CONTENTS x

Appendix B Matrix Notation

B.1 Introduction B-1
B.2 Definition of Matrix Notation B-2
B.3 Matrix Transpose and Scalar Multiplication B-4
B.4 Definition of a Numerical Operation B-6
B.5 Programming Matrix Multiplication B-6
B.6 Order of Matrix Multiplication B-7
B.7 Summary B-7

Appendix C Solution or Inversion of Linear Equations

C.1 Introduction C-1
C.2 Numerical Example C-2
C.3 The Gauss Elimination Algorithm C-3
C.4 Solution of a General Set of Linear Equations C-6
C.5 Alternative to Pivoting C-6
C.6 Matrix Inversion C-9
C.7 Physical Interpretation of Matrix Inversion C-11
C.8 Partial Gauss Elimination, Static Condensation and Substructure
Analysis C-13
C.9 Equations Stored in Banded or Profile Form C-15
C.10 LDL Factorization C-16
C10.1 Triangularization or Factorization of the A Matrix C-17
C10.2 Forward Reduction of the b Matrix C-18
C10.3 Calculation of x by Backsubstitution C-19
C.11 Diagonal Cancellation and Numerical Accuracy C-20
C.12 Summary C-20
C.13 References C-21

Appendix D The Eigenvalue Problem

D.1 Introduction D-1
D.2 The Jacobi Method D-2
D.3 Calculation of 3d Principal Stresses D-4
D.4 Solution of the General Eigenvalue Problem D-5
D.5 Summary D-6
CONTENTS xi

Appendix E Transformation of Material Properties

E.1 Introduction E-1
E.2 Summary E-4

Appendix F A Displacement-Based Beam Element With Shear

Deformations
F.1 Introduction F-1
F.2 Basic Assumptions F-2
F.3 Effective Shear Area F-5

Appendix G Numerical Integration

G.1 Introduction G-1
G.2 One-Dimensional Gauss Quadrature G-2
G.3 Numerical Integration in Two Dimensions G-4
G.4 An Eight-Point Two-Dimensional Rule G-5
G.5 An Eight-Point Lower Order Rule G-6
G.6 A Five-Point Integration Rule G-7
G.7 Three-Dimensional Integration Rules G-8
G.8 Selective Integration G-11
G.9 Summary G-11

Appendix H Speed of Computer Systems

H.1 Introduction H-1
H.2 Definition of One Numerical Operation H-1
H.3 Speed of Different Computer Systems H-2
H.4 Speed of Personal Computer Systems H-3
H.5 Paging Operating Systems H-3
H.6 Summary H-4

Appendix I Method of Least Square

I.1 Simple Example I-1
I.2 General Formulation I-3
I.3 Calculation Of Stresses Within Finite Elements I-4
CONTENTS xii

Appendix J Consistent Earthquake Acceleration and Displacement

Records
J.1 Introduction J-1
J.2 Ground Acceleration Records J-2
J.3 Calculation of Acceleration Record From Displacement Record J-3
J.4 Creating Consistent Acceleration Record J-5
J.5 Summary J-8

Index
 1.

MATERIAL PROPERTIES

 Material Propertie s Must Be Evaluate d 

By Labo ratory o r Field Tests

1.1 INTRODUCTION

The fundamental equations of structural mechanics can be placed in three

categories[1]. First, the stress-strain relationship contains the material property
information that must be evaluated by laboratory or field experiments. Second,
the total structure, each element, and each infinitesimal particle within each
element must be in force equilibrium in their deformed position. Third,
displacement compatibility conditions must be satisfied.

If all three equations are satisfied at all points in time, other conditions will
automatically be satisfied. For example, at any point in time the total work done
by the external loads must equal the kinetic and strain energy stored within the
structural system plus any energy that has been dissipated by the system. Virtual
work and variational principles are of significant value in the mathematical
derivation of certain equations; however, they are not fundamental equations of 
mechanics.

1.2 ANISOTROPIC MATERIALS

The linear stress-strain relationships contain the material property constants,

which can only be evaluated by laboratory or field experiments. The mechanical
material properties for most common material, such as steel, are well known and
are defined in terms of three numbers: modulus of elasticity E , Poisson’s ratio
1-2 STATIC AND DYNAMIC ANALYSIS

ν  and coefficient of thermal expansion α . In addition, the unit weight w and the
unit mass ρ are considered to be fundamental material properties.

Before the development of the finite element method, most analytical solutions in
solid mechanics were restricted to materials that were isotropic (equal properties
in all directions) and homogeneous (same properties at all points in the solid).
Since the introduction of the finite element method, this limitation no longer
exists. Hence, it is reasonable to start with a definition of anisotropic materials,
which may be different in every element in a structure.

The positive definition of stresses, in reference to an orthogonal 1-2-3 system, is

shown in Figure 1.1.

σ 3

τ 13 τ 23
τ 32
τ 31 σ 2
τ 21 τ 12 2
σ 1

 Figure 1.1 Definition of Positive Stresses

All stresses are by definition in units of force-per-unit-area. In matrix notation,

the six independent stresses can be defined by:

T
f  = [σ 1 σ 2 σ 3 τ 21 τ 31 τ 23 ] (1.1)
MATERIAL PROPERTIES 1-3

From equilibrium, τ 12 = τ 21 , τ 31 = τ 13 and τ 32 = τ 23 . The six corresponding
engineering strains are:

d
T
= [ε 1 ε 2 ε 3 γ 21 γ 31 γ 23 ] (1.2)

The most general form of the three dimensional strain-stress relationship for
linear structural materials subjected to both mechanical stresses and temperature
change can be written in the following matrix form[2]:

 1 ν 12 ν 13 ν 14 ν 15 ν 16 

  E  − − − − −
 E 2  E 3  E 4  E 5  E 6 
 1 
− ν 21 1
−
ν 23
−
ν 24
−
ν 25
−
ν 26 
 ε 1    E   E 2  E 3  E 4  E 5  E 6   1 
σ   α 1 
 ε    1    α  
 2  − ν 31 ν 32 1 ν 34 ν 35 ν 36  σ 2   2
− − − − 
 ε 3    E 1  E 2  E 3  E 4  E 4  E 6  σ 3   α 3 
  =  ν  ν 42 ν 43 1 ν 45 ν 46  τ 21 
+ ∆T   (1.3)
γ 21  − 41 − − − −   α 21 
γ 31    E 1  E 2  E 3  E 4  E 5  E 6  τ   α 31 
   ν 51 ν 56   
31
ν 52 ν 53 ν 54 1  
γ 23  − − − − −   23 
τ  α 23 
 E 1  E 2  E 3  E 4  E 5  E 6
 
− ν 61 −
ν 62
−
ν 63
−
ν 64
−
ν 65 1 
  E 1  E 2  E 3  E 4  E 5  E 6 

Or, in symbolic matrix form:

d = Cf + ∆T a (1.4)

The C matrix is known as the compliance matrix and can be considered to be the
most fundamental definition of the material properties because all terms can be
evaluated directly from simple laboratory experiments. Each column of the C
matrix represents the strains caused by the application of a unit stress. The
temperature increase ∆T is in reference to the temperature at zero stress. The a
matrix indicates the strains caused by a unit temperature increase.

Basic energy principles require that the C matrix for linear material be
symmetrical. Hence,
1-4 STATIC AND DYNAMIC ANALYSIS

ν ij ν  ji
= (1.5)
 E  j  E i

However, because of experimental error or small nonlinear behavior of the

material, this condition is not identically satisfied for most materials. Therefore,
these experimental values are normally averaged so that symmetrical values can
be used in the analyses.

1.3 USE OF MATERIAL PROPERTIES WITHIN COMPUTER

PROGRAMS

Most of the modern computer programs for finite element analysis require that
the stresses be expressed in terms of the strains and temperature change.
Therefore, an equation of the following form is required within the program:

f  = Ed + f 0 (1.6)

in which E = C-1 . Therefore, the zero-strain thermal stresses are defined by:

f 0 = - ∆T  Ea (1.7)

The numerical inversion of the 6 x 6 C matrix for complex anisotropic materials

is performed within the computer program. Therefore, it is not necessary to
calculate the E matrix in analytical form as indicated in many classical books on
solid mechanics. In addition, the initial thermal stresses are numerically
evaluated within the computer program. Consequently, for the most general
anisotropic material, the basic computer input data will be twenty-one elastic
constants, plus six coefficients of thermal expansion.

Initial stresses, in addition to thermal stresses, may exist for many different types
of structural systems. These initial stresses may be the result of the fabrication or
construction history of the structure. If these initial stresses are known, they may
be added directly to Equation (1.7).
MATERIAL PROPERTIES 1-5

1.4 ORTHOTROPIC MATERIALS

The most common type of anisotropic material is one in which shear stresses,
acting in all three reference planes, cause no normal strains. For this special case,
the material is defined as orthotropic and Equation (1.3) can be written as:

 1 ν 12 ν 13 
  E  − − 0 0 0 
 E 2  E 3
 1 
− ν 21 1 ν 23
 ε 1    E  − 0 0 0   σ 1  α 1 
 E 2  E 3 
 ε    1
 σ 2  α  
 2  − ν 31 −
ν 32 1
0 0 0    2
 ε 3    E   E 2  E 3  σ 3  + ∆T α 3 
 = 1
1  τ 21    (1.8)
γ 21   0 0 0 0 0   0
γ 31   G4  τ 31  0
   1     
γ 23   0 0 0 0 0 τ 23   0 
G5 
 
 0 1 
0 0 0 0
 G6 

For orthotropic material, the C matrix has nine independent material constants,
and there are three independent coefficients of thermal expansion. This type of 
material property is very common. For example, rocks, concrete, wood and many
fiber reinforced materials exhibit orthotropic behavior. It should be pointed out,
however, that laboratory tests indicate that Equation (1.8) is only an
approximation to the behavior of real materials.

1.5 ISOTROPIC MATERIALS

An isotropic material has equal properties in all directions and is the most
commonly used approximation to predict the behavior of linear elastic materials.
For isotropic materials, Equation (1.3) is of the following form:
1-6 STATIC AND DYNAMIC ANALYSIS

 1 ν  ν  
  E  − − 0 0 0
 E   E 
 
 ε 1  − ν  1
−
ν 
0 0 0   σ 1  1
 ε     E   E   E   σ   1
   ν  ν     
0  σ  
2 1 2
− −
 ε 3    E   E   E  0 0 1
 3
 = 1  τ   + α  ∆T 0 (1.9)
γ 21   0 0 0 0 0   21   
γ 31   G  τ 31  0
   1    
γ 23   0 0 0 0 0  τ 23  0
G  
 1
 0 0 0 0 0 
 G

It appears that the compliance matrix has three independent material constants. It
can easily be shown that the application of a pure shear stress should result in
pure tension and compression strains on the element if it is rotated 45 degrees.
Using this restriction, it can be shown that:

 E 
G= (1.10)
2(1 + ν )

Therefore, for isotropic materials only Young's modulus E and Poisson's ratio ν 
need to be defined. Most computer programs use Equation (1.10) to calculate the
shear modulus if it is not specified.

1.6 PLANE STRAIN ISOTROPIC MATERIALS

If  ε 1 , γ 13 , γ 23 , τ 13 , and τ 23 are zero, the structure is in a state of plane strain. For
this case the compliance matrix is reduced to a 3 x 3 array. The cross-sections of 
many dams, tunnels, and solids with a near infinite dimension along the 3-axis
can be considered in a state of plane strain for constant loading in the 1-2 plane.
For plane strain and isotropic materials, the stress-strain relationship is: