THEORETICAL

GEOMECHANICS
Marian IVAN
University of Bucharest
Department of Geophysics
6 Traian VUIA str.
70138 BUCHAREST o.p.37
ROMANIA
e-mail : ivam@gg.unibuc.ro
ivam@math.math.unibuc.ro

Samizdat Press
 2

In memoriam

Professor Marin DOROBANÞU

INTRODUCTION

These course notes present some theoretical problems related to the Mechanics of a Continuum Solid Body, of particular
importance to Applied Geomechanics, Geological Engineering and Structural Geology. In most cases, only static aspects are
discussed, but some dynamic cases are also presented.

As a rule, the modern tensorial approach is used. Some basic elements are reviewed at the beginning of the notes. More details
can be found into the excellent books of George Backus and Brian Kennett, also available at Samizdat Press. The linear
elasticity and the homogeneity of the continuum solid body are almost thoroughly assumed to be valid, but some elements of
Rheology are also presented.

In most cases, the semi-inverse method is used to solve the problems. According to it, the solution is supposed to be of a
particular form, as a consequence of the simplified hypothesis previously assumed. It is verified that solution checks both the
corresponding equations and the boundary conditions. Based on the Uniqueness Theorem of the Linear Elasticity, it follows the
assumed particular solution is just the general solution of the problem. In all the cases discussed here, the assumed simplified
hypotheses allow one to obtain simple, analytical solutions. At a first glance, the importance of such solutions is minor with
respect to the real cases, where mainly the non-homogeneity of the medium plays a great role. However, the analytical solutions
are the basis for deriving finite element algorithms, allowing one to model satisfactory the complex real cases. Such examples
are also presented.

The author will kindly appreciate any critical remarks on these notes, being very grateful to Samizdat Press for the possibility
to shear his work.
 3

CONTENTS

INTRODUCTION
page

Chapter A: BASIC ELEMENTS. ………………………………………………………………………………………….... 5

A.1) The displacement vector. Lagrangean (material) and Eulerian (spatial) co-ordinates. …………………………………..5
A.2) Invariants of a tensor. Tensor deviator. ………………………………………………………………………….………..5
A.3) Strain tensor. Stress tensor. Equation of motion / equilibrium. …………………………………………………………..5
A.4) HOOKE’s law. ………………………………………………………………………………………………………...…..8

Chapter B: DEFORMATION OF A CYLINDRICAL BODY IN THE PRESENCE OF GRAVITY … … … … … ....… 9

B.1) The model… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .9
B.2) Equations of equilibrium. Boundary conditions. Simplifying hypothesis. … … … … … … … … ..… … … … … … … … ..… ..9
B.3) The final shape of the body. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… ..12

Chapter C: LÉVY’s PROBLEM - the triangular dam… … … … … … … … … … … … … … … … … … … … … … … … … … . 14

C.1) The SAINT-VENANT’s equations. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… ..14

C.2) The model. Simplifying hypothesis. The planar deformation state. … … … … … … … … … … … … … … … … … … … … ...14
C.3) Equations of equilibrium. AIRY’s potential. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..15
C.4) Boundary conditions. The final shape of the dam..… … … … … … … … … … … … … … … … … … … … … … … … … ..… ..16

Chapter D: KIRSCH’s PROBLEM - the circular bore hole / tunnel… … … … … … … … … … … … … … … … … … ..… … 19

D.1) The model. … … ... … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… 19

D.2) The planar state of deformation in cylindrical co-ordinate system. … … … … … … … … … … … … … … … … … … … .… ..20
D.3) The circle of MOHR. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… … … … … … … … 21
D.4) AIRY’ s potential in cylindrical co-ordinates. The bi-harmonic equation.… … … … … … … … … … … … … … … … … .… 21
D.5) The divergence of a tensor in cylindrical co-ordinates. … … … … … … … … … … … … … … … … … … … … … … … … … ..22
D.6) The gradient of a vector and the strain tensor in cylindrical co-ordinates. … … … … … … … … … … … … … … … … ..… .23
D.7) The bi-harmonic equation cylindrical co-ordinates. … … … … … … … … … … … … … … … … … … … … … … … … .… … .24
D.8) The stress elements. Conditions at infinity for the stress elements. … .… … … … … … … … … … … … … … … ..… … … … 25
D.9) Strain and displacement vector. Conditions at infinity. … … … … … … … … … … … … … … … … … … … … … … .… … … 27
D.10) Boundary conditions for the stress elements on the wall of the circular cavity. … … … … … … … … … … … … … .… … .28
D.11) The final shape of the wall. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… … … 31

Chapter E: BOUSSINESQ’s PROBLEM - concentrated load acting on an elastic semi-space… … … … … … … … … ....33

E.1) The equations of BELTRAMI and MITCHELL. … … … … … … … … … … … … … … … … … … … … … … … … … … … … 33

E.2) The model. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… … … … .… 35
E.3) The equations of equilibrium and strain tensor in spherical co-ordinates. … … … … … .… … … … … … … … … … … … … 35
E.4) LAPLACE operator in spherical co-ordinates. LEGENDRE’ s polynomials. … … … … … … … … … … … … … … .… … … 36
E.5) The displacement field. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… ..… … … … … 37
E.6) Boundary conditions for the stress elements. The final solution. … … … … … … … ..… … … … … … … … … … … .… … … 39

Chapter F: ELEMENTS OF THIN PLATE THEORY… … … … … … … … … … … … … … … … … … … … … … … … … ....41

F.1) The model of a thin elastic plane plate… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .41

F.2) The planar state of a plate. The bending state… … … … … … … … … … … … … … … … … … … … … … … … … … … … .… .41
 4
F.3) Loads acting on the plate… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..42
F.4.) Odd and even functions for the planar state and for the bending state. … … … … … … … … … … … … … … … … … … … .42
F.5) Mean value of a function. Equilibrium equations for thin plates. … … … … … … … … … … … … … … … … … … … … … ..43
F.6) Thin plate in the bending state. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… ..45
F.7) BERNOULLI’ s hypothesis. … … … … … … … … … … … … … … … … … … … … … … … … … ..… … … … … … … … … .… ..45
F.8) HOOKE’ s law for a thin plate. … … … … … … .… … … … … … … … … … … … … … … … … … … … … … … … … … … .… ..46
F.9) The infinite, 1-dimensional (1-D) plate. The flexure of the lithosphere. … … … … ..… … … … … … … … … … … … ..… ..47
F.10) Exterior forces on the lateral surface of the plate. Buckling. … … … … … … … … … … … … … … … … … … … … … … ....48
F.11) The buckling of a simply leaning thin plate. … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… ..50
F.12) The infinite extended 1-D plate. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… … .… ..51
F.13) FOURIER transforms. Properties. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… .… ..51
F.14) Solution of the flexure equation by using FOURIER transforms. … … … ..… … … … … … … … … … … … … … … … .… ..52
F.15) Finite plates. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… … … … … … … … … ..… ..55
a) Significance of Σ ij and Μ ij for the bending state. … … .… … … … … … … … … … … … … … … … … … … … ..… … 55
b) The rectangular plate. Boundary conditions. LÉVY’ s solution. … … .… … … … … … … … … … … … … … … ..… … … 55
F.16) Vibrations of a plate laying on a viscous substratum. … … … … … … … … … … … … … … … … … … … … … … ..… … … 57
a) The differential equation. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ...… 57
b) The rectangular plate with 3 embedded sides. … … … … … … … … … … … … … … … … … … … … … … … … … … ...… 58

Chapter G: THE SPHERICAL SHELL… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … . 60

G.1) The model. BERNOULLI’ s hypothesis. Displacement vector and strain tensor. … … … … … … … … … … … … ..… … … 60
G.2) Quasi-mean values. Equation of motion. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… .61
G.3) Integrals of the stress elements. Quasi-moments. … … … ..… … … … … … … … … … … … … … … … … … … … … … ...… .62
G.4) Integrals of displacement vector. … … … … … … … … … … … … … … … … … … … … … .… … … … … … … … … … … ..… .62
G.5) Equation of motion in quasi-mean values. … … … … … … .… … … … … … … … … … … … … … … … … … … … … … ...… .63
G.6) Quasi-mean value of the shell density. The differential equation. … … ..… … … … … … … … … … … … … … … … … ..… .63
G.7) The buckling of a spherical shell. … … … … … ..… … … … … … … … … … … … … … … … … … … … … … … … … … ...… .64
G.8) Load on the upper face. Stress on the lower surface of the shell. … … … … … … … … … … … … … … … … … … … … ..… .66
G.9) The differential equation of time dependent flexure. … … … ..… … … … … … … … … … … … … … … … … … ..… … … … .67
G.10) Spherical effects with respect to the plane plate. … … … … .… … … … … … … … … … … … … … … … … … … … … ....… .68

Chapter H: ELEMENTS OF RHEOLOGY… … … … … … … … … … .… … … … … … … … … … … … … … … … … … … ..… 69

H.1) Introduction. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… ..69

H.2) Linear models. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… … … 69
a) KELVIN-VOIGT (strong viscous) model. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .69
b) MAXWELL (viscous-elastic) model. … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… … … 70
c) BURGERS (general linear) model. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .70
d) Remarks on the linear models. … … … … … … … … … … … … … … … … … … … … … … … .… … … … … … … … … … ..71
H.3) Non-linear models. … … … … … .… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .… 73
H.4) Brittle. Creep. Empirical criteria. … … … ..… … … … … … … … … … … … … … … … … … .… … … … … … … … … … … … 73
H.5) Empirical criteria for shear-faulting. TRESCA criterion. COULOMB-NAVIER criterion. … … … … … … … … … … .....74
H.6) Von MISES-HENCKY criterion for ductile flow (plasticity).. … … … … … … … … … ..… … … … … … … … … … … … … 76
H.7) Rheological models. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… … … .77
a) SAINT VENANT body (elastic-plastic material). … … … … … … … … … … … … … … … … … … … … … … … … … … .77
b) BINGHAM body (visco-plastic material). … … … … … … … … … … … … … … … … … … … … … … … … … … … … … .77

Chapter I: THE ACCRETION WEDGE… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… . 78

I.1) The model. … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … 78

I.2) Equations of equilibrium. Yield condition. Stress field. … … … .… … … … … … … … … … … … … … … … … … … … … .… 78
I.3) Boundary conditions. Final results. … … ..… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … 79

References… … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … … ..… … .81

 5
A) BASIC ELEMENTS.
A.1) The displacement vector. Lagrangean (material) and Eulerian (spatial) co-ordinates.

Consider an arbitrary material point inside a continuum body, subject to a deformation process. At the initial time t 0 , that
→
point has the position vector denoted by X , with respect to the origin of a co-ordinate system (Fig.A1). At a time t ≥ t 0 , the
→ → →
new position vector is to be x . The difference x − X represents the displacement vector. Taking into account that the
→ →
components (x1 , x 2 , x 3 ) of the vector x are all functions of the components (X1 , X 2 , X 3 ) of X , the displacement
vector can be written as
→ →
U = U (X1 , X 2 , X 3 , t ) . (a1)
This represents a Lagrangean (material) description of the deformation process. Here, (X1, X 2 , X3 ) are representing the
→
Lagrangean (material) co-ordinates. Alternately, the components (X1, X 2 , X3 )of X can be seen as functions of the
→
components ( 1 2 3)
x , x , x of the vector x . Consequently, the displacement vector can be written as
→ →
u = u (x1 , x 2 , x 3 , t ) . (a2)
This represents a Eulerian (spatial) description of the deformation process, where (x1 , x 2 , x 3 ) are the Eulerian (spatial)
co-ordinates. A basic supposition assumed thoroughly in that notes is that the deformation process is a continuous one, i.e. all
→ →
the components of u or U are continuous functions together their derivatives with respect to both their spatial co-ordinates or
to time.
The Lagrangean co-ordinates are usual in the Solid Mechanics, while the Eulerian co-ordinates are commonly used in Fluid
Mechanics. However, in the Linear Elasticity, the distinction between these two kinds of co-ordinates is not important, as it will
be seen in the next chapters. More details on such aspects can be found in (Aki and Richards 1980; Ranalli 1987).

Fig.A.1. The continuum deformed body and the displacement vector.

A.2) Invariants of a tensor. Tensor deviator.

A second order tensor represents mainly a 3x3 matrix. The elements of the tensor are changing according to a certain rule
with respect to a change of the co-ordinate system. Such a change with respect to a rotation will be discussed later. For
simplicity, only symmetric tensors will be considered. A symmetric tensor is equal to its transpose

t
T= T (a3)

(or Tij = Tji ). The superscript “t” shows the transposed tensor (matrix).
Let the components of the tensor be real numbers
 6
 T11 T12 T13 
 
T =  T12 T22 T23 

, (a4)
 T13 T23 T33 
→
The scalar λ and the vector u are representing the eigen-value and the eigen-vector respectively of that tensor if
→ → → →
T u = λ u , u ≠ 0 . (a5)
It’s easy to see that the eigen-values are not changing with respect to a rotation of the co-ordinates system.
→
Suppose now that the eigen-value λ and the components of the eigen-vector u are complex numbers. By taking the complex
conjugate (denoted by an asterisk) into (a5), it follows
→ →
T u = λ u*
* (a6)
Taking into account the symmetry of the tensor, the next inner product is evaluated into two different ways
→ → → → →2
* *
< T u , u >=< λ u , u >= λ u , (a7)

and
→ → → → → → → → →2
* t * * * * *
< T u , u >=< u , T u >=< u , T u >=< u , λ u >= λ u (a8)

From (a7) and (a8) it follows that the eigen-values (and the components of the eigen-vectors) of a symmetric tensor are real
numbers.
Eq.(a5) can be written as
 → → → →
 −λ  u = 0 , u ≠ 0
T 1 , (a9)
 

where 1 denotes the unit tensor. From (a9) it follows that the next determinant vanishes
T11 − λ T12 T13
T12 T22 − λ T23 = 0 (a10)
T13 T23 T33 − λ
Hence the eigen-values are the roots of the third degree equation
− λ3 + I1λ2 − I 2 λ + I 3 = 0 , (a11)
where

I1 = T11 + T22 + T33 = tr (T) ,

2 − T2 − T2
I 2 = T11T22 + T22 T33 + T33 T11 − T12 , (a12)
23 13

I 3 = T11T22 T33 + T12 T23 T13 +... = det (T)

With respect to a rotation of the co-ordinates system, the elements of the tensor are generally changing. Because the quantities
defined by (a12) can also be expressed as functions of the roots of (a11), it follows their values are not changing with respect to
a rotation. They represent the main invariants of the tensor. The first invariant is the trace of the tensor, while the third one
represent just its determinant.
The tensor defined by

T = T − 3 tr (T)1
* 1
, (a13)
 7
represents the tensor deviator, having its trace equal to zero. Elementary computations show its second invariant is

I *2 = − (T11 − T22 )2 + (T22 − T33 )2 + (T33 − T11 )2 + 6 T12

2 + T 2 + T 2  
1
(a14)
6  23 13  
That invariant is especially important to define constitutive equations for plasticity.

A.3) Strain tensor. Stress tensor. Equation of motion / equilibrium.

By using the spatial co-ordinates, the strain tensor is defined as
1 → →
ε =  grad u + grad t u  .
2  

(a15)

where “grad” denotes the gradient. In Cartesian co-ordinates, that symmetric tensor has the elements

1  ∂u i ∂uj  1
ε ij =  + (
 = u i, j + u j, i
2  ∂x j ∂x i  2
) . (a16)

According to the CAUCHY’s hypotheses there are two kinds of forces acting at an arbitrary point placed inside a body or on
→
its boundary. The first ones are represented by the mass forces, characterised by a mass density b . For the problems
→
discussed in that book, such mass forces are ignored. Or, they are represented by the gravity, when b is just the gravitational
→
acceleration g . Suppose now a mechanical state of tension (stress) is present inside the deformed body, e.g. as a result of the
→
action of a pair forces ± T . An arbitrary cross section is considered through a certain point of the body, dividing it into a part
denoted by D l at left and a pert D r at the right respectively (Fig.A.2). A surface element dS is considered on the boundary
→
of D l , having the outer pointing normal vector denoted by n . The material points of the boundary of D r are acting on dS
→
by an elementary force df . It follows (e.g. Aki and Richards 1980; Ranalli 1987) that the next relation is valid
→
df →
=σ n , (a.17)
dS
where the tensor σ represents the CAUCHY stress tensor, spatial co-ordinates being used. According to the Principle of

the Kinetic Momentum Balance, stress is a symmetric tensor. It can be shown too that the Principle of Impulse Balance leads to
the next equation of motion /equilibrium
→
→ d2 u
div σ +ρ b =ρ , (a18)
dt 2
→
d2 u
That equation is valid at an arbitrary point inside the body , where ρ is the density and represents the acceleration. By
dt 2
projecting eq.(a18) on the co-ordinates system axes, three scalar equations are obtained.

Fig.A.2. An imaginary cross section through the deformed body.

 8
A.4) HOOKE’s law.
Neglecting the initial stress (in most cases), it is further assumed a linear relation between the stress and strain tensors, i.e.

σ =H ε , (a19)

or
σ ij = H ijkl ε kl , (a20)

where H is a fourth-order tensor. Eq.(a19) represents HOOKE’s law. In the usual cases discussed here, an elastic,

homogeneous, isotropic medium is considered. Then eq.(a19) takes the particular form

σ = λtr ε 1 + 2µ ε . (a21)

Here, tr denotes the trace of the tensor, 1 is the unit tensor(matrix) and λ , µ are the elastic coefficients of LAMÉ. Hence

σ11 = λ(ε11 + ε 22 + ε 33) + 2µ ε11 ,

σ 22 = λ(ε11 + ε 22 + ε 33) + 2µ ε 22 , , (a.22)
σ 33 = λ(ε11 + ε 22 + ε 33) + 2µ ε 33
σ12 = σ 21 = 2µ ε12 , σ13 = σ 31 = 2µ ε13 , σ 23 = σ 32 = 2µ ε 23
Alternately, HOOKE’s law ( a21) can be reversed to give

1+ ν ν
ε =
E σ − tr σ
E 1 , (a23)

where the modulus of YOUNG is

3λ + 2µ
E=µ (a24)
λ+µ
and the transverse contraction coefficient of POISSON is
λ
ν= . (a25)
2(λ + µ )
By reversing (a24) and (a25), it follows

ν 1
λ= E , µ= E . (a26)
(1 + ν)(1 − 2ν) 2(1 + ν)
The parameter defined by

3λ + 2µ E
χ= = (a27)
3 3(1 − 2 ν )

represents the incompressibility or bulk modulus. For (theoretical) incompressible rocks, that modulus approaches infinity.
Other constitutive equations will be discussed in relation to the rheological bodies.
 9
B) DEFORMATION OF A CYLINDRICAL BODY IN THE PRESENCE OF
GRAVITY
B.1) The model.
An elastic homogeneous isotropic body is considered (Fig.B1). Its initial shape is a right, vertical, very thin cylinder of
radius equal to r and height equal to H . The base of the body is placed on the horizontal, absolutely rigid, plane x1Ox 2 .
The deformation of the body due to its own weight follows to be studied and the final shape of the body into the final
equilibrium state will be found. The approximations of the linear theory are assumed and the variation of the density is ignored.
The problem is solved by following the next steps:
i) - the equations of equilibrium are used, the unknowns here being the components of the stress tensor σ ; these equations

are processed according to the simplifying hypothesis of the problem;

ii) - by using the reversed HOOKE’s law, the equations of equilibrium are processed in order to have only the components of
the strain tensorε as unknowns;

→
iii) - by using the definition of the strain tensor, the components of the displacement vector u are obtained and the final
shape of the body is found.

Fig.B.1. (a) A vertical cylinder lying on a rigid plane; (b) The final shape of a vertical cross section (solid line) with respect to
the initial shape (dashed line). [NO SCALE]

B.2) The equations of equilibrium. Boundary conditions. Simplifying hypothesis.

A simplified approach can be derived by using cylindrical co-ordinates. However, the problem here is an introductory one. So
these co-ordinates will be used later, in relation to other problems. The equations of equilibrium in Cartesian co-ordinates are
σ111, + σ12,2 + σ13,3 = 0
σ12,1 + σ 22,2 + σ 23,3 = 0 (b1)

σ13,1 + σ 23,2 + σ 33,3 − ρg = 0

Here, δ is the density and g is the gravitational acceleration. The forces acting upon the body are the reaction force of the
horizontal plane and the gravity of the cylinder. The boundary conditions are:
-on the lateral surface of the cylinder:
→ →
σ n = 0 , for x 3 ∈[0, H], x1 , x 2 ∈Γ (b2)

-on the upper base of the cylinder

→ →
σ n = 0 , for x 3 = H, x1 , x 2 ∈ ∆ , (b3)

Here, ∆ is the disc of radius equal to r , having the centre at the origin of the co-ordinate system and the boundary denoted by
Γ . The outer pointing normal at the lateral surface of the body is a linear combination with variable coefficients of the
horizontal unit vectors, i.e.
→ → →
n = C1 ( x1 , x 2 ) e1 + C 2 ( x1 , x 2 ) e 2 (b4)
 10
For x 3 ∈[0, H ], x1 , x 2 ∈ Γ eq.(b2) becomes
→ → → → → → →
C1 ( x1 , x 2 )[σ11 e1 + σ12 e 2 + σ13 e3 ] + C 2 ( x1 , x 2 )[σ12 e1 + σ 22 e 2 + σ 23 e3 ] = 0 (b5)
→
The outer pointing normal at the upper base of the body is the unit vector e . For x 3 = H, x1 , x 2 ∈ ∆ , eq. (b3) gives
3
→ → → →
σ13 e1 + σ 23 e 2 + σ 33 e3 = 0 (b6)

Eq. (b5) is satisfied if the stress tensor has the form

0 0  0
 
 = 0
σ
0 0  (b7)
  
0 0 σ 33
on the lateral surface of the body.
Because the cylinder is a very thin one, the stress at its inner points is approximately the same one to the stress on the lateral
surface. So, it is assumed that eq.(b7) holds inside the whole volume of the body. It follows eqs.(b1a)-(b1b) are identical
verified. From eq. (b1c) it follows that
∂ σ 33
∂x 3
= ρg, σ 33 ( x1 , x 2 , x 3 = H ) = 0 (b8)

The problem represented by eq.(b8) has the next immediate solution

σ 33 (x1, x 2 , x 3 ) = ρg( x 3 − H ) (b9)

The reversed HOOKE’s law is

ε11 = E [(1 + ν) σ11 − ν(σ11 + σ 22 + σ 33)]

1

ε 22 = E [(1 + ν) σ 22 − ν(σ11 + σ 22 + σ 33)]

1

ε 33 = E [(1 + ν) σ 33 − ν(σ11 + σ 22 + σ 33)]

1
(b10)

1+ ν 1+ ν 1+ ν
ε12 = E σ 12 , ε 23 =
E σ 23 , ε13 =
E σ13

Because

1  ∂ u i ∂ u j
ε ij = 2  ∂x j
+
∂x i 
, (b11)

eqs. (b10) lead to

∂ u1 νρg ∂ u2 νρg ∂ u3 ρg
= (H − x3 ) , = (H − x3 ) , = ( x3 − H)
∂x1 E ∂x 2 E ∂x 3 E
(b12)
∂ u1 ∂ u2 ∂ u2 ∂ u3 ∂ u3 ∂ u1
+ =0 , + =0 , + =0
∂x 2 ∂x1 ∂x 3 ∂x 2 ∂x1 ∂x 3

By integrating eq.(b12) it follows that

 11
νρg
u1 = E
( H − x 3 )x1 + f1 ( x 2 , x 3 ) ,

νρg
u 2 = E ( H − x 3 )x 2 + f 2 ( x1, x 3 ) ,
2 (b13)
ρg x 3
u 3 = E ( 2 − Hx 3 ) + f3 ( x1, x 2 ) ,
∂f1 ∂f 2 ∂f 2 ∂f νρg ∂f3 ∂f1 νρg
+ =0 , + 3 = x2 , + = x1
∂x 2 ∂x1 ∂x 3 ∂x 2 E ∂x1 ∂x 3 E

Hence the displacement field is found if the unknown functions f1 , f 2 , f 3 are finally obtained. Differentiating (b13e) with
respect to x1 and (b13f) with respect to x 2 and adding the results, it follows

∂ 2 f3 ∂  ∂f1 ∂f 2 
2 +  + =0 . (b14)
∂x1∂x 2 ∂x 3  ∂x 2 ∂x1 
So, using eq.(b13d) it follows
∂ 2 f3
=0 (b15)
∂x1∂x 2
From eq.(b15) it follows that
f 3 ( x 1 , x 2 ) = h 1 ( x1 ) + h 2 ( x 2 ) , (b16)

where h1 , h 2 are two unknown functions, following to be found. Eqs.(b13e) and (b13f) give

∂f 2 ( x1 , x 3 ) νρg dh ( x ) ∂f1 ( x 2 , x 3 ) νρg dh ( x )

= x2 − 2 2 , = x1 − 1 1 (b17)
∂x 3 E dx 2 ∂x 3 E dx1

The left side of eq.(b17a) is represented by a function depending on x1 , x 3 only, while the right side is a function of x 2 .
Hence both sides are equal to a constant, i.e.

∂f 2 ( x1 , x 3 ) dh 2 νρg
= −a 2 , = x2 + a2 (b18)
∂x 3 dx 2 E
It follows that
νρg 2
f 2 ( x1 , x 3 ) = − a 2 x 3 + g 2 ( x1 ) , h 2 ( x 2 ) = x + a 2x2 + b2 (b19)
2E 2
In a similar manner, eq.(b17b) gives

νρg 2
f1 ( x 2 , x 3 ) = − a1x 3 + g1 ( x 2 ) , h1 ( x1 ) = x + a1x1 + b1 (b20)
2E 1
From eq. (b13d) it follows that
dg1 ( x 2 ) dg ( x )
=− 2 1 =K (b21)
dx 2 dx1
where K is a constant. Then

g1 ( x 2 ) = Kx 2 + C1 , g 2 ( x1 ) = − Kx1 + C 2 (b22)

For simplicity, material co-ordinates are used to obtain the final expression of the displacement field
 12
 
 
   ν( H − X 3 )X1   − a1X 3 − KX 2   C 2 
→ →  u1  ρg      
u ( X ) =  u 2 =  ν( H − X 3 )X 2  +  − a 2 X 3 + KX1  +  C1  (b23)
  a1X1 + a 2 X 2   b1 + b 2 
E
   2
 u 3 X
 3 ν 
 − HX 3 + ( X12 + X 2 )
 2 2 2 
The first term in eq.(b23) is the true displacement, the last one is a translation while the second term is the rigid rotation
 a2  →
 
 − a1  × X (b24)
 
K 
B.3) The final shape of the body.
a) The final shape of the upper base

Consider an arbitrary point of co-ordinates equal to ( X1 , X 2 , X 3 = H ) . In the initial stage, it is placed on the upper base of
the cylinder. Finally, the co-ordinates of the point are
 
 
 x1   X 1   0 
    ρg  
 x2 =  X2 +
E
0

(b25)
   
 x3   H  ν H 
2
 (X 2 + X 2 ) − 
2 1 2 2 
Hence

x1 = X1

x 2 = X 2 (b26)
 νρg 2 ρg 2
x 3 = ( X1 + X 22 ) − H +H
 2E 2E
From eq.(b26) it follows that
x12 + x 2 2 2
2 = X1 + X 2 = r
2 . (b27)

Hence the circle representing the contour of the upper base remains a circle of the same radius. The plane of the circle is
2 2
moving downward by a quantity equal to ρg( H − νr ) / ( 2 E ) . The surface of the disc representing the upper base of the
body is no longer a plane one. It becomes a rotational parabolic surface having the equation

νρg 2 ρg 2
x3 = ( x1 + x 2
2 )− H +H (b28)
2E 2E
b) The final shape of the lower base

Consider now an arbitrary point initially placed on the lower base of the body. The point has the co-ordinates equal to
( X1 , X 2 , X 3 = 0) . The final co-ordinate of the point are
 
 x1   X1   HX1 
    νρg  
=
 2  2
x X + HX 2 (b29)
    E  
 x3   0   ( X 2 + X 2 ) / 2
 1 2 
Hence
 13

x1 = (1 + νρgH / E )X1

x 2 = (1 + νρgH / E )X 2 (b30)
 νρg 2
x 3 = ( X1 + X 22 )
 2E
From eq.(b30) it follows that

x12 + x 22 = (1 + νρgH / E) 2 ( X12 + X 2 2 2

2 ) = (1 + νρgH / E) r , (31)

i.e. the circle representing the contour of the lower base remains a circle. The new radius is increased by a quantity equal to
νρgH / E . The initial horizontal plane of the circle is uplifted by a quantity equal to νρgr 2 / ( 2 E ) . The surface of the disc
representing the lower base becomes a rotational paraboloid having the equation

νρg
x3 = ( x12 + x 2
2) (b32)
2E(1 + νρgH / E ) 2

c) The final shape of the lateral surface

Consider now a point initially placed on a generatrix line of the cylinder. Because of the cylindrical symmetry of the problem,
the point having the initial co-ordinates equal to ( X1 = 0, X 2 = r, X 3 ) is considered. Finally, that point has the position
characterised by the co-ordinates



x1 = 0
 νρg
x 2 = r + ( H − X 3 )r (b33)
 E
  X2 
x = X + ρg  3 − HX  + νρg r 2
E 2
 3 3 3
2E
  
From (b33), it follows that the generatrix remains into the initial vertical plane. Its shape is changed from a straight line
segment to a convex parabolic segment, having the equation

E E ρg 2 νρg 2
x3 = (x 2 / r − 1)2 − (x 2 / r − 1) + H − H + r (b34)
2 ν 2 ρg νρg 2E 2E

OBSERVATION. On the lower base of the cylinder it is acting the reaction force of the rigid plane, equal to the weight of the
body. When the surface of the base is decreasing, approaching the paraboloid of eq.(b32), the normal unit effort (equal to the
weight divided by the contact area) is increasing. At a certain moment, its magnitude will exceed a yielding value of the
material. Then, HOOKE’s law, valid in the elastic domain, will be no longer appropriate here.
 14
C) LÉVY’s PROBLEM - THE TRIANGULAR DAM
C.1) The SAINT-VENANT ‘s equations.
Differentiating a certain element of the strain tensor

ε ij =  u i, j + u j, i / 2 , (c1)

it follows, for example, that

ε11,22 + ε 22,11 = (u11, ),22 + (u 2,2),11 = u1122

, + u 2,211

(u1,2),12 (u 2,1),12 ( ),12

(c2)
= + = u1,2 + u 2,1 = 2 ε12,12
Hence
ε11,22 + ε 22,11 = 2 ε12,12 (c3)
Also,
ε 22,33 + ε 33,22 = 2 ε 23,23 (c4)

ε 3311
, + ε11,33 = 2 ε 31,31 (c5)
In a similar way, it follows that

(ε12,3 + ε 23,1 − ε 31,2),2 = ε 22,31

(ε 23,1 + ε 31,2 − ε12,3),3 = ε 33,12 (c6)-(c8)

(ε 31,2 + ε12,3 − ε 23,1),1 = ε11,23

The above equations (c3)-(c8) represent the SAINT-VENANT’s equations of compatibility.

C.2) The model . Simplifying hypothesis. The planar deformation state.

A horizontal dam of infinite length is considered. The cross-section is represented by a rectangular triangle OAB (Fig.C1).
The length of the base is AB=l and the height is OA=h. On OA catheter is acting the hydrostatic pressure of a liquid (water)
having the specific weight equal to γ. As a result, the dam is deformed. The dam is represented by an elastic homogeneous,
isotropic material. Its specific weight is equal to Γ and its elastic constants are E and ν.

Fig.C1. A vertical cross section through the dam. N.Hs. is the free surface of the water, acting on OA side by a pressure linearly
increasing with depth.

Because the shape of the dam, the displacement vector has the components like
 15
 u = u ( x1 , x 2 )
 1 1
 u 2 = u 2 ( x1 , x 2 ) (c9)

u 3 = 0
It follows the strain tensor components are like
ε11 = u11, = ε11(x1, x 2 )
ε12 = 2 (u1,2 + u 2,1) = ε12 ( x1 , x 2 )
1

ε13 = 2 (u1,3 + u 3,1) = 0

1
(c10)
ε 22 = u 2,2 = ε 22 ( x1 , x 2 )
ε 23 = 2 (u 2,3 + u 3,2) = 0
1

ε 33 = u 3,3 = 0
Hence the strain matrix is
 ε11 ε12 0
 
 = (x , x ) = 
ε  ε  1 2  ε12 ε 22 0 , (c11)
0 
 0 0
It corresponds to a planar state of the strain (the plane here being 1-2).
The components of the stress tensor are

σ11 = λ(ε11 + ε 22) + 2µ ε11

σ12 = 2µ ε12
σ13 = 2µ ε13 = 0
σ 22 = λ(ε11 + ε 22) + 2µ ε 22
(c12)

σ 23 = 2µ ε 23 = 0
σ 33 = λ(ε11 + ε 22) + 2µ ε 33 = λ(ε11 + ε 22) = 2(λ + µ ) (σ11 + σ 22) = ν(σ11 + σ 22)
λ

Hence the stress matrix is

 
 σ11 σ12 0 
 = (x , x ) =  
σ  σ  1 2  σ12 σ 22 0  (c13)

 0 0 ( 
ν σ11 + σ 22  )
Because the component 33 of the stress has a non-zero value, eq.(c13) shows that the stress state corresponding to a planar state
of the strain is not generally a planar one too.

C.3) Equations of equilibrium. AIRY’s potential.

The only body force acting on the dam is its weight. The equations of equilibrium are
σ111, + σ12, 2 + Γ = 0

 (c14)
σ12,1 + σ 22, 2 = 0
Because the presence of Γ , eqs.(c14) represent a non-homogeneous system. In the beginning, the homogeneous system is
solved, i.e.
 16
Σ 111, + Σ12,2 = 0
 (c15)
Σ 12,1 + Σ 22,2 = 0
Using an unknown function ϕ , the first equation of (c15) is verified for
∂ϕ ∂ϕ
Σ11 = ∂x 2 , Σ12 = − ∂x1 (c16)

In the same way, the second equation of (c15) is verified for

∂ψ ∂ψ
Σ12 = ∂x 2 , Σ12 = − ∂x1 (c17)

It follows that
∂ϕ ∂ψ
+ =0 (c18)
∂x1 ∂x 2
i.e. the unknown functions are
∂A ∂A
ϕ= , ψ=− (c19)
∂x 2 ∂x1
The unknown function A = A ( x1 , x 2 ) represents the AIRY’s potential. It allows one to obtain the next expressions for
the components of the stress tensor when the body force are absent:

Σ11 = A ,22 , Σ12 = − A ,12 , Σ 22 = A ,11 (c20)

From (c13), the trace of the stress tensor can be written using LAPLACE’s operator in 1-2 co-ordinates
tr Σ = (1 + ν ) (Σ11 + Σ 22) = (1 + ν)∆* A (c21)

The components of the strain tensor are obtained using the reversed HOOKE’s law
1+ ν  1+ ν  1+ ν
ε11 =  − ν∆* A  , ε 22 =  − ν∆* A  , ε12 = −
E  A ,22 E  A ,11 E A ,12
. (c22)
 
Using (c22) and (c3) it follows

A ,2222 − ν ∆ A  − ν ∆* 

* + = −2 A ,1212
 ,22 A ,1111  A  ,11
, (c23)

i.e.
(1 − ν)∆* ∆* A = 0 (c24)

Because ν < 0.5 , it follows that AIRY’s potential is a solution of the bi-harmonic equation
∆* ∆* A =0 (c25)

Because the trace of a tensor is an invariant, eq.(c25) holds too in the general case of the orthogonal curvilinear co-ordinates.
However, eq.(c20) has to be modified.

C.4) Boundary conditions. The final shape of the dam.

On the side OA of the dam is acting the hydrostatic pressure. It follows that
→ →
σ ( − e 2) = γx1 e 2 (c26)

On the side OB of the dam is acting the negligible atmospheric pressure. It follows that
→ →
σ n= 0 (c27)

where the outer pointing normal at the dam is

→ → →
n = − sin α e1+ cos α e 2 (c28)
On the side OA, for x1 ∈[0, h ], x 2 = 0 , it follows that
 17
σ12 = 0,
 (c29)
σ 22 = − γx1
On the side OB it follows for x1 ∈[0, h ], x 2 = x1 tan α that
σ12 − σ11 tan α = 0,
 (c30)
σ 22 − σ12 tan α = 0
Eqs.(c29)-(c30) represent 4 boundary conditions, suggesting a solution of the bi-harmonic equation (c25) which depends on 4
unknown coefficients denoted by a , b , c , d , i.e.

a b c d
A (x1, x 2 ) = 6 x13 + 2 x12 x 2 + 2 x1x 22 + 6 x 32 (c31)

Using (c20), the solution of the homogeneous system is

Σ = cx1 + dx 2
 11
Σ 12 = − ( bx1 + cx 2 ) (c32)

Σ 22 = ax1 + bx 2
A particular solution of the non-homogeneous system (c14) is
σ11 = σ 22 = 0
 (c33)
σ12 = − Γx 2
It follows the general solution of (c14) is
σ = cx1 + dx 2
 11
σ12 = −( bx1 + cx 2 ) − Γx 2 . (c34)

σ 22 = ax1 + bx 2
Replacing (c34) into (c29)-(c30) it follows that
−( bx1 + cx 2 ) − Γx 2 = 0 , for x1 ∈[0, h ] , x 2 = 0
ax + bx = − γx , for x ∈[0, h ] , x = 0
 1 2 1 1 2
 (c35)
−( cx1 + dx 2 ) tan α − ( bx1 + cx 2 ) − Γx 2 = 0 , for x1 ∈[0, h ] , x 2 = x1 tan α
( bx1 + cx 2 + Γx 2 ) tan α + ax1 + bx 2 = 0 , for x1 ∈[0, h ] , x 2 = x1 tan α
It follows that
a = − γ
b = 0

 2 (c36)
c = − Γ + γ / tan α
 3
d = Γ / tan α − 2γ / tan α
and
σ = Ax1 + Bx 2
 11
σ12 = − Cx 2 (c37)

σ 22 = − γx1
where
A = γh 2 / l 2 − Γ , B = Γh / l − 2γh 3 / l 3 , C = − γh 2 / l 2 (c38)
Hence
u11, = ε11 = C1x1 + C 2 x 2 , (c39)
where
 18

C1 = (1 + ν )[A − ν( A − γ )] / E , C 2 = (1 − ν 2 )B / E (c40)
It follows that
2
u1 = C1x1 / 2 + C 2 x1x 2 + f1( x 2 ) (c41)
where the unknown function f1 follows to be found. In the same manner,
2
u 2 = C 3 x1x 2 + C 4 x 2 / 2 + f 2 ( x1 ) (c42)
But

ε12 = 2 (u1,2 + u 2,1) = 2 (C 2 x1 + f1’( x 2 ) + C3x 2 + f 2’ ( x1 ) =

1 1 1+ ν 1+ ν
E σ 12 =−
E
Cx 2 (c43)

Hence
C x + f ’ ( x ) = K
 2 1 2 1
 , (c44)
C 3 x 2 + f1’( x 2 ) = − K
where K is an arbitrary constant. It follows
f1 ( x 2 ) = − C 3 x 22 / 2 − Kx 2 + K1 , f 2 ( x1 ) = − C 2 x12 / 2 + Kx1 + K 2 (c45)
Hence, the displacement field is
 = C x 2 / 2 + C x x − [C + 2(1 + ν)C / E ]x 2 / 2 − Kx + K
u 1 1 1 2 1 2 3 2 2 1
 (c46)
u = − C 2 x 2 / 2 + C 3 x1x 2 + C 4 x 2 / 2 + Kx1 + K 2
 2 1 2
The last terms into (c46) represent a rigid roto-translation.
It should be outlined that the above boundary conditions on stress values on the sides OA and OB are not complete ones. As a
result, the unknown constants C 3 , C 4 are present in (c46). Boundary conditions on stress values (or displacements) on the
side AB are required in order to obtain an unique solution of the problem
For example, consider the case when the points A and B are fixed ones. It follows
 = C  x 2 − h 2  / 2 + C x − h x − [C + 2(1 + ν)C / E]x l − x / 2
 u1 1 1  2( 1 )2 3 2( 2)
 (c47)
u = C 2  h 2 − x 2  / 2 + C 3 x 2 (x1 − x 2 h / l) + {C 2 h − [C 3 + 2(1 + ν)C / E]l / 2}( x1 − h )
 2  1
An arbitrary point placed initially on the side AB has the initial co-ordinates (X1 = h; X 2 ) . Its final position is
x1 = X1 + u 1 ( X1 , X 2 ) = h + [C 3 + 2(1 + ν)C / E]X 2 ( l − X 2 ) / 2
 (c48)
x 2 = X 2 + u 2 ( X1 , X 2 ) = X 2 + C 3 hX 2 ( l − X 2 ) / l
Elementary computations show that

2(1 + ν) 1+ ν  h2 
C3 + C= −  γ (1 − ν) − νΓ − ( 2 − ν)γ  (c49)
E E  l 2 

If
2(1 + ν)
C3 + C< 0 , (c50)
E
the final shape of the side AB is a concave parabolic segment. Because the possibility of the water to flow below the dam, that
situation is not recommended in real cases. Therefore, it is asked to
γ (1 − ν ) − νΓ − ( 2 − ν )γ ( h / l ) 2 ≤ 0 , (c51)
i.e.
h / l≥ [γ (1 − ν) − νΓ ] / ( 2 − ν) / γ (c52)

For example, assuming that γ = 1000 Kgs / m

3 , Γ = 2400 Kgs / m3 , ν = 0.25 it follows that h ≥ 0.29 l .
EXERCISE. Obtain the final shape of the dam in the above hypothesis.
 19
D) KIRSCH’s PROBLEM - THE CIRCULAR BORE HOLE / TUNNEL

D.1) The model.

It is assumed that the whole 3-dimensional space is represented by an elastic, homogeneous, isotropic medium, having the
elastic constants denoted by E and ν , respectively λ and µ . A co-ordinate system having the third axis positive upward will

σ
be used. The initial state of stress is represented by the homogeneous tensor
0 , corresponding to a planar state of
deformation, i.e.
 0 
 σ11 σ12
0 0 
 
σ  σ12 σ 22
0= 0 0 0 , (d1)

 0 0 ν σ11 + σ 22 
 0 0 
  

σ ij have constant values. The mass forces are ignored, hence the equilibrium equation
where the components
0

→
div σ 0 = 0 (d2)
is identically satisfied.
Suppose that a circular, infinite bore hole / tunnel is performed along the third axis, its material being instantly removed. The
origin of the co-ordinate system is placed at the centre of the cavity. On the wall of the bore hole is acting now the atmospheric
pressure (or the pressure of the drilling mud), denoted by p 0 . Consequently, a new (non-homogeneous) stress value is
σ
obtained and the circular shape of the bore hole is changing too. It follows to obtain the new stress, denoted by
f , and the
new shape of the bore hole in the final equilibrium stage, where
→
div σ f = 0 . (d3)

It is also assumed that the deformation is an elastic one, i.e. the stress perturbation σ = σ f − σ 0 is related to the strain
tensor by
σ = λtr ε 1 + 2µ ε . (d4)

The unknown components of the displacement vector are supposed to correspond to a planar deformation state, i.e.
u1 = u1 ( x1 , x 2 ) , u 2 = u 2 ( x1 , x 2 ) , u 3 = 0 . (d5)
Because the symmetry of the problem, the cylindrical co-ordinate system (r, θ, z) will be used, having the unit vectors
→ → →
denoted by  e , e , e  (see Fig.D1).
 r θ z
 

Fig.D1. The cylindrical co-ordinate system.

D.2) The planar sate of deformation in cylindrical co-ordinate system.

 20

With respect to Fig.D1 it follows that

→ → →
e = cos θ e + sin θ e ,
 r 1 2
→ → →

e θ = − sin θ e1 + cos θ e2 , (d6)

→ →
e z = e3 .

Hence the matrix for passing from the Cartesian co-ordinates to cylindrical co-ordinates is

 cos θ sin θ 0 
Q = − sin θ cos θ 0
 (d7)
 
 0 0 1

It represents a rotation of angle equal to θ in a positive (counter clockwise) sense. From (d4) and (d5) it follows that the stress
matrix in Cartesian co-ordinates is

 
 σ11 σ12 0 
crt  
[σ] =  σ12

σ 22 0  .

(d8)

 0 0 ν σ11 + σ 22 
 
  
Let the stress matrix in cylindrical co-ordinates be

σ σ rθ σ rz 
cyl 
rr
[σ] 
= σ rθ
 σ θθ σ θz . (d9)

 σ rz σ θz σ zz 
It follows that
 
 cos θ sin θ 0   σ11 σ12 0   cos θ − sin θ 0
[σ] [σ]
cyl crt t    
=Q Q =  − sin θ cos θ 0  σ12 σ 22 0   sin θ cos θ 0
    0 
 0 0 1  
 0 0 (
ν σ11 + σ12  ) 0 1

(d10)
By performing the computations in (d10), it follows

σ11 + σ 22 + σ11 − σ 22 cos 2θ

σ rr = 2 σ12 sin 2θ + 2
, (d11)

σ11 + σ 22 σ11 − σ 22
σ θθ = 2
− σ12 sin 2θ −
2
cos 2θ , (d12)

σ11 − σ 22 sin 2θ +
σ rθ = − 2 σ12 cos 2θ , (d13)

σ rz = σ θz = 0 , σ zz = ν(σ11 + σ 22) = ν(σ rr + σ θθ) . (d14)

D.3) The circle of MOHR.

 21

Suppose the Cartesian co-ordinate system is selected in order its axes to be along the first two eigen vectors of the stress
tensor. In that case, σ11 and σ 22 are eigenvalues of the stress tensor and σ12 = 0 . From equations (d11)-(d13) it follows
that
2 2
 σ − σ11 σ 22 
 +   σ11 − σ 22 
 rr  + σ rθ = 
2  ,
 (d15)
 2   2 
and an identical relation obtained by replacing σ rr with σ θθ . Eq.(d15) shows that σ rr and σ rθ are placed on a circle
of radius equal to σ − σ
11 22 / 2 . Suppose now that σ rr (or σ θθ ) (i.e. the radial stress component, usually denoted
by σ ) and σ
rθ (i.e. the tangential stress, usually denoted by τ ) are obtained at various angles θ and the MOHR’s
circle represented by eq.(d15) is obtained. Its radius and its position of the centre allow one to obtain graphically the
eigenvalues of the stress tensor. Further discussion will be presented in relation to the empirical failure criteria of materials.

D.4) AIRY’s potential in cylindrical co-ordinates. The bi-harmonic equation.

Consider the representation of the stress components with the AIRY’s potential in Cartesian co-ordinates, i.e.
σ11 = A ,22 , σ 22 = A ,11 , σ12 = − A ,12 , (d16)
where the AIRY’s potential verifies the bi-harmonic equation

∆*∆* A = 0 . (d17)

In the beginning, the derivatives in eq.(d16) will be evaluated by using the polar co-ordinates

x1 = r cos θ r = x 2 + x 2

 , 1 2 (d18)
x 2 = r sin θ θ = a tan( x 2 / x1 )
Than, a representation of the stress components in cylindrical co-ordinates with the help of the AIRY’s potential will be
obtained from (d11)-(d13).
But
∂r x ∂r x ∂θ x sin θ ∂θ x cos θ
= 1 = cos θ , = 2 = sin θ , =− 2 =− , = 1= (d19)
∂x1 r ∂x 2 r ∂x1 r2 r ∂x 2 r 2 r
It follows
∂A ∂A ∂r ∂ A ∂θ sin θ
A ,1 = ∂x1
=
∂r ∂x1
+
∂θ ∂x1
= cos θ A , r −
r A ,θ
(d20)

In the same way,

cos θ
A ,2 = sin θ A , r + r A ,θ
. (d21)

Also,

A ,12 = (A ,1),2 =  cos θ A , r −

 sin θ 
A θ
r ,  ,2

 sin θ  cos θ  sin θ 

= sin θ cos θ A , r − A θ +  cos θ A , r − A θ (d22)
 r ,  ,r r  r ,  ,θ

 A , r − A ,θθ  sin 2θ +  A , rθ − A ,θ  cos 2θ

=  A , rr −
   
 r r2  2  r r2 
In the same way
 A ,θθ  + 2 sin θ cos θ − A , rθ + A ,θ 
= 2θ + 2 θ A , r +
A ,11 cos A , rr sin
 r   
, (d23)
 r2   r r2 
 22
 A , θθ  − 2 sin θ cos θ − A , rθ + A , θ 
= 2θ + 2 θ A , r +
A ,22 sin A , rr cos
 r   
, (d24)
 r2   r r2 
Hence
A , r + A , θθ
σ11 + σ 22 = A ,22 + A ,11 = A , rr + , (d25)
r r2
 A , r + A ,θθ  cos 2θ − 2 sin 2θ − A , rθ + A , θ 
σ11 σ 22 A ,22 A ,11  − A , rr +
− = − =
  
, (d26)
 r r2   r r2 
From eqs.(d11)-(d13) it follows

A , r + A , θθ A , rθ + A , θ
σ rr = , σ θθ = A , rr , σ rθ = − . (d27)
r r2 r r2

D.5) The divergence of a tensor in cylindrical co-ordinates.

In the case of the cylindrical co-ordinates, the square of the elementary arc is equal to

ds 2 = dx12 + dx 2 2 2 2 2
2 + dx 3 = dr + r dθ + dz
2 (d28)

Hence the differential parameters of Lamé are

h1 = 1 , h 2 = r , h 3 = 1 , (d29)

the orthogonal curvilinear co-ordinates are equal to

c1 = 1 , c 2 = θ , c 3 = z , (d30)
and the unit vectors are
→ → → → → →
n1 = e r , n 2 = eθ , n 3 = ez . (d31)
It follows that
∂h q
= δ q 2δ β1 . (d32)
∂c β

Substituting the above results in the formula (a2.77) (Ivan 1996) it follows the next formula for the divergence of a tensor in
cylindrical co-ordinates

→
 ∂T ∂T ∂T T − Tθθ 
divT =  rr + rθ + rz + rr  er
 ∂r r∂θ ∂z r 
(d33)
→ →
 ∂Trθ ∂Tθθ ∂Tθz Trθ   ∂Trz ∂Tθz ∂Tzz Trz 
+ + + +2 e + + + + e
 ∂r r∂θ ∂z r  θ  ∂r r∂θ ∂z r  z

D.6) The gradient of a vector and the strain tensor in cylindrical co-ordinates.
Substituting the above results in the formula (a2.68) (Ivan 1996) it follows the next formula for the gradient of a vector in
cylindrical co-ordinates
 23
→ ∂u → → ∂u → → ∂u → →
grad u = r e r ⊗ e r + r e r ⊗ eθ + r e r ⊗ e z
∂r r∂θ ∂z
→ → → → → →
∂u θ ∂u θ ∂u θ
+ e ⊗ e r + r∂θ eθ ⊗ eθ + ∂z eθ ⊗ e z . (d34)
∂r θ
→ → → → → → → → → →
∂u z ∂u z ∂u z uθ ur
+ e ⊗ e r + r∂θ e z ⊗ eθ + ∂z e z ⊗ e z − r e r ⊗ eθ + r eθ ⊗ eθ
∂r z

 →  → t 
= grad u +  grad u   are
1
The components of the strain tensor ε 2   
 
∂u r 1  ∂u ∂u  1  ∂u r ∂u z 
, ε rθ =  r − θ + θ 
u
ε rr = ∂r 2  r∂θ r ∂r 
, ε rz = 2  ∂z
+
∂r 

. (d35)
∂u 1  ∂u ∂u  ∂u
ε θθ = r∂θθ + rr , ε θz =  θ + z 
u
, ε zz = z
2  ∂z r∂θ  ∂z
→
For the particular displacement field represented by (d5), the components of the vector u are

u r = u r ( r, θ ) , u θ = u θ ( r , θ ) , u z = 0 . (d36)
Hence

ε rz = ε θz = ε zz = 0 . (d37)
It is the case of a planar state of deformation, i.e.

 ε rr ε rθ 0
 
ε =  ε rθ

ε θθ 0

, (d38)

 0 0 0
and all the strain elements are functions of r and θ .
Consequently, the stress is
 
 σ rr σ rθ 0 
 
σ =  σ rθ σ θθ 0  , (d39)
 
 0
 0 (
ν σ rr + σ θθ  )
all the components of the stress being too functions of r and θ , according to HOOKE’ s reversed law. It follows from (d33)
the next equations of equilibrium are obtained in the absence of mass forces:

∂ σ rr ∂ σ rθ σ rr − σ θθ = 0
+ + (d40)
∂r r∂θ r
∂ σ rθ ∂ σ θθ σ rθ = 0
+ +2 (d41)
∂r r∂θ r

D.7) The bi-harmonic equation in cylindrical co-ordinates.

By using (a2.69) and (a2.43) (Ivan 1996), it follows the LAPLACE operator in cylindrical co-ordinates is
 24

1 ∂  ∂f  1 ∂ 2 f ∂ 2 f
∆f = div( grad f ) = r  + + . (d42)
r ∂r  ∂r  r 2 ∂θ 2 ∂z 2
The AIRY’ s potential is also a function of r and θ . Hence the AIRY’ s potential is the solution of the bi-harmonic equation
∆*∆* =0
A , (d43)

where the LAPLACE operator in polar co-ordinates is

1 ∂  ∂  1 ∂2 ∂2 1 ∂ 1 ∂2
∆* =  r  + = + + . (d44)
r ∂r  ∂r  r 2 ∂θ 2 ∂r 2 r ∂r r 2 ∂θ 2
It should be noted that the singular point r = 0 is avoided in (d44) because r ≥ R > 0 , where R is the radius of the bore hole.
It follows to solve (d43) by using (d44) in order to derive the AIRY’ s potential. The stress components will be obtained from
(d27), imposing the boundary conditions on the wall of the bore hole. The components of the strain will be derived by using the
HOOKE’ s reversed law. The displacement vector will be obtained from the definition of strain elements, allowing one to find
the final shape of the deformed bore hole wall.
Consider the FOURIER expansion of the AIRY’ s potential, having the coefficients equal to functions of r
∞
A ( r, θ ) = A 0 ( r ) + ∑ [A n ( r ) cos nθ + Bn ( r) sin nθ] (d45)
n =1
It follows
∂A ∞
∂r
= A ’0 + ∑  A ’n cos nθ + B’n sin nθ , (d46)
n =1
∂2 A ∞
= A ’’
0+ ∑  A ’’n cos nθ + B’’n sin nθ , (d47)
∂r 2 n =1
∂2 A ∞
=− ∑ n 2 (A n cos nθ + Bn sin nθ) . (d48)
∂θ 2 n =1
Hence
∞     
1 n2  cos nθ +  ’’+ 1 ’− n2
∆ A = ∆ A0 + ∑  A ’’+ r A ’− A  B r B B  sin nθ
 
(d49)
n =1  r2   r2  
and
∞  2 2 4 2 
’’’’ + 2 ’’’ − 2n + 1 ’’ + 2n + 1 ’ + n − 4n
∆∆ A = ∆∆ A 0 + ∑  A r A

A A A  cos nθ

n =1  r2 r3 r4 
. (d50)
 2 2n 2 + 1 ’’ 2n 2 + 1 ’ n 4 − 4n 2  
+  B’’’’ + B’’’ − B + 3 B+ B  sin nθ
  
 r r2 r r4  
By using (d50), the bi-harmonic (d43) is verified if
∆∆ A 0 = 0 , (d51)

and if the functions A ,B are the solutions of the next differential equation

2 2 4 2
’’’’ + 2 ’’’ − 2n + 1 ’’ + 2n + 1 ’ + n − 4n
Φ rΦ Φ Φ Φ =0 . (d52)
r2 r3 r4
Because A is a function of r only, eq.(d51) is
0
1 d  d  1 d  d A 0   
r  r   = 0 .
r dr  dr  r dr  dr   
(d53)
  
 25
Hence
d 1 d  d A 0   d  d A 0
r  r   = a0 , r  = a 0 r ln r + b 0 r
dr  r dr  dr   dr  dr 
. (d54)
 
But
r2  1
∫ r ln rdr = 2  ln r − 2  + const. (d55)

Hence
d A0 a  1 b
r = 0 r 2  ln r −  + 0 r 2 + c 0 (d56)
dr 2  2 2
Finally, denoting again the constants, it follows
2 ln r + b r 2 + c ln r + d .
A 0 (r) = a 0 r 0 0 0 (d57)
In order to solve eq.(52), a solution of the form

Φ = rm (d58)

is considered. Substituting (d58) in (d52), it follows that the exponent m is the solution of the algebraic equation
m ( m − 1)( m − 2)( m − 3) + 2m ( m − 1)( m − 2) − ( 2n 2 + 1)m ( m − 1) + ( 2n 2 + 1)m + n 4 − 4n 2 = 0 ,(d59)
having the roots
m1 = − n , m 2 = − n + 2 , m 3 = n , m 4 = n + 2 . (d60)
Hence the AIRY’ s potential is
∞
A ( r, θ ) = a 0 r 2 ln r + b 0 r 2 + c 0 ln r + d 0 + ∑  a n r n + 2 + b n r n + c n r − n + 2 + d n r − n  cos nθ
n =1
,(d61)
∞
+ ∑  α n r n + 2 + β n r n + γ n r − n + 2 + δ n r − n  sin nθ
n =1
where the unknown coefficients a i , b i , c i , d i , α j , β j , γ j , δ j , i = 0,1,2,... , j = 1,2,... follows to be obtained.

D.8) The stress elements. Conditions at infinity for the stress elements.
By using (d61) and (d27), it follows
c0
σ θθ = A , rr = a 0 (2 ln r + 3) + 2b 0 −
r2
∞
+ ∑  a n ( n + 2)( n + 1)r n + b n n( n − 1)r n − 2 + c n ( n − 2)( n − 1)r − n + d n n( n + 1)r − n − 2  cos nθ (d62)
n =1
∞
+ ∑  α n ( n + 2)( n + 1)r n + β n n( n − 1)r n − 2 + γ n ( n − 2)( n − 1)r − n + δ n n( n + 1)r − n − 2  sin nθ
n =1
At great distances from the cylindrical cavity, the elastic perturbation has to vanish, i.e.
lim σ θθ = 0 . (d63)
r→∞
It follows
a 0 = 0 , 2 b 0 + 2 b 2 + 2β 2 = 0 , a n = α n = 0 , n = 1,2,...
(d64)
b n = β n = 0 , n = 3,4,...
Hence the AIRY’ s potential is
 26
∞
A ( r, θ ) = c 0 ln r + d 0 + (b1 + β1 )r + ∑ (c n cos nθ + γ n sin nθ)r − n + 2
n =1
(d65)
∞
+ ∑ (d n cos nθ + δ n sin nθ)r − n
n =1
and
∞
∑ ( n − 2)( n − 1)(c n cos nθ + γ n sin nθ)r − n
c0
σ θθ = −
2
+
r n =1
. (d66)
∞
+ ∑ n( n + 1)(d n cos nθ + δ n sin nθ)r − n − 2
n =1
From (d65), it follows that
∞
A , r = r + b1 + β1 + ∑ ( − n + 2)(c n cos nθ + γ n sin nθ)r − +
c0 n 1
n =1
, (d67)
∞
+ ∑ ( − n )(d n cos nθ + δ n sin nθ)r − n −1
n =1
∞ ∞
−n + 2 +
A ,θ ∑ n
= n (− c sin nθ + γ n cos nθ )r ∑ n(−d n sin nθ + δ n cos nθ)r − n , (d68)
n =1 n =1
∞ ∞
2 c cos nθ + γ sin nθ r − n + 2 −
A ,θθ ∑
= − n ( n n ) ∑ n 2 (d n cos nθ + δ n sin nθ)r − n . (d69)
n =1 n =1
Hence
A , r + A ,θθ = c 0 + b1 + β1 − ∞
σ rr = ∑  n 2 + n − 2 (c n cos nθ + γ n sin nθ)r − n
r r2 r2 r
n =1
(d70)
∞
− ∑  n 2 + n (d n cos nθ + δ n sin nθ)r − n − 2
n =1
From (d70), it follows that
lim σ rr = 0 . (d71)
r →∞
Also,
∞ ∞
− n +1 +
A , rθ = ∑ n ( n − 2 )(c n sin nθ − γ n cos nθ )r ∑ n 2 (d n sin nθ − δ n cos nθ)r − n −1 . (d72)
n =1 n =1
Hence
A , rθ + A ,θ
σ rθ = − r r2
(d73)
∞ ∞
=− ∑ ( n 2 − n )(c n sin nθ − γ n cos nθ)r − n − ∑  n 2 + n (d n sin nθ − δ n cos nθ)r − n − 2
n =1 n =1
From (d73), it follows that
lim σ rθ = 0 . (d74)
r →∞
 27

D.9) Strain and displacement vector. Conditions at infinity.

From (d39) and HOOKE’ s reversed law it follows that

ε =
1
E 
(1 + ν ) σ − νtr σ 1  =
1+ ν 
E σ
−ν (σ rr + σ θθ)1  (d75)

i.e.
1 − ν2  ν  ∂u
ε rr = E σ rr − 1 − ν σ θθ = ∂rr , (d76)

Hence
∂u r ∞
1 b1 + β1  ν 
( n − 2 )(c n cos nθ + γ n sin nθ)r − n
E c
= 0 + − ∑ ( n − 1) n + 2 +
1 − ν ∂r 1 − ν r
2 2 r
n =2
 1− ν 
(d77)
∞
∑ n(n + 1)(d n cos nθ + δ n sin nθ)r − n − 2
1
−
1− ν
n =1
Integrating (d77) it follows
∞
 ν 
n − 2)(c n cos nθ + γ n sin nθ)r − n +1
E c0 1
ur = − + (b1 + β1 ) ln r + ∑  n + 2 + (
1 − ν2 1− ν r  1− ν 
n=2
(d78)
∞
n(d n cos nθ + δ n sin nθ)r − n −1 + ϕ(θ )
1
+
1− ν
∑
n =1
From
lim u r = 0 , (d79)
r →∞
it follows that
b1 + β 1 = 0 , ϕ ( θ ) ≡ 0 . (d80)
Hence
∞
 ν 
n − 2)(c n cos nθ + γ n sin nθ)r − n +1
E c0 1
ur = − + ∑ n + 2 + (
1 − ν2 1− ν r  1− ν 
n=2
. (d81)
∞
∑ n(d n cos nθ + δ n sin nθ)r − n −1
1
+
1− ν
n =1
Finally
∞
1 + ν  c0
ur = − + ∑ (n + 2 − 4 ν)(c n cos nθ + γ n sin nθ)r − n +1
E  r
 n =2
. (d82)
∞ 
+ ∑ n(d n cos nθ + δ n sin nθ)r − n −1 

n =1 
In the same way,
1 − ν2  ν  ∂u
ε θθ = E σ θθ − 1 − ν σ rr  = r∂θθ + rr
u
. (d83)

It follows
E ∂u θ E ν
u r + r σ θθ −
1 − ν σ rr
=− r . (d84)
1 − ν 2 ∂θ 1 − ν2
Substituting (d82), (d66) and (d70) into (d84) and integrating with respect to θ , it follows after some computations that
 28
∞
1+ ν 
uθ = (1 − ν)ψ ( r ) + ∑ (n − 4 + 4 ν)(c n sin nθ − γ n cos nθ)r − n +1
E 
 n=2
, (d85)
∞ 
+ ∑ n (d n sin nθ − δ n cos nθ)r − n −1 

n =1 
Form the condition
lim u θ = 0 , (d86)
r →∞
it follows the unknown function ψ = ψ ( r ) is subject to the condition
lim ψ (r ) = 0 . (d87)
r →∞
But
1+ ν 1  ∂u ∂u 
=  r − θ + θ ,
u
ε rθ = E σ rθ 2  r∂θ r ∂r 
(d88)

or
E  ∂u r ∂u 
2r σ rθ =  − uθ + r θ  . (d89)
1 + ν  ∂θ ∂r 
Substituting (d73), (d82) and (d85) into (d89), it follows after some computations that
dψ
r =ψ , (d90)
dr
i.e. ψ = Cr . From (d87) it follows that ψ( r ) ≡ 0 and, finally,
∞ ∞
1+ ν  
uθ =  ∑ (n − 4 + 4 ν)(c n sin nθ − γ n cos nθ)r − n +1 + ∑ n (d n sin nθ − δ n cos nθ)r − n −1 
E  
n = 2 n =1 
(d91)

D.10) Boundary conditions for the stress elements on the wall of the circular cavity.
Using the previous results, the final expressions of the plane elements of the stress are equal to
∞
σ rr 2 ∑  n 2 + n − 2 (c n cos nθ + γ n sin nθ)r − n
c0
= −
r n =1
(d92)
∞
− ∑  n 2 + n (d n cos nθ + δ n sin nθ)r − n − 2
n =1
∞
∑ ( n − 2)( n − 1)(c n cos nθ + γ n sin nθ)r − n
c0
σ θθ = − +
r2 n =1
(d93)
∞
+ ∑ n( n + 1)(d n cos nθ + δ n sin nθ)r − n − 2
n =1
and
∞ ∞
2 −
σ rθ = − ∑ ( n − n )(c n sin nθ − γ n cos nθ)r − ∑  n 2 + n (d n sin nθ − δ n cos nθ)r − n − 2 (d94)
n
n =1 n =1
→
The cavity wall has the outer normal (with respect to the rock domain) equal to − e and radius equal to r = R . In the case
r
of a bore hole, let ∆p be the difference between the mud pressure and the pressure of the fluid contained by the porous rock
 29
(usually, because the atmospheric pressure is negligible, it follows in the case of a tunnel that ∆p = 0 ). It follows the final

σ
stress
f satisfies the next boundary condition

 → →
f  −  = ∆p
σ  er  er , for r=R (d92)
 
Hence

σ rr r=R
= − σ 0rr − ∆p
 . (d93)
σ = − σ 0rθ
 rθ r = R

σ12 = 0 . Hence
With no loss of generality, it can be assumed that the stress at infinity is along its main axes, i.e.
0

c ∞ ∞
 0 − ∑  n 2 + n − 2 (c n cos nθ + γ n sin nθ)R − n − ∑  n 2 + n (d n cos nθ + δ n sin nθ)R − n − 2
R2    
 n = 2 n =1

σ σ σ σ 22 cos 2θ − ∆p
0 + 0 0 − 0
 =− 11 22 − 11
 2 2

 ∞ ∞
 n 2 − n (− c sin nθ + γ cos nθ)R − n +
∑   n n ∑  n 2 + n (−d n sin nθ + δ n cos nθ)R − n − 2
n = 2 n =1

σ σ 22 sin 2θ
 0 − 0
= 11

 2
(d94)
Hence

 c 0 = − σ11 σ 22 − ∆p
0 + 0

R2 2 , (d95)

d1 = δ1 = 0
and
∞    γ δ  

∑  n 2 + n − 2 nn +  n 2 + n nn+ 2  cos nθ +  n 2 + n − 2 nn +  n 2 + n nn+ 2  sin nθ
c d

n = 2  R R   R R  

σ − σ 22 cos 2θ
0 0
= 11
2
(d96)

∞     γ δ  

∑ −  n 2 − n nn +  n 2 + n nn+ 2  sin nθ +  n 2 − n nn +  n 2 + n nn+ 2  cos nθ
c d

n=2   R R   R R  
. (d97)

σ σ 22 sin 2θ
0 − 0
= 11
2
It follows that
 30

 c 0 = − σ11 σ 22 − ∆p
0 + 0

R2 2

 c d 2 σ11 − σ 22
0 0
 2
4 +6 = . (d98
 R2 R4 2

d 2 σ11 − σ 22
0 0
 c2
 −2 2 − 6 4 = 2
 R R
and
 2 c dn
( n + n − 2) n + ( n 2 + n ) = 0 , n = 3,4,...
 Rn R n +2
 2 γ δn
( n + n − 2) n + ( n 2 + n ) = 0 , n = 2,3,...
 Rn R n +2
 (d99)
( n 2 c dn
− n) n + (n 2 + n) = 0 , n = 3,4,...
 Rn R n+2

( n 2 γ δn
− n) n + (n 2 + n) = 0 , n = 2,3,...
 n n + 2
 R R
Hence
  0 
c = − σ11 + σ 22 + ∆p R 2
0
 0  2 
  

σ σ 22 R 2 , c = 0 , n = 3,4,...
0 − 0
 11
c 2 = n . (d100)
 2
 4
d = − 0 − 0  R
 σ11 σ 22 4
, d n = 0 , n = 3,4,...
 2

γ n = 0 , δ n = 0 , n = 2,3,...
σ12 = 0 , the expressions of the final stress are equal to
By using (d11-d13) for
0

σ σ σ σ 22 cos 2θ
0 + 0 0 − 0
σ rr
f = 11 22 + 11
2 2
 0 + 0
σ 22 + ∆p R 2 − 2 0 − 0  R 2 cos 2θ + 3  0 − 0  R 4 cos 2θ
(d101)
σ
 σ11 σ 22 2 2  σ11 σ 22 r 4
−  11  2
2
  r r
and

σ11 + σ 22 − σ11 − σ 22 cos 2θ +  σ11 + σ 22 + ∆p R 2 − 3 

0 0 0 0 0 0 4
0 − 0  R cos 2θ (d102)
σ θθ = 2  σ11 σ 22 r 4
f
2 2  2  2
  r

f = − σ11 − σ 22 sin 2θ −  0 − 0  R sin 2θ + 3  0 − 0  R sin 2θ

0 0 2 4
σ rθ 2  σ11 σ 22 2 2  σ11 σ 22 r 4
(d103)
r
In real cases, the value of the stress at infinity are positive ones for compression.
 31
D.11) The final shape of the wall.
Consider again the case when the direction of the horizontal axes of the co-ordinate system is along the corresponding eigen

σ σ12 = 0 . Using the above results, it follows the displacement vector for the
vectors of the initial stress
0 . In that case, 0
points initially placed on the wall of the circular cavity is

  0 
u ( r = R, θ ) = 1 + ν R  σ11 + σ 22 + ∆p + (3 − 4ν ) σ11 − σ 22 cos 2θ
0 0 0
 r E  2 2 
  
 . (d104)

σ σ 22 sin 2θ
0 − 0
 1+ ν 11
u θ ( r = R, θ ) = − E R(3 − 4 ν) 2

Consider an arbitrary point on the wall of the bore hole. In the initial state, it has the polar co-ordinates (r = R, θ) . Its
position vector with respect to the centre of the circle is
→  → →
X = R cos θ e1 + sin θ e2 
 (d105)
 
Using (d6) the position vector in the final stage is
→ → →  → → →
χ = X + u r ( R, θ) e r + u θ ( R, θ) eθ = R cos θ e1 + sin θ e2 

 
 0  → →
1 + ν  σ11 + σ 22 σ σ
0 0 − 0
+ R + ∆p + (3 − 4 ν) 11 22 cos 2θ cos θ e1 + sin θ e 2  .
  (d106)
E  2 2  
  
 → →
σ σ
0 − 0
1+ ν
− R (3 − 4 ν ) 11 22 sin 2θ − sin θ e1 + cos θ e 2 

E 2  
 
Hence
  0 
x = R cos θ + 1 + ν R  σ11 + σ 22 + ∆p + (3 − 4 ν) σ11 − σ 22 cos 2θ cos θ
0 0 0
 1 E  2 2 
  

σ σ 22 sin 2θ sin θ
0 − 0
 1+ ν 11
 + R (3 − 4 ν )
 E 2
 . (d107)
 0
 1 + ν  σ11 + σ 22 σ σ 22 cos 2θ sin θ
0 0 − 0
x 2 = R sin θ + R + ∆p + (3 − 4ν) 11
 E  2 2 
 
 

σ σ 22 sin 2θ cos θ
0 − 0
 1+ ν 11
 − R( 3 − 4 ν )
E 2

After elementary computations, it follows

 32
   0 0 
x = 1 + 1 + ν  σ11 + σ 22 + ∆p + (3 − 4 ν) σ11 − σ 22  
0 0
 1  E  2 2   X1
    
 . (d108)
   0 + 0 0 − 0 

x 2 = 1 + 1 + ν  σ11 σ 22 + ∆p − (3 − 4 ν)
σ11 σ 22  
  E  2 2   X2
    
2 2 2 , it follows the final shape of the cavity is an ellipse of equation
Taking into account that X1 + X 2 = R
x12 x 2
+ 2 =1 , (d109)
a 2 b2
where the semi-axes of the ellipse are equal to

   0 0 
a = R 1 + 1 + ν  σ11 + σ 22 + ∆p + (3 − 4 ν) σ11 − σ 22  
0 0
  E  2 2 
    
 . (d110)
   0 + 0 0 − 0 
b = R 1 + 1 + ν  σ11 σ 22 + ∆p − (3 − 4 ν) σ11 σ 22  
  E  2 2 
    

Fig.D2. The shape of the borehole (tunnel) in the initial state (the circle) and in the final state (the ellipse), corresponding to a
compressive stress.

σ
In real cases, the initial stress
0 is usually a compressive one, i.e. (see Fig.D2)

σ11 = −S1 , σ 22 = −S3

0 0 (d111)
where the maximum compressive stress S1 and the minimum compressive stress S 3 have positive values 0 ≤ S 3 ≤ S1 . It
follows here that the major semi-axis a corresponds to the minimum stress and a ≤ b .
 33
E) BOUSSINESQ’S PROBLEM - the concentrated force acting on the elastic semi-
space
E.1) The equations of BELTRAMI and MITCHELL.

It follows to obtain the partial derivative equations for the stress tensor σ in the particular case of an elastic, homogeneous,

isotropic media. In the beginning, the next symmetric tensor is evaluated

 → →
S = − grad (ρ b ) + grad t (ρ b ) , (e1)
 
Using the equilibrium equation, it follows that
→
div σ = − ρ b (e2)

Hence
S = grad div σ  + grad t  div σ 
   
.
(e3)

The HOOKE’s reversed law gives

1+ ν ν
ε =
E σ − Θ1
E
, (e4)

so
E ν
σ =
1+ ν ε + Θ
1+ ν 1
, (e5)

where
1 − 2ν
Θ = tr σ , ϑ = tr ε = Θ . (e6)
E
But
div ( f 1 ) = grad f . (e7)

Hence
E ν
div σ = div ε + grad Θ (e8)
1+ ν 1+ ν
Expression (e1) becomes
E  ν 
S = grad ( div ε ) + grad t ( div ε ) + grad ( grad Θ ) + grad t ( grad Θ ) , (e9)
1+ ν    
 1+ ν  
Because the tensor grad( grad Θ ) is a symmetric one, it follows that
2ν
S = 1 + ν grad( div ε ) + grad ( div ε ) + 1 + ν grad(grad Θ ) . (e10)
E t

The tensor in the first parenthesis of (e10) has the ij - component equal to
grad( div ) + grad t ( div ) = ( div ) + ( div ) =
 ε ε  ij ε i, j ε j, i ε iq ,qj + ε jq ,qi =
1
u + + +  = 1  +  +    =
2  i,qqj u q , iqj u j, qqi u q , jqi 2  i, j

u u j, i , qq 2 u
 q , q  , ij

(e11)

 ∆  +  tr  =  ∆  + 1 − 2 ν Θ =  ∆  + 1 − 2 ν [grad ( gradΘ )]
 ε  ij  ε  , ij  ε  ij E
, ij  ε 
ij E ij
Hence
1 − 2ν
grad ( div ε ) + grad t ( div ε ) = ∆ ε + grad ( gradΘ ) (e12)
E
Using (e12) and (e10), the expression (e1) becomes
 34
E 1
S = ∆ε + grad gradΘ (e13)
1+ ν 1+ ν
Applying the 3-dimensional LAPLACE operator ∆ in (e4), it follows
1+ ν ν
∆ε = ∆ σ − ∆Θ1 (e14)
E E
Replacing (e14) into (e13) gives
ν 1
S = ∆σ − ∆Θ1 + grad gradΘ , (e15)
1+ ν 1+ ν
i.e.
ν 1  → →
∆σ − ∆Θ1 + grad gradΘ = − grad(ρ b ) + grad t (ρ b ) (e16)
1+ ν 1+ ν  
But
tr ( ∆ σ ) = ∆( tr σ ) = ∆ Θ ,
tr ( grad grad Θ ) = ∆ Θ ,
tr ( ∆Θ1 ) = 3∆ Θ , (e17)

 → →  → →
tr grad(ρ b ) + grad t (ρ b ) = 2tr grad (ρ b ) = 2 div(ρ b )
   
Applying the trace operator in (e16) and using eqs.(e17) gives
1 3ν →
∆Θ + ∆Θ − ∆Θ = − 2 div(ρ b ) (e18)
1+ ν 1+ ν
It follows the trace of the stress tensor verifies the relation
1+ ν →
∆Θ = − div(ρ b ) (e19)
1− ν
Replacing (e19) into (e16) gives
ν → 1  → →
∆σ + div (ρ b )1 + grad gradΘ = − grad (ρ b ) + grad t (ρ b ) (e20)
1− ν 1+ ν  
Eqs.(e19)-(e20) represents the equations of BELTRAMI and MITCHELL, having as unknowns only the elements of the stress
tensor . Together with appropriate conditions (in tensions) on the boundary of the elastic body, they allow one to solve the
corresponding linear static problem.
Particular cases.
a) Suppose that
→
div (ρ b ) = 0 , (e21)
→
i.e. a vector potential ψ exists having the property that
→ →
ρ b = rot ψ , (e22)
From (e19) it follows that the traces of both stress and strain tensors are harmonic functions
∆Θ = ∆θ = 0 , (e23)
b) Suppose that
→
ρ b = grad ϕ , where ∆ϕ = 0 , (e24)
it follows that
→
div (ρ b ) = ∆ϕ = 0 , (e25)
i.e. eq. (e23) is verified and eq.(20) gives
1
∆σ + grad gradΘ = − 2grad ( gradϕ ) (e26)
1+ ν
 35
Applying the LAPLACE operator ∆ to eq.(e26), it follows that the stress tensor is the solution of the þi-harmonic equation
∆∆ σ = 0 . (e27)

In most real cases, the volume forces are neglected (or they are represented only by the weight of the body, satisfying eq.(24). It
follows the elastic linear problem involve solving harmonic and bi-harmonic equations.

E.2) The model.

A co-ordinate system having the third vertical axis positive downward is used. The semi-space x 3 ≥ 0 is represented by an
elastic, homogeneous, isotropic medium having the elastic coefficients λ and µ (or E and ν respectively). In the origin of the
co-ordinate system is acting a vertical force having the magnitude equal to P . It follows to find the stress and the
displacements. Spherical co-ordinates will be used (Fig.E1):
x1 = r sin ϑ cos λ , x 2 = r sin ϑ sin λ , x 3 = r cos ϑ (e28)

Fig.E1 The spherical co-ordinate system and a vertical force of magnitude equal to P acting at the origin.

Because symmetry, the displacement vector has the components like

u r = u r ( r, ϑ ) , u λ = 0 , u ϑ = u ϑ ( r, ϑ ) (e29)
It follows the components of both stress and strain tensors are functions of r and ϑ only.

E.3) The equations of equilibrium and strain tensor in spherical co-ordinates.

The LAMÉ differential parameters for spherical co-ordinates are
1 = 1, 2=r, 3 = r sin ϑ
h h h (e30)
The generalised curvilinear co-ordinates are
1= r , 2=ϑ, 3= λ
c c c (e31)
The unit vectors of the axis are
→ → → → → →
1= 2 3
n e r , n = eϑ , n = eλ (e32)

By using, for example, (IVAN, 1996), the divergence of a symmetric tensor T in spherical co-ordinates is

div T =

 ∂ T rr 1 ∂ T rϑ
 + +
1 ∂ T rλ 2 T rr
+ +
Trϑ − Tϑϑ + Tλλ  → +
 ∂r r ∂ϑ r sin ϑ ∂ λ r tan ϑ  er
 r r 
 ∂ T rϑ 1 ∂ Tϑϑ →
1 ∂ T ϑλ 3 T rϑ Trϑ + Tϑϑ − Tλλ  (e33)
 + + + +
 ∂r r ∂ϑ r sin ϑ ∂ λ r tan ϑ r tan ϑ  eϑ +
 r 
 ∂ T rλ 1 ∂ Tϑλ 1 ∂ Tλλ 3 T rλ 2Tϑλ  →
 + + + + e
 ∂r r ∂ϑ r sin ϑ ∂ λ r tan ϑ  λ
 r 
In the same way, the gradient of a vector in spherical co-ordinates is
 36
→ ∂ v r → → 1 ∂ v r vϑ → →  1 ∂ v r vλ → →
v = ⊗ + −  ⊗ + −  ⊗ +
∂ r e r e r  r ∂ ϑ r  e r e ϑ  r sin ϑ ∂ λ r  e r e λ
grad

∂ vϑ → → 1 ∂ vϑ v r → → 
1 ∂ vϑ v λ  → ⊗ → +
⊗ +  +  ⊗ +  −
∂ r e ϑ e r  r ∂ ϑ r  e ϑ e ϑ  r sin ϑ ∂ λ r tan ϑ  e ϑ e λ
(e34)

∂ vλ → → 1 ∂ vλ → → 
1 ∂ vλ vr v ϑ  → →
⊗ + ⊗ +  + +  ⊗
∂ r e λ e r r ∂ ϑ e λ e ϑ  r sin ϑ ∂ λ r r tan ϑ  e λ e λ
The components of the strain tensor are
∂ ur 1 1 ∂ u r uϑ ∂ u ϑ
ε rr = ∂r
, ε rϑ = 
2  r ∂ ϑ
−
r
+ ,
∂ r 

1  1 ∂ ur uλ ∂ uλ 1 ∂ uϑ ur
ε rλ 2  r sin ϑ ∂ λ r
=  − +
∂ r  ε ϑϑ r ∂ ϑ
 , = +
r
, (e35)

1  1 ∂ uϑ u λ + 1 ∂ u λ  , 1 ∂ uλ ur uϑ
ε ϑλ = 

2  r sin ϑ ∂ λ
−
r tan ϑ r ∂ ϑ   ε λλ =
r sin ϑ ∂ λ
+
r
+
r tan ϑ
In the particular case of eqs.(e29), it follows
∂ ur 1 1 ∂ u r u ϑ ∂ uϑ
ε rr = ∂r
, ε rϑ 2  r ∂ ϑ − r + ∂ r  ,
= ε rλ = 0
(e36)
1 ∂ uϑ ur , ur + uϑ
ε ϑϑ = r ∂ϑ
+
r ε ϑλ = 0, ε λλ = r r tan ϑ
Hence
σ rλ = σ ϑλ = 0 , (e37)
By using (e33), the equilibrium equations in the absence of the volume forces are

∂ σ rr 1 ∂ σ rϑ 3 σ rr σ rϑ = σ rr + σ ϑϑ + σ λλ ,
+ + + (e38)
∂r r ∂ϑ r r tan ϑ r
∂ σ rϑ 1 ∂ σ ϑϑ 3 σ rϑ σ rϑ + σ ϑϑ − σ λλ = 0
+ + +
∂r r ∂ϑ r r tan ϑ r tan ϑ
i.e.
∂  3 ∂  2
 r sin ϑ σ  +  r sin ϑ σ  = r 2 sin ϑ Θ ,
∂r  rr  ∂ϑ rϑ
(e39)
∂  3 ∂  2
 r sin ϑ σ  +  r sin ϑ σ  = r 2 cos ϑ σ
∂r  rϑ  ∂ϑ ϑϑ λλ

E.4) LAPLACE operator in spherical co-ordinates. LEGENDRE’s polynomials.

Using, for example, (IVAN, 1996) the gradient of a scalar function in spherical co-ordinates is
∂f → 1 ∂f → 1 ∂f →
grad f = + +
∂ r e r r ∂ ϑ e ϑ r sin ∆ ∂ λ e λ
. (e40)

Also, the LAPLACE operator is

1 ∂  2 ∂f ∂  ∂f  1 ∂2f 
∆ f = div grad f = sin ϑ r +  sin ϑ +  (e41)
2 
r sin ϑ  ∂ r  ∂ r  ∂ ϑ  ∂ ϑ  sin ϑ ∂λ 
2
Consider the LAPLACE equation
 37
∆f =0 (e42)
for the particular case when the unknown function has the form f = f (r, ϑ ) . Equation (e42) becomes

∂  2 ∂f 1 ∂  ∂f 
r +  sin ϑ  = 0. (e43)
∂ r  ∂ r  sin ϑ ∂ ϑ  ∂ ϑ
By using the method of separation of the variables, a solution for eq.(43) has the form
( )
f r , ϑ = R ( r ) Y( ϑ ) , (e44)
where R and Y are two unknown functions. Eq.(e43) becomes
d  2 dR  d  dY 
r   sin ϑ 
dr  dr  dϑ  dϑ 
=− = k, (e45)
R Y sin ϑ
where k is a constant. From (e45) it follows that
d 2R dR
r2 + 2r − kR = 0 (e46)
dr 2 dr
In a general case, the function R can be developed in power series. Let’s look for a particular solution having the form
R n ( r ) = r n where n is a natural number. It follows from (e46) that
k = n ( n + 1) (e47)
Then the particular solution of eq. (e46) can be expressed with the aid of two arbitrary constants as
R n (r) = A n r n + Bn / r n (e48)
Because f has to approach finite values for r → ∞ , it has to take A n = 0 .
The second relation (e45) gives
d 2Y dY
sin ϑ + cos ϑ + n( n + 1)Y sin ϑ = 0 (e49)
dϑ 2 dϑ
By performing the substitution z = cos ϑ eq.(e49) becomes
d  dY 
1 − z 2  + n ( n + 1)Y = 0 ,

dz   dz 
(e50)

The solution of (e50) is represented by the LEGENDRE polynomials denoted by Pn ( z ) , n = 0,1,2,... So,

Po ( z) = 1 , P1 ( z ) = z , P2 ( z ) = (3z 2 − 1) / 2 (e51)
Because the trace Θ of the stress tensor is a solution of the harmonic equation (e23), it follows that the general solution for that
trace in the case of the BOUSSINESQ problem is
∞ an
Θ= ∑ n + 1
Pn (cos ϑ ) . (e52)
n=0 r
According to eq.(e6), a similar solution exists for the trace of strain tensor.

E.5) The displacement field.

We look for a displacement field having the form
1 1
u r = r ϕ( ϑ ) , u ϑ = r ψ( ϑ ) , uλ = 0 , (e53)

where ϕ and ψ are two unknown functions following to be obtained. Substituting (e53) into (e36) it follows that

ε rr = − ϕ / r 2 , ε rϑ = (dϕ / dϑ − 2ψ ) /  2r 2  , ε rλ = 0

(e54)
ε ϑϑ = (ϕ + dψ / dϑ) / r 2 , ε ϑλ = 0 , ε λλ = (ϕ + ψ / tgϑ) / r 2
It follows that the trace of the strain tensor is
θ = tr ε = ε rr + ε ϑϑ + ε λλ = (ϕ + dψ / dϑ + ψ / tgϑ ) / r 2 (e55)

A comparison of (e55) to (e52) shows that

 38
ϕ + dψ / dϑ + ψ / tgϑ = a cos ϑ , (e56)
where a is a constant remaining to be obtained. Hence
θ = a cos ϑ / r 2 , Θ = (3λ + 2µ )θ = (3λ + 2µ ) a cos ϑ / r 2 ,

( )
(e57)
ε λλ = θ − ε rr + ε ϑϑ = (a cos ϑ − dψ / dϑ) / r 2
By using HOOKE’s law, it follows that

σ rr = (λ a cos ϑ − 2µϕ) / r 2 , σ rϑ = µ(dϕ / dϑ − 2ψ ) / r 2 ,

σ ϑϑ = [λ a cos ϑ + 2µ(ϕ + dψ / dϑ )] / r 2 , (e58)

σ λλ = [( λ + 2µ ) a cos ϑ − 2µdψ / dϑ] / r 2

Substituting (e58) into the first equilibrium equation (e39) and using (e56), it follows after elementary computations that
d  dϕ 
 sin ϑ  = a ( 2 + λ / µ ) sin 2ϑ . (e59)
dϑ  dϑ 
Hence
dϕ a a
sin ϑ = − ( 2 + λ / µ ) cos 2ϑ + b = b − ( 2 + λ / µ ) + a ( 2 + λ / µ ) sin 2 ϑ ,
dϑ 2 2
(e60)
dϕ b − a ( 2 + λ / µ ) / 2
= + a ( 2 + λ / µ )sin ϑ
dϑ sin ϑ
where b is an integration constant. Because
dx x
∫ sin x = ln tg
2
+C , (e61)

where C is a new integration constant, it follows that

 a  ϑ
ϕ = b − ( 2 + λ / µ ) ln tg − a ( 2 + λ / µ ) cos ϑ + C . (e62)
 2  2
For ϑ → π / 4 , the logarithmic term into (e62) leads to infinite radial displacements. That can be avoided by taking
a
b − (2 + λ / µ ) = 0 . (e63)
2
Hence
dϕ
ϕ = − a ( 2 + λ / µ ) cos ϑ + C , = a ( 2 + λ / µ ) sin ϑ . (e64)
dϑ
and

σ rr = [(3λ + 4µ)a cos ϑ − 2µC] / r 2 . (e65)

Substituting eq.(e64) into eq.(e56), it follows that

dψ ψ
+ = a (3 + λ / µ ) cos ϑ + C ,
dϑ tgϑ
d a
(ψ sin ϑ) = (3 + λ / µ ) sin 2ϑ − C sin ϑ ,
dϑ 2 (e66)
a
ψ sin ϑ = − (3 + λ / µ ) cos 2ϑ + C cos ϑ + D
4
a a
= D − (3 + λ / µ ) + (3 + λ / µ )sin 2 ϑ + C cos ϑ
4 2
Hence
D − a (3 + λ / µ ) / 4 a
ψ= + (3 + λ / µ )sin ϑ + C ctgϑ . (e67)
sin ϑ 2
Because the tangential displacements has to be finite ones for ∆ → 0 , it follows that
 39
cos ϑ − 1 a ϑ a
ψ=C + (3 + λ / µ )sin ϑ = − C tg + (3 + λ / µ )sin ϑ ,
sin ϑ 2 2 2
(e68)
dψ C a
=− + (3 + λ / µ ) cos ϑ
dϑ 2
2 cos ( ϑ / 2 ) 2
By substituting eqs.(e64) and (e68) into eq.(e58) it follows that
 ϑ    ϑ 
σ rϑ = µ 2C tg 2 − a sin ϑ / r 2 , σ ϑϑ = µ C1 − tg 2 2  − a cos ϑ  / r 2 ,
(e69)
  ϑ 
σ λλ = µ C1 + tg 2 2  − a cos ϑ  / r 2
Substituting eqs.(e69) into the second equilibrium equation (e39) it can be seen that the last one is identically verified.

E.6) Boundary conditions for stress elements. The final solutions.

Eqs.(e65) and (e69) contain the unknown coefficients a and C . These constants follow to be obtained taking into account
that the force P concentrated in the origin of the co-ordinate system is acting on the elastic semi-space. It can be seen that the
points of the horizontal plane x 3 = 0 have the co-latitude ϑ = π / 2 .
The unit vectors of the spherical co-ordinate system are related to the same vectors of the rectangular co-ordinate system by
→  →  →  → 
       
er  e 1  e 1 er
→  →  →  t → 
  =Q   ,   =Q   , (e70)
 e ϑ   e 2  e 2  e ϑ
→  →  →  → 
       
 e λ  e 3  e 3  e λ
where the orthogonal matrix is
 sin ϑ cos λ sin ϑ sin λ cos ϑ
 
Q =  cos ϑ cos λ cos ϑ sin λ − sin ϑ . (e71)
 
 − sin λ cos λ 0 
The outer pointing unit vector normal to the elastic semi-space is equal to
→ → →
− e3 = − cos ϑ e r + sin ϑ e ϑ . (e72)
The resulting exterior force acting on the elastic semi-space is vanishing for all the points of the horizontal plane ϑ = π / 2 ,
excepting the origin, i.e.
 → →
−  = 0 .
σ  e3
(e73)
 
Substituting (e72) into (e73) for ϑ = π / 2 gives
σ ϑϑ = 0 , σ rϑ = 0 . (e74)
By using (e69), the first eq.(e74) becomes an identity and the second one leads to
a = 2C . (e75)
So, eqs.(e65) and (e69) give

σ rr = 2C[(3λ + 4µ) cos ϑ − µ] / r 2 , (e76)

ϑ  ϑ 
σ rϑ = −2Cµ tg 2 cos ∆ / r 2 , σ ϑϑ = Cµ cos ϑ tg 2 2 − 1 / r 2 . (e77)

Consider an elastic hemisphere having the centre at the origin of the co-ordinate system. The curved surface S of the
→
hemisphere has the outer pointing normal equal to e r . On that surface, the rest of the elastic body (i.e. the semi-space minus
the hemisphere) is acting on the hemisphere with a total force equal to
 40
→  → →
∫∫ σ e r dA = ∫∫  σ rr e r + σ rϑ e ϑ dA ,
 (e78)
S S 
where dA is the surface element and the unit vectors are obtained with (e70). It follows that
→ 2π π / 2 →
∫∫ σ e r dA = ∫ dλ ∫ σ e r r sin ϑdϑ
2
S 0 0
(e79)
π/2 → →
 ϑ 2 
= 4πC ∫ [(3λ + 4µ) cos ϑ − µ]cos ϑ sin ϑ + µ tg 2 cos ϑ sin ϑ dϑ e3 = 4πC( λ + µ ) e3
0
Because the hemisphere is into an equilibrium state, it follows that
→ → →
∫∫ σ er dA + P e3 = 0 , (e80)
S
Hence
P
C=− . (e81)
4π( λ + µ )
Finally, the non-zero components of the stress tensor are equal to

σ rr = − 2π( λ + µ ) [(3λ + 4µ) cos ϑ − µ] / r 2 ,

P
(e82)

P ϑ
σ r∆ = 2π( λ + µ ) µ tg 2 cos ϑ / r 2 , (e83)

P  ϑ
σ ϑϑ = 4π( λ + µ ) µ cos ϑ1 − tg 2 2  / r 2 . (e84)

The non-zero components of the displacement vector are

P
u r = 4π( λ + µ ) [2(2 + λ / µ ) cos ϑ − 1] / r ,
(e85)
P  ϑ 
u ϑ = 4π( λ + µ ) tg 2 − (3 + λ / µ )sin ϑ  / r
Using eq.(e71), the components of the stress tensor into the Cartesian base can be obtained as
σ
 rr σ rϑ
σ σ12 σ13  0 
 11 
  = Qt  Q
 σ 22 σ 23  σ rϑ σ ϑϑ
0

(e86)
   0
 σ 33  0 σ λλ
Also, the components of the displacement vector into the Cartesian base are
u  ur 
 1  
u  = Qt u  . (e87)
 2  ϑ
   0 
 u 3
Of particular importance in real life are the components

3P 3 5 1+ ν P  1 z2 
σ 33 = − 2π
z /r , u3 = E 2π 
2(1 − ν ) +  . (e88)

r r 
3
The BOUSSINESQ problem has a great importance in Geomechanics, in relation to the computation of a building foundation.
The above solution derived for a concentrated vertical force can be used in the case of arbitrary vertical forces (spread on a
certain domain) by assuming the principle of the superposition.
 41
F) ELEMENTS OF THIN PLATES THEORY
F.1) The model of a thin elastic plane plate.
The thin plane plate is a cylindrical body having an arbitrary horizontal cross section in the initially non-deformed state. Its
height, denoted by H = 2 h , is much smaller than the other dimensions (usually, around 7-10 times). The material of the plate
is an elastic, homogenous, isotropic one, having the constants denoted by E and ν , respectively λ and µ . The third axis of
the co-ordinate system, denoted by Oz , is a vertical one, positive downward. Let be x1 = x , x 2 = y . In the initial, non-
deformed state, the plate is a plane horizontal one. The upper face has the equation z = − h and the down face has z = h
respectively. The plane of equation z = 0 represents the median plane. After the plate is deformed, it becomes a median
surface.

F.2) The planar state of a plate. The bending state.

Two particular situation for a deformed plate are considered (Fig.F1).

Fig.F1. (a) The non-deformed plate; (b) The plate into a planar state; (c) The plate into a bending state. Here, the median plane
is denoted by m.p. while m.s. is the median surface.

In the first case, the horizontal components of the displacement vector are symmetrical ones with respect to the median plane,
being even functions with respect to the z -variable, while the vertical component of the displacement vector is an anti-
symmetrical one (odd function with respect to z ), i.e.

u k ( x, y,− z) = u k ( x, y, z), k = 1,2 (f1)

u 3 ( x, y,− z) = − u 3 ( x, y,− z).
From eq.(f1b) it follows that
u 3 ( x, y,0) = 0 , (f2)

i.e. the material points placed initially into the median plane have no vertical displacement as a consequence of the deformation
of the plate (the median plane holds a horizontal plane one). This kind of deformation represents the planar state of the plate.
In the second case, the horizontal components of the displacement vector are anti-symmetrical ones (odd functions with respect
to z ), while the vertical component of the displacement vector is a symmetrical one with respect to the median plane (even
function with respect to z ), i.e.
u k ( x, y,− z) = − u k ( x, y, z), k = 1,2 (f3)
u 3 ( x, y,− z) = u 3 ( x, y,− z).
By deformation, the material points initially placed into the median plane are displaced on the vertical direction, having no
horizontal movement. The median plane becomes a median surface. This state represents the bending state of a plate.

F.3) Loads acting on the plate.

 42
For simplicity, the forces acting on the lateral surface of the plate are neglected. On the upper surface of the plate, having the
→ →
→ → s s
equation z = − h and the outward pointing normal vector n = − e3 , it is acting the surface force q = q ( x, y ) . In the
→ →
same way, on the down surface, having the equation z = h and the normal vector n = e3 , it is acting the surface force
→ →
j j
q = q ( x, y ) . Hence
 → →
 −  = s , for z = − h ,
σ  e3  q
(f4)
 
i.e.
s s s
σ13 ( x, y,− h ) = − q 1 ( x, y) , σ 23 ( x, y,− h ) = − q 2 ( x, y) , σ 33 ( x, y,− h ) = − q 3 ( x, y) (f5)
Also
→ →
σ e3 = q j , for z = h , (f6)

i.e.
j j j
σ13 ( x, y, h ) = q 1( x, y) , σ 23 ( x, y, h ) = q 2 ( x, y) , σ 33 ( x, y, h ) = q 3 ( x, y) . (f7)
We shell see that the above presented deformation states are compatible only to certain distributions of volume or surface
forces applied to the plate.

F.4) Odd and even functions for the planar state and for the bending state.

Let f = f ( x, y, z) be a function of three variables, supposed to be smooth enough. It can be seen that
f ( x, y, z) + f ( x, y,− z ) f ( x, y, z ) − f ( x, y,− z )
f ( x, y, z ) = + (f8)
2 2
Let
f ( x, y, z ) + f ( x, y,− z )
f + ( x, y, z ) = , (f9)
2
f ( x, y, z ) − f ( x, y,− z )
f − ( x, y, z ) = (f10)
2
It follows that
f + ( x, y,− z ) = f + ( x, y, z ) , f − ( x, y,− z ) = − f − ( x, y, z) . (f11)
+ −
The function f represents the even part of f (with respect to z -variable), while the function f represents its odd part. It
follows that
+ − − +
 f +  = f+ ,  f −  = f − ,  f +  =  f −  = 0 , (f12)
 
i.e. the even part of the even part is equal to the even part too. A similar relation holds for the odd part. The even part of an odd
part (and the odd part of an even part) is vanishing. For k = 1,2 , x1 = x , x 2 = y , it follows that

 f +  = ∂ f + ( x, y, z ) = 1  ∂ f ( x, y, z ) + ∂ f ( x, y,− z ) = f ( x, y, z ) +
  ,k ∂ x
k

2 ∂ xk ∂ xk


,k [ ] (f13)

Hence, the partial derivative (with respect to a horizontal co-ordinate) of the even part of a certain function is equal to the even
part of that derivative. This property holds for the odd part. So,
+ −
  ,k ( )
 f +  = f, k ( )
,  f −  = f, k
,k
(f14)

Also,
 43

 f +  = 1  ∂ [f ( x, y, z ) + f ( x, y,− z )] = 1 f ( x, y, z ) − f ( x, y,− z ) = f − (f15)

  ,3 2  ∂ z


 2
,3 [ ,3 ,3 ] ( )
Hence the partial derivative of the even part with respect to the vertical co-ordinate is equal to the odd part of the partial
derivative of the function itself with respect to same co-ordinate. In the same way it follows that
+
 f −  = f,3
  ,3 ( ). (f16)

By using the HOOKE’s law and the definition of the strain, it can be resumed that the planar state and the bending state are
characterised by the next components:
+ + −
-for the planar state: - the displacement vector: u1 , u 2 , u 3 ,
- strain tensor: ε
+ + − + − + +
11 , ε 12 , ε 13 , ε 22 , ε 23 , ε 33 , ϑ ,
+ + − + − +
- stress tensor: σ 11 , σ 12 , σ 13 , σ 22 , σ 23 , σ 33 .
− − +
-for the bending state: - the displacement vector: u1 , u 2 , u 3 ,
- strain tensor: ε
− − + − + − −
11 , ε 12 , ε 13 , ε 22 , ε 23 , ε 33 , ϑ ,
− − + − + −
- stress tensor: σ 11 , σ 12 , σ 13 , σ 22 , σ 23 , σ 33 .
F5) Mean value of a function. Equilibrium equations for thin plates.

Let f = f ( x, y, z ) an integrable function with respect to z -variable. The mean value of f computed on the thickness of the
plate is denoted by
+h
1
f ( x, y ) =
2h ∫ f ( x, y, z)dz . (f17)
−h
For an odd function f , its mean value vanishes, i.e.

f − ( x, y ) = 0 , (f18)
It follows that

f = f+ +f− = f+ +f− = f+ (f19)

For a function C = C( x, y ) depending only on the horizontal co-ordinates, eq.(f17) leads to
C = C , zC = 0 . (f20)
Differentiating eq.(f17) with respect to the horizontal co-ordinates, it follows that:
f, k = f ( ), k , k = 1,2 . (f21)

For the vertical derivative, it follows that

+h
∂f
dz = [f ( x, y, h ) − f ( x, y,− h )] = f − ( x, y, z = h ) . (f22)
1 1 1
f,3 =
2h ∫ ∂z 2h h
−h
Consider the function zf ( x, y, z ) . It follows that
zf + = (zf )− , zf − = (zf )+ . (f23)
Its mean value is

zf = z( f + + f − ) = zf + + zf − = zf − = (zf )+ . (f24)
Hence
 44
+h +h +h
1 ∂f 1 ∂  1  +  1
∫ ∫  ∂ z ( zf ) − f dz = 2h ( zf ) − h  − 2h ∫ fdz =
zf,3 = z dz = h
2h ∂z 2h
−h −h −h . (f25)
f ( x, y, h ) + f ( x, y,− h )
− f ( x, y) = f + ( x, y, z = h ) − f
2
Neglecting the volume forces (the weight of the plate itself, for example), the equilibrium equations for the planar state are
 +  +  +  +  −  = 0
 σ11 ,1  σ12 ,2  σ13 ,3

 σ +  +  σ +  +  σ −  = 0 . (f26)
 12 ,1  22 ,2  23 ,3

 −  +  −  +  +  = 0
 σ13 ,1  σ 23 ,2  σ 33 ,3
Let the mean values of the stress components be denoted by
+h
1
Σ ij = σ ij = 2h ∫ σ ij ( x, y, z)dz i , j = 1,2,3 , (f27)
−h
Applying the mean value operator to eqs. (f26) gives

(Σ11),1 + (Σ12),2 + 21h [σ13( x, y, h ) − σ13 ( x, y,− h )] = 0

, (f28)
(Σ12),1 + (Σ 22),2 + 21h [σ 23( x, y, h ) − σ 23( x, y,− h )] = 0
Using eqs.(f5) and (f7) it follows

(Σ11),1 + (Σ12),2 + 21h q1j( x, y) + q1s ( x, y) = 0

. (f29)
( ) ( )1  j s 
Σ12 ,1 + Σ 22 ,2 + 2h q 2 ( x, y) + q 2 ( x, y) = 0
Let
+h
1
Μ ij = z σ ij = 2h ∫ z σ ij ( x, y, z)dz i , j = 1,2,3 . (f30)
−h
Multiplying eqs.(f26) by z and using again the mean value operator, it follows

(Μ13),1 + (Μ 23),2 + σ33

( x, y, h ) + σ 33 ( x, y,− h )
− Σ 33 = 0 . (f31)
2
Hence, using again eqs.(f5) and (f7),
j j
q 3 ( x, y) − q 3 ( x, y)
(Μ13),1 + (Μ 23),2 + 2
− Σ 33 = 0 , (f32)

So, the equilibrium of a thin plate into the planar state leads to eqs.(f29) and (f32).
Consider the weight of the plate, the equation of equilibrium for the plate into the bending state are
 −  +  −  +  +  = 0
 σ11 ,1  σ12 ,2  σ13 ,3

 −  +  −  +  +  = 0
 σ12 ,1  σ 22 ,2  σ 23 ,3
. (f33)

 +  +  +  +  −  + ρg = 0
 σ13 ,1  σ 23 ,2  σ 33 ,3
Here, the density of the plate is ρ and the acceleration of gravity is g .
Proceeding in a similar manner to above, it follows the equations of equilibrium for the thin plate into a bending state are
 45

(Σ13),1 + (Σ 23),2 + 21h q 3j ( x, y) + q 3s ( x, y) + ρg = 0,

(Μ11),1 + (Μ12),2 + 12 q1j( x, y) − q1s ( x, y) − Σ13 = 0 . (f34)

(Μ12),1 + (Μ 22),2 + 12 q 2j ( x, y) − q s2 ( x, y) − Σ 23 = 0

Eq.(f34b) is differentiated with respect to x and eq.(f34c) is differentiated with respect to y. The results are substituted in
eq.(f34a). Hence

(Μ11),11 + 2(Μ12),12 + (Μ 22),22 + 21h q 3j ( x, y) + q s3( x, y) + ρg

. (f35)
1 j s  1 j s 
+ q ( x, y) − q ( x, y) + q ( x, y) − q ( x, y) = 0
2 1 1  ,1 2  2 2  ,2

F.6) Thin plane plate in the bending state.

The component u1 of the displacement vector is developed in power series with respect to z -variable. It follows
∂u
u1( x, y, z) = u1( x, y,0) + z ∂ z1 ( x, y,0)+... (f36)

For the bending state, the first term in eq.(f36) vanishes. Because the thickness of the plate is a small one, only the second term
is kept. So the horizontal components of the displacement vector are
∂ u1 ∂ u2
u1( x, y, z) = z ∂z
( x, y,0) , u 2 ( x, y, z) = z ∂z
( x, y,0) . (f37)

The vertical displacement of the points placed into the median plane is denoted by
w = w( x, y) = u 3 ( x, y,0) , (f38)
i.e. the mean surface has the equation z = w( x, y ) . Here, w represents the arrow of the plate.

F7) BERNOULLI’s hypothesis.

According to BERNOULLI, it is assumed that an arbitrary material segment of the plate, initially perpendicular on the
median plane in the non-deformed state, rests perpendicular on the mean surface in the deformed state. Let A ( x, y, z ) a
certain point of the plate (not placed in the mean plane) and A 0 ( x, y,0 ) its projection on the median plane. So, the segment
→
A 0 A is perpendicular on the median plane. After deformation, the material point A is moving at the point having the co-
( 1 2 3 )
ordinates M x + u ( x, y, z ), y + u ( x, y, z ), z + u ( x, y, z ) , while the point A 0 is moving at the point

( )
M 0 x + u1( x, y,0 ), y + u 2 ( x, y,0 ), w( x, y ) . The horizontal displacements of the points placed in the median plane
are vanishing. So, writing the vector along the line, it follows that
→
M 0M = (u1( x, y, z), u 2 ( x, y, z), z + u 3 ( x, y, z) − w( x, y)) (f39)

Expanding in power series the even function u 3 with respect to z -variable, it follows
2
u 3 ( x, y, z) = w( x, y) + z β( x, y)+... (f40)
Using eqs.(f37) and (f40), it follows that
 46
→  ∂ u1 ∂ u2 
M 0 M =  z ( x, y,0), z ( x, y,0), z + z 3β( x, y)+...
 ∂z ∂z 
(f41)
 ∂ u1 ∂ u2 
= z ( x, y,0), ( x, y,0), 1 + z 2β( x, y)+...
 ∂z ∂z 
Hence the direction is
 ∂ u1 ∂ u2 2 β( x, y )+...
→  ( x , y, 0 ), ( x , y, 0 ), 1 + z
 ∂z ∂z 
M 0M  
= (f42)
→ 2 2
 ∂ u1  ∂ u 2 2
+ 1 + z 2 β( x, y )+...
|| M 0 M||
  +  
 ∂z  
   ∂z 
The normal vector on the median surface is
→  ∂w ∂w   ∂ w
2
 ∂ w
2
n = − ,− ,1   +  +1 . (f43)
 ∂x ∂y   ∂x  ∂y
Comparing eq.(f42) to eq.(f43), the supplemental hypothesis of BERNOULLI is satisfied if
∂ u1 ∂w ∂ u2 ∂w
( x, y,0) = − , ( x, y,0) = − , u 3( x, y, z) = w( x, y) , (f44)
∂z ∂x ∂z ∂y
i.e., by using eq.(f37), the displacement vector has the horizontal components equal to
∂w ∂w
u1( x, y, z) = − z ∂ x , u 2 ( x, y, z) = − z ∂ y , u 3( x, y, z) = w( x, y) . (f45)

From eq. (f45), the components of the strain tensor are

∂2w ∂2w ∂2w
ε11 = − z 2 , ε12 = − z
∂ x∂ y ε 22
, = −z , ε 33 = 0 . (f46)
∂x ∂ y2
The LAPLACE operator in horizontal components is denoted by

* ∂2 ∂2
∆ = + . (f47)
∂ x 2 ∂ y2
Hence the trace of the strain tensor is
θ = tr ε = ε11 + ε 22 + ε 33 = − z∆* w . (f48)

F8) HOOKE’s law for a thin plate.

By using eqs.(f46) and (f48) it follows that
 2 
 λ∆* w + 2µ ∂ w  .
σ11 = λ tr ε + 2 µ ε11 
= − z

(f49)
 ∂ x2 
But
+h 2
1 2 dz = h
z2 =
2h ∫ z
3
(f50)
−h
Hence

h2
Μ11 = z σ11 = − 3
In the same way,
 2  2  2 
 λ∆* w + 2µ ∂ w  = − H  λ∆* w + 2µ ∂ w  .
   
(f51)
 ∂ x2  12  ∂ x2 
 47

h 2 ∂2w H2 ∂2w
Μ12 = z σ12 = −2µ 3 ∂ x∂ y = −µ 6 ∂ x∂ y , (f52)

h2  2  2  2 
 λ∆* w + 2µ ∂ w  = − H  λ∆* w + 2µ ∂ w 
Μ 22 = z σ 22 = −    
(f53)
3  ∂ y2  12  ∂ y2 
Substituting eqs.(f51)-(f53) into eq.(f35) it follows the equation of Sophie GERMAIN:

j s
D∆*∆* w = ρgH + q ( x, y) + q ( x, y)
3 3
H  j  
, (f54)
s   j s
+ q ( x, y) − q ( x, y) + q ( x, y) − q ( x, y) 
2  1 1  ,1  2 2  ,2 
where
2h 3 E(1 − ν) H3
D = ( λ + 2µ ) = (f55)
3 (1 + ν)(1 − 2 ν) 12
represents the flexural rigidity of the plate.

So, obtaining the median surface asks someone to solve a bi-harmonic equation, with certain boundary conditions on the upper/
lower faces of the plate. The equation (f55) will be solved in same cases of particular importance in real cases.

EXERCISE. Find the expressions of Σ13 and Σ 23 for a thin plate into a bending state.
F.9) The infinite, 1-dimensional (1-D) plate. The flexure of the lithosphere.

Consider an infinite extended plate alongy -co-ordinate. The component u 2 of the displacement vector is equal to zero,
and the rest of the components does not depend on y -co-ordinate. In this case, the plate is assumed to be in a cylindrical
bending state. Neglecting the horizontal loads, eq.(f55) becomes
d 4w j s
D = ρgH + q ( x ) + q ( x ) . (f56)
d x4 3 3
For the case presented in Fig.F2, on the upper face of the plate is acting the load P due to the relief and the lithostatic pressure,
i.e.
s
q 3 ( x ) = P + ρ r g ( w( x ) − h ) , (f57)

where ρ r is the density of the filling sediments (assumed to be homogeneous ones) placed between the reference plane of
elevation equal to zero and the upper surface of the plate. On the down face of the plate, it is acting the pressure of the liquid of
density equal to ρ m ,i.e.
j
q 3 ( x ) = p 0 − ρ m g( w( x ) + h ) , (f58)

where p o is an unknown constant. Eq.(f59) becomes

d 4w
D = P − (ρ m − ρ r )g w( x ) + p − ρ r gh − ρ m gh + ρgH . (f59)
dx 4 0
It is assumed that in the absence of the relief (i.e. P=0), the non-deformed plate (i.e. w ( x ) ≡ 0 ) is in an equilibrium state under
the action of its own weight and of the pressure of the liquid, i.e.

p 0 − ρ r gh − ρ m gh + ρgH = 0 (f60)
 48

Fig.F2. The flexure of the lithosphere under the action of the relief (P), of the lithostatic pressure ΓL and of the pressure of the
liquid ΓM .

It results the flexure equation of the 1-D plate :

d 4w 1
= [P − (ρ m − ρ r )g w ( x )] (f61)
d x4 D
Let

(ρ m − ρ r )g / D = 4 / α 4 , (f62)

where α is the flexural parameter of the plate. Eq.(f61) becomes

d 4 w( x ) P 4
= − w( x ) (f63)
d x4 D α4

F.10) Exterior forces on the lateral surface of the plate. Buckling.

To derive Sophie GERMAIN equation (f56) for the bending state, exterior forces acting on the lateral surface of the plate
have been ignored, especially those placed into the median plane. Consider a very thin plate simply leaning like in Fig.F3. The
load P is absent and the plate is infinite developed in a direction perpendicular on the plane of the figure. An element of the
plate having the length equal to unit along that direction is considered too. Let h be the thickness of the plate. The forces per
→
unit length along the above direction are denoted by ± N , being derived from the stress σ c ( positive for compression ) by
N = σch (f64)

Fig.F3. A plate simply leaning, subject to the action of forces placed into the median plane.
(a)A lateral view. (b) A view from above.
→
± N are small ones, the plate will be deformed according to a plane state, attempting a final configuration similar
If the forces
→
to Fig.1b. If the forces ± N are above a certain critical value, the plate loses suddenly its equilibrium state, attending a
deformation state like Fig.F4a, (or in the contrary sense, i.e. symmetrically with respect to the line of its supports. The
displacement field in this case is similar to the cylindrical bending. That phenomenon represents the buckling of the plate. It
characterises very thin plates or bars.
 49

→
Fig.F4. (a) The equilibrium of a buckled plate acted by the forces ± N and by the reactions of the supports.
(b) The equilibrium of a buckled plate acted by a vertical load p and by the reactions of the supports.
→
(c)The equilibrium of a buckled plate acted by the forces ± N and by a load q having the same magnitude but a
contrary sense with respect to the load presented in Fig.F4 (b).
In order to use the previous results, it is necessary to find when the mechanical state of stress / strain corresponding to the
presence of the lateral forces is identical to the mechanical state of stress / strain corresponding to an unknown vertical load
p = p( x ) (Fig.F4b). The equilibrium condition for the case shown in Fig.4a is
→ → →
∑ N + ∑ R (1) = 0 , (f65)
and the equilibrium condition for the case shown in Fig.F4b is
→ → →
∑ p + ∑ R ( 2) = 0 , (f66)
→ →
where ∑ R ∑ R ( 2) are the reactions of the supports in the above cases. Because the mechanical state of stress is
(1) and
identical in both cases, particularly in the neighbourhood of the supports, the reactions will be the same, i.e.
→ →
∑R (1) =
∑ R ( 2) . (f67)
It follows that
→ → →
∑ ∑ p)= 0 .
N + ( − (f68)
Hence for the corresponding state of stress / strain, the plate is in an equilibrium state if it is acted by the lateral forces and by a
→ →
vertical load q = − p , in the absence of the supports (Fig.F4c). Consider a plate element having the horizontal length equal
to dx and the ends denoted by A and B (Fig.F5).

Fig.F5. The equilibrium of a plate element due to the load q and to the internal tensions.
 50
Because the plate is a thin one, the tangential efforts are neglected. At the point A, it is acting a force (per unit length) denoted
→ →
by t , representing a normal effort, tangent to the plate. The angle between t and the horizontal axis is denoted by ϕ. At the
→
point B, it is acting the effort t + dt , making an angle equal to ϕ+dϕ with the horizontal axis. The equilibrium conditions are
t cos ϕ − (t + dt ) cos(ϕ + dϕ ) = 0
(f69)
t sin ϕ − (t + dt ) sin(ϕ + dϕ ) − qdx = 0
For small angles ϕ, it follows that
cos ϕ ≅ cos(ϕ + dϕ) ≅ 1
sin ϕ ≅ ϕ , sin (ϕ + dϕ) ≅ ϕ + dϕ (f70)
dw
ϕ ≅ tgϕ = −
dx
The first equation in (f69) gives
dt
=0 (f71)
dx
Hence t has no variation along x-axis, i.e.
t=N (f72)
The second equation in (f69) gives
tϕ − t(ϕ + dϕ ) − qdx = 0 (f73)
By using (f74) and (f70c) it follows that
dϕ d 2w
q = −N =N . (f74)
dx dx 2
Hence the forces on the lateral faces of the plate are mechanical equivalent to a vertical load equal to
d2w
p = p( x ) = − N . (f75)
dx 2
It follows that eq. (f63) has the next general form
d 4w N d 2w 4 P
+ + w= (f76)
d x4 D d x2 α 4 D

F.11) The buckling of a simply leaning thin plate.

Consider the plate in Fig.F3. For simplicity, it is assumed that c = 0 (i.e. the plate is leaning just at its ends). The buoyancy
force and the vertical loads are neglected. Equation (f76) becomes
d 4w N d 2w
+ =0 , (f77)
d x4 D d x2
together with the next conditions:
-at the end point having x=0: w = 0 , d 2 w / dx 2 = 0 (f78)

-at the end point having x=a: w = 0 , d 2 w / dx 2 = 0 (f79)

The equations (f77)-(f79) has the trivial solution w ≡ 0 . It follows to find a critical buckling value N = N* in order the
system (f77)-(f79) to have further non-trivial solutions. Successively, equation (f77) can be written as

d2  d 2w N 
 + w = 0 , (f80)
 2 D 
d x2 dx 
d 2w N
+ w = C1x + C 2 , (f81)
d x2 D
where C1 , C 2 are two integration constants, vanishing according to (f78)-(f79). Hence (f81) is
 51

d 2w N
+ w=0 , (f82)
dx 2 D
having the solution
w( x ) = C3 sin( N / Dx ) + C 4 cos( N / Dx ) . (f83)
From (f78) it follows that C 4 = 0 , while from (f79) it follows the critical values
N *k = D( kπ / a )2 , k = 1,2,... (f84)
The lowest critical value is obtained for k = 1 .
EXERCISES.
(1)Perform a study for the buckling of a 1-D plate having an embedded end point, the other being free.
(2)Perform a study for the buckling of a 1-D plate having both end points free. The plate is simply leaning at 1/3 from its length
with respect to its left end.
(3)Perform a study of the simply leaning 1-D plate in the presence of the buoyancy force.
(4)Modify the equation of Sophie GERMAIN for the 2-D plate in the presence of lateral forces.
(5)Perform a study for the buckling a 2-D rectangular plate, simply leaning at all its sides.

F.12) The infinite extended 1-D plate.

By integrating both sides of eq.(f76) it follows
+∞ 4 +∞ 2 +∞ +∞
d w d w
D ∫ dx + N ∫ dx + (ρ m − ρ)g ∫ w( x )dx = ∫ P( x )dx , (f85)
dx 4 dx 2
−∞ −∞ −∞ −∞
Because w and its derivatives of any order are vanishing at infinite, the first two integrals in (f85) are vanishing too. It follows
that the area bounded by the median curve (the flexure) and the horizontal x-axis is proportional to the load due to the relief,
irrespective the presence of the lateral forces:
+∞ +∞
1
∫ w( x )dx = (ρ m − ρ)g ∫ P( x )dx . (f86)
−∞ −∞
Let an approximation of the relief be a set of m steps, each one of height equal to h j and density equal to ρ j , i.e.
m
P( x ) = ∑ P j ( x ) , P j ( x ) = 
[
ρ j gh j , pentru x ∈ a j , b j
.
] (f87)
j =1 
0 , in rest
Equation (f86) becomes
+∞ m
1
∫ w( x )dx = ρm − ρ
∑ ρ jh j ( b j − a j ) . (f88)
−∞ j =1
Hence the area bounded by the flexural curve and the horizontal axis is a linear combination of the areas approximating the
relief. In real cases, the flexural curve can be outlined along a finite interval denoted by [−L , L], hence an upper bound for
the difference of the densities can be obtained as
m +L
ρm − ρ < ∑ ρ jh j ( b j − a j ) ∫ w( x )dx . (f89)
j =1 −L
Equation (f76) will be solved by using FOURIER transforms.

F.13) FOURIER transforms. Properties.

The direct FOURIER transform ( t.F.d. ) of a function f ( x ) is the new function Φ of variable u , defined as
+∞
Φ[f ]( u ) = ∫ f ( x ) exp( − iux )dx , i = −1 (f90)
−∞
The inverse FOURIER transform ( t.F.i. ) of a function Φ( u ) is the function f ( x ) defined as
 52
+∞
f ( x ) = Φ −1[Φ[f ]( u )]( x ) =
1
2π ∫ Φ( u ) exp( +iux )du (f91)
−∞
df
Differentiating both sides of eq. (f91) with respect to x, it follows that the t.F.d. of the first derivative can be obtained by
dx
d4w
multiplying the the t.F.d. of f ( x ) by iu . Hence the t.F.d. of the derivative can be obtained by multiplying the t.F.d.
dx 4

of w( x ) by ( iu ) 4 = u 4 . Consider two functions f and g of one variable. Their convolution product is

+∞
( f * g )( x ) = ∫ f ( y)g( x − y)dy . (f92)
−∞
Permuting the integrals, it follows that the direct FOURIER transform of the convolution product is the product pf the
transforms of both factors of the product, i.e.
+∞ +∞ 
Φ[f * g]( u ) = ∫ ∫ f ( y)g( x − y)dy exp( − iux )dx

 
−∞ −∞ 
. (f93)
+∞ +∞
= ∫ f ( y) exp( − iuy)dy ∫ g( z ) exp( − iuz)dz = Φ[f ] Φ[g]
−∞ −∞

F.14) Solution of the flexure equation by using FOURIER transforms.

The solution of eq.(f76) is the sum of two terms, a term corresponding to the homogeneous equation and a term corresponding
to a particular solution, i.e.
w( x ) = w h ( x ) + w p ( x ) (f94)
p p
A particular solution w = w ( x ) will be obtained applying the direct FOURIER transform to eq.(f76) and by using the
above presented properties of the FOURIER transform
N 2 4 1
u 4 Φ[w p ]( u ) − u Φ[w p ]( u ) + Φ[w p ]( u ) = Φ[P]( u ) , (f95)
D α4 D
i.e.
1 Φ[P]( u )
Φ[w p ]( u ) = . (f96)
D u 4 − Ku 2 + 4 / α 4
It is assumed that the value of the positive constant K = N / D is small enough. Consider the particular case when the load
due to the relief is a load concentrated at the origin of the axes, having the magnitude equal to unit. The direct FOURIER
transform of this load is equal to unit too. The corresponding solution, denoted by w G , represents the elastostatic GREEN
function. It allows one to obtain the solution corresponding to an arbitrary load of magnitude equal to P . Hence

1 1
Φ[w G ]( u ) = (f97)
D 2 2
 u − K / 2 + 4 / α 4 − ( K / 2 )2
 
It is assumed that the next condition is satisfied
K < 4 / α2 (f98)
Let

A = 1 / α2 + K / 4 , B = 1 / α2 − K / 4 (f99)
But
 53

1 α2  u u 
=  − 
u 4 − Ku 2 + 4 / α 4 8 A  (u + A )2 + B2 (u − A )2 + B2 
. (f100)
α2  1 1 
+  + 
4  (u + A )2 + B2 (u − A )2 + B2 
 
The next result is valid (Rîjic and Gradstein 1955)
∞
z sgn Re( z ) π
∫ z2 + x 2
cos xdx = exp( − z ) , z ≠ 0 .
2
(f101)
0
By using (f101), the next inverse FOURIER transforms are obtained
 
Φ −1  ( x ) = 1 exp(− B x m iAx) .
1
(f102)
 (u ± A )2 + B2  2B
 
 
Φ −1  ( x ) = 1 exp(− Bx ) cos( A x ) .
1 1
+ (f103)
 (u + A )2 + B2 (u − A )2 + B2  B
 
By using the property of the derivative, it follows
 1 d
Φ [ 
exp(− B x − iAx) ( u ) =] iu
. (f104)
 2 B dx  (u + A )2 + B2
In the same way
 1 d
Φ [ 
exp(− B x + iAx) ( u ) =] iu
. (f105)
 2 B dx  (u − A )2 + B2
Subtracting eq.(f104) from eq.(105), it follows that
1 d
Φ [ 
exp(− B x ) sin(Ax) ( u ) = ] u
−
u
. (f106)
 B dx  (u − A )2 + B2 (u + A )2 + B2
Hence
 
Φ − 1 
u
−
u

 (u − A )2 + B2 (u + A )2 + B2 
  . (f107)

=
1 d
B dx
[ ] A
B

exp(− B x ) sin (Ax) = exp(− B x ) cos( A x ) − sin ( A x )

Using the above results, it follows after some elementary computations that

α 2   sin( A x ) cos( A x )  
Φ[w G ]( u ) = Φ exp(− B x ) +  , (f108)
8D   A B  

α2  sin (A x ) cos( A x ) 
w G ( x) = exp(− B x ) +  , (f109)
8D  A B 
It can be observed that w G → ∞ for B → 0 , corresponding to the buckling of the infinite plate in the presence of a lateral
compressive stress. From (f96) and (f97) it follows that
Φ[w p ] = Φ[P] Φ[w G ] . (f110)
Hence the solution for an arbitrary load is the convolution of the load due to the relief and the function given by (f109),
representing a general property of the GREEN function:
+∞
w p ( x) = ( P * w G )( x ) = ∫ P( y) w G ( x − y)dy . (f111)
−∞
 54
For the approximation of the relief represented by eq.(f87), it follows that

∑ ρ jh j[I(b j − x) − I(a j − x)]

m
1
w p (x) = . (f112)
4(ρ m − ρ)
j =1
where
   
 
I( z ) = sgn( z )exp( − B z )−2 cos( A z ) + sin( A z ) + 2
K
(f113)
 
  (ρ m − ρ)g / D − K 2 / 4  
 
The solution of the homogeneous equation can be immediately derived as
w h ( x ) = [C1 cos( Ax ) + C 2 sin( Ax )]exp( − Bx ) + [C 3 cos( Ax ) + C 4 sin( Ax )]exp( Bx ) (f114)
Hence the general solution is

∑ ρ jh j[I(b j − x) − I(a j − x)]

m
1
w( x ) =
4(ρ m − ρ) (f115)
j=1
+ [C1 cos( Ax ) + C 2 sin( Ax )]exp( − Bx ) + [C3 cos( Ax ) + C 4 sin( Ax )]exp( Bx )
It follows to find the unknown coefficients C1 , C 2 , C 3 and C 4 in some particular cases. For the infinite plate, the flexure
W is subject to the next conditions:
lim W( x ) = 0 (f116)
x →±∞
Hence the coefficients C1 , C 2 , C 3 and C 4 are vanishing and the general solution is just the particular solution represented
by eq.(116). In the case of the semi-infinite plate the flexure W is subject, for example, to the next conditions:
lim W( x ) = 0 (f117)
x →∞
W( 0 + 0 ) = W0 (f118)

d 2W
( 0 + 0) = W0’’ = − M 0 / D (f119)
dx 2
’’
where W0 , W0 and M (positive when acting into a clockwise sense) are the values of the flexure, that of the second
0
derivative of the flexure and that of the bending moment respectively at the left end of the plate where the origin of the x-axis is
selected. It follows

∑ ρ jgh j[I(b j ) − I(a j )]

m
C1 = W0 − ,
j =1
(f120)
 
[ ( ) ( )]
m
C2 =
1  1
2AB  4 ABD
) ( ) (
∑ ρ jgh j exp − Ba j sin Aa j − exp − Bb j sin Ab j − 2 C1 − W0 
K ’’
 j =1 
and
C3 = C 4 = 0 . (f121)
A finite plate of variable thickness can be approximated in real cases by a sum of n elements having constant thickness and
homogeneous elastic properties. To obtain the values of the unknown coefficients C1 , C 2 , C 3 and C 4 for each element,
proper conditions have to be verified at the ends of each element. A finite element algorithm based on the continuity of the
values of the flexure, of its first derivative, of the bending moment and of the share force has been derived by Ivan (1997).
EXERCISE. Derive the expression of w ( x ) for a load due to a relief having the equation
ρgh 0 sin( 2π x / λ ) , pentru x ∈[− λ / 2, + λ / 2]
P = P( x ) = ρ gh ( x ) =  , (f122)
0 , in rest
where h 0 is the amplitude of the relief and λ is its wave-length.
 55
F.15) Finite plates.
In the case of the finite plates, the boundary conditions on the contour of the plate are essential ones in order to obtain the
values of the integration constants. Some particular, most common cases will analysed in detail. As a consequence, a previous
examination of the mean values Σ ij and Μ ij , i, j=1,2,3. it is necessary.

F.15 a) Significance of Σ ij and Μ ij for the bending state.

According to the definition of the stress tensor elements, σ ij is the projection along j -axis of the surface force acting on
→
the surface having the outer pointing normal equal to n i . Because the elements σ13 , σ 23 are even functions, the
mean values Σ13 , Σ 23 are representing share forces, acting like in Fig.F3.

Fig.F3. The mean values Σ 13 (a) and Σ 23 (b). Both of them are share forces.

Similar considerations allow one to conclude that the mean values M11 and M 22 are bending moments, while
Μ12 , Μ 21 are torsion moments (Fig.F4).

Fig.F4. The bending moments M11 (a) and M22 (b). The torsion moment M12 acting on the side having the outer pointing
normal 1 is presented in Fig.4c. A similar torsion moment is acting on the side having the normal 2, but it is not presented in the
figure.

F.15 b) The rectangular plate. Boundary conditions. LÉVY ‘s solution.

Let consider a rectangular plate having the sides equal to 2a and 2b respectively (Fig.F5). Consider, for example, the side
AB, having the equation x = a , y ∈[− b, b ] ). Among most commonly used boundary conditions are
-the embedded side: the flexure of the plate and the derivative of the flexure are both equal to zero
w ( x, y ) = 0 , ∂ w ∂ x = 0 ; (f123)
-the rotating side: the flexure of the plate and the bending moment are both equal to zero:
Μ11
w ( x, y ) = 0 , =0 , (f124)
i.e.:
 56

* ∂2w
w ( x, y) = 0 , λ ∆ w + 2µ =0 . (f125)
∂x 2
-the free side: the share force, the bending moment and the torsion moment are all equal to zero.

Fig.F5. The rectangular plate.

Similar conditions can be derived for plates of arbitrary shape.

As an example, consider the rectangular plate with two opposite articulated sides. Neglecting the lateral forces and the
gravity, it follows to solve the simplified equation of the flexure
∆ ∆ w = q ( x, y ) / D , (f126)
with the boundary conditions (f125) written for x = ± a and a 2-D LAPLACE operator. Consider a particular solution having
the form
∞
w( x, y) = ∑ f k ( y)sin( kπ x / a ) , (f127)
k =1
it follows that
w ( ± a, y ) = 0 ,
∞
∂2w π2
∑
. (f128)
=− k 2 f k sin( kπ x / a )
∂ x2 a 2 k =1
Hence
∂ 2 w / ∂ x 2 = ∂ 2 w / ∂ y 2 = 0 , for x = ± a . (f129)
It follows the conditions (f125) are fulfilled for the two opposite articulated sides. Substituting (f127) into (f126), it follows that
∞  2 4 
"" ( y ) − 2 kπ  f " ( y ) +  kπ  f ( y ) sin kπ x = q ( x, y ) / D .
∑ k  f  
 a  k
 
 a  k 
(f130)
k =1 
a

The function q ( x, y ) is expanded in FOURIER series and the coefficients are identified. Both sides of eq.(f130) are
jπ x
multiplied by sin , j = 1,2,... The result is integrated on the interval [− a, a] , taking into account that
a
+a
kπ x jπ x 0 , pentru j ≠ k
∫ sin
a
sin
a
dx = 
 a , pentru j = k
. (f131)
−a
Hence
2 4 +a
 kπ  "  kπ  1 kπ x
∫
""
f k ( y) − 2  f k ( y) +   f k ( y ) = q ( x, y )sin dx , k = 1,2,... . (f132)
 a   a Da a
−a
In the beginning, the next homogeneous equation is solved, i.e.
2 4
 kπ   kπ 
Fk"" ( y) − 2  Fk" ( y) +   Fk ( y) = 0 , k = 1,2,... . (f133)
 a   a 
The characteristic equation is

2 4
4  kπ  2  kπ 
r − 2  r +   = 0 , deci r1 = r2 = − kπ / a , r3 = r4 = kπ / a , (f134)
 a   a 
hence the solution of the homogeneous equation is
 57
kπ y kπ y  kπ y kπ y 
Fk ( y) = A k cosh + B k sinh + y C k cosh + D k sinh  , (f135)
a a  a a 
where A k , B k , C k , D k , k = 1,2,... are some constants following to be obtained from the boundary conditions on the
other two (non-articulated) sides of the plate. The general solution of equation (f132) is the sum of (f135) and a particular
solution. The last one can be obtained by the usual techniques (e.g. CAUCHY method). The above approach is due to LÉVY.

F.16) Vibrations of a plate laying on a viscous substratum.

In this chapter, the flexure is considered as a function of both spatial co-ordinates and time. The corresponding differential
equation is derived, being solved in the case of the 2-dimensional plate.

a) The differential equation.

As usually, a co-ordinate system is used having the horizontal axes x and y. The z -axis is positive downward, having the unit
→
vector denoted by e z . By applying the mean-value operator, the equations of motion for the bending state are
••
 d u 
∂ Μ xx ∂x + ∂ Μ xy ∂y + q ( x, y, t ) − q ( x, y, t ) / 2 − Σ xz = ρ z u x , (f136)
 x x 
••
 d u 
∂ Μ xy ∂x + ∂ Μ yy ∂y + q ( x, y, t ) − q ( x, y, t ) / 2 − Σ yz = ρ z u y , (f137)
 y y 
and
••
1  d u 
∂ Σ xz ∂x + ∂ Σ yz ∂y + q ( x , y, t ) + qz ( x, y, t ) + ρg = ρ uz
2h  z  . (f138)

It will be assumed that Bernoulli’s hypothesis is valid for all time. Hence the displacement vector has the elements
∂w ∂w
u x ( x, y, z, t ) = − z ∂ x , u y ( x, y, z, t ) = − z ∂ y , u z ( x, y, z, t ) = w( x, y, t ) . (f139)

Substituting (f139) into (f136)-(f138) it follows after elementary computations that

•• ρ ••
D∆*∆* w = ρgH − ρH w + H 3∆* w +q ( x, y, t ) + q ( x, y, t )
l u
12 z z
, (f140)
H  l u   l u  
+ ∂ q ( x, y, t ) − q ( x, y, t ) ∂x + ∂ q ( x, y, t ) − q ( x, y, t ) ∂y
2  x x   y y  

As usually, the horizontal loads for the upper face of the plate are neglected, the surface forces being assumed to be
u u u ••
qx = 0 , qy = 0 , q z = ρFgw + P(1 − w / g) , (f141)

F
where ρ is the density of the filling sediments and P is the load. The last term in (f141) is an inertial one. For the material
below the plate, the next constitutive equation is assumed
•
σ = [ p0 − ρMg( H + w )]1 + λM tr ε 1 + 2 µ M ε + 2 ηM ε (f142)

where p 0 is a reference pressure and ρM is the density of the material below the plate. The LAMÉ elastic coefficients are
λM , µ M and ηM is the viscosity. Hence the loads on the lower face of the plate are
 58
• •
l ∂w ∂ w l ∂w ∂ w
q x = µ M ∂x + ηM ∂x , q = µM
y ∂y
+ ηM
∂x
,
. (f143)
d H
q z = p0 − 2 λM∆*w + ρMg( H + w)
Again, a correction due to the compressive horizontal stresses σ cx , σ cy acting along the x- and y-axes respectively at the ends
of the plate follows to be considered further. The reference pressure p 0 is selected in order the flexure w to vanish in the
absence of the load P . Finally, a generalisation of the Sophie GERMAIN equation for a time dependent flexure is obtained as
 ∂2 w ∂2 w 
D∆*∆*w +  λM − µ M  ∆*w + H σ cx
H
+ σ cy + (ρM − ρF )gw
 2 2 
2  ∂x ∂y  (f144)
H *• •• ρ ••
= P + ηM ∆ w − (ρH + P / g) w + H 3∆* w
2 12

b) The rectangular plate with 3 welded sides.

M M
For usual materials λ = µ , hence the second term on the left side of (f144) vanishes. Because the load is mainly
represented by the relief, having the elevations much smaller then the thickness of the plate, the inertial term is usually
negligible on the right side of (15). Let w e = w e ( x, y) be the equation of the flexure corresponding to the state of
equilibrium in the presence of the load, i.e.

D∆*∆*w e + H  σcx∂2 w e / ∂x2 + σ cy ∂2 w e / ∂y2  + (ρM − ρ F )gw e = P (f145)

Consider the difference

δ = δ( x, y, t ) = w( x, y, t ) − w e ( x, y ) (f146)
It follows that

* * 12 1  c ∂2δ c ∂2δ  ρM − ρ F g 1  * •• 6ηM * • 12 ••

∆ ∆ δ+ σ + σy + 12 δ= ∆ δ+ ∆ δ− δ (f147)
 x 2  ρ  
ρa 2 H 2  ∂x ∂y2  H 3a 2 a2  ρH 2 H2 

where a denotes the velocity of P-waves through the plate.

Consider the lengths of the sides are L x , L y respectively. Suppose the plate is welded according to
δ( x,0, t ) = 0, with 0 ≤ x ≤ Lx
δ( 0, y, t ) = 0, with 0 ≤ y ≤ Ly , (f148)

δ( L x , y, t ) = 0, with 0 ≤ y ≤ Ly
at any time. A particular solution satisfying (f148) is

δ mn ( x, y, t ) = sin(mπx / L x )sin[( n − 0.5)πy / L y ] τ n ( t ) , m , n = 1,2,... (f149)

Substituting (f149) in (f147) it follows the modes are damped harmonic, i.e.

•• • 4π 2
τ + 2ζ mn τ + τ=0 , (f150)
2
Tmn
For m , n = 1, 2,... the periods are
H A mn + 12
Tmn = 2π , (f151)
a 2 2 c 2 2 c 2 2 2 *
A mn − 12π [σ x m H x + σ y ( n − 0.5) H y ] / (ρa ) + 12ρ gH / a 2
 59
and the decay constants are
3ηM A mn
ζ mn = , (f152)
ρH 2 A mn + 12
where
H x = H / Lx , H y = H / Ly , ρ* = (ρM − ρ F ) / ρ (f153)
and

A mn = π 2  m 2 H 2x + ( n − 0.5) 2 H 2y  . (f154)
 
According to (f139), the modes are both toroidal and spheroidal.
For each mode, a critical viscosity can be found from
ζ cr
mn = 2π / Tmn , (f155)
i.e.
2πρH 2 A mn + 12
η cr
mn = . (f156)
3Tmn A mn
For values of viscosity less than the critical value (f156), the motion of the plate is represented by a sum of damped oscillations.
The decay constants are obtained from (f152) and the periods are

* =T
Tmn / 1 −  ζ / ζ cr  2 (f157)
mn  mn mn 
For values of viscosity greater than the critical value, the motion of the plate is aperiodic and the characteristic roots of (21) are
 2  and − ζ 2 
− ζ mn + ζ 2mn − 4π 2 / Tmn 
2 2
 mn − ζ mn − 4π / Tmn  . It follows that for large scale of times, the
   
solution of (f150) behaves like exp( − Ct ) , where the decay constant is

2 − 4π 2 / T 2
C = ζ11 − ζ11 , (f158)
11
all the other terms being faster attenuated. In many real applications, an approximate value of (f159) is

4π 2 / T11
2
2π 2
C= ≅ (f159)
2
ζ11 + ζ11 − 4π / T11 ζ11T11
2 2 2

With regard to (f149), the decay constant (f158) or (f159) can be obtained as the ratio between the amplitude of the velocity
∂δ / ∂t and the amplitude of δ . By using eq.(f152), that ratio can be used to estimate the mean viscosity of the material below
the plate
ρH 2 2 π 2 A mn + 12
ηM = (f160)
3C T2 A mn
11
A numerical application with respect to the Moesian Platform, a micro-plate bounded by Carpathians and Balkan Mountains
and by the Black Sea is presented in detail by (Ivan 1997a,b).
 60
G) THE SPHERICAL SHELL

G.1) The model. BERNOULLI’s hypothesis. Displacement vector and strain tensor.

A spherical elastic, homogeneous shell having the elastic moduli equal to λ , µ is considered and the usual spherical co-
ordinate system (r, θ, ϕ) is used. However, some of the derived results are also valid in the case of a more general, non-elastic
spherical shell. In the initial state, the homogeneous density is denoted by ρ 0 and the median spherical surface of the shell has
the equation
r=R , (g1)
where R is the radius of the sphere. At a certain time during the deformation, the median surface will be

r = R − w (θ , ϕ , t ) , (g2)
where w = w(θ, ϕ, t ) is the flexure of the shell, positive downward. Hence the unit vector normal to the median surface at a
certain point of co-ordinates (r = R − w, θ, ϕ ) is

→  → ∂w →  2
1 ∂w →  1  ∂w  2 1  ∂w  
n =  e r+ +
eθ r sin θ ∂ϕ eϕ / 1 +   +   . (g3)
 ∂θ 2  ∂θ  2  ∂ϕ  
 r  r  sin θ 
Neglecting the quantities of the second order, it follows that
→ → ∂w → 1 ∂w →
n ≅ er+ +
R∂θ eθ R sin θ ∂ϕ e ϕ
. (g4)

Consider the initial, non-deformed state of the shell and two points of co-ordinates A 0 ( R , θ, ϕ ) and B 0 ( R + h , θ, ϕ ) ,
→ →
where H=2h is the thickness of the shell. It follows that the segment A 0 B 0 has the unit vector equal to e . At an arbitrary
r
time, in the deformed state of the shell, the point A 0 is displaced to a point A having the position vector equal to
→ → →
rA = R e r + u ( R , θ, ϕ, t ) , (g5)
while the point B 0 is displaced to the point B having the position vector equal to
→ → →
rB = ( R + h ) e r + u ( R + h, θ, ϕ, t ) . (g6)
→
Here, u is the displacement vector at a point of certain spherical co-ordinates. Assuming the shell is thin, quantities of the
→
2
order h are neglected and the segment AB has the unit vector equal to
→ → → →  ∂u → ∂u → ∂u ϕ →
 rB − rA  / rB − rA ≅ 1 + r  e + θ e +
∂r  r ∂r θ ∂r e ϕ
. (g7)
  
 
The partial derivatives in (g7) are computed at the point ( R , θ, ϕ ) . It is assumed that Bernoulli’s hypothesis is valid for all
time. It follows that a segment inside the shell, which is initially normal to the median spherical surface, will be always normal
to the median surface during the deformation. From (g4) and (g7), it is supposed that the displacement vector has the elements

 u ( r , θ, ϕ , t ) = − w (θ, ϕ , t )
 r
 r − R ∂w
 u θ ( r, θ, ϕ , t ) = + x(θ, ϕ, t ) , (g8)
 R ∂θ
 r − R 1 ∂w
u ϕ ( r, θ, ϕ, t ) = R sin θ ∂ϕ + y(θ, ϕ, t )

where x and y are two unknown functions representing the horizontal displacements of the points initially placed on the
median sphere. A further hypothesis on x and y will be later considered. It follows the elements of the strain tensor are
 61
∂u r
∂r ε rr =
=0 , (g9)

1  1 ∂u ∂u 
ε rθ = 2  r ∂θr − rθ + ∂rθ  = − 2r ,
u x
(g10)

1  1 ∂u r u ϕ ∂u ϕ  y
ε rϕ 2  r sin θ ∂ϕ − r + ∂r  = − 2r ,
= (g11)

1  ∂u θ  r − R ∂ 2 w 1  ∂x 
ε θθ = r  ∂θ + u r  = rR 2 + r  ∂θ − w , (g12)
∂θ
1  1 ∂u uϕ ∂u ϕ  r − R ∂  1 ∂w  1  1 ∂x ∂y y 
ε θϕ = 2r  sin θ ∂ϕθ − tan θ + ∂θ  = rR ∂θ  sin θ ∂ϕ  + 2r  sin θ ∂ϕ + ∂θ − tan θ  , (g13)

and

1  1 ∂u ϕ u θ  r − R  1 ∂ 2 w 1 ∂w  1  1 ∂y x 
ε ϕϕ r  sin θ ∂ϕ r tan θ  rR  2
=  + u +  = +

+  −w+  (g14)
 sin θ ∂ϕ 2 tan θ ∂θ  r  sin θ ∂ϕ tan θ 
Hence the trace of the strain tensor is
R(r − R ) ∗ 1  ∂ ∂y 
tr ε = ε rr + ε θθ + ε ϕϕ = ∆ w+  −2w sin θ + (x sin θ) +  , (g15)
r r sin θ  ∂θ ∂ϕ 
where the two-dimensional LAPLACE operator is
∂ ∂w  1 ∂ 2w 
∆∗ w =   sin θ
1
+  . (g16)
2 
R sin θ  ∂θ  ∂θ  sin θ ∂ϕ 
2

G.2) Quasi-mean values. Equations of motion.

Consider the stress tensor in spherical co-ordinates. For each element of the tensor, the corresponding quasi-mean value is
defined, for example, by
R+h
1 r
Σ rr = σ rr = H ∫ R σ rr
dr . (g17)
R−h
In the same way, the corresponding quasi-moment is defined by
R+h
1 r
M rr = H ∫ ( r − R ) σ rr dr . (g18)
R
R−h
The equations of motion in spherical co-ordinates are
∂ σ rr σ ∂ σ rθ + σ rθ 1 ∂ σ rϕ σ θθ + σ ϕϕ ∂ 2u r
+ 2 rr + + − − ρg = ρ
∂r r r∂θ r tan θ r sin θ ∂ϕ r ∂t 2
∂ σ rθ σ ∂ σ θθ + 1 ∂ σ θϕ σ θθ − σ ϕϕ ∂ 2uθ
+ 3 rθ + + =ρ . (g19)
∂r r r∂θ r sin θ ∂ϕ r tan θ ∂t 2
∂ σ rϕ σ rϕ ∂ σ θϕ 2σ θϕ 1 ∂ σ ϕϕ ∂ 2u ϕ
+3 + + + =ρ
∂r r r∂θ r tan θ r sin θ ∂ϕ ∂t 2
In the above equations, ρ denotes the density of the shell in the deformed state. Assumed to be a negligible second order
effect, that density will be replaced by the initial density ρ 0 at the right side of (g19).
 62
2
The first equation in (g19) will be multiplied by r and the quasi-mean operator defined by (g17) will be applied. The next
two equations in (g19) will be multiplied by r and the quasi-moment operator defined by (g18) will be applied. In order to do
that, some intermediary results are necessary.

G.3) Integrals of the stress elements. Quasi-moments.

Let the stress values on the upper/lower faces of the shell be denoted as

σ σ
U= ( R + h , θ, ϕ , t ) (g20)

and, respectively, by

σ σ
L= ( R − h , θ, ϕ, t ) . (g21)

Elementary computations show that

R+h  ∂
1 1 2 σ rr + 2r  R  h2 U  h  2 L
HR ∫  ∂r
 r σ rr H  R  σ rr  R  σ rr .
 dr =  1 +  −  1 − (g22)
R−h  
In the case of the elastic shell, the HOOKE’s law leads to
σ rθ = 2µ ε rθ = −µx / r , Σ rθ = −µx / R . (g23)
Integrating by parts, it follows that
R+h 2 ∂ σ rθ  R+h R+h 
1 r 1  2  
H ∫ R ( r − R )
∂r
dr = 
HR 
r ( r − R ) σ rθ 
R−h
+ µx ∫ ( 3r − 2 R )dr 
R−h  R−h 
. (g24)

R  h 2
 h 2 
= 1 +  σ U θ + 1 −  σ L rθ − R Σ rθ

2  R r  R
 
Also,
R+h 2 3 σ rθ
1 r
H ∫ R ( r − R )
r
dr = 0 . (g25)
R−h
Similar relations are derived for σ rϕ .
Also,
 R+h
 ∗ 2µ ∂ 2 w  1 2 2 
2 dr = h  λ∆∗ w + 2µ ∂ w  .
Μ θθ 
= λ∆ w +
 ∫ ( r − R )
3  
(g26)
 R 2 ∂θ 2  H R − h R 2 ∂θ 2 
In the same way, it follows that
h 2 2µ ∂  1 ∂w 
Μ θϕ = 3 2 ∂θ  sin θ ∂ϕ  , (g27)
R
and

h2  2µ ∂ 2 w 
Μ ϕϕ = ( λ + 2µ )∆∗ w −  . (g28)
3  2 ∂θ 2 
 R 

G.4) Integrals of the displacement vector.

Using (g8), it follows
R+h 2  1  h  2  ∂2w
1 r ∂2u r
H ∫ R 0 2
ρ dr = − ρ 0 1 +    2 .
R
 3 R  ∂t
(g29)
R−h ∂t
Also,
 63
R+h 2 2  2 4  2
1 r (r − R) ∂2uθ 2 ∂  h + h  ∂w + 2 x  h   ,
H ∫ R
ρ 0 2
dr = ρ 0 R
2  3R 2 5R 4  ∂θ
 
3  R 
(g30)
R−h ∂t ∂t  
and
R+h 2 ∂2u ϕ 2  2 4  2
1 r (r − R) 2 ∂  h + h  1 ∂w + 2 y  h   . (g31)
H ∫ R
ρ 0 2
dr = ρ 0 R
2  3R 2 5R 4  sin θ ∂ϕ
 
3  R 
R−h ∂t ∂t  

G.5) Equations of motion in quasi-mean values.

r 2 . By applying the quasi-mean operator defined by (g17), it follows

Let the first equation in (g19) be multiplied by

1 ∂ ∂ Σ rϕ  R+h

sin θ  ∂θ
(
sin θ Σ rθ + ) ∂ϕ 
 
−  Σ θθ + Σ ϕϕ −
 

1
RH ∫ ρr 2 g( r )dr
  R−h . (g32)
 2  2
+ (1 + h / R ) 2 σ U
R
− (1 − h / R ) 2 σ L  = −ρ R 1 + 1  h   ∂ w
H  rr rr  0
 3  R   ∂t 2
The next two equations in (g19) are multiplied by r and the quasi-moment operator defined by (g18) is applied. Hence

1  ∂ Μ θθ 1 ∂ Μ θϕ  Μ θθ − Μ ϕϕ 1 2 L
(1 + h / R )2 σ U
rθ + (1 − h / R ) σ rθ
sin θ ∂ϕ  Σ rθ
+ − + +
R  ∂θ R tan θ 2 

∂2  1 h 2 1 h 4  ∂w 2x h 2 
       
= ρ0 R    +    +   
∂t 2  3 R 5 R  ∂θ 3  R 
 
(g33)
and

1  ∂ Μ θϕ 1 ∂ Μ ϕϕ  2 Μ θϕ 1
+ − Σ rϕ + + (1 + h / R )2 σ Uϕ + (1 − h / R )2 σ L 
rϕ 
R  ∂θ sin θ ∂ϕ  
R tan θ 2  r
. (g34)
∂ 2  1  h  2 1  h  4  1 ∂w 2 y  h  2 
= ρ0 R    +    +   
∂t 2  3  R  5 R  sin θ ∂ϕ

3  R 


G.6) Quasi-mean value of the shell density. The differential equation.

According to (g17), the quasi-mean of a constant is equal to that constant. Applying the quasi-mean operator to the mass
balance equation in the linear approximation gives
ρ = ρ 0 (1 − tr ε ) . (g35)

It will be further assumed that the quasi-mean of the density is constant and equal to its initial value. From (g35) it follows that
tr ε = 0 , (g36)

i.e., using (g15),

∂ ∂y
( x sin θ ) + = 2w sin θ . (g37)
∂θ ∂ϕ
For the elastic shell,
σ = λtr ε 1 + 2µ ε . (g38)
 64
By using (g9), it follows
σ θθ + σ ϕϕ = 2(λ + µ )tr ε , (g39)

or, using (g36),

Σ θθ + Σ ϕϕ = 0 . (g40)

The term (h / R )4 will be neglected. Using (g40) and substituting Σ rθ and Σ rϕ between (g32)-(g34) it follows

 2
(
1  ∂ sin θ Μ θθ
+
) 2
∂ 2  sin θ Μ θϕ
+
1 ∂ Μ ϕϕ ∂ 
2
−  cos θ Μ  


 sin θ ∂θ∂ϕ sin θ ∂ϕ 2 ∂θ  ϕϕ
R2  ∂θ 2 
 
sin θ  2 L  + 1 ∂ sin θ  1 + h / R 2 U + 1 − h / R 2 L  
+ (1 + h / R )2 σ U − (1 − h / R ) σ rr  2R ∂θ  ( ) σ rθ ( ) σ rθ
H  rr 
(g41)
R+h
1 ∂  − 1
+
(1 + h / R )2 σ U ϕ + (1 − h / R )2 σ L ϕ  ∫ ρr 2 gdr
2R ∂ϕ  r r  HR
R−h
h2  ∂ 2 w    h  2  ∂ 2 w 
 ∗
= ρ 0 sin θ ∆   − 1 − 
3  2    R   2 
  ∂t    ∂t 

σ σ
A correction due to the compressive horizontal stresses
NS , WE acting at the ends of the shell, approximately along
the θ − and ϕ − axes respectively, follows to be further considered.

G.7) The buckling of a spherical shell.

Consider the element ABCD of the mean surface of the deformed shell from Fig.G1. It is centred at the point M, where the
→ → →
local unit vectors are e r , eθ , eϕ . The centres of the lateral sides are denoted by M k , k = 1,2,3,4 . On the meridian
cross section M 4 MM 2 is acting a normal compressive stress σ WE , having the approximate direction from West to East,
and a tangential stress τ
NS , having the approximate direction from North to South. On the parallel cross section M MM
1 3
σ
is acting a normal compressive stress
NS , having the approximate direction from North to South, and a tangential stress

τ
WE , having the approximate direction from West to East. Let Φ be the angle between the normal vector to the meridian
→
section M 4 MM 2 and the unit vector e . Also, let Θ be the angle between the normal vector to the parallel section
ϕ
→
M 1MM 3 and the unit vector eθ . The concentrated force acting at the point M1 is
→   → →
 WE ∂ σ dϕ   
WE
∂Φ dϕ   ∂Φ dϕ  
Σ1 =  σ − ∂ϕ 2  cos Φ − ∂ϕ 2  eϕ + sin Φ − ∂ϕ 2  e r 
   
 (g42)
  → →  
dϕ   
+  τ NS − τ
∂ NS
∂Θ dϕ   ∂Θ dϕ     ∂w dϕ 
cos Θ −  eθ + sin  Θ −  er  ×  R − w +  Hdθ
 ∂ϕ 2   ∂ϕ 2   ∂ϕ 2    ∂ϕ 2 
  
 65

Fig.G1. A rectangular element of the mean surface of the deformed shell

In the same way, the concentrated force acting at the point M 3 is equal to

→  WE   → →
 WE ∂ σ dϕ    ∂Φ dϕ   ∂Φ dϕ  
Σ 3 = −  σ + ∂ϕ 2  cos Φ + ∂ϕ 2  eϕ + sin Φ + ∂ϕ 2  e r 
   
 (g43)
   → → 
+  τ NS + τ
∂ NS
dϕ    ∂Θ dϕ   ∂Θ dϕ     ∂w dϕ 
cos Θ +  eθ + sin  Θ +  er  ×  R − w −  Hdθ
 
∂ϕ 2    ∂ϕ 2   ∂ϕ 2     ∂ϕ 2 
  
The concentrated forces acting at the points M 2 and M 4 are respectively equal to

→  NS   → →
 NS ∂ σ dθ    ∂Θ dθ   ∂Θ dθ 
Σ 2,4 = m  σ ± ∂θ 2  cos Θ ± ∂θ 2  eθ + sin Θ ± ∂θ 2  e r 
  
  → → 
 ∂ τ WE dθ    ∂Φ dθ   ∂Φ dθ  
+ τ  e r 
WE ± cos Φ ±  e ϕ + sin Φ ± (g44)
 ∂θ 
2   ∂θ 2   ∂θ 2  
   
 ∂w dθ 
× R − w m  sin(θ ± dθ )Hdϕ
 ∂θ 2 
But
1 ∂w 1 ∂w
cos Φ ≅ 1 , cos Θ ≅ 1 , sin Φ ≅ Φ ≅ , sin Θ ≅ Θ ≅ . (g45)
R sin θ ∂ϕ R ∂θ
Let p = p(θ, ϕ ) a surface density of forces normal to the element of the shell, having the same mechanical effect as the
presence of the compressive stress. The force due to that density is equal to
→ →
Σp = p(θ, ϕ )( R − w ) 2 sin θdθdϕ er . (g46)
It follows the deformed element of the shell is into an equilibrium state due to the action of the lateral stress and to the opposite
→
force − Σ p , i.e.
→ → → → → →
Σ1 + Σ 2 + Σ 3 + Σ 4 − Σ p = 0 . (g47)
Hence the next three equations of equilibrium are obtained:
 66

sin θ ∂w   NS ∂w τ ∂w 
WE

R ∂θ   σ
 2 cos θ −  +
 ∂θ sin θ ∂ϕ 
, (g48)
∂  NS ∂w τ ∂w  ∂  σ ∂w 
WE WE
∂w R(R − w )
σ +τ
+ sin θ + + NS = − p(θ, ϕ ) sin θ
∂θ  ∂θ  
sin θ ∂ϕ  ∂ϕ  sin θ ∂ϕ ∂θ  H

NS  2 cos θ − sin θ ∂w  + sin θ ∂ σ + τ

NS ∂ NS
σ  R ∂θ

 ∂θ ∂ϕ
=0 , (g49)

and
∂ σ WE  sin θ ∂w  ∂ τ WE
+ τ WE  2 cos θ −  + sin θ =0 . (g50)
∂ϕ  R ∂θ  ∂θ
For the particular case when σ τ τ
WE is a constant and NS = WE = 0 , eqs.(g48)-(g50) show that the presence of the
lateral compressive stress is equivalent to a supplemental load placed on the upper face of the shell, having the value
 2 2 
H  NS ∂ w + WE 1 ∂ w  ,
σ σ
p(θ, ϕ ) = − (g51)

R2  ∂θ 2 sin 2 θ ∂ϕ 2 
a result similar to the case of the plane plate (Timoshenko and Woinowsky-Krieger 1959; Nowacki 1961). Quantities of the
second order, like (∂w / ∂θ)2 have been neglected again.
G.8) Load on the upper face. Stress on the lower surface of the shell.
Consider now the differential equation (g41). Usually, the horizontal loads for the upper face of the shell are neglected, and it
is assumed that
 2 
U = −ρ F gw − P 1 − 1 ∂ w 
σ rr σ rθ = 0 σ rϕ = 0
, U , U , (g52)
 g ∂t 2 

F
where ρ is the density of the filling sediments, P is the load and an inertial term is considered. For the material below the
shell, the next constitutive equation is assumed

M = p − ρ M g( R − h − w ) + λM tr M + 2µ M M + 2η M ∂ ε
M
σ  0 1 ε 1 ε ∂t
, (g53)

M
where p 0 is a reference pressure, ρ is the density of that material and 1 is the unit tensor. The elastic coefficients are

ε
λM , µ M and the viscosity is denoted by η M . The strain tensor inside the material is M and the strain rate here is
∂ ε M / ∂t .
The first boundary condition assumed on the lower face of the shell, having the equation r = R − h , is the continuity of the
displacement vector. Hence the elements of the displacement vector inside the material placed immediately below the lower
face of the shell are assumed to be equal to the same elements at the points of the shell placed on the lower face. By using eqs.
(g8), it follows

u ( r, θ, ϕ, t ) = − w(θ, ϕ, t )
 r
 − h ∂w
u θ ( r, θ, ϕ, t ) = + x( θ, ϕ, t ) . (g54)
 R ∂θ
 − h 1 ∂w
u ϕ ( r, θ, ϕ, t ) = R sin θ ∂ϕ + y(θ, ϕ, t )

Hence the next values for the strain tensor immediately below the shell are obtained
 67
1 ∂w x 1 1 ∂w y
ε rr ε rθ = − 2R ε rϕ = − 2R sin θ ∂ϕ − 2( R − h )
M=0 , M − , M . (g55)
∂θ 2( R − h )
→
The normal vector at the lower face of the shell is e r . From Newton’s third low, it follows a relation between the stress
σ σ
S inside the shell and the stress M inside the material below the shell, i.e.

 →  →
S−  = − M   ,
σ  er  σ  er  (g56)
   
on the lower surface of the shell. Here, the next elements of the stress are obtained after some elementary computations

σ rr = p 0 − ρ M g( R − h − w ) − R − h λM ∆∗ w
L Rh
, (g57)

 1 ∂w x  M  1 ∂  ∂w  + 1 ∂x  ,
σ rθ = −µ M  R
L + −η     (g58)
∂θ R − h   R ∂θ  ∂t  R − h ∂t 
and
 1 ∂w y   1 ∂  ∂w  1 ∂y 
σ rϕ = −µ M  R sin θ ∂ϕ + R − h  − η M  R sin θ ∂ϕ  ∂t  + R − h ∂t 
L . (g59)

G.9) The differential equation of time dependent flexure.

The reference pressure p 0 in (g53) is selected in order the flexure of the shell to vanish in the absence of the load.
Substituting the loads on the upper/lower faces of the shell in eq.(s41) and taking into account the presence of the lateral
gstress, a generalisation of the Sophie GERMAIN plain plate static equation in the case of a time dependent flexure of a
spherical elastic shell is obtained as

H3  ∆∗ w   h2  1 ∂ 2 w  
( λ + 2µ )∆∗ ∆∗ w + 2µ  − 1 +  ρ F gw + P1 − 
12  2   R    g 2 
 R    ∂t  

 h  2 Rh M ∗  H  h2 M ∗ 2w 
+1 −   ρ M gw + λ ∆ w −  1 −  µ  R∆ w + 
 R  R−h  2R  R  R − h
. (g60)
H  h  2 M  ∗  ∂w  2 ∂w  H  NS ∂ 2 w 1 ∂ 2w 

−
2R 
1 −  η  R∆ 
R  ∂
+
 − ∂  + 2 σ + σ WE

 t R h t  R  ∂θ 2 2 2
sin θ ∂ϕ 
H2  ∂ 2 w    h  2  ∂ 2 w 

= ρ0 H ∆∗   − 1 − 
12  2    R   2 
  ∂t    ∂t 
2
If quantities of the order ( h / R ) are neglected with respect to unit, it follows finally that

D∆∗ ∆∗ w +  λM − µ M  ∆∗ w +  ρ M − ρ F − ρ∗  gw

H
2
 1 ∂ 2 w  WE ∂2w 
+ H  σ NS + σ ∗ + σ ∗
1
+ σ  , (g61)
  2
R ∂θ 2
  2
R sin 2 θ ∂ϕ 2 

∗ H M ∂ ∗ 2w   P ∗  ∂ 2 w ρ 3 ∂ 2 ∗
=P + η ∆ w +  − ρH + + H (∆ w )
∂t   g  ∂t 2 12
2 R2   ∂t 2
where
 68

(λ + 2µ )H 3 2 H3
D= = ρa (g62)
12 12
is the flexural rigidity and a is the velocity of the P-wave through the shell. Also,
 µ M  H 2
ρ∗ =  ρ M + ρ F + ∗= µ H  H
, P∗ = P1 + 
 gR  R
, σ 6 2  R
. (g63)
 R

G.10) Spherical effects with respect to the plane plate.

For the usual materials, λM = µ M . In that case, the corresponding equation for the plane plate (Ivan 1997) is.
 ∂2w ∂ 2 w   M
D∆∗ ∆∗ w + H σ cx + σ cy +  ρ − ρ F  gw
 2 2  
 ∂x ∂y 
. (g64)
∂  P ∂2w ρ 3 ∂2 ∗
= P + η M  ∆∗ w −  ρH + 
H
+ H (∆ w)
2 ∂t  g  ∂t 2 12 ∂t 2

With respect to (g64), a change of the density difference ρ

M F
− ρ according to (g63a) and a substitution of the real lateral
∗ ∗ can be observed in eq.(s61). A supplemental
σ σ σ σ ,σ σ
stresses
NS , WE by their apparent values NS + WE +
load is present according to (g63c). For usual values (e.g. Ivan 1997a) like H / R ∝1 / 100 ,
µ, µ M ∝1011 Pa , ρ M , ρ F ∝ 3000 kgs / m 3 , σ NS , σ WE ∝ 30 MPa , all these effects are usually negligible and
2 ∗
difficult to be observed in real life. To compare 2w / R to ∆ w in the left side of (g61), the case of a rectangular plate
having the sides equal to L x , L y is considered. Here, the flexure is proportional to a product of sins (cosines) functions, i.e.
w ∝ sin( mπx / L x ) sin( nπy / L y ) . It follows the LAPLACE operator of the flexure is proportional

to ∆
∗ w ∝ π 2  m 2 / L2 + n 2 / L2  sin( mπx / L )sin( nπy / L ) . For the fundamental mode ( m = n = 1 ) it
 x y x y
follows that
2w / R 2 1
= . (g65)
∆∗ w π 2 R 2 (1 / L2x + 1 / L2y )
That ratio is negligible too in the usual cases.
It can be concluded that in the usual cases, the sphericity of the crustal plates can be ignored and the equation derived for the
flat plate can be used.
 69
H) ELEMENTS OF RHEOLOGY
H.1) Introduction.
Especially for geological processes at a large time scale and great values of the stress, the internal friction of the material
cannot be ignored. Consequently, the HOOKE’s law has to be replaced by assuming different constitutive equations (models).
Their expressions are mainly depending on the time scale of the geophysical process to be modelled. In relation to seismic or
seismological applications, for example, short periods and short stresses are required (usually, seconds up to minutes, with a
maximum value around one hour for the fundamental mode of the free oscillations of the Earth). Here, the non-elasticity is
related to the very short period irreversible changes in the crystal defect structures of the medium (e.g. opening/closing of pre-
existing cracks) and/or to the energy lost by friction at the two sides of a crack or on the non-elastic boundary coupling grain
particles to the adjacent material (Aki and Richards, 1980; Ranalli, 1987; Wahr, 1996).
With respect to the mathematical relation between stress and strain, there are two kinds of constitutive equations (models).

H.2) Linear models.

Simplified models involves a linear relation between stress (and its derivatives of various orders with respect to time) and
strain (together with its time derivatives).
In the beginning, only the 1-D case will be discussed. More general examples follow to be presented in relation with the
dynamic aspects of the flexure of a plate (shell) and to the accretion prism. For each constitutive equation, a mechanical
analogue can be considered. The elastic part will be represented by a spring, while the inelastic (viscous) behaviour is
associated to a dashpot. Both parts are supposed to be linear ones, i.e. a linear relation σ = 2µ ε is valid for the spring and

•
a similar linear relation holds for the dashpot σ = 2 η ε.

Fig.H1. (a) KELVIN-VOIGT model; (b) MAXWELL model; (c) BURGERS model

a) KELVIN-VOIGT (strong viscous) model.

The mechanical analogue of the first model to be considered is represented in Fig.H1a. The total stress is the sum between the
stress in the spring and the stress in the dashpot, while the total deformation is equal to both the deformation of the spring and
the deformation of the dashpot. It follows that KELVIN-VOIGT model has the next constitutive equation
•
σ = 2µ K ε + 2η K ε , (h1)

where the dot shows the (total, material) derivative with respect to time. Here, the second term is the inelastic one, η K being
the viscosity. Suppose now a constant stress equal to σ0 is applied. Elementary computations show that the differential
equation
 70
dε
σ 0 = 2µ K ε + 2η K
dt
(h2)

with the initial condition

ε (t = t 0 ) = 0 (h3)

has the solution

σ0   µK 
ε = 1 − exp −
2µ K   ηK
( t − t 0 )  ,

(h4)

for t ≥ t 0 . Hence, for very great values of time, the strain approaches a limiting value equal to
σ
ε ∞ = 2µ K0 . (h5)

The “flowage function”, denoted by

1   µK 
J( t ) = 1 − exp − ( t − t 0 )  , (h6)
2µ K   ηK 
shows that for a constant stress (equal to unit here), there is a temporal variation of the strain.
Suppose now at a certain moment t = t1 , the constant stress σ 0 is removed, the corresponding strain at that moment being
equal to ε1 . It follows now the corresponding solution decreases towards zero as
 µ 
ε = ε1 exp − K ( t − t1 ) .
 ηK 
(h7)

b) MAXWELL (viscous-elastic) model.

Consider the mechanical analogue represented in Fig.H1b. The total stress is equal to both the stress of the spring and to the
stress of the dashpot, while the total deformation is the sum between the deformation of the spring and the deformation of the
dashpot. It follows that MAXWELL model has the next constitutive equation
•
• σ
ε = 2µσM +
2η M
, (h8)

with the initial condition represented by (h3).

Suppose again a constant stress equal to σ 0 is applied. The spring is instantly deformed to a value equal corresponding to
the first term in the right hand of (h8), i.e.
σ
ε 0 = 2µ M0 . (h9)

and the solution of (h8) (for a constant stress σ 0 ) with the initial condition (h9) is the straight line
σ σ
ε = 2η M0 (t − t 0 ) + 2µ M0 , (h10)

σ 0 and to the viscosity of the dashpot. Suppose now at a certain moment t = t1 , the
having the slope related to the stress

constant stress σ is removed, the corresponding strain at that moment being equal to ε . It follows from (h8) that the
0 1
strain remains constant.
c) BURGERS (general linear) model.
The third model to be considered has the mechanical analogue represented in Fig.H1c.
EXERCISE. Show that the corresponding differential equation is
•• • η ••  η µ  • µ
2η1 ε + 2µ1 ε = 1 σ +  1 + 1 + 1 σ + 1
µ2  η2 µ 2  η2 σ . (h11)

Consider now the same initial condition (h3) and suppose again a constant stress equal to σ 0 is applied. In a similar manner
to MAXWELL model, the system is instantly deformed to a value equal to
 71

σ
ε 0 = 2µ 02 (h12)

and the solution of (h11) (for a constant stress σ 0 ) with the initial condition (h12) is
σ σ   µ 
ε = 2µ 02 + 2η02 ( t − t 0 ) + C1 − exp − η11 ( t − t 0 )  , (h13)
 
where C is an unknown coefficient (because (h11) is a second order differential equation, two initial conditions are required to
obtain the complete solution). However, differentiating (h13) it follows
• σ µ  µ 
ε = 2η02 + C η11 exp − η11 ( t − t 0 ) . (h14)

Hence, for great values of time, the solution (h13) approaches asymptotically to a straight line having the slope equal to
σ 0 / 2η 2 . Suppose now at a certain moment t = t1 , the constant stress σ 0 is removed, the corresponding strain at that
moment being equal to ε . It follows from (h13) that the strain decreases exponentially towards zero, i.e
1
 µ 
ε = ε1 exp − 1 ( t − t1 ) .
 η1 
(h15)

EXERCISE. Show that two (or more) springs / dashpots connected in series (or parallel) sequence are equivalent to a single
spring / dashpot. Justify that the BURGERS model is the general linear model.
d) Remarks on the linear models.
In the most general case, the linear relation between stress and strain can be written as
P( D ) σ = Q( D ) ε , (h16)

where
P( D ) = A 0 + A1 D + A 2 D 2 +...+ A n D n (h17)
and
Q( D ) = B0 + B1 D + B2 D 2 +...+ Bm D m (h18)
d
are formal polynomials of the variable D=
representing the derivative with respect to time, applied to stress and strain
dt
0 1 n 0 1 m
respectively. Here, A , A ,..., A , B , B ,..., B are fourth rank tensors. For example, with respect to the
MAXWELL body having the constitutive equation (h8), it follows that
1 1
P( D ) = + D , Q( D ) = D (h19)
2 η M 2µ M
A common way to solve (h16) is by using the LAPLACE transform (e.g. Sokolnikoff and Redheffer, 1958). Consider a certain
function of one real variable f ( t ) , providing that
1. f ( t ) = 0 , for t < 0 ;
2. f ( t ) is piecewise continuous on every finite interval;
3. there are two constants 0 < M , a ≤ 0 in order to have f ( t ) ≤ M exp( at ) , for an arbitrary t .
Under the above conditions, the LAPLACE transform of f ( t ) is a new function of the variable p , defined by
∞
L[f ]( p ) = ∫ f ( t ) exp( − pt )dt (h20)
0
EXERCISE. Show that:
(a) The LAPLACE transform of the first derivative is
 df 
L ( p ) = pL[f ] − f ( 0 + 0 ) (h21)
 dt 
1
(b) L[exp( − at )]( p ) = , 0≤a (h22)
p+a
 72
(c) The convolution theorem. Consider the functions f , g vanishing for negative values of their argument and a new function
(also vanishing for negative values of the argument defined) by the convolution product
+∞ t
h ( t ) = (f ∗ g )( t ) = ∫ f ( τ )g( t − τ )dτ = ∫ f ( τ )g( t − τ )dτ (h23)
−∞ 0
Show that
L[f ∗ g] = L[f ] L[g] (h24)
As an example, consider again the MAXWELL model by applying the LAPLACE transform to both sides of (h8). It follows
that
1   
pLε  − ε 0 =  pL σ − σ  + Lσ 
0 1
, (h25)
  2µ M  
    2 η M  

σ σ ( 0 + 0) , ε 0 = ε ( 0 + 0) are the initial stress and strain respectively.

where
0=
EXERCISE. By using the above properties of the LAPLACE transform, show that
2µ 2M t  µ  0 exp − µ M t (h26)
σ ( t ) = 2µ M ε (t) − ∫ε ( t ) exp − M ( t − τ )dτ +  σ 0 − 2µ M ε  
ηM  
0  ηM   ηM 
If the initial conditions are elastically coupled, i.e.

σ ε
0 = 2µ 0 (h27)
M
it follows an integral representation of the stress which is independent on the initial conditions
2µ 2M t  µ 
σ ( t ) = 2µ M ε (t) −
ηM ∫ε ( t ) exp − M ( t − τ )dτ
 ηM 
(h28)
0
Hence, the actual value of the stress is related both to the actual value of the deformation and to the previous values of the
strain (i.e. the stress depends on the “history” of the deformation). If a constant strain ε ( t ) = ε 0 , for t ≥ 0 is applied to
the MAXWELL body, it follows from (h28) that the stress decreases as
0 exp − µ M t ,
σ ( t ) = 2µ M ε 
 ηM 
 (h29)

representing a “relaxation phenomenon” (stress decreases in time if a constant deformation is present). Here, the function
G(t ) = 2µ M exp(− t / τ M ) (h30)
is the “relaxation” kernel and the parameter τ M = η M / µ M is the relaxation time.
A very similar approach (e.g. Wahr 1996) is based on the use of FOURIER transform (see Chapter F.13). Formally, the
results derived by using FOURIER transform are derived from the same results obtained with LAPLACE transform by
performing the substitution p = iu .
The above 1-D models can be generalised for the 3-D case. For example, consider again the MAXWELL body. There is a
strong experimental evidence that the MAXWELL Rheology applies only to the dissipation of the shear energy, i.e. the stress
and strain tensors in (h8) are the deviatoric tensors
1 1
σ ↔ σ − tr σ
3 1 , ε ↔ ε − tr ε
3 1 (h31)

It will be further assumed that there is no dissipation of the compressional energy, i.e. a proportionality like
tr σ = (3λ M + 2µ M )tr ε , (h32)
is valid. By differentiating with respect to time
• •
tr σ = (3λ M + 2µ M )tr ε (h33)
Substituting (h31) and (h33) in (h8) it follows
• 1 •  • 3λ + 2µ •   
1  − M M tr  + 1  − 3λ M + 2µ M tr 
ε 3 tr ε1 = 2µ M
−
σ 3 ε1 2η M σ 3 ε1 (h34)
   
Hence
 73
• µM • • µ 3λ M + 2µ M
σ+ ηM σ = 2µ M ε M ε1 + η M
+ λ tr
M 3
tr ε 1 (h35)

By applying the FOURIER transform to both sides of (h35), it follows

  ~   ~  
Φ σ  = λ tr Φ ε 1 + 2 µ Φ ε  , (h36)
     
     
i.e. a relation similar to HOOKE’s law is valid between the FOURIER (or LAPLACE) transforms of stress and strain. Here, the
coefficients similar to LAMÉ parameters are
µM  2 µM 
iu +  1+ 
~ ηM  3 λM  ~ iu
λ = λM , µ = µM (h37)
µ µ
iu + M iu + M
ηM ηM
At short periods T correspond high values of the pulsation u = 2π / T . From (h37) it follows
~ ~
λ = λM , µ = µM (h38)
and the behaviour of the MAXWELL body is an elastic one.
At long periods T correspond low values of the pulsation and
~ 3λ + 2µ ~
λ= M M =K
M , µ=0 (h39)
3
and the MAXWELL body is a fluid having the compressional coefficient denoted by KM .

H.3) Non-linear models.

For tectonic applications, large stresses and periods of thousands to millions of years are appropriate. Here, the non-elastic
behaviour is probably related to the diffusion or dislocation creep of the molecules, a major factor being the high temperatures.
There are great difficulties to consider constitutive equations with non-linear relations between stress and strain, but some
attempts have been made. A very common non-linear model is the work-hardening plasticity (e.g. Ranalli, 1994)
n −1 *
 *
3  σ E σ ij
ε ij = 2   , (h40)
σ
 0 σ 0
3 * *
σ ij is the component of the deviatoric stress, σ E = 2 σ ij σ ij
where
* * is related to the second invariant of the deviatoric

stress, and σ , n are material parameters.

0

H.4) Brittle. Creep. Empirical criteria.

The usual materials are reacting in an elastic manner only for small values of the (deviatoric) stress, i.e. for stress values
smaller than a limiting value representing the yield strength (or the yield stress), denoted by σ Y . The yield stress is a
function of the nature of the material, of the temperature, pressure, the chemical composition of the adjacent rocks and, finally,
of the history of the deformation (i.e. the intermediate steps followed to attend the yield value). When the yield stress is
attended, there are two possibilities of behaviour of the material:
• a rupture deformation of the rock, when the continuity of the deformation is lost, usually along a fault surface; this is the
case of the brittle materials. The process is illustrated in Fig.H2a and b.
• a plastic, irreversibly flow of the material (creep), when, apparently, the continuity holds. The phenomenon is quite
similar to the usual viscous flow, but it can be observed only when the yield stress is attended. This is the case of the
ductile materials. The process is illustrated in Fig.H2c. An usual non-linear constitutive equation is the BYERLEE
power-law creep
•
ε = Aσ
n exp( − H / RT) , (h41)
 74
where A, n are material parameters, H is the activation enthalpy, R is the gas constant and T is the absolute
temperature. It should be noted that the same material can act as a brittle or a ductile one according to the external
conditions.

Fig.H2. (a) Faulting of a brittle material; (b) Increasing of the yield stress due to the history of deformation;
(c) Creep of a ductile material.

H.5) Empirical criteria for shear-faulting. TRESCA criterion. COULOMB-NAVIER criterion.

A first criterion also (valid for plasticity) is due to TRESCA. It assumes that faulting (for brittle materials) or creep (for
ductile ones) is attended at that points of the material where the maximum value of the shear (tangential) stress is equal to a
yield value denoted by σ Y . Consider now a homogeneous (constant) stress field inside the material. Such a case can be
obtained either by considering an infinitesimal volume of material or by taking into account a prismatic body with very large
(infinite) sides. Consider the eigen-values σ1 , σ 2 , σ 3 of the stress tensor. It will be assumed that they are denoted in order
to have σ 3 < σ 2 < σ1 , with the remark that ,in real life, stress is assumed to be positive for compression. Hence, with
respect to Fig.H3, let σ = − σ , σ
11 1 22 = − σ 3 and σ12 = 0 in eqs. (d11) and (d13). It follows that
σ1 + σ 3 − σ1 − σ 3 cos 2ψ
σ −σ rr
= =
2 2
, (h42)
σ − σ
τ = − σ rθ = 1 2 3 sin 2ψ
where σ , τ are so-called “ normal stress” and “ tangential (shear) stress” respectively, acting on a plane inside the material.

π
The plane is at an angle ψ=θ− with the direction of the maximum compressive stress. The outer-pointing normal at that
2
plane makes an angle θ with the direction of the maximum compressive stress. With respect to a σ − τ reference system,

eqs.(h42) are the parametric equations of the MOHR circle (see Section D4), plotted in Fig. H3. In the most general case, stress
field is varying from point to point inside the material. Hence both the eigen-vectors of the stress tensor (i.e. the local directions
of the maximum / minimum compressive stress) and the eigen-values of that tensor (i.e. the magnitudes of the maximum /
minimum compressive stress) are also changing from point to point. For a fixed point inside the material, both normal stress
and shear (tangential) stress are varying with the angle between the plane (with respect to normal and tangential stresses are
defined) and the direction of the local maximum compressive stress. Consider a certain point inside the material and imagine
various planes passing through that point. Hence, according to TRESCA empirical criterion, failure of the material is produced
here if
max τ = σY (h43)
ψ
Using eq.(h42b), it follows
σ1 − σ 3 sin 2ψ = σ1 − σ 3
max 2 2 max sin 2ψ = σ Y (h44)
ψ ψ
 75
Consider now a homogeneous stressed material subject to progressive increasing values of the difference σ1 − σ 3 . The
material is characterised by a yield value denoted by σ Y . Eq.(h44) shows that:
• if (σ1 − σ 3) / 2 < σ Y , there is no failure inside the material;
• when the equality
σ1 − σ 3 = 2 σ Y (h45)
π
is attended, a failure is produced along the planes at angles ψ=± with the direction of the maximum compressive stress.
4

Fig.H3.(a) The MOHR circle;

(b) A plane inside the material, at an angle ψ with the direction of the maximum compressive stress;
(c) TRESCA criterion. E.d and F.D. denote the elastic domain and the failure domain respectively;
(d) COULOMB - NAVIER criterion.

Eq.(h45) represents the TRESCA criterion in terms of the eigenvalues of the stress tensor. In the case of a material subject to a
non-homogeneous stress field, eq.(h45) is a local condition. Here, the eigen-values are obtained (see Section A). For a
compressive stress, that values are expected to be negative ones. They have to be denoted by − σ1 ,− σ 2 ,− σ 3 , where

σ 3 < σ 2 < σ1 .
A second criterion is due to COULOMB and NAVIER. It can be used to describe only the shear fracture. According to it, a
material is characterised by the cohesive strength denoted by S and by the coefficient of friction, denoted by µ * = tan φ .
Here, φ is the angle of internal friction ( φ = 30 in most rocks). According to COULOMB - NAVIER empirical criterion,
o
shear failure of the material is produced at its points where
 
 
max  τ − µ* σ = S (h46)
ψ  
Using eqs.(h42), it follows
 76
σ − σ
3 sin 2ψ − tan φ σ1 + σ 3 − σ1 − σ 3 cos 2ψ   = S

 1
max  2 
 2 2 

(h47)
ψ 

If ψ is a solution of (h47), π − ψ (or just −ψ ) is a solution too. Hence, without loss of generality, the values of the angle
ψ will be limited to the first quadrant, where eq.(h47) is

σ1 − σ 3 σ σ +
max [sin(2ψ + φ)] = S + tan φ 1 2 3
1
(h48)
2 cos φ
ψ
Eq.(h48) shows that:

• if
σ1 − σ 3 1 σ + σ 3 < S , there is no failure of the material;
− tan φ 1
2 cos φ 2
• when the equality
σ1 − σ 3
1 σ + σ3 = S
− tan φ 1 (h49)
2 cos φ 2
 π φ
is attended, a share fracture is produced along the planes at angles ψ = ± −  with the direction of the maximum
 4 2
compressive stress. Eq.(h49) represents the COULOMB - NAVIER criterion in terms of the eigenvalues of the stress tensor.
Taking into account that µ * = tan φ , eq.(h49) can be written as

 2   2 
 µ *  + 1 − µ *  −  µ *  + 1 + µ *  = 2S
σ1   


σ3   


, (h50)
   
outlining that COULOMB - NAVIER criterion is a generalisation of TRESCA criterion for a non-zero internal friction.

H.6) Von MISES-HENCKY criterion for ductile flow (plasticity).

Because the ductile (plastic) flow is independent of the co-ordinate system used, it depends only on the invariants of the stress
tensor (Section A, eq.(a22)). Hence an equation like

f (I1 , I 2 , I 3 ) = 0 , (h51)

will be valid. There are strong experimental evidence that the plastic flow does not depend on the hydrostatic pressure, being
also similar for compressive and tensile states of stress. It follows the function f in (h51) depends only on the second invariant
of the stress deviator (see eq.(a23)). Hence, ductile flow occurs only at those points of material where the second invariant of
the deviator stress reaches a certain value, depending on the nature of the material. Using eq.(a24), the criterion of Von MISES
- HENCKY assumes that ductile flow occurs at those points of the material where

(σ11 − σ 22)2 + (σ 22 − σ 33)2 + (σ 33 − σ11)2 + 6 σ122 + σ 223 + σ132 = 6k 2 (h52)

In terms of the principal stresses, eq.(h52) is

(σ1 − σ 2)2 + (σ 2 − σ3)2 + (σ3 − σ1)2 = 6k 2 (h53)

Hence the criterion of Von MISES - HENCKY can be regarded too as a generalisation of TRESCA criterion, by taking into
account the presence of the intermediate stress.
 77
H.7) Rheological models.

a) SAINT-VENANT body (elastic-plastic material).

The behaviour of that material is characterised by linear elasticity for stress values below the yield strength. When the yield
stress is attended, the body exhibits a pure plasticity. Its constitutive equation (using deviator tensors) has the symbolic form

σ * = 2µ S ε * , σ < σY

 (h54)
 * *
f ( I 2 , I 3 ) = σ Y , σ = σY
The above material has the mechanical analogue presented in Fig. (h4a), being referred as a SAINT-VENANT body.

b) BINGHAM body (visco-plastic material).

Similar to the SAINT-VENANT body, that material exhibits linear elasticity for stress values lower than the yield strength.,
but flows linearly above that value. The strain rate is proportional to the difference between the deviatoric stress and the yield
strength. Its constitutive equation is

σ * = 2µ B ε * , σ < σY

 (h55)
 * d ε*
σ = σ Y + 2 η B dt , σ ≥ σY

The above material has the mechanical analogue presented in Fig. (h4b), being referred as a BINGHAM body.

Fig.H4. (a) The SAINT-VENANT body; (b) The BINGHAM body.

 78
I) THE ACCRETION WEDGE.
I1) The model.
Consider a 2-D prismatic body having a triangular vertical section (Fig.I1), in the presence of gravity. The wedge rests on a
rigid basement having the slope equal to θ 0 . Both the compressional force acting on the left side of the wedge and the friction
to the basement cause thickening of the incompressible material and the development of a topographical slope equal to α . It is
assumed that the material is into a state of plastic yielding according to the VON MISES-HENCKY criterion. It follows to
obtain a condition relating the slopes of the topography and that of the basement to the geometry of the wedge, its yield strength
and the friction coefficient to the basement.

Fig.I1. The 2-D accretion wedge.

I2) Equations of equilibrium. Yield condition. Stress field.

Taking into account that the 2-D case is discussed, the stress in polar co-ordinates is
 
 σ rr σ rθ 0 
 
σ =  σ rθ σ θθ
0  . (i1)
 
 0
 (
0 ν σ rr + σ θθ  )
Because the material is assumed to be incompressible, the POISSON coefficient is ν = 1 / 2 .
Using polar co-ordinates, the equilibrium equation (d40)-(d41) in the presence of gravity are
∂ σ rr ∂ σ rθ σ rr − σ θθ + ρg sin θ = 0
+ + (i2)
∂r r∂θ r
∂ σ rθ ∂ σ θθ σ rθ + ρg cos θ = 0
+ +2 (i3)
∂r r∂θ r
Using (i1), the yield condition (r54) is
σ1 − σ 3 = 2 k (i4)
Taking into account again that the compressive stress is assumed to have positive sign, eqs.(d11)-(d13) give
σ = − f ( r , θ ) − k cos 2ψ
 rr
σ θθ = − f ( r , θ ) + k cos 2ψ , (i5)

σ rθ = k sin 2ψ

1
2 tr σ
where the trace of the stress has been denoted by f ( r, θ ) = ( ) and ψ = ψ ( r, θ ) is the angle between the radius and
the local direction of the maximum compressive stress (variable inside the wedge). However, it will be assumed that
ψ = ψ (θ ) only. After some manipulations, substituting (i5) into (i2)-(i3) gives
 ∂f 2k dψ 2 k
 ∂r = r cos 2ψ dθ − r cos 2ψ + ρg sin θ
 (i6)
 ∂f = −2k sin 2ψ dψ + 2k sin 2ψ + ρgr cos θ
 ∂θ dθ
 79
Eqs.(i6a) and (i6b) are differentiated with respect to θ , r respectively, the results of the differentiation being equal each other.
After some elementary manipulations, it follows that
d  dψ 
 cos 2ψ − cos 2ψ  = 0 , (i7)
dθ  dθ 
i.e.
dψ C
=1+ , (i8)
dθ cos 2ψ
where C is a constant of integration. Substituting (i8) into (i6a) gives
∂f 2 Ck
= + ρg sin θ , (i9)
∂r r
i.e.
f = 2Ck ln r + ρgr sin θ + g(θ ) , (i10)
To find the unknown function g = g(θ ) , eq.(i10) is substituted into (i6b) to obtain
dg
= −2Ck tan 2ψ . (i11)
dθ
Using (i8), it follows that
dg sin 2ψ
= −2Ck , (i12)
dψ cos 2ψ + C
i.e.
g = Ck ln( C + cos 2ψ ) + A , (i13)
where A is another constant of integration. Hence the final stress inside the wedge is
σ = −2 Ck ln r − Ck ln( C + co2ψ ) − ρgr sin θ − A − k cos 2ψ
 rr
σ θθ = −2 Ck ln r − Ck ln( C + co2ψ ) − ρgr sin θ − A + k cos 2ψ . (i14)

σ rθ
= k sin 2ψ

I3) Boundary conditions. Final results.

Consider the segment AC placed on the side OA of the wedge, having θ = 0 and OC ≤ r ≤ OA , where the point C is very
→ →
closed to the point A. The outward pointing normal vector is n = − e θ . Here, is acting the lithostatic pressure due to the
topography. Hence
 → →
σ  θ 
 − e  = ρgr tan α e θ , (i15)

i.e.
σ rθ (θ = 0) = 0
 . (i16)
σ θθ (θ = 0) = −ρgr tan α
Because the angle θ 0 has very small values, it will be assumed for all angles θ that

σ θθ = −ρgr tan α , (i17)

or
∂ σ θθ
= −ρg tan α . (i18)
∂r
However, all the next derivations are supposed to be valid at the rear of the wedge, where the topography is generated due to
the horizontal compression, i.e. the radius r is a mean value of the lengths OA and OC, the point C being closed to A. By
differentiating (i14b), eq.(i18) leads to
2Ck
= ρg tan α , (i19)
r
where
 80
h
r≅ 0 . (i20)
θ0
→ →
Consider now the side OB of the wedge, having θ = θ 0 and the outward pointing normal vector n = e θ . Here, is acting
the friction force due to the basement, assumed to have the magnitude equal to λk , where λ is a friction coefficient. Hence
→ →
σ θe = λ k e r , (i21)
i.e.
σ rθ (θ = θ 0 ) = λk . (i22)
The next partial derivative follows to be evaluated in two ways. In the first approach, eqs.(i22), (i16a) and (i20) are used to
give
∂ σ rθ 1 σ rθ (θ = θ 0 ) − σ rθ (θ = 0) λk
≈ = . (i23)
r∂θ r θ0 − 0 h0
The dominant stress is into the wedge is the horizontal compression. Hence ψ ≅ θ , both angles having small values. It follows
cos 2ψ ≅ 1 . Eqs.(i5c), (i8), (i19) and (i20) give
∂ σ rθ 2k dψ 2k 2Ck 2k 2kθ 0
= cos 2ψ = ( C + cos 2ψ ) ≈ + = ρg tan α + (i24)
r∂θ r dθ r r r h0
From (i23) and (i24) it follows
ρgh 0 tan α + 2 kθ 0 ≅ λk , (i25)
showing that the friction force (resistance to sliding of the wedge onto the basement) is balanced by tqo forces. The first one is
due to the topography and the second force is related to the compressive stress and to the slope of the basement. Further details
related to the application of eq.(i25) in real cases are presented by Ranalli (1987).
 80

General references

Aki, K., Richards, P.G., 1980. Quantitative Seismology, Freeman and Co., New York

Backus, G., Continuum Mechanics, Samizdat Press (http://samizdat.mines.edu)

Bayly, B., 1992, Mechanics in Structural Geology, Springer - Verlag, N.Y.

Filonenco-Borodici, M.M., 1952, Theory of Elasticity, Ed.tehnicã, Bucharest

Ivan, M., 1996, Seismology (in Romanian), v.I, Ed.Universitãtii Bucuresti

Ivan, M., 1997a. A finite element algorithm for computing the flexure of the crustal, in press, Studii si
cercetari de GEOFIZICA 35

Ivan, M., 1997b. Aperiodic motion of a crustal microplate, in press, Revue Roumaine de Géophysique 41

Jaeger, J.C., Cook, N.G.W., 1969, Fundamental of Rock Mechanics, Methuen and Co. London

Kennett, B., Continuum Mechanics, Samizdat Press (http://samizdat.mines.edu)

Mandl, G., 1988, Mechanics of tectonic faulting, Elsevier, Amsterdam

Meissner, R., 1986, The Continental Crust - A Geophysical Approach, Academic Press, Inc, N.Y.

Nowacki, W., 1961. Dynamika budowli (in Polish), Arkadi, Warszawa

Ranalli, G., 1987, Rheology of the Earth, Allen & Uniwin Inc., Boston

Rîjic, I.M., Gradstein, I.S., 1955, Tables of integrals, sums, series and products, -in Romanian, Ed.tehnicã,
Bucharest

Sokolnikoff, I.S., Redheffer, R.M., 1958, Mathematics of Physics and Modern Engineering, McGraw-Hill
Book Co., Inc., New York

Timoshenko, S., Woinowsky-Krieger, S. 1959. Theory of plates and shells, Mc-Graw-Hill, New York

Timoshenko, S.P., Gere, J.M., 1961, Theory of elastic stability, McGraw-Hill, New York

Turcotte, D.L., Schubert, G. 1982. Geodynamics, John Wiley and Sons, New York

Wahr, J., 1996, Gravity and Geodesy - Class Notes, Samizdat Press