2
13
21
11
23
31
t1
32
V2
X2
22
V1
X1
X2
12
Stress Vector/Tensor
Strain Tensor
Engineering Strain
Equilibrium
Boundary Conditions
Energy Potential
Hookes Law
ti = Tij nj
1 ui
uj
uk uk
Eij =
+
2 xj
xi
xi xj
1
1
11
2 12
2 13
1
1
= 2 12 22
2 23
1
1
13 2 23 33
2
(8.1-a)
(8.1-b)
(8.1-c)
(8.1-d)
(8.1-e)
(8.1-f)
(8.1-g)
(8.1-h)
Draft
82
INTERMEZZO
xx
yy
zz
xy
yz
xz
Plane Stress
Plane Strain
Victor Saouma
zz = 0;
zz = 0;
1
E
0
0
0
zz = 0
zz = 0
E
1
E
0
0
0
0
0
0
0
0
0
1
G
1
E
0
0
0
0
0
0
1
G
0
0
0
0
0
1
G
xx
yy
zz
xy
yz
xz
(8.1-i)
(8.1-j)
(8.1-k)
Draft
Part II
ELASTICITY/SOLID
MECHANICS
Draft
Draft
Chapter 9
Preliminary Considerations
3 Equations of Motion (Equilibrium): i.e. Equations relating the applied tractions and body forces
to the stresses (3)
2 ui
Tij
+ bi = 2
(9.1)
Xj
t
6 Stress-Strain relations: (Hookes Law)
T = IE + 2E
(9.2)
6 Geometric (kinematic) equations: i.e. Equations of geometry of deformation relating displacement to strain (6)
1
(9.3)
E = (u x + x u)
2
2 Those 15 equations are written in terms of 15 unknowns: 3 displacement ui , 6 stress components Tij ,
and 6 strain components Eij .
In addition to these equations which describe what is happening inside the body, we must describe
what is happening on the surface or boundary of the body. These extra conditions are called boundary
conditions.
9.2
4
Boundary Conditions
Draft
92
Properly specied boundary conditions result in well-posed boundary value problems, while improperly specied boundary conditions will result in ill-posed boundary value problem. Only the former
can be solved.
Thus we have two types of boundary conditions in terms of known quantitites, Fig. 9.1:
u
Figure 9.1: Boundary Conditions in Elasticity Problems
Displacement boundary conditions along u with the three components of ui prescribed on the
boundary. The displacement is decomposed into its cartesian (or curvilinear) components, i.e.
u x , uy
Traction boundary conditions along t with the three traction components ti = nj Tij prescribed
at a boundary where the unit normal is n. The traction is decomposed into its normal and shear(s)
components, i.e tn , ts .
Mixed boundary conditions where displacement boundary conditions are prescribed on a part of
the bounding surface, while traction boundary conditions are prescribed on the remainder.
We note that at some points, traction may be specied in one direction, and displacement at another.
Displacement and tractions can never be specied at the same point in the same direction.
Various terms have been associated with those boundary conditions in the litterature, those are suumarized in Table 9.1.
u, u
Dirichlet
Field Variable
Essential
Forced
Geometric
t, t
Neuman
Derivative(s) of Field Variable
Non-essential
Natural
Static
Victor Saouma
Draft
93
C
AB
BC
CD
DE
EA
E
A
u
ux uy
?
0
?
?
?
?
0
?
?
?
tn ts
?
0
0
0
0
?
0
0 0
9.3
9
2 ui
in
t2
(9.4)
1
(u x + x u)
(9.5)
2
T = IE + 2E in (9.6)
(9.7)
u = u in u
(9.8)
t = t in t
9.4
Compacted Forms
10 Solving a boundary value problem with 15 unknowns through 15 equations is a formidable task.
Hence, there are numerous methods to reformulate the problem in terms of fewer unknows.
9.4.1
Navier-Cauchy Equations
11 One such approach is to substitute the displacement-strain relation into Hookes law (resulting in
stresses in terms of the gradient of the displacement), and the resulting equation into the equation of
motion to obtain three second-order partial dierential equations for the three displacement components
known as Naviers Equation
( + )
2 uk
2 ui
+
+ bi
Xi Xk
Xk Xk
( + ) ( u) +
u + b
2 ui
t2
(9.9)
2u
t2
(9.10)
or
=
(9.11)
Victor Saouma
Draft
94
Essential B.C.
ui :
?
Body Forces
Displacements
bi
ui
Equilibrium
Tij
xj
Kinematics
E =
+ bi = dvi
dt
1
(u
2
?
Stresses
Tij
x u)
?
-
Constitutive Rel.
T = IE + 2E
Strain
Eij
Natural B.C.
ti :
Victor Saouma
Draft
9.4.2
95
Beltrami-Mitchell Equations
12 Whereas Navier-Cauchy equation was expressed in terms of the gradient of the displacement, we can
follow a similar approach and write a single equation in term of the gradient of the tractions.
Tij +
1
Tpp,ij
1+
1
Tpp,ij
Tij,pp +
1+
9.4.3
9.5
ij
1
or
=
(9.12)
(9.13)
13 For the isotropic Hookes law, we saw that there always exist a strain energy function W which is
positive-denite, homogeneous quadratic function of the strains such that, Eq. 7.20
W
Eij
(9.14)
1
Tij Eij
2
(9.15)
Tij =
hence it follows that
W =
14
The external work done by a body in equilibrium under body forces bi and surface traction ti is
bi ui d +
equal to
becomes
Tij nj ui d =
(Tij ui ),j d =
(9.16)
but Tij ui,j = Tij (Eij + ij ) = Tij Eij and from equilibrium Tij,j = bi , thus
bi ui d +
ti ui d =
(Tij Eij bi ui )d
bi ui d +
(9.17)
or
bi ui d +
Tij Eij
d
2
ti ui d = 2
External Work
(9.18)
that is For an elastic system, the total strain energy is one half the work done by the external forces
acting through their displacements ui .
9.6
Because the equations of linear elasticity are linear equations, the principles of superposition may be
(1)
(1)
used to obtain additional solutions from those established. Hence, given two sets of solution Tij , ui ,
15
(2)
(2)
(2)
(1)
(2)
(1)
ui
(1)
ti
(2)
= 0 on u , and ui = ui
(2)
with bi = bi
(1)
bi
ti ui d = 2
(1)
ui
= 0 must also be a
u d = 0.
= 0 on t , thus
Draft
96
But u is positive-denite and continuous, thus the integral can vanish if and only if u = 0 everywhere,
and this is only possible if Eij = 0 everywhere so that
17
(2)
(1)
(2)
(9.19)
hence, there can not be two dierent stress and strain elds corresponding to the same externally imposed
body forces and boundary conditions1 and satisfying the linearized elastostatic Eqs 9.1, 9.14 and 9.3.
9.7
18 This famous principle of Saint Venant was enunciated in 1855 and is of great importance in applied
elasticity where it is often invoked to justify certain simplied solutions to complex problem.
dx
t
F=tdx
9.8
Cylindrical Coordinates
20 So far all equations have been written in either vector, indicial, or engineering notation. The last two
were so far restricted to an othonormal cartesian coordinate system.
21 We now rewrite some of the fundamental relations in cylindrical coordinate system, Fig. 9.5, as
this would enable us to analytically solve some simple problems of great practical usefulness (torsion,
pressurized cylinders, ...). This is most often achieved by reducing the dimensionality of the problem
from 3 to 2 or even to 1.
1 This
Victor Saouma
Draft
97
9.8.1
22
Strains
With reference to Fig. 9.6, we consider the displacement of point P to P . the displacements can be
y
uy
P*
ur
ux
=
=
ur cos u sin
ur sin + u cos
(9.20-a)
(9.20-b)
substituting into the strain denition for xx (for small displacements) we obtain
xx
ux
ux
r
x
r
x
xx
Victor Saouma
ux
ux r
ux
=
+
x
x
r x
ur
u
=
cos ur sin
sin u cos
ur
u
=
cos
sin
r
r
sin
=
r
(9.21-a)
(9.21-c)
(9.21-d)
cos
(9.21-b)
(9.21-e)
ur
u
cos + ur sin +
sin + u cos
sin
r
Draft
98
(9.21-f)
ur
r
(9.22)
r =
24
u
1 ur
1
u
xy |0 =
+
2
r
r
r
(9.24)
In summary, and with the addition of the z components (not explicitely derived), we obtain
rr
rz
9.8.2
zz
ur
r
ur
1 u
+
r
r
uz
z
u
ut heta
1 1 ur
+
2 r
r
r
1 uz
1 u
+
2 z
r
ur
1 uz
+
2 r
z
(9.25)
(9.26)
(9.27)
(9.28)
(9.29)
(9.30)
Equilibrium
Trr
dr (r + dr)d Trr (rd)
r
d
T
+ T dr sin
T +
2
d
Tr
d Tr dr cos
+ fr rdrd
+ Tr +
2
Trr +
(9.31-a)
(9.31-b)
1+
dr
r
T
T d 1 Tr
+
+ fr = 0
r
dr
r
(9.32)
27 Similarly we can take the summation of forces in the direction. In both cases if we were to drop the
dr/r and d/r in the limit, we obtain
Victor Saouma
Draft
99
T
T + d
Trr
d
fr
T
r d r
r
Tr
r d r
r
r +
rr +
T
r d
T
r
r +
r
r+dr
(9.33)
(9.34)
28 It is often necessary to express cartesian stresses in terms of polar stresses and vice versa. This can
be done through the following relationships
cos
sin
sin
cos
Trr
Tr
Tr
T
cos
sin
sin
cos
Txy
Tyy
Txx
Tyy
=
=
(9.36-a)
(9.36-b)
Txy
Txx
Txy
(9.36-c)
(9.35)
yielding
(recalling that sin2 = 1/2 sin 2, and cos2 = 1/2(1 + cos 2)).
9.8.3
Stress-Strain Relations
29 In orthogonal curvilinear coordinates, the physical components of a tensor at a point are merely the
Cartesian components in a local coordinate system at the point with its axes tangent to the coordinate
curves. Hence,
Trr
T
=
=
e + 2rr
e + 2
(9.37)
(9.38)
Tr
Tzz
=
=
2r
(Trr + T )
(9.39)
(9.40)
with e = rr + . alternatively,
Victor Saouma
Draft
910
Err
E
Erz
30
Er
9.8.3.1
1
(1 2 )Trr (1 + )T
E
1
(1 2 )T (1 + )Trr
E
1+
Tr
E
Ez = Ezz = 0
(9.41)
(9.42)
(9.43)
(9.44)
Plane Strain
(1 )
rr
zz (1 + )(1 2)
r
0
(1 )
0
0
0
12
2
rr
(9.45)
and zz = rz = z = rz = z = 0.
31
Inverting,
9.8.3.2
32
1 2
rr
1 (1 + )
E
r
0
(1 + )
1 2
0
rr
zz
0
r
2(1 +
(9.46)
Plane Stress
1
0
rr
rr
E
1
0
1 2
r
r
0 0 1
2
1
(rr + )
zz =
1
(9.47-a)
(9.47-b)
and rz = z = zz = rz = z = 0
33
Inverting
rr
Victor Saouma
1
1
E
0
1
0
0
rr
2(1 + )
r
(9.48-a)
Draft
Chapter 10
10.1
Semi-Inverse Method
Often a solution to an elasticity problem may be obtained without seeking simulateneous solutions to
the equations of motion, Hookes Law and boundary conditions. One may attempt to seek solutions
by making certain assumptions or guesses about the components of strain stress or displacement while
leaving enough freedom in these assumptions so that the equations of elasticity be satised.
If the assumptions allow us to satisfy the elasticity equations, then by the uniqueness theorem, we
have succeeded in obtaining the solution to the problem.
This method was employed by Saint-Venant in his treatment of the torsion problem, hence it is often
referred to as the Saint-Venant semi-inverse method.
10.1.1
Let us consider the elastic deformation of a cylindrical bar with circular cross section of radius a and
length L twisted by equal and opposite end moments M1 , Fig. 10.1.
From symmetry, it is reasonable to assume that the motion of each cross-sectional plane is a rigid body
rotation about the x1 axis. Hence, for a small rotation angle , the displacement eld will be given by:
(10.1)
or
u1 = 0;
u2 = x3 ;
u3 = x2
(10.2)
where = (x1 ).
8
E12
E22 = E33 = 0
1
x3
2 x1
(10.3-a)
(10.3-b)
Draft
102
X2
MT
a
n
X3
X1
MT
L
Figure 10.1: Torsion of a Circular Bar
1
x2
2 x1
E13
(10.3-c)
The non zero stress components are obtained from Hookes law
T12
T13
x1
x2
x1
x3
(10.4-a)
(10.4-b)
10 We need to check that this state of stress satises equilibrium Tij /xj = 0. The rst one j = 1 is
identically satised, whereas the other two yield
d2
dx2
1
d2
x2 2
dx1
x3
(10.5-a)
(10.5-b)
thus,
d
= constant
(10.6)
dx1
Physically, this means that equilibrium is only satised if the increment in angular rotation (twist per
unit length) is a constant.
11 We next determine the corresponding surface tractions. On the lateral surface we have a unit normal
1
vector n = a (x2 e2 + x3 e3 ), therefore the surface traction on the lateral surface is given by
0 T12 T13 0
x T
1
1 2 12
x2
T21
0
0
0
{t} = [T]{n} =
(10.7)
=
a
a
0
0
x3
T31
0
12
Substituting,
(x2 x3 + x2 x3 )e1 = 0
(10.8)
a
which is in agreement with the fact that the bar is twisted by end moments only, the lateral surface is
traction free.
t=
Victor Saouma
Draft
13
103
(10.9)
this distribution of surface traction on the end face gives rise to the following resultants
R1
T11 dA = 0
(10.10-a)
R2
T21 dA =
x3 dA = 0
(10.10-b)
R3
T31 dA =
x2 dA = 0
(10.10-c)
M1
M2
(x2 + x2 )dA = J
2
3
M3 = 0
(10.10-d)
(10.10-e)
We note that (x2 + x3 )2 dA is the polar moment of inertia of the cross section and is equal to
2
3
J = a4 /2, and we also note that x2 dA = x3 dA = 0 because the area is symmetric with respect to
the axes.
14
M
J
(10.11)
which implies that the shear modulus can be determined froma simple torsion experiment.
15
0
MJx3 MJx2
[T] = MJx3
0
0
M x2
0
0
J
10.2
10.2.1
(10.12)
16 If the deformation of a cylindrical body is such that there is no axial components of the displacement
and that the other components do not depend on the axial coordinate, then the body is said to be in a
state of plane strain. If e3 is the direction corresponding to the cylindrical axis, then we have
u1 = u1 (x1 , x2 ),
u2 = u2 (x1 , x2 ),
u3 = 0
(10.13)
E22
E12
E13
u1
x1
u2
x2
1 u1
u2
+
2 x2
x1
E23 = E33 = 0
(10.14-a)
(10.14-b)
(10.14-c)
(10.14-d)
and the non-zero stress components are T11 , T12 , T22 , T33 where
T33 = (T11 + T22 )
Victor Saouma
(10.15)
Draft
104
17
Considering a static stress eld with no body forces, the equilibrium equations reduce to:
T12
T11
+
x1
x2
T22
T12
+
x1
x2
T33
=0
x1
(10.16-a)
(10.16-b)
(10.16-c)
we note that since T33 = T33 (x1 , x2 ), the last equation is always satised.
18 Hence, it can be easily veried that for any arbitrary scalar variable , if we compute the stress
components from
T11
T22
T12
2
x2
2
2
x2
1
2
x1 x2
(10.17)
(10.18)
(10.19)
then the rst two equations of equilibrium are automatically satised. This function is called Airy
stress function.
19 However, if stress components determined this way are statically admissible (i.e.
they satisfy
equilibrium), they are not necessarily kinematically admissible (i.e. satisfy compatibility equations).
20 To ensure compatibility of the strain components, we obtain the strains components in terms of
from Hookes law, Eq. 5.1 and Eq. 10.15.
E11
E22
E12
21
1
1
2
2
(1 2 )T11 (1 + )T22 =
(1 2 ) 2 (1 + ) 2
E
E
x2
x1
2
1
1
2
(1 2 )T22 (1 + )T11 =
(1 2 ) 2 (1 + ) 2
E
E
x1
x2
2
1
1
(1 + )T12 = (1 + )
E
E
x1 x2
(10.20-a)
(10.20-b)
(10.20-c)
For plane strain problems, the only compatibility equation, 4.140, that is not automatically satised
is
2 E22
2 E12
2 E11
2 + x2 = 2 x x
x2
1
2
1
(10.21)
4
4
4
4 + 2 x2 x2 + x4
x1
1
2
1
=0
(10.22)
=0
(10.23)
or
4
4
4
4 + 2 x2 x2 + x4 = 0 or
x1
1
2
1
Hence, any function which satises the preceding equation will satisfy both equilibrium and kinematic
and is thus an acceptable elasticity solution.
Victor Saouma
Draft
105
22 We can also obtain from the Hookes law, the compatibility equation 10.21, and the equilibrium
equations the following
2
2
+
x2
x2
1
2
(T11 + T22 ) = 0 or
(T11 + T22 ) = 0
(10.24)
23 Any polynomial of degree three or less in x and y satises the biharmonic equation (Eq. 10.23). A
systematic way of selecting coecients begins with
Cmn xm y n
(10.25)
m=0 n=0
24
Txx
n(n 1)Cmn xm y n2
(10.26-a)
(10.26-b)
m=0 n=2
Tyy
m=2 n=0
Txy
mnCmn xm1 y n1
(10.26-c)
m=1 n=1
25
(10.27)
but since the equation must be identically satised for all x and y, the term in bracket must be equal to
zero.
(m+2)(m+1)m(m1)Cm+2,n2 +2m(m1)n(n1)Cmn +(n+2)(n+1)n(n1)Cm2,n+2 = 0 (10.28)
Hence, the recursion relation establishes relationships among groups of three alternate coecients which
can be selected from
0
0
C02 C03 C04 C05 C06
0
C11 C12 C13 C14 C15
(10.29)
C50 C
51
C60
(10.30)
Victor Saouma
Draft
106
10.2.1.1
26
(10.31)
=
=
=
(10.32-a)
(10.32-b)
(10.32-c)
These can be used for the end-loaded cantilever beam with width b along the z axis, depth 2a and
length L.
28
6C13 xy
(10.33-a)
Tyy
Txy
=
=
0
3C13 y 2
(10.33-b)
(10.33-c)
29 This will give a parabolic shear traction on the loaded end (correct), but also a uniform shear traction
Txy = 3C13 a2 on top and bottom. These can be removed by superposing uniform shear stress Txy =
+3C13 a2 corresponding to 2 = 3C13 a2 xy. Thus
(10.34)
P =b
a
Txy dy = 3bC13
hence
C13 =
a
a
(a2 y 2 )dy
P
4a3 b
(10.35)
(10.36)
Txx
Txy
Tyy
3P
P
xy 3 xy 3
4ab
4a b
3P
3 xy
2a b
3P
3 (a2 y 2 )
4a b
0
(10.37-a)
(10.37-b)
(10.37-c)
(10.37-d)
We observe that the second moment of area for the rectangular cross section is I = b(2a)3 /12 = 2a3 b/3,
hence this solution agrees with the elementary beam theory solution
31
Txx
Txy
Tyy
Victor Saouma
P
3P
xy 3 xy 3
4ab
4a b
y
M
P
xy = M =
I
I
S
P 2
2
(a y )
2I
0
C11 xy + C13 xy 3 =
(10.38-a)
(10.38-b)
(10.38-c)
(10.38-d)
Draft
107
10.2.2
Polar Coordinates
10.2.2.1
32
In polar coordinates, the strain components in plane strain are, Eq. 9.46
Err
E
Er
Erz
1
(1 2 )Trr (1 + )T
E
1
(1 2 )T (1 + )Trr
E
1+
Tr
E
Ez = Ezz = 0
(10.39-a)
(10.39-b)
(10.39-c)
(10.39-d)
r r
r
r
1 T
1 Tr
+
r2 r
r
33
(10.40-a)
(10.40-b)
Again, it can be easily veried that the equations of equilibrium are identically satised if
Trr
Tr
1 2
1
+ 2 2
r r
r
2
r2
1
r r
(10.41)
(10.42)
(10.43)
34 In order to satisfy the compatibility conditions, the cartesian stress components must also satisfy Eq.
10.24. To derive the equivalent expression in cylindrical coordinates, we note that T11 + T22 is the rst
scalar invariant of the stress tensor, therefore
(10.44)
We also note that in cylindrical coordinates, the Laplacian operator takes the following form
2
36
1 2 2
1
+ 2 2 +
r r
r
r2
1 2
2
1
+ 2 2
+
r2
r r r
(10.45)
Victor Saouma
1 2
1
2
+ 2 2
+
2
r
r r r
= 0 or
=0
(10.46)
Draft
108
10.2.2.2
37
and
38
1 d
;
r dr
T =
d2
;
dr2
Tr = 0
(10.47)
1 d2
1 d
d 4 2 d3
=0
+
2 2 + 3
dr4
r dr3
r dr
r dr
(10.48)
DSolve[phi[r]+2 phi[r]/r-phi[r]/r^2+phi[r]/r^3==0,phi[r],r]
= A ln r + Br2 ln r + Cr2 + D
39
(10.49)
A
+ B(1 + 2 ln r) + 2C
r2
A
= 2 + B(3 + 2 ln r) + 2C
r
= 0
=
(10.50)
(10.51)
(10.52)
Err
Er
40
1 (1 + )A
ur
=
+ (1 3 4 2 )B + 2(1 2 2 )B ln r + 2(1 2 2 )C
r
E
r2
ur
1
(1 + )A
1 u
+
=
+ (3 4 2 )B + 2(1 2 2 )B ln r + 2(1 2 2 )C
r
r
E
r2
0
Finally, the displacement components can be obtained by integrating the above equations
ur
10.2.2.3
(1 + )A
1
(10.56)
(10.57)
41 If we consider a circular cylinder with internal and external radii a and b respectively, subjected to
internal and external pressures pi and po respectively, Fig. 10.2, then the boundary conditions for the
plane strain problem are
Trr
Trr
Victor Saouma
=
=
pi at r = a
po at r = b
(10.58-a)
(10.58-b)
(10.53)
(10.54)
(10.55)
Draft
109
Saint Venant
po
a
p
i
42
These Boundary conditions can be easily shown to be satised by the following stress eld
Trr
Tr
A
+ 2C
r2
A
2 + 2C
r
0
(10.59-a)
(10.59-b)
(10.59-c)
These equations are taken from Eq. 10.50, 10.51 and 10.52 with B = 0 and therefore represent a
possible state of stress for the plane strain problem.
We note that if we take B = 0, then u = 4rB (1 2 ) and this is not acceptable because if we were
E
to start at = 0 and trace a curve around the origin and return to the same point, than = 2 and the
displacement would then be dierent.
43
44
Tr
(b2 /r2 ) 1
1 (a2 /r2 )
p0
2 /a2 ) 1
(b
1 (a2 /b2 )
(b2 /r2 ) + 1
1 + (a2 /r2 )
p0
pi 2 2
(b /a ) 1
1 (a2 /b2 )
0
pi
(10.60)
(10.61)
(10.62)
45 We note that if only the internal pressure pi is acting, then Trr is always a compressive stress, and
T is always positive.
46
If the cylinder is thick, then the strains are given by Eq. 10.53, 10.54 and 10.55. For a very thin
Victor Saouma
Draft
1010
Ezz
Er
1
du
= (Trr T )
dr
E
1
u
= (T Trr )
r
E
dw
= (Trr + T )
dz
E
(1 + )
Tr
E
(10.63-a)
(10.63-b)
(10.63-c)
(10.63-d)
It should be noted that applying Saint-Venants principle the above solution is only valid away from
the ends of the cylinder.
47
10.2.2.4
48 We consider next a hollow sphere with internal and xternal radii ai and ao respectively, and subjected
to internal and external pressures of pi and po , Fig. 10.3.
ai
ao
49
ur = ur (r),
10.2.2.5
u = u = 0
(10.64)
50 Analysing the innite plate under uniform tension with a circular hole of diameter a, and subjected
to a uniform stress 0 , Fig. 10.4.
51 The peculiarity of this problem is that the far-eld boundary conditions are better expressed in
cartesian coordinates, whereas the ones around the hole should be written in polar coordinate system.
52 First we select a stress function which satises the biharmonic Equation (Eq. 10.23), and the far-eld
boundary conditions. From St Venant principle, away from the hole, the boundary conditions are given
by:
Tyy = Txy = 0
(10.65)
Txx = 0 ;
2
Recalling (Eq. 10.19) that Txx = , this would would suggest a stress function of the form =
y 2
0 y 2 . Alternatively, the presence of the circular hole would suggest a polar representation of . Thus,
substituting y = r sin would result in = 0 r2 sin2 .
Victor Saouma
Draft
1011
y
r
rr
rr
b
rr
b
II
Since sin2 = 1 (1 cos 2), we could simplify the stress function into
2
= f (r) cos 2
(10.66)
Substituting this function into the biharmonic equation (Eq. 10.46) yields
1 2
1 2
1
2
2 1
+ 2 2
+ 2 2
+
+
r2
r r r
r2
r r
r
2
4
4f
1 d
1 df
d
d2 f
2
2
+
+
2
2
dr
r dr r
dr
r dr
r
54
= 0
(10.67-a)
= 0
(10.67-b)
The general solution of this ordinary linear fourth order dierential equation is
1
+D
r2
(10.68)
1
+ D cos 2
r2
(10.69)
Ar2 + Br4 + C
1 2
1
6C
4D
+ 2 2 = 2A + 4 + 2 cos 2
r r
r
r
r
2
6C
=
= 2A + 12Br2 + 4 cos 2
r2
r
1
6C
2D
= 2A + 6Br2 4 2 sin 2
=
r r
r
r
(10.70-a)
(10.70-b)
(10.70-c)
55 Next we seek to solve for the four constants of integration by applying the boundary conditions. We
will identify two sets of boundary conditions:
1. Outer boundaries: around an innitely large circle of radius b inside a plate subjected to uniform
stress 0 , the stresses in polar coordinates are obtained from Eq. 9.35
Trr
Tr
Tr
T
cos
sin
sin
cos
0
0
0
0
cos
sin
sin
cos
(10.71)
yielding (recalling that sin2 = 1/2 sin 2, and cos2 = 1/2(1 + cos 2)).
(Trr )r=b
(Tr )r=b
(T )r=b
Victor Saouma
0 cos2 =
1
0 (1 + cos 2)
2
1
0 sin 2
2
0
(1 cos 2)
2
(10.72-a)
(10.72-b)
(10.72-c)
Draft
1012
For reasons which will become apparent later, it is more convenient to decompose the state of
stress given by Eq. 10.72-a and 10.72-b, into state I and II:
(Trr )I
r=b
(Tr )I
r=b
(Trr )II
r=b
(Tr )II
r=b
1
0
2
0
1
0 cos 2
2
1
0 sin 2
2
(10.73-a)
(10.73-b)
(10.73-c)
(10.73-d)
Where state I corresponds to a thick cylinder with external pressure applied on r = b and of
magnitude 0 /2. This problem has already been previously solved. Hence, only the last two
equations will provide us with boundary conditions.
2. Around the hole: the stresses should be equal to zero:
(Trr )r=a
(Tr )r=a
=
=
0
0
(10.74-a)
(10.74-b)
56 Upon substitution in Eq. 10.70-a the four boundary conditions (Eq. 10.73-c, 10.73-d, 10.74-a, and
10.74-b) become
4D
b2
2D
2
b
4D
+ 2
a
2D
2
a
6C
b4
6C
2A + 6Bb2 4
b
6C
2A + 4
a
6C
2A + 6Ba2 4
a
2A +
57
0
;
4
a
b
B = 0;
1
0
2
1
0
2
(10.75-a)
(10.75-b)
(10.75-c)
(10.75-d)
a4
0 ;
4
D=
a2
0
2
(10.76)
58 To this solution, we must superimpose the one of a thick cylinder subjected to a uniform radial
traction 0 /2 on the outer surface, and with b much greater than a. These stresses were derived in Eqs.
10.60 and 10.61 yielding for this problem (carefull about the sign)
Trr
0
2
0
2
a2
r2
a2
1+ 2
r
1
(10.77-a)
(10.77-b)
Tr
Victor Saouma
a2
a4
4a2 1
0 cos 2
+ 1+3 4 2
r2
r
r
2
a2
3a4 1
1+ 2 1+ 4
0 cos 2
r
r
2
3a4
2a2 1
0 sin 2
1 4 + 2
r
r
2
0
2
0
2
(10.78-a)
(10.78-b)
(10.78-c)
Draft
1013
59 We observe that as r , both Trr and Tr are equal to the values given in Eq. 10.72-a and 10.72-b
respectively.
60
which for =
and
Victor Saouma
3
2
(10.79)
Draft
Victor Saouma
1014
Draft
Chapter 11
THEORETICAL STRENGTH OF
PERFECT CRYSTALS
This chapter (taken from the authors lecture notes in Fracture Mechanics) is of primary
interest to students in Material Science.
11.1
Introduction
In Eq. ?? we showed that around a circular hole in an innite plate under uniform traction, we do
have a stress concentration factor of 3.
Following a similar approach (though with curvilinear coordinates), it can be shown that if we have
an elliptical hole, Fig. ??, we would have
( )=0, = 0 1 + 2
=0
a
b
(11.1)
We observe that for a = b, we recover the stress concentration factor of 3 of a circular hole, and that for
a degenerated ellipse, i.e a crack there is an innite stress. Alternatively, the stress can be expressed in
x2
= o
2b
2a
Draft
112
Strength (P/A)
Theoretical Strength
Diameter
(11.2)
From this equation, we note that the stress concentration factor is inversely proportional to the radius
of curvature of an opening.
This equation, derived by Inglis, shows that if a = b we recover the factor of 3, and the stress
concentration factor increase as the ratio a/b increases. In the limit, as b = 0 we would have a crack
resulting in an innite stress concentration factor, or a stress singularity.
Around 1920, Grith was exploring the theoretical strength of solids by performing a series of experiments on glass rods of various diameters.
He observed that the tensile strength ( t ) of glass decreased with an increase in diameter, and that
1
for a diameter 10,000 in., t = 500, 000 psi; furthermore, by extrapolation to zero diameter he
obtained a theoretical maximum strength of approximately 1,600,000 psi, and on the other hand for very
large diameters the asymptotic values was around 25,000 psi.
5
Area A1
Failure Load P1
t
Failure Strength (P/A) 1
<
<
>
A2
P2
t
2
<
<
>
A3
P3
t
3
<
>
>
A4
P4
t
4
(11.3)
Furthermore, as the diameter was further reduced, the failure strength asymptotically approached a
limit which will be shown later to be the theoretical strength of glass, Fig. 11.2.
Clearly, one would have expected the failure strength to be constant, yet it was not. So Grith was
confronted with two questions:
Victor Saouma
Draft
113
11.2
Theoretical Strength
We start, [?] by exploring the energy of interaction between two adjacent atoms at equilibrium separated by a distance a0 , Fig. 11.3. The total energy which must be supplied to separate atom C from C
is
(11.4)
U0 = 2
8
where is the surface energy1 , and the factor of 2 is due to the fact that upon separation, we have
two distinct surfaces.
11.2.1
We shall rst derive an expression for the ideal strength in terms of physical parameters, and in the
next section the strength will be expressed in terms of engineering ones.
Solution I: Force being the derivative of energy, we have F = dU , thus F = 0 at a = a0 , Fig. 11.4, and
da
is maximum at the inection point of the U0 a curve. Hence, the slope of the force displacement
curve is the stiness of the atomic spring and should be related to E. If we let x = a a0 , then
x
F
the strain would be equal to = a0 . Furthermore, if we dene the stress as = a2 , then the
0
curve will be as shown in Fig. 11.5.
From this diagram, it would appear that the sine curve would be an adequate approximation to
this relationship. Hence,
x
theor
(11.5)
= max sin 2
theor
and the maximum stress max would occur at x = . The energy required to separate two atoms
4
is thus given by the area under the sine curve, and from Eq. 11.4, we would have
2 = U0
=
=
1 From
x
dx
0
theor
2x
2
max [ cos (
)] |0
2
theor
max sin 2
(11.6)
(11.7)
watching raindrops and bubbles it is obvious that liquid water has surface tension. When the surface of a liquid
is extended (soap bubble, insect walking on liquid) work is done against this tension, and energy is stored in the new
surface. When insects walk on water it sinks until the surface energy just balances the decrease in its potential energy. For
solids, the chemical bonds are stronger than for liquids, hence the surface energy is stronger. The reason why we do not
notice it is that solids are too rigid to be distorted by it. Surface energy is expressed in J/m2 and the surface energies
of water, most solids, and diamonds are approximately .077, 1.0, and 5.14 respectively.
Victor Saouma
Draft
114
Energy
Interatomic
Distance
Repulsion
Attraction
Force
Interatomic
Distance
2
theor
) + cos(0)]
[ cos (
2 max
2
2
theor
max
=
=
(11.8)
(11.9)
Also for very small displacements (small x) sin x x, thus Eq. 11.5 reduces to
theor
max
Ex
2x
a0
(11.10)
elliminating x,
theor
max
E
a0 2
(11.11)
theor
max
E
a0
(11.12)
Solution II: For two layers of atoms a0 apart, the strain energy per unit area due to (for linear elastic
systems) is
U
= 1 ao
2
= E
U=
2 ao
2E
(11.13)
If is the surface energy of the solid per unit area, then the total surface energy of two new fracture
surfaces is 2.
For our theoretical strength, U = 2
Victor Saouma
theor
(max )2 a0
2E
theor
= 2 or max = 2
E
a0
Draft
115
theor
max =
(11.14)
(2 1011 )(1)
2 1010
N
3.16 1010 2
m
E
6
(11.15)
(11.16)
(11.17)
11.2.2
10 We note that the force to separate two atoms drops to zero when the distance between them is a0 + a
where a0 corresponds to the origin and a to . Thus, if we take a = or = 2a, combined with Eq.
2
2
11.11 would yield
E a
theor
(11.18)
max
a0
11
theor
max
(11.19)
12
E
a0
(11.20)
However, since as a rst order approximation a a0 then the surface energy will be
Victor Saouma
Ea0
10
(11.21)
Introduction to Continuum Mechanics
Draft
116
(11.22)
11.3
13 In his quest for an explanation of the size eect, Grith came across Ingliss paper, and his strike of
genius was to assume that strength is reduced due to the presence of internal aws. Grith postulated
that the theoretical strength can only be reached at the point of highest stress concentration, and
accordingly the far-eld applied stress will be much smaller.
14
theor
act
max = cr 1 + 2
(11.23)
act
is the stress at the tip of the ellipse which is caused by a (lower) far eld stress cr . Asssuming a0
a
and since 2 a0
1, for an ideal plate under tension with only one single elliptical aw the strength
theor
act
max = 2cr
(11.24)
a
=
ao
E
a0
(11.25)
Micro
Macro
From this very important equation, we observe that
1. The left hand side is based on a linear elastic solution of a macroscopic problem solved by Inglis.
2. The right hand side is based on the theoretical strength derived from the sinusoidal stress-strain
assumption of the interatomic forces, and nds its roots in micro-physics.
Finally, this equation would give (at fracture)
act
cr =
E
4a
(11.26)
=
=
E
4a
Ea0
10
act
cr
a
a0
=
=
E 2 ao
40 a
2, 500
act
cr
E2
100,000
100 10
(11.27)
Thus if we set a aw size of 2a = 5, 000a0 in Ea0 this is enough to lower the theoretical fracture
10
E
a
theor
act
max = 2cr
ao
106
theor
act
act
6
= 200cr
(11.28)
max = 2cr
a
= 10 m = 1
1010
= = 1010 m
ao
= 1A
Victor Saouma
Introduction to Continuum Mechanics
Draft
117
Therefore at failure
act
cr
theor
max
=
=
Victor Saouma
theor
max
200
E
10
E
2,000
act
cr
30,000
2,000
E
2, 000
(11.29)
= 15 ksi
Draft
Victor Saouma
118
Draft
Chapter 12
BEAM THEORY
This chapter is adapted from the Authors lecture notes in Structural Analysis.
12.1
Introduction
In the preceding chapters we have focused on the behavior of a continuum, and the 15 equations and
15 variables we introduced, were all derived for an innitesimal element.
In practice, few problems can be solved analytically, and even with computer it is quite dicult to
view every object as a three dimensional one. That is why we introduced the 2D simplication (plane
stress/strain), or 1D for axially symmetric problems. In the preceding chapter we saw a few of those
solutions.
Hence, to widen the scope of application of the fundamental theory developed previously, we could
either resort to numerical methods (such as the nite dierence, nite element, or boundary elements),
or we could further simplify the problem.
Solid bodies, in general, have certain peculiar geometric features amenable to a reduction from three
to fewer dimensions. If one dimension of the structural element1 under consideration is much greater
or smaller than the other three, than we have a beam, or a plate respectively. If the plate is curved,
then we have a shell.
4
For those structural elements, it is customary to consider as internal variables the resultant of the
stresses as was shown in Sect. ??.
Hence, this chapter will focus on a brief introduction to beam theory. This will however be preceded
by an introduction to Statics as the internal forces would also have to be in equilibrium with the external
ones.
Beam theory is perhaps the most successful theory in all of structural mechanics, and it forms the
basis of structural analysis which is so dear to Civil and Mechanical engineers.
12.2
Statics
12.2.1
Equilibrium
far we have restricted ourselves to a continuum, in this chapter we will consider a structural element.
Draft
122
10
BEAM THEORY
11
=
=
Fy
My
=
=
Fz
Mz
=
=
0
0
(12.1)
Fy
Mz
(12.2)
12 All the externally applied forces on a structure must be in equilibrium. Reactions are accordingly
determined.
13
For reaction calculations, the externally applied load may be reduced to an equivalent force3 .
14
Summation of the moments can be taken with respect to any arbitrary point.
15 Whereas forces are represented by a vector, moments are also vectorial quantities and are represented
by a curved arrow or a double arrow vector.
16
Equations
Fx
Fy
Fy
Mz
Mz
Fz
Fx
Fy
Fz
Alternate Set
A
B
Mz
Mz
A
B
Fx
Mz
Mz
A
B
C
Mz
Mz
Mz
Mx
Mx
My
My
Mz
19
If your reaction is negative, then it will be in a direction opposite from the one assumed.
20 Summation of external forces is equal and opposite to the internal ones (more about this below).
Thus the net force/moment is equal to zero.
21
The external forces give rise to the (non-zero) shear and moment diagram.
2 In
3 However
Victor Saouma
Draft
12.2 Statics
12.2.2
22
123
Reactions
23 Once the reactions are determined, internal forces (shear and moment) are determined next; nally,
internal stresses and/or deformations (deections and rotations) are determined last.
24
Depending on the type of structures, there can be dierent types of support conditions, Fig. 12.1.
12.2.3
Equations of Conditions
25 If a structure has an internal hinge (which may connect two or more substructures), then this will
provide an additional equation (M = 0 at the hinge) which can be exploited to determine the reactions.
26 Those equations are often exploited in trusses (where each connection is a hinge) to determine reactions.
27 In an inclined roller support with Sx and Sy horizontal and vertical projection, then the reaction
R would have, Fig. 12.2.
Sy
Rx
=
Ry
Sx
Victor Saouma
(12.3)
Draft
124
BEAM THEORY
12.2.4
Static Determinacy
28 In statically determinate structures, reactions depend only on the geometry, boundary conditions and
loads.
29 If the reactions can not be determined simply from the equations of static equilibrium (and equations
of conditions if present), then the reactions of the structure are said to be statically indeterminate.
The degree of static indeterminacy is equal to the dierence between the number of reactions
and the number of equations of equilibrium (plus the number of equations of conditions if applicable),
Fig. 12.3.
30
12.2.5
Geometric Instability
33 The stability of a structure is determined not only by the number of reactions but also by their
arrangement.
34
Victor Saouma
Draft
12.2 Statics
125
12.2.6
Examples
Solution:
The beam has 3 reactions, we have 3 equations of static equilibrium, hence it is statically determinate.
(+ - ) Fx = 0; Rax 36 k = 0
)
(+ 6 Fy = 0; Ray + Rdy 60 k (4) k/ft(12) ft = 0
c
(+ ) Mz = 0; 12Ray 6Rdy (60)(6) = 0
1 0
0 1
0 12
(on your
calculator)
Rax 36
Rax 36 k
Ray
Ray
56 k
108
=
=
Rdy
Rdy
52 k
360
a
(+ ) Mz = 0;
M d = 0;
(+ )
z
Check:
Victor Saouma
Rdy = 52 k 6
(Ray )(18) (60)(12) (48)(6) = 0 Ray = 56 k 6
)
(+ 6 Fy = 0; ; 56 52 60 48 = 0
Draft
126
BEAM THEORY
12.3
12.3.1
36
Before we derive the Shear-Moment relations, let us arbitrarily dene a sign convention.
4
37 The sign convention adopted here, is the one commonly used for design purposes . With reference to
Fig. 12.5
12.3.2
38 Let us derive the basic relations between load, shear and moment. Considering an innitesimal length
dx of a beam subjected to a positive load5 w(x), Fig. 12.6. The innitesimal section must also be in
equilibrium.
39 There are no axial forces, thus we only have two equations of equilibrium to satisfy Fy = 0 and
Mz = 0.
40 Since dx is innitesimally small, the small variation in load along it can be neglected, therefore we
assume w(x) to be constant along dx.
To denote that a small change in shear and moment occurs over the length dx of the element, we add
the dierential quantities dVx and dMx to Vx and Mx on the right face.
41
4 Note
5 In
that this sign convention is the opposite of the one commonly used in Europe!
this derivation, as in all other ones we should assume all quantities to be positive.
Victor Saouma
Draft
127
42
or
dV
= w(x)
dx
(12.4)
The slope of the shear curve at any point along the axis of a member is given by
the load curve at that point.
43
Similarly
dx
(Mx + dMx ) = 0
(+ ) Mo = 0 Mx + Vx dx wx dx
(12.5)
The slope of the moment curve at any point along the axis of a member is given
by the shear at that point.
44
=
=
w(x)dx
Vx2 Vx1 =
(12.6)
x2
w(x)dx
(12.7)
x1
The change in shear between 1 and 2, V21 , is equal to the area under the load
between x1 and x2 .
and
M
V (x)dx
M21
M2 M1 =
(12.8)
x2
V (x)dx (12.9)
x1
The change in moment between 1 and 2, M21 , is equal to the area under the
shear curve between x1 and x2 .
Victor Saouma
Draft
128
BEAM THEORY
45
Note that we still need to have V1 and M1 in order to obtain V2 and M2 respectively.
46
It can be shown that the equilibrium of forces and of moments equations are nothing else than
12.3.3
d
dt
Tij
xj
(rt)dS +
S
Examples
Solution:
The free body diagram is drawn below
Victor Saouma
Draft
(+ ) MA = 0;
)
(+ 6 Fy = 0;
129
Victor Saouma
Draft
1210
BEAM THEORY
12.4
Beam Theory
12.4.1
Fig.12.7 shows portion of an originally straight beam which has been bent to the radius by end
couples M . support conditions, Fig. 12.1. It is assumed that plane cross-sections normal to the
47
O
+ve Curvature, +ve bending
d
-ve Curvature, -ve Bending
Neutral Axis
F
Y
dA
dx
and
d =
49
dx
(12.11)
dx
(12.12)
dx y dx dx
E F EF
=
EF
dx
(12.13)
or after simplication
x =
(12.14)
where y is measured from the axis of rotation (neutral axis). Thus strains are proportional to the
distance from the neutral axis.
Victor Saouma
Draft
1211
50 (Greek letter rho) is the radius of curvature. In some textbook, the curvature (Greek letter
kappa) is also used where
1
(12.15)
=
thus,
x = y
(12.16)
51 It should be noted that Galileo (1564-1642) was the rst one to have made a contribution to beam
theory, yet he failed to make the right assumption for the planar cross section. This crucial assumption
was made later on by Jacob Bernoulli (1654-1705), who did not make it quite right. Later Leonhard
Euler (1707-1783) made signicant contributions to the theory of beam deection, and nally it was
Navier (1785-1836) who claried the issue of the kinematic hypothesis.
12.4.2
Stress-Strain Relations
52 So far we considered the kinematic of the beam, yet later on we will need to consider equilibrium in
terms of the stresses. Hence we need to relate strain to stress.
53
x = Ex
where E is Youngs Modulus.
54
12.4.3
(12.18)
55 Just as external forces acting on a structure must be in equilibrium, the internal forces must also
satisfy the equilibrium equations.
The internal forces are determined by slicing the beam. The internal forces on the cut section must
be in equilibrium with the external forces.
56
12.4.3.1
57
Fx = 0; Neutral Axis
58 Since there are no external axial forces (unlike a column or a beam-column), the internal axial forces
must be in equilibrium.
Fx = 0
x dA = 0
(12.19)
Victor Saouma
EydA = 0
(12.20-a)
Draft
1212
BEAM THEORY
But since the curvature and the modulus of elasticity E are constants, we conclude that
(12.21)
ydA = 0
A
or the rst moment of the cross section with respect to the z axis is zero. Hence we conclude that the
neutral axis passes through the centroid of the cross section.
12.4.3.2
M = 0; Moment of Inertia
59 The second equation of internal equilibrium which must be satised is the summation of moments.
However contrarily to the summation of axial forces, we now have an external moment to account for,
the one from the moment diagram at that particular location where the beam was sliced, hence
Mz = 0; +ve; M =
x ydA
(12.22)
Ext.
Int.
x
61
=
=
Ey
x ydA
A
y 2 dA
M = E
(12.23)
We now pause and dene the section moment of inertia with respect to the z axis as
def
y 2 dA
I =
(12.24)
I
c
(12.25)
S =
12.4.4
Beam Formula
62 We now have the ingredients in place to derive one of the most important equations in structures, the
beam formula. This formula will be extensively used for design of structural components.
63
y 2 dA
E
A
y 2 dA
M
EI
==
(12.26)
which shows that the curvature of the longitudinal axis of a beam is proportional to the bending moment
M and inversely proportional to EI which we call exural rigidity.
64
Victor Saouma
=
=
Ey
M
EI
x = M y
I
(12.27)
Draft
1213
Hence, for a positive y (above neutral axis), and a positive moment, we will have compressive stresses
above the neutral axis.
65 Alternatively, the maximum ber stresses can be obtained by combining the preceding equation with
Equation 12.25
x =
M
S
12.4.5
12.4.6
(12.28)
Example
1 k/ft
r
0.25
20
Solution:
1. Steel has E = 29, 000 ksi, and from above Mmax =
wL2
8 ,
max =
wL4
185EI ,
and I = r3 t.
Mmax =
(12.29)
=
=
wL4
185EI
3
=
=
=
M
S
I
r
r3 t
=
=
=
r t
S
I
wL4
185Er 3 t
4
3
3
(1) k/ft(20)4 ft (12)3 in / ft
3 (0.25) in
(185)(29,000) ksi(3.14)r
65.65
r3
=
=
M
r 2 t
(50) k.ft(12) in/ft
(3.14)r 2 (0.25) in
764
r2
(12.30)
(12.31)
Victor Saouma
65.65
= 4.61 in (12.32-a)
0.67
(12.32-b)
Draft
BEAM THEORY
Victor Saouma
1214
Draft
Chapter 13
VARIATIONAL METHODS
Abridged section from authors lecture notes in finite elements.
Variational methods provide a powerful method to solve complex problems in continuum mechanics
(and other elds as well).
As shown in Appendix C, there is a duality between the strong form, in which a dierential equation
(or Eulers equation) is exactly satised at every point (such as in Finite Dierences), and the weak
form where the equation is satised in an averaged sense (as in nite elements).
Since only few problems in continuum mechanics can be solved analytically, we often have to use
numerical techniques, Finite Elements being one of the most powerful and exible one.
At the core of the nite element formulation are the variational formulations (or energy based methods)
which will be discussed in this chapter.
For illustrative examples, we shall use beams, but the methods is obviously applicable to 3D continuum.
13.1
6
Preliminary Denitions
def
F.ds
(13.1-a)
Fx dx + Fy dy
(13.1-b)
dW
The change in energy is proportional to the amount of work performed. Since only the change of
energy is involved, any datum can be used as a basis for measure of energy. Hence energy is neither
created nor consumed.
+ U ) = We + H
(13.2)
Draft
132
VARIATIONAL METHODS
U0
A
U0
A
U0
A
U0
A
Nonlinear
Linear
We = U
13.1.1
11
(13.3)
The strain energy density of an arbitrary material is dened as, Fig. 13.1
def
U0 =
:d
(13.4)
12
U0 =
def
13
:d
(13.5)
def
U0 d
(13.6)
U0 d
(13.7)
def
14 To obtain a general form of the internal strain energy, we rst dene a stress-strain relationship
accounting for both initial strains and stresses
= D:( 0 ) + 0
(13.8)
where D is the constitutive matrix (Hookes Law); is the strain vector due to the displacements u;
is the initial strain vector; 0 is the initial stress vector; and is the stress vector.
15 The initial strains and stresses are the result of conditions such as heating or cooling of a system or
the presence of pore pressures in a system.
Victor Saouma
Draft
16
133
(13.9)
U=
1
2
T :D:d
T :D:0 d +
T : 0 d
(13.10)
17
U=
18
1
2
(13.11)
E d
When this relation is applied to various one dimensional structural elements it leads to
Axial Members:
d
2
U=
=P
A
P
= AE
d = Adx
1
2
U=
L
0
P2
dx
AE
(13.12)
Flexural Members:
Mz y
x = Iz
Mz y
U=
= EIz
d = dAdx
2
y dA = Iz
U=
1
2
1
2
L
0
M2
dx
EIz
(13.13)
13.1.2
19
External Work
uT bd +
We =
uT
td
(13.14)
where b is the body force vector; is the applied surface traction vector; and t is that portion of the
t
boundary where is applied, and u is the displacement.
t
20
We =
0
Victor Saouma
P d +
M d
(13.15)
Draft
134
21
VARIATIONAL METHODS
P = K
f
1
f
d = K2
We = K
f
2
P d
We =
0
(13.16)
1
Pf f
2
(13.17)
We =
1
Mf f
2
(13.18)
13.1.3
Virtual Work
23 We dene the virtual work done by the load on a body during a small, admissible (continuous and
satisfying the boundary conditions) change in displacements.
def
:d
(13.19)
def
tud +
bud
(13.20)
where all the terms have been previously dened and b is the body force vector.
Note that the virtual quantity (displacement or force) is one that we will approximate/guess as long
as it meets some admissibility requirements.
24
13.1.3.1
Next we shall derive a displacement based expression of U for each type of one dimensional structural
member. It should be noted that the Virtual Force method would yield analogous ones but based on
forces rather than displacements.
25
26 Two sets of solutions will be given, the rst one is independent of the material stress strain relations,
and the other assumes a linear elastic stress strain relation.
Elastic Systems In this set of formulation, we derive expressions of the virtual strain energies which
are independent of the material constitutive laws. Thus U will be left in terms of forces and
displacements.
Axial Members:
U =
0
d = Adx
Victor Saouma
U = A
dx
(13.21)
Draft
135
Flexural Members:
U =
x x d
M
=
y
A
= y y =
x ydA
M=
d =
dAdx
0
x dA
U =
A
M dx
(13.22)
Axial Members:
U =
x = Ex = E du
dx
= d(u)
dx
d = Adx
U =
E
0
du d(u)
Adx
dx dx
(13.23)
Flexural Members:
My
x = Iz
d v
x = 2 Ey
2
d v
U =
dx
M = dx2 EIz
(v)
x = x = d dx2 y
d = dAdx
U =
x x d
or:
Eq. 13.24
13.1.3.2
27
d2 (v)
d2 v
Ey
ydAdx
2
dx2
A dx
U =
y dA = Iz
EIz
0
(13.24)
d2 v d2 (v)
dx
dx2 dx2
(13.25)
W =
qdx +
(i )Pi +
(13.26)
(i )Mi
13.1.4
28 We dene the complementary virtual work done by the load on a body during a small, admissible
(continuous and satisfying the boundary conditions) change in displacements.
def
def
Victor Saouma
:d
(13.27)
utd
(13.28)
Draft
136
13.1.5
29
VARIATIONAL METHODS
Potential Energy
We =
uT bd +
uT + uP
td
(13.29)
where u are the displacements, b is the body force vector; is the applied surface traction vector; t is
t
that portion of the boundary where is applied, and P are the applied nodal forces.
t
30
Note that the potential of the external work (W) is dierent from the external work itself (W )
31
def
U We
(13.30)
U0 d
u + uP
td
ubd +
(13.31)
32 Note that in the potential the full load is always acting, and through the displacements of its points
of application it does work but loses an equivalent amount of potential, this explains the negative sign.
13.2
The principles of Virtual Work and Complementary Virtual Work relate force systems which satisfy
the requirements of equilibrium, and deformation systems which satisfy the requirement of compatibility:
33
1. In any application the force system could either be the actual set of external loads dp or some
virtual force system which happens to satisfy the condition of equilibrium p. This set of external
forces will induce internal actual forces d or internal hypothetical forces compatible with the
externally applied load.
2. Similarly the deformation could consist of either the actual joint deections du and compatible
internal deformations d of the structure, or some hypothetical external and internal deformation
u and which satisfy the conditions of compatibility.
34
Thus we may have 2 possible combinations, Table 13.1: where: d corresponds to the actual, and
1
2
Force
External Internal
p
dp
d
Deformation
External Internal
du
d
u
Formulation
U
U
13.2.1
35 Derivation of the principle of virtual work starts with the assumption of that forces are in equilibrium
and satisfaction of the static boundary conditions.
Victor Saouma
Draft
36
137
(13.32)
(13.33)
where b representing the body force. In matrix form, this can be rewritten as
0 y xx
bx
x
=0
+
by
0 y x yy
xy
(13.34)
or
LT + b = 0
(13.35)
= t + u
t = on t Natural B.C.
t
u = u on u Essential B.C.
(13.36-a)
(13.36-b)
(13.36-c)
The principle of virtual work (or more specically of virtual displacement) can be stated as
A deformable system is in equilibrium if the sum of the external virtual work and the internal
virtual work is zero for virtual displacements u which are kinematically admissible.
uT bd
Wi
uT
td
(13.37)
= L:u
u = 0
in
on
(13.38)
(13.39)
t
We
39 Note that the principle is independent of material properties, and that the primary unknowns are the
displacements.
Victor Saouma
Draft
138
VARIATIONAL METHODS
Figure 13.2: Tapered Cantilivered Beam Analysed by the Vitual Displacement Method
The approximate solutions proposed to this problem are
v
v
x
v2
2l
x
x 2
2
3
L
L
1 cos
=
=
(13.40)
3
v2
(13.41)
2. These equations do indeed satisfy the essential B.C. (i.e kinematic), but for them to also satisfy
equilibrium they must satisfy the principle of virtual work.
3. Using the virtual displacement method we evaluate the displacements v2 from three dierent
combination of virtual and actual displacement:
Solution
1
2
3
Total
Eqn. 13.40
Eqn. 13.40
Eqn. 13.41
Virtual
Eqn. 13.41
Eqn. 13.40
Eqn. 13.41
Where actual and virtual values for the two assumed displacement elds are given below.
Trigonometric (Eqn. 13.40)
1 cos
v2
1 cos x v2
2l
x
2l
6
12x
L2 L3
6
12x
L2 L3
x
4L2 cos 2l v2
2
x
4L2 cos 2l v2
v
v
x 2
x 3
2 L
L
x 2
x 3
2 L
L
v2
v2
v2
v2
v EIz v dx
(13.42)
= P2 v2
(13.43)
Solution 1:
L
=
=
=
Victor Saouma
12x
2
6
x
v2
cos
3
4L2
2l
L2
L
0
3EI1
10 16
+ 2 v2 v2
1
2L3
P2 v2
v2 EI1 1
x
dx
2L
(13.44)
Draft
139
which yields:
v2 =
P2 L3
2.648EI1
(13.45)
Solution 2:
L
=
=
=
x
4
x
v2 v2 EI1 1
dx
cos2
4
2l
2l
0 16L
4 EI1 3
1
+ 2 v2 v2
3
32L
4
P2 v2
(13.46)
which yields:
v2 =
P2 L3
2.57EI1
(13.47)
Solution 3:
L
=
0
=
=
12x
6
3
L2
L
x
EI1 v2 v2 dx
2l
9EI
v2 v2
L3
P2 v2
(13.48)
which yields:
v2 =
13.2.2
P2 L3
9EI
(13.49)
40 Derivation of the principle of complementary virtual work starts from the assumption of a kinematicaly
admissible displacements and satisfaction of the essential boundary conditions.
41 Whereas we have previously used the vector notation for the principle of virtual work, we will now
use the tensor notation for this derivation.
42
43
1
(ui,j + uj,i )
2
ui = u on u
44
(13.50)
(13.51)
The principle of virtual complementary work (or more specically of virtual force) which can be stated
as
A deformable system satises all kinematical requirements if the sum of the external complementary virtual work and the internal complementary virtual work is zero for all statically
admissible virtual stresses ij .
The major governing equations are summarized
Victor Saouma
Draft
1310
VARIATIONAL METHODS
Figure 13.3: Tapered Cantilevered Beam Analysed by the Virtual Force Method
ij ij d
ui ti d
(13.52)
in
on
(13.53)
(13.54)
Wi
We
ij,j = 0
ti = 0
45 Note that the principle is independent of material properties, and that the primary unknowns are the
stresses.
46
Expressions for the complimentary virtual work in beams are given in Table 13.3
Example 13-2: Tapered Cantilivered Beam; Virtual Force
Exact solution of previous problem using principle of virtual work with virtual force.
L
M
0
M
dx =
EIz
Internal
(13.55)
External
Note: This represents the internal virtual strain energy and external virtual work written in terms
of forces and should be compared with the similar expression derived in Eq. 13.25 written in terms of
displacements:
L
d2 v d2 (v)
EIz 2
dx
(13.56)
U =
dx dx2
0
M
EIz
L
(1)
x
0
=
=
P2
EI1
P2 2L
EI1
x2
x dx
L)
L+x
2l
L
2
0
dx
x
dx
L+x
(13.57)
1 1
x2
= 3
(a + bx)2 2a(a + bx) + a2 ln(a + bx)
a + bx
b 2
0
Victor Saouma
(13.58)
Draft
1311
=
=
=
=
2P2 L 1
(L + x)2 2L(L + x) + L2 ln(L + x) |L
0
EI1 2
2P2 L
L2
+ 2L2 + L2 log L
2L2 4L2 + L2 ln 2L
EI1
2
1
2P2 L 2
L (ln 2 )
EI1
2
3
P2 L
2.5887EI1
(13.59)
Similarly:
L
=
=
2M L
M (1)
2M L L 1
=
ln(L + x) |L
=
0
x
EI1 0 L + x
EI1
0 EI1 .5 + L
2M L
2M L
ML
(ln 2L ln L) =
ln 2 =
EI1
EI1
.721EI1
13.3
Potential Energy
13.3.1
(13.60)
Derivation
47
dij
dU0 =
ij
dU0 =
48
(13.61-a)
(13.61-b)
U0
ij dij
(13.62-a)
dU0
ij dij
(13.62-b)
ij
(13.63)
ij
(13.64)
thus,
U0
ij
U0
ij
49
50
U
ij = ij ij
ij
(13.65)
Applying the principle of virtual work, Eq. 13.37, it can be shown that
1 Note
U =
that the variation of strain energy density is, U0 = ij ij , and the variation of the strain energy itself is
U0 d.
Victor Saouma
Draft
1312
VARIATIONAL METHODS
k= 500 lbf/in
(13.66)
def
U We
(13.67)
U0 d
51
u + uP
td
ubd +
(13.68)
We have thus derived the principle of stationary value of the potential energy:
Of all kinematically admissible deformations (displacements satisfying the essential boundary conditions), the actual deformations (those which correspond to stresses which satisfy
equilibrium) are the ones for which the total potential energy assumes a stationary value.
52
For problems involving multiple degrees of freedom, it results from calculus that
53
1 +
2 + . . . +
n
1
2
n
(13.69)
It can be shown that the minimum potential energy yields a lower bound prediction of displacements.
54 As an illustrative example (adapted from Willam, 1987), let us consider the single dof system shown
in Fig. 13.4. The strain energy U and potential of the external work W are given by
We
1
u(Ku) = 250u2
2
mgu = 100u
(13.70-a)
(13.70-b)
(13.71)
Victor Saouma
d
= 0 500u 100 = 0 u = 0.2 in
du
(13.72)
Draft
1313
Energy [lbfin]
20.0
0.0
20.0
40.0
0.00
0.10
0.20
Displacement [in]
0.30
= 250(0.2)2
= 100(0.2)
= 10 20
= 10 lbf-in
= 20 lbf-in
= 10 lbf-in
(13.73)
13.3.2
Rayleigh-Ritz Method
55 Continuous systems have innite number of degrees of freedom, those are the displacements at every
point within the structure. Their behavior can be described by the Euler Equation, or the partial
dierential equation of equilibrium. However, only the simplest problems have an exact solution which
(satises equilibrium, and the boundary conditions).
56 An approximate method of solution is the Rayleigh-Ritz method which is based on the principle of
virtual displacements. In this method we approximate the displacement eld by a function
u1
c1 1 + 1
i i
0
(13.74-a)
c2 2 + 2
i i
0
(13.74-b)
c3 3 + 3
i i
0
(13.74-c)
i=1
n
u2
i=1
n
u3
i=1
Victor Saouma
Draft
1314
VARIATIONAL METHODS
2. Must be admissible, i.e. satisfy the essential boundary conditions (the natural boundary conditions
are included already in the variational statement. However, if also satisfy them, then better
results are achieved).
3. Must be independent and complete (which means that the exact displacement and their derivatives
that appear in can be arbitrary matched if enough terms are used. Furthermore, lowest order
terms must also be included).
In general is a polynomial or trigonometric function.
58
We determine the parameters cj by requiring that the principle of virtual work for arbitrary variations
i
or
n
1 2 3
(13.75)
c + 2 ci + 3 ci = 0
(u1 , u2 , u3 ) =
c1 i
ci
ci
i
i=1
cj .
i
cj
i
=0
i = 1, 2, , n; j = 1, 2, 3
(13.76)
Thus we obtain a total of 3n linearly independent simultaneous equations. From these displacements,
we can then determine strains and stresses (or internal forces). Hence we have replaced a problem with
an innite number of d.o.f by one with a nite number.
59
(13.77)
for this particular solution, let us retain only the rst term:
v = a1 x(L x)
(13.78)
We observe that:
Victor Saouma
Draft
1315
Figure 13.6: Uniformly Loaded Simply Supported Beam Analyzed by the Rayleigh-Ritz Method
1. Contrarily to the previous example problem the geometric B.C. are immediately satised at both
x = 0 and x = L.
2. We can keep v in terms of a1 and take
a1
=U W =
Recalling that:
M
EIz
d2 v
dx2 ,
L
2)
and take
M2
dx
2EIz
vmax
= 0.
wv(x)dx
(13.79)
=
0
EIz
2
d2 v
dx2
wv(x) dx
(13.80)
EIz
(2a1 )2 a1 wx(L x) dx
2
EIz 2
L3
L3
4a1 L a1 w
+ a1 w
2
2
3
a1 wL3
2a2 EIz L
1
6
0
=
=
If we now take
a1
(13.81)
= 0, we would obtain:
wL3
6
a1
4a1 EIz l
wL2
24EIz
wL4
24EIz
(13.82)
L
2:
x2
x
2
L L
a1
Victor Saouma
Draft
1316
VARIATIONAL METHODS
wL4
96EIz
(13.83)
4
5
wL
exact
This is to be compared with the exact value of vmax = 384 wLz = 76.8EIz which constitutes 17%
EI
error.
wL2
w
Note: If two terms were retained, then we would have obtained: a1 = 24EIz & a2 = 24EIz and vmax
exact
would be equal to vmax . (Why?)
13.4
60
Summary
Virtual Work U
Complimentary Virtual Work U
Starts with
KAD
SAS
Ends with
SAS
KAD
In terms of virtual
Displacement/strains
Forces/Stresses
Solve for
Displacement
Displacement
61
A summary of the various methods introduced in this chapter is shown in Fig. 13.7.
U
Axial
Flexure
L
0
1
2
P2
dx
AE
1
2
M
dx
EIz
Virtual Displacement U
General
Linear
L
L
du d(u)
dx
E
Adx
dx dx
0
0
L
M dx
0
d v d (v)
EIz 2
dx
dx dx2
W
1
i 2 Pi i
1
i 2 Mi i
P
M
M dx
0
Virtual Displacement W
i Pi i
i Mi i
w(x)v(x)dx
0
Virtual Force U
General
Linear
L
L
P
dx
P
dx
AE
0
0
Virtual Force W
i Pi i
i Mi i
w(x)v(x)dx
0
M
M
dx
EIz
w(x)v(x)dx
0
Table 13.3: Summary of Variational Terms Associated with One Dimensional Elements
Victor Saouma
Draft
13.4 Summary
1317
+ b = 0
Natural B.C.
Essential B.C.
D:u = 0
u = 0 u
t t = 0 t
def
U0 =
ij
1
2
(ui,j + uj,i ) = 0
u i u = 0 u
def
U0 =
:d
ij,j = 0
ti = 0 t
6
Gauss
Gauss
?
?
Principle of Virtual Work
T :d
uT bd t uT td = 0
Wi We = 0
?
?
Principle of Complementary
Virtual Work
ij ij d u ui ti d = 0
Wi We = 0
?
Principle of Stationary
Potential Energy
= 0
def
= U We
U0 d ( ui bi d +
ui ti d)
?
Rayleigh-Ritz
n
cj j + j
0
i i
uj
i=1
cj
i
=0
i = 1, 2, , n;
j = 1, 2, 3
Victor Saouma
Draft
1318
VARIATIONAL METHODS
?
Principle of Stationary
Complementary Energy
Principle of Complementary
Principle of Stationary
Potential Energy
Virtual Work
6
?
Statically Admissible Stresses
Stresses satisfy the equilibrium conditions
and the static boundary conditions
Victor Saouma
Draft
Chapter 14
INELASTICITY (incomplete)
F
t
Creep
t
Relaxation
Draft
2
INELASTICITY (incomplete)
Strain Hardening
Relaxation
Creep
Creep
Creep
Perfectly Elastic
Relaxation
Viscoelastic
Relaxation
Elastoplastic Hardeing
Draft
E
0
E
E1
Ei
En
Linear Elasticity
Linear Visosity
.
=
Nonlinear Viscosity
. 1/N
Stress Threshold
Strain Threshold
s < < s
s< <
s
0
Figure 14.7: epp
Victor Saouma
Draft
4
INELASTICITY (incomplete)
pi
Si
Ei
Sj
Ej
Em
Victor Saouma
Draft
Appendix A
w L
R
R
L / 2
L / 2
Shear
M max.
R
Vx
at center Mmax
Mx
max
x
= V
= w L x
2
2
= wL
8
= wx (L x)
2
4
5
= 384 wL
EI
wx
= 24EI (L3 2Lx2 + x3 )
Moment
R1 = V1
Max R2 = V2
Vx
at x = .577L Mmax
Mx
at x = .5193L max
x
=
=
=
=
=
=
=
W
3
2W
3
W
W x2
3 L2
.1283W L
Wx
2
x2
3L2 (L 3 )
.01304 W L
EI
W x3
4
2 2
4
180EIL2 (3x 10L x + 7L )
Draft
A2
R=V
for x <
Vx
at center Mmax
=
=
=
L
2
L
2
Mx
x
max
=
=
L
2
for x <
for x <
W
2
W
2
2
2L2 (L 4x )
WL
6
x2
W x 1 2 L2
2
3
Wx
2
480EIL2 (5L
W L3
60EI
4x2 )2
R1 = V1
R2 = V2
Vx
Mx
Mx
Mx
Mmax
=
=
=
=
=
=
=
wb
2L (2c + b)
wb
2L (2a + b)
R1 w(x
a)
R1 x
R1 x w (x a)2
2
R2 (L x)
R1
R1 a + 2w
max
at x =
when x <
whenx <
at x =
L
2
L
2
L
2
L
2
R 1 = V1
R=V
Mmax
Mx
x
max
= wa (2L a)
2L
= 2P
= P4L
= P2x
Px
= 48EI (3L2 4x2 )
P L3
= 48EI
Victor Saouma
Draft
A3
a(a+2b)
3
R1 = V1
R2 = V2
Mmax
Mx
a
x
=
=
=
=
=
=
=
Pb
L
Pa
L
P ab
L
P bx
L
P a2 b2
3EIL
P bx
2
2
2
6EIL (L b x )
P ab(a+2b) 3a(a+2b)
27EIL
when x < a
when a < x < L a
R=V
Mmax
max
x
x
= P
= Pa
Pa
= 24EI (3L2 4a2 )
Px
= 6EI (3La 3a2 x2 )
Pa
= 6EI (3Lx 3x2 a2 )
R1 = V1
R 2 = V2
Vx
M1
M2
Mx
Mx
=
=
=
=
=
=
=
P
L (L a + b)
P
L (L b + a)
P
L (b a)
R1 a
R2 b
R1 x
R1 x P (x a)
Victor Saouma
Draft
A4
at x = 3 L
8
at x = .4215L
R1 = V1
R2 = V2
Vx
Mmax
M1
Mx
x
max
=
=
=
=
=
=
=
=
3
8 wL
5
8 wL
R1 wx
wL2
8
9
2
128 wL
2
R1 x wx
2
wx
3
48EI (L
wL4
185EI
3Lx+ 2x3 )
at x = L
when x < L
2
when L < x
2
at x = .4472L
R1 = V1
R2 = V2
Mmax
Mx
Mx
max
=
=
=
=
=
=
5P
16
11P
16
3P L
16
5P x
16
P L
2
11x
163
.009317 P L
EI
at x = a
at x = L
at x = a
2
L2
when a < .414L at x = L 3L2+a 2
a
when .414L < a at x = L
a
2L+a
R1 = V1
R 2 = V2
M1
M2
a
max
=
=
=
=
=
=
max
P b2
2L3 (a + 2L)
Pa
2
2
2L3 (3L a )
R1 a
P ab
(a
2L2 2 3 + L)
Pa b
(3L + a)
12EIL32
2 3
P a (L a )
3EI (3L2 a2 )2
P ab2
a
6EI
2L+a2
Victor Saouma
Draft
A5
at x = 0 and x = L
at x = L
2
at x = L
2
R=V
Vx
Mmax
M
max
x
= wL
2
= w L x
2
2
= wL
122
= wL
24
wL4
= 384EI
wx2
= 24EI (L x)2
at
when
at
when
x=
x<
x=
x<
L
2
L
2
L
2
L
2
R=V
Mmax
Mx
max
x
=
=
=
=
=
P
2
PL
8
P
8 (4x L)
P L3
192EI
P x2
48EI (3L
4x)
at x = L
at x = 0
R=V
Vx
Mmax
Mx
x
max
= 8W
3
x2
= W L2
= WL
3 2
= W x2
3L
W
= 60EIL2 (x5 5L2 x + 4L5 )
W L3
= 15EI
Victor Saouma
Draft
A6
at x = L
at x = 0
R=V
Vx
Mx
Mmax
x
max
= wL
= wx
2
= wx
22
= wL
2
w
= 24EI (x4 4L3 x + 3L4 )
wL4
= 8EI
at x = L
when a < x
at x = 0
at x = a
when x < a
when a < x
R=V
Mmax
Mx
max
a
x
x
= P
= Pb
= P (x a)
P b2
= 6EI (3L b)
P b3
= 3EI
P b2
= 6EI (3L 3x b)
2
= P (Lx) (3b L + x)
6EI
at x = L
at x = 0
R=V
Mmax
Mx
max
x
= P
= PL
= Px
3
= PL
3EI
P
= 6EI (2L3 3L2 x + x3 )
Victor Saouma
Draft
A7
at x = 0 and x = L
at x = 0
Victor Saouma
R=V
Mx
Mmax
max
x
= P
= P L x
2
= P2L
P L3
= 12EI
2
= P (Lx) ((L + 2x)
12EI
Draft
Appendix B
SECTION PROPERTIES
Section properties for selected sections are shown in Table B.1.
Draft
B2
SECTION PROPERTIES
Y
x
A
x
y
Ix
Iy
X
y
= bh
b
= 2
h
= 2
3
= bh
12
3
= hb
12
h h
A = bh b h
b
x = 2
h
y = 2
3
3
Ix = bh b h
12 3
3
Iy = hb h b
12
X
y
b
b
A
y
=
=
Ix
h(a+b)
2
h(2a+b)
3(a+b)
h3 (a2 +4ab+b2
36(a+b)
A
x
y
X
Ix
y
Iy
=
=
=
=
=
bh
2
b+c
3
h
3 3
bh
36
bh 2
36 (b
bc + c2 )
r
X
A =
Ix = Iy =
r2 =
r 4
4 =
d2
4
d4
64
A = 2rt = dt
3
Ix = Iy = r3 t = d t
8
b
X
b
a
A
Ix
Iy
= ab
3
= ab
33
= ba
4
Victor Saouma
Draft
Appendix C
MATHEMATICAL
PRELIMINARIES; Part IV
VARIATIONAL METHODS
Abridged section from authors lecture notes in finite elements.
C.1
1
Euler Equation
The fundamental problem of the calculus of variation1 is to nd a function u(x) such that
b
F (x, u, u )dx
(3.1)
is stationary. Or,
= 0
(3.2)
We dene the domain of a functional as the collection of admissible functions belonging to a class of
functions in function space rather than a region in coordinate space (as is the case for a function).
Letting u to be a family of neighbouring paths of the extremizing function u(x) and we assume that
at the end points x = a, b they coincide. We dene u as the sum of the extremizing path and some
(3.3)
= (x)
(3.4-a)
(3.4-b)
1 Dierential calculus involves a function of one or more variable, whereas variational calculus involves a function of a
function, or a functional.
Draft
C2
u, u
u(x)
C
B
u(x)
du
dx
x=a
x=c
x=b
with u if = 0
6 The variational operator and the dierential calculus operator d have clearly dierent meanings. du
is associated with a neighboring point at a distance dx, however u is a small arbitrary change in u for
a given x (there is no associated x).
For boundaries where u is specied, its variation must be zero, and it is arbitrary elsewhere. The
variation u of u is said to undergo a virtual change.
(u + ) = () =
F (x, u + , u + )dx
(3.5)
a
9
d()
d
10
=0
(3.6)
=0
From Eq. 3.3 and applying the chain rule with = 0, u = u, we obtain
d()
d
=
=0
F
F
+
u
u
dx = 0
(3.7)
11 It can be shown (through integration by part and the fundamental lemma of the calculus of variation)
that this would lead to
d F
F
=0
u
dx u
(3.8)
12 This dierential equation is called the Euler equation associated with and is a necessary condition
for u(x) to extremize .
13
Generalizing for a functional which depends on two eld variables, u = u(x, y) and v = v(x, y)
=
Victor Saouma
(3.9)
Draft
C3
2
F
xy u,xy
F
2
+ xy v,xy
+
+
2 F
y 2 u,yy
2 F
y 2 v,yy
(3.10)
14 We note that the Functional and the corresponding Euler Equations, Eq. 3.1 and 3.8, or Eq. 3.9 and
3.10 describe the same problem.
15 The Euler equations usually correspond to the governing dierential equation and are referred to as
the strong form (or classical form).
16 The functional is referred to as the weak form (or generalized solution). This classication stems
from the fact that equilibrium is enforced in an average sense over the body (and the eld variable is
dierentiated m times in the weak form, and 2m times in the strong form).
17 Euler equations are dierential equations which can not always be solved by exact methods.
An
alternative method consists in bypassing the Euler equations and go directly to the variational statement
of the problem to the solution of the Euler equations.
18 Finite Element formulation are based on the weak form, whereas the formulation of Finite Dierences
are based on the strong form.
19
=
=
F
F
u u + u
b
F dx
a
F
F
u +
u
u
u
=
a
dx
(3.11)
u
a
d F
F
u
dx u
(3.12)
dx
20 We have just shown that nding the stationary value of by setting = 0 is equivalent to nding
the extremal value of by setting d()
equal to zero.
d
=0
Similarly, it can be shown that as with second derivatives in calculus, the second variation 2 can
be used to characterize the extremum as either a minimum or maximum.
21
22
Revisiting the integration by parts of the second term in Eq. 3.7, we obtain
b
F
F
dx =
u
u
d F
dx
dx u
(3.13)
We note that
1. Derivation of the Euler equation required (a) = (b) = 0, thus this equation is a statement of the
essential (or forced) boundary conditions, where u(a) = u(b) = 0.
2. If we left arbitrary, then it would have been necessary to use
the natural boundary conditions.
F
u
23 For a problem with, one eld variable, in which the highest derivative in the governing dierential
equation is of order 2m (or simply m in the corresponding functional), then we have
Essential (or Forced, or geometric) boundary conditions, involve derivatives of order zero (the eld
variable itself) through m-1. Trial displacement functions are explicitely required to satisfy this
B.C. Mathematically, this corresponds to Dirichlet boundary-value problems.
Victor Saouma
Draft
C4
Nonessential (or Natural, or static) boundary conditions, involve derivatives of order m and up.
This B.C. is implied by the satisfaction of the variational statement but not explicitly stated in
the functional itself. Mathematically, this corresponds to Neuman boundary-value problems.
These boundary conditions were already introduced, albeit in a less formal way, in Table 9.1.
24
Table C.1 illustrates the boundary conditions associated with some problems
Problem
Axial Member
Distributed load
2
AE d u + q = 0
dx2
1
u
Dierential Equation
m
Essential B.C. [0, m 1]
Natural B.C. [m, 2m 1]
du
dx
or x = Eu,x
Flexural Member
Distributed load
4
EI d w q = 0
dx4
2
w, dw
dx
d2 w
d3 w
dx2 and dx3
or M = EIw,xx and V = EIw,xxx
=
0
EA
2
du
dx
dx P u(L)
(3.14)
=
0
EA
2
2
du
dx
du
dx
dx P u(L)
(3.15)
d
dx
u
0
EA
d
dx
du
udx + EA u P u(L)
dx
0
du
du
EA
P u(L)
dx + EA
dx
dx x=L
EA
du
dx
du
dx
u(0)
(3.16-a)
(3.16-b)
x=0
The last term is zero because of the specied essential boundary condition which implies that
u(0) = 0. Recalling that in an arbitrary operator which can be assigned any value, we set the
coecients of u between (0, L) and those for u at x = L equal to zero separately, and obtain
Euler Equation:
Victor Saouma
d
dx
EA
du
dx
=0
0<x<L
(3.17)
Draft
C5
du
P =0
dx
at x = L
Solution II We have
F (x, u, u ) =
EA
2
du
dx
(3.18)
(3.19)
(note that since P is an applied load at the end of the member, it does not appear as part of
F (x, u, u ) To evaluate the Euler Equation from Eq. 3.8, we evaluate
F
F
=0 &
= EAu
u
u
(3.20-a)
u
dx u
du
d
EA
dx
dx
0 Euler Equation
(3.21-a)
0 B.C.
(3.21-b)
=
0
1
M pw dx =
2
1
(EIw )w pw dx
2
(3.22)
0
L
F
F
w dx
w +
w
w
F dx =
=
=
(EIw w pw)dx
(3.23-a)
(3.23-b)
=
=
(EIw w )|0
[(EIw ) w pw] dx
(3.23-c)
[(EIw ) + p] wdx = 0
(3.23-d)
Or
(EIw ) = p
for all x
or
or
Natural
EIw = M = 0
(EIw ) = V = 0
at x = 0 and x = L
Victor Saouma
Draft
C6
Victor Saouma
Draft
Appendix D
1. (2 pts) Write in matrix form the following 3rd order tensor Dijk in R2 space. i, j, k range from 1
to 2.
2. (2 pts) Solve for Eij ai in indicial notation.
3. (4 pts) if the stress tensor at point P is given by
10 2 0
= 2 4 1
0
1 6
determine the traction (or stress vector) t on the plane passing through P and parallel to the plane
ABC where A(6, 0, 0), B(0, 4, 0) and C(0, 0, 2).
4. (5 pts) For a plane stress problem charaterized by the following stress tensor
=
6
2
2
4
use Mohrs circle to determine the principal stresses, and show on an appropriate gure the orientation of those principal stresses.
5. (4 pts) The stress tensor throughout a continuum is given with respect to Cartesian axes as
0
3x1 x2 5x2
2
0
2x2
= 5x2
2
3
0
2x2
0
3
(a) Determine the stress vector (or traction) at the point P (2, 1,
to the cylindrical surface x2 + x2 = 4 at P ,
2
3
Draft
D2
3
1
x1
2
2x1 x2
0
x2 + (x2 x2 )
x
1
x2 + (x2 x2 )
0
2x1 x2
1
2
1
2
2
0
0
(x1 + x2 )
where is a constant, satisfy equilibrium in the X1 direction?
13. (2 pts) From which principle is the symmetry of the stress tensor derived?
14. (2 pts) How is the First principle obtained from the equation of motion?
15. (4 pts) What are the 1) 15 Equations; and 2) 15 Unknowns in a thermoelastic formulation.
16. (2 pts) What is free energy ?
17. (2 pts) What is the relationship between strain energy and strain?
18. (5 pts) If a plane of elastic symmetry exists in an anisotropic
T22
c2222 c2233 c2212 c2223
T33
c3333 c3312 c3323
=
c1212 c1223
T12
T23
SYM.
c2323
T31
Victor Saouma
material,
c1131
E11
c2231
E22
c3331
E33
c3131
2E31 (31 )
Draft
then,
D3
1
aj = 0
i
0
0
1
0
0
0
1
6 2 0
T = 2 3 0 MP a
0 0 0
with E = 207 GPa, = 80 GPa, and = 0.3.
(a) Determine the engineering strain components
(b) If a ve centimer cube of structural steel is subjected to this stress tensor, what would be the
change in volume?
Victor Saouma
Draft
Victor Saouma
D4
Draft
Appendix E
MATHEMATICA ASSIGNMENT
and SOLUTION
Connect to Mathematica using the following procedure:
1. login on an HP workstation
2. Open a shell (window)
3. Type xhost+
4. type rlogin mxsg1
5. On the newly opened shell, enter your password rst, and then type setenv DISPLAY xxx:0.0
where xxx is the workstation name which should appear on a small label on the workstation itself.
6. Type mathematica &
and then solve the following problems:
1. The state of stress through a continuum is given with respect to the cartesian axes Ox1 x2 x3 by
3x1 x2 5x2
0
2
0
2x3 MPa
Tij = 5x2
2
0
2x3
0
Determine the stress vector at point P (1, 1, 3) of the plane that is tangent to the cylindrical
surface x2 + x2 = 4 at P .
2
3
2. For the following stress tensor
6 3
Tij = 3 6
0
0
0
0
8
(a) Determine directly the three invariants I , II and III of the following stress tensor
(b) Determine the principal stresses and the principal stress directions.
(c) Show that the transformation tensor of direction cosines transforms the original stress tensor
into the diagonal principal axes stress tensor.
(d) Recompute the three invariants from the principal stresses.
(e) Split the stress tensor into its spherical and deviator parts.
(f) Show that the rst invariant of the deviator is zero.
3. The Lagrangian description of a deformation is given by x1 = X1 +X3 (e2 1), x2 = X2 +X3 (e2 e2 ,
and x3 = e2 X3 where e is a constant. SHow that the Jacobian J does not vanish and determine
the Eulerian equations describing this motion.
2
2
2
4. A displacement eld is given by u = X1 X3 e1 + X1 X2 e2 + X2 X3 e3 . Determine independently the
material deformation gradient F and the material displacement gradient J and verify that J = FI.
Draft
E2
Determine
Determine
Determine
Determine
Determine
the
the
the
the
the
100
Tij = 0
0
1
= 4X1 , x2 = 2 X2 , x3 = 1 X3 and the Cauchy
2
0 0
0 0 MPa
0 0
SYM.
c2323 c2331
c3131
reduces to
ciso
ijkm
c1111
c1122
c2222
SYM.
c1133
c2233
c3333
0 0
0 0
0 0
a 0
b
0
0
0
0
0
c
30 50 20
Eij = 50 40 0 106
20 0 30
and the material is steel with = 119.2 GPa and = 79.2 GPa.
Victor Saouma
Draft
E3
11. Determine the strain tensor at a point where the Cauchy stress tensor is given by
100 42 6
Tij = 42 2 0 MPa
6
0 15
with E = 207 GPa, = 79.2 GPa, and = 0.30
12. Determine the thermally induced stresses in a constrained body for a rise in temerature of 50oF ,
= 5.6 106 / 0F
13. Show that the inverse of
1
xx
0
0
0
xx
yy
0
0
0
1
yy
zz
zz
1
0
0
0
=
(5.1)
0
0
0
1+
0
0
xy (2xy ) E
xy
(2 )
yz
yz yz
0
0
0
0
1+
0
zx (2zx )
is
xx
yy
zz
xy
yz
zx
E
(1+)(12)
zx
1+
1
0
0
0
1
0
0
0
1
xx
yy
zz
xy (2xy )
yz (2yz )
zx (2zx )
(5.2)
and then derive the relations between stresses in terms of strains, and strains in terms of stress, for
plane stress and plane strain.
14. Show that the function = f (r) cos 2 satises the biharmonic equation ( ) = 0 Note: You
must <<CalculusVectorAnalysis, dene , and SetCoordinates[Cylindrical[r,,z]], and
nally use the Laplacian (or Biharmonic) functions.
15. Solve for
Trr
Tr
Tr
T
cos
sin
sin
cos
0
0
0
0
cos
sin
sin
cos
(5.3)
p
1
r
p
show that for = r sin we have the following stress tensors:
2p cos
r
0
0
0
2p cos
r
2p sinrcos
2p sinrcos
2
2p sinr cos
(5.4)
Determine the maximum principal stress at an y arbitrary point, (contour) plot the magnitude of
this stress below p. Note that D[,r], D[,{,2}] would give the rst and second derivatives of
with respect to r and respectively.
Victor Saouma
Draft
Victor Saouma
E4