Chapter 3

Numerical Model

Development

3.1 Introduction

The finite element methods are widely used in solution of engineering problems. This
method was initially developed on physical basis for the analysis of problems in engineering
mechanics. However it is recognized that it can be vastly used in many other classes of
problems. The development of finite element method for solution of practical engineering
problems began with the initiation of digital computers. Fundamental nature of finite element
method is to establish a set of governing algebraic equations and solve, and it is only through
computers this method is rendered effective.

The physical problem involves an actual structure or components of structure subjected to

loadings. The idealization of physical problem to mathematical model requires certain
assumption that leads to differential equations governing mathematical model. The finite
element analysis solves the mathematical model. Since finite element solution technique is a
numerical procedure, it is necessary to judge the accuracy (Figure 3.1). If the accuracy

1
criterion is not met the numerical solution has to be repeated with refined parameters i.e. finer
mesh until a sufficient accuracy is not reached (Klaus-Jurgen Bathe).

2
 Finite Element Methodology
A computer based finite element method for solution of problems of continuum
mechanics was proposed in 1956. Even in the early stage it became apparent that the method
possessed inherent capabilities of accommodating highly complex characteristics of the
continuum into analysis. For example the method could easily tackle the aspects such as:
complex geometrical configurations, non-homogeneity and anisotropy of the continuum,
nonlinear behavior of the phenomenon being analyzed, unusual boundary conditions etc.
Consequently, the method could trigger a technological revolution. In the past 30 years the
finite element method has been developed to such an extent that it has become an accepted
tool of analysis for all types of engineering problems. For example the problems of structural
mechanics, fluid mechanics, fracture mechanics, heat transfer etc.

Physical Problem Change of Physical problem

Mathematical Model governed by

differential equations Assumptions on
Geometry
Kinematics Improve Mathematical Model
Material law
Loading
Boundary conditions

FE solution choice of
Finite Elements
Mesh density
Solution parameters representation
of
Loading
Boundary conditions Refine mesh, solution of
parameters etc.

Assessment of accuracy of FE
solution of mathematical model

Interpretation of Results Refine Analysis

Design improvements structural

optimization

Figure 3.1. The process of Finite Element Analysis (Source: Finite Element analysis by Klaus
Jurgen Bathe, Sec.1.2, pp. 3)

3
 Finite Element Methodology
Finite element method is derived through three principle steps. These are:

1. Finite element idealization of the problem being analyzed: The structure being considered
is transformed into a finite element idealization. The same is referred to as idealized
system.
2. Formulation and solution of equations related to the phenomenon governing the
idealized problems.
3. Evolution of structural response of the element used in the idealization system.

3.2 Definition and Notations

The structural system is referred to global (x, y, z) reference axes shown in Figure 3.2 and
constitute the horizontal plane and z - axis is
oriented in vertical direction. At any point over the structural system, there exists six degrees
of freedom. These are translations (u, v, w) along (x, y, z) axes and rotations ,

about (x, y, z) axes as illustrated in Figure 3.3. A load at a point over a structural domain is
expressed in six components. They are forces ( FX , FY , FZ ) in (x, y, z) direction and couples

( CX , CY , CZ
) about (x, y, z) axes.

(0, 0, 0) y,(v)

x,(u)

z,(w)
x,

Free edge

Figure 3.2. Notation of problem

The spatial structural systems have presence of two types of members in their structural
constitution. These are 1 - D member i.e. the member whose structural behavior could be
derived through their meridian chords. For example segment of standard I - section has its
structural behavior closely related to the behavior of its neutral axis. Two dimensional
4
 Finite Element Methodology
members, whose structural behavior could be derived through their meridian surfaces, for
example segment of a slab having its structural behavior closely related to the behavior of its
neutral surface. Finite element solution technique deals with the analysis of the finite element
idealization of the structural systems and in case of spatial structural systems; the idealization
would require application of 1 - D and 2 - D finite elements. This is because in case of one-
dimensional members, it is sufficient to represent the meridian chord whereas the two
dimensional members could be adequately represented through the idealization of the
meridian surface.

29
28 24
27 23
19
Y 26 22 18
25 21 17 14
20 13 X
16 9
12 8
15 11 7
10
6
5

3
2

1
Figure 3.3. Illustration of basic elements

For deriving an algorithm for a general purpose finite element analysis program, the
element should be simple as well as versatile. In other words, the element should have wide
range of applicability. In this context a 2 - node line element and a 3 - node triangular plate
element hold bright promise. In Figure 3.4 these basic elements along with their illustrative
and application are presented.

In this chapter, we shall discuss the characteristics of the proposed line, triangular plate,
four node quadrilateral plate elements and eight node hexahedral elements. The details are
derived in two stages. In the first stage, the local reference axis is used to derive the element
behavior and in the next stage the local characteristics are transformed into the global

5
 Finite Element Methodology
characteristics i.e. the required characteristics. In the analysis the structural system is
transformed into assembly of line elements and plate elements. The basic aim of the analysis
is to derive the nodal displacement and the nodal reaction at the constrained nodes. The
constraints vary from the need to maintain the static structural system. The details eventually
permit evaluation of the element structural response.

3.3 Line Element

The line element is denoted through its node numbers (1 - 2), and it is referred to local set
of orthogonal axes (x, y, z) as shown in Figure 3.4 (a - e). In this x - axis is aligned in the axial
direction (1 - - axes are oriented in the directions
parallel to the principal flexural axes of the section of the member. A line element displays
four independent modes of deformation. These are axial mode, torsion mode and flexural
modes around each of the principal flexural axes of the element section.

(a) 2- Node line element (b) 3- Node triangular plate element

(c) Idealization of a one-dimensional member through assembly of line elements

(d) Idealization of triangular element

6
 Finite Element Methodology

Y Y
Principle
X
1 X 2 Flexural
Axis
Z
Z

(e) 2 - Node line element- Local reference axis

Figure 3.4 (a e). Basic features of finite element idealization

3.3.1 Axial Mode

Axial mode of deformation is characterized by translation - direction and force

Fx also in x - direction. Consequently, the element nodal displacements and element nodal

loads are as shown in Figure 3.5 (a c). These are denoted through the vectors [ a
] and [ Fa

] as defined in Eq.3.1.

U2 Fx Fx
U1
X 1 2
1 2

a) Element nodal displacement b) Element nodal loads

Px /unit length

1 2

c) Member force
Figure 3.5 (a c). Characteristics of axial mode of deformation

= and = (3.1)

Equilibrium of the element in axial mode is defined by Eq.3.2;

(3.2)

Whereas, is the stiffness matrix in the axial node defined by Eq.3.3.

Assuming linear variation of over the element length, it can be shown that;

7
 Finite Element Methodology

(3.3)

Where, E =modules of elasticity of the element material.

A = area of the element section

L= length of element.

For uniformly distributed load of intensity PX over the element length as shown in Figure
3.5 (c).

(3.4)

3.3.2 Torsional Mode

Torsional mode of deformation is characterized by rotation ( x

) about x - axis and couple

( Cx ) around x - axis. Consequently the element nodal displacements and the element
nodal loads are as shown in Figure 3.6 (a) and Figure 3.6 (b) respectively. These are
denoted
through vectors t and Ft and defined in Eq.3.5

Cx1 Cx2
X
X
1 2 1 2
a) Element nodal displacement b) Element nodal loads

Figure 3.6 (a b). Characteristics of torsional mode of deformation

= (3.5)

Equilibrium of the element in torsional mode is defined by Eq.3.6.

(3.6)

Whereas; Ka is the stiffness matrix in the axial mode.

Assuming linear variation of ( x ) over the element length, it can be shown through Eq.3.7.
8
 Finite Element Methodology

(3.7)

Wherein, G = the shear modulus of the element material,

J = torsional constant of the element section,

L = the element length.

G is given by Eq.3.8.

G= (3.8)

Wherein; E and element

material for the types of structural systems under consideration Ft 0.

3.3.3 Flexural Mode around Y - axis

Flexural mode of the deformation with the curvature around y - axis characterized by

translation in z - direction, rotation around y - axis, force in z - direction and

couple around y - axis. Hence, the element nodal displacement and element nodal loads
and uniformly distributed load are as shown in Figure 3.7 (a - c). These are denoted through

the vectors defined in Eq.3.9.

x 1 2

a) Element nodal displacement

9
 Finite Element Methodology

2 1 2
1

c) Member force

b) Element nodal loads

Figure 3.7 (a c). Characteristics of flexural mode of deformations around Y axis

= (3.9)

Equillibrium of the element in the flexural mode around Y- axis is defined by Eq.3.10.

[ K y ][ y
] = [ Fy ] (3.10)

Wherein; [ K y ] is the stiffness matrix in flexural mode around Y- axis.

Assuming pure flexural action defined by y dw dx and also assuming cubic variation of

w over the length, it can be shown that (Eq.3.11);

12 6L 12 6L
EI y 6L 4L 2
6L 2L2
[ Ky ] (3.11)
L3 12 6L 12 6L
=
6L 2L2 6L 4L2

Wherein; E is the modulus of elasticity of the element material, I is the moment of inertia of
y

the element section around Y-axis and L is element length.

For the case of uniformly distributed load of intensity
Pz over the element length shown in
Figure 3.7(c); Fy is given by Eq.3.12.

10
 Finite Element Methodology
L2
L2 2
[ Fy ] = (3.12)
L2
Pz
L2 2

3.3.4 Flexure Mode around Z axis

Flexural mode of deformation with the curvature around z - axis characterized by

- direction, rotation ( z ) around z - axis, force ( Fy ) in y - direction and

Cz - axis. Hence, the nodal displacement, the nodal loads and uniformly

distributed load are as shown in Figure 3.8 (a - c). These are denoted through the vectors [ z ]
and [ Fz ] defined in Eq. 3.13.

v Fy1
1 1 Cz1
[ z ]= ; Fy1 (3.13)
z F
y2
v2
2 Cz 2

1 2
1 2

b) Member force

a) Element nodal loads

X 1 2

Z
c) Element nodal displacement

Figure 3.8 (a c). Characteristics of flexural mode of deformations around Z axis

11
 Finite Element Methodology
Equillibrium of the element in the flexural mode around Y- axis is defined by Eq.3.14.

[ K z ][ z ] = [ Fz ] (3.14)

Wherein; [ K z ] is the stiffness matrix in flexural mode around Z- axis given by Eq.3.15

Assuming pure flexural action defined by z dv dx and also assuming cubic variation of
v over the element length, it can be shown that;

EI z 12 6L 12 6L
[K ]= 6L 4L2 6L 2L2
z
L3 (3.15)
12 6L 12 6L
6L 2L2 6L 4L2

Wherein; E is the modulus of elasticity of the element material, I is the moment of inertia of
z

the element section around Z - axis and is element length.

For the case of uniformly distributed load of intensity Py over the element length shown in
Figure 3.8 (c), Fz is given by Eq.3.16.
L2
L2 2
[ Fz ] = Py (3.16)
L2
L2 2

3.3.5 Element Local Equilibrium

In local reference system (x, y, z) a point over an elelement (1 - 2) has six degrees of
freedom represented by translation ( u, v, w ) in (x, y, z) directions and rotation
x , , )
( y z

around (x, y, z) axes. Consequently, the elelement nodal degrees of freedom in local system
are as shown in Figure 3.9 (a). A point over the elelement could be subjected to six load
components represented by forces (fx, fy, fz) in (x, y, z) directions and couples (Cx, Cy, Cz)
around (x, y, z) axes. Consequently the element nodal load components in the local system
are as shown in Figure 3.9 (b).

In the local system of reference the elelement nodal displacements and elelemnt nodal
loads are represented through vectors [ L
] and [ F L ] respectively, and the same are as
in e e
12
 Finite Element Methodology
Eq.3.17.

13
 Finite Element Methodology

L
1 F1L
L
e
; L
Fe (3.17)
L L
2 F2

L
Such that the sub vectors ( , F L ) are as shown in Eq. 3.18;
1 1

u1 f x1
v1 f y1
L
w f z1
1 1
; fL (3.18)
1
c x1
y1
cy1
z1
cz1

Equillibrium of the element in the local system of reference is as given in Eq.3.19

L L L
Ke e fe (3.19)

Wherein; [ K eL ] is the element local stiffness matrix.

It may be noted that the vectors L

and L
could be constructed from appropriate
e
fe
model vectors considered above. Therefore [ K L ] could be built up through appropriate
e
placement of the coefficient of the model stiffness matrices. The details are presented in Table
3.1.(Annexure I).

3.3.6 Element Global Equilibrium

In global reference system (x, y, z) a point over an element (1 - 2) has six degrees of

freedom represented by translation (u, v, w) in (x, y, z) directions and rotations ( x y

,
,
z

) around (x, y, z) axes. A point over the element could be subjected to six load components
represented by forces (Fx, Fy, Fz) in (x, y, z) directions and couples (Cx, Cy, Cz) around (x, y, z)
axes.

In global system of reference the element nodal displacements and element nodal loads
are represented through vectors [ e ] and [Fe] respectively and the same are as defined in

14
 Finite Element Methodology
Eq.3.20.

15
 Finite Element Methodology
F1
e
1
; Fe (3.20)
2 F2

a) Element nodal displacement and local reference system

b) Element nodal loads and local reference system

X
Y

Z
Z c) Arbitrary placement of element in
the global space

Figure 3.9 (a c). Characteristics of two node line element.

16
 Finite Element Methodology
Such that the sub vectors ( 1 , F1 ), etc are shown in Eq. 3.21

U1 FX 1
V1 FY 1
W1 FZ 1
1
; 1
(3.21)
X1 CX 1
Y1 CY 1
Z1 CZ 1

Equilibrium of the element in the global system of reference is guven by Eq.3.22;

= [Fe] (3.22)

Wherein; is the element global stiffness matrix.

3.3.7 Transformation

The element with local axes ( x, y, z) is placed arbitrarily in the global (X, Y, Z) space as
X, PY, PZ ) in (X,
Y, Z) directions and components (px, py, pz ) in (x, y, z) directions.

It follows from the vector notation that (Eq.3.23);

PX px
PY T py (3.23)
PZ pz

Wherein; T is the mechanical transformation matrix as shown defined in Eq. 3.24.

Xx Xy Xz
T Yx Yy Yz (3.24)
Zx Zy Zz

- axis with (x, y, z) axes etc.

It could be shown that (Eq.3.25);

Fe R . f eL

T
K R
e K eL . R (3.25)

Wherein; R is rotation matrix defined by the diagonal placement of T matrix as defined in

17
 Finite Element Methodology
Eq. 3.26.

18
 Finite Element Methodology
T
T
R (3.26)
T
T

3.4 Triangular Plate Element

The triangular plate element is denoted through its nodes numbered (1 2 - 3) in
anticlockwise fashion. It is refered to local set of orthogonal axes (x, y, z) as shown in Figure
3.10. In this the plate element is confined to the (x, y) plane and the z-axis is orientated in a
direction normal to the plane of the plate. A plate element displays three independent modes
of deformation. Therse are membrane or plane stress mode, flexural mode and drilling mode.

Y
2

1 X

Figure 3.10. Local reference system 3 - node triangular plate element

3.4.1 Membrane Action

Plane stress deformation associted with the membrane action as characterized by;
translation (u, v) in (x, y) direction and forces (f x, fy) in ( x, y) directions. Consequently, the
element nodal displacements and element nodal loads are as shown in Figure 3.11(a) and
Figure 3.11(b) respectively. These are denoted through vectors m and f m as defined in

Eq. 3.27.

19
 Finite Element Methodology
u
f x1
1

v1 f y1
f fx
u2 2 (3.27)
v2 ; f y2
m m

u3 f x3
v3 f y3

Equilibrium of the lement in the membrane mode of deformation defined by Eq.3.28.

Km . m fm (3.28)

Y Y
3 3

X 1 X
1 2 2
a) Element nodal displacement b) Element nodal loads

Figure 3.11 (a b). Membrane Action (Plane stress)

3.4.1.1 Interpolation function

The displacement at a point (x, y) over the element is given by Eq.3.29

m
Nm m
(3.29)

Wherein; [Nm] is the matrix of the interpolation functions,and given by Eq.3.30

Now;

m U
= V

Nm 0 Nm 0 Nm 0
N m 1 2 3 (3.30)
0 Nm 0 Nm 0 Nm
1 2 3

20
 Finite Element Methodology
Assuming linear variation of and over the element domain, it could be shown
that(Eq.3.31);

Nm L ; Nm L ; Nm L (3.31)
1 1 2 2 3 3

Wherein; L1, L2, L3 are the area co-ordinates as defined in Eq. 3.32, Eq.3.33.

= (3.32)

So that;

a1 = X2Y3 X3Y2

b1 = Y2 Y3

c1 = (X2 X3) (3.33)

= Area of the element (1 - 2 - 3)

3.4.1.2 Strains

At a point (X, Y) the membrane action includes normal strain in X - direction, normal

strain in Y - direction, and shear strain xy as shown in Figure 3.11(c).The strain

components are denoted by element strain vector m] defined in Eq. 3.34 and Eq.3.35.

c) Strain component d) Stress

component Figure 3.11 (c d). Membrane
21
 Finite Element Methodology
Action

22
 Finite Element Methodology

(3.34)

= = (3.35)

In view of the above details, it follows that; (Eq.3.36)

(3.36)
Wherein;

= Element strain transformation matrix as defined in Eq. 3.37.

(3.37)

It may be noted that [Bm] is composed of only the constant coefficients. Hence in the
membrane mode the element happens to be a strain element.

3.4.1.3 Stresses

Associated with the strains considered above a point (X, Y) over the element, the normal
stresses in X direction, normal stress in Y - direction and shear stress as shown in
Figure 3.11 (d) are introduced. The stress components are denoted by an element stress vector
as defined in Eq. 3.38.

= (3.38)

= = (3.39)

Wherein; is the matrix of elastic constant as defined in Eq.3.40.

23
 Finite Element Methodology

E
cm (3.40)
(1 2
)

3.4.1.4 Stiffness matrix

Element stiffness matrix is given by Eq.3.41.

T
=t (3.41)

Wherein;

t = element thickness. The integral is taken over the entire element area. In view of the fact
that and are composed of coefficient:Eq.3.42.

= t. T (3.42)

3.4.1.5 Element load vector

Element may be subjected to the pressure of intensities (P X, PY) in (X, Y) directions as

shown in Figure 3.12 (a). Let these be represented by a vector given by Eq.3.43 and
3.44.

[Pm]= (3.43)

T
=t (3.44)

The integral in Eq.3.44 is taken over the element area and could be evaluated through
numerical integration. Details depend upon the order of the integral terms. Normally three

case of uniform (PX , PY) even one integration point located at the C.G. of the triangle (1 2-
3) as shown in Figure 3.12 (c) is sufficient. This leads to Eq.3.45.

24
 Finite Element Methodology

Y 3 3
X-Integration points (I, II, III)

(1/2, 0, 1/2)X IIX (0, 1/2, 1/2)

III

1 X
2 I
1 X 2
(1/2, 1/2, 0)
a) Distributed load over element b) L1, L2, L3 Integration points in case of
domain three point integration

X
(1/3, 1/3, 1/3)

1 2
c) L1, L2, L3 of point Integration scheme

Figure 3.12. Load vector evaluation for membrane action

3.4.2 Flexure Action

Flexural action is carried by the biaxial bending deformations. This kind of deformation

involves at a point over the element direction, and rotations

around (X, Y) axes. Consequently, the element nodal displacements and element nodal loads
are as shown in Figure 3.13 (a) and Figure 3.13 (b) respectively. These are denoted through
vectors and as defined in Eq. 3.46

; (3.46)

The sub vectors etc are as shown in Eq. 3.47,

25
 Finite Element Methodology
; (3.47)

26
 Finite Element Methodology
Equilibrium of the element subjected to flexural action is defined by,

(3.48)

Wherein; - the stiffness matrix associated with the flexural

action. and are derived in the following manner.

Y W3
2

1 W2
X
W1

Z a) Element nodal displacements

Y FZ3
2

1 FZ2
X
FZ1

Z
b) Element nodal loads

Figure 3.13 (a b). Flexural Action- Element nodal displacements and nodal loads

3.4.2.1 Interpolation function

27
 Finite Element Methodology

W= (3.49)

Wherein; is the matrix of interpolation function.

X =- and Y = it could be shown that (Eq.3.50)

(3.50)

Wherein;

( ) etc are interpolation functions defined by Eq.3.51.

(3.51)

are the area coordinates.

3.4.2.2 Generalized strains

Consider a slice at distance Z from the neutral surface of the plate. The slice is subjected
to normal strains and shear strain that could be schematically represented by
Figure 3.14 (a). The components
the curvature developed by the deformed surface. These curvatures constitute generalized
strains that could be defined by a vector [ ] as defined in Eq. 3.52

(3.52)

Wherein; and are the curvatures around Y axis and X axis

respectively and represents the twist of the deformed surface.

In view of Eq.3.49 to Eq.3.52 it follows that;

(3.53)

Wherein; is the element strain transformation matrix.

28
 Finite Element Methodology
From Eq.3.52, it follows that matrix would comprise the coefficients due to the
d Eq.3.50 requires the second derivatives
of interpolation functions. The interpolation functions are defined in Eq.3.50 and Eq.3.51
requires the second derivatives of the interpolation function. The interpolation functions are
defined in Eq.3.51 from which it follows that their second derivatives would in turn required
the second derivatives of linear terms etc. and cubic terms etc.
Obviously, while derivatives in turn require the second derivatives, the linear terms would
vanish, whereas the second derivatives of the cubic terms are as presented in Table 3.2.

t/2

Z
t/2

a) Various strains across plate thickness

t/2

t/2

b) Various stresses across plate thickness

Figure 3.14 (a b). Flexure action- Various strains and stresses across plate thickness

29
 Finite Element Methodology

Table 3.2. The second derivatives of cubic terms

Term Second Derivative

30
 Finite Element Methodology

3.4.2.3 Generalized stresses

At a point over a slice at a distance Z from the neutral surface of plate, there prevails
normal stress ( ) and shear stress , that could schematically represented by Figure

the element thickness shown in Figure 3.14 (b). The components ( ) besides being
function of Z are related to the bending and twisting moments suffered by plate. These
moments constitute generalized stresses that could be denoted by a vector as defined in
Eq.3.54

(3.54)

Herein; and are the bending moments in X and Y directions i.e. around Y and X axis
respectively, whereas is the twisting moment.

These are illustrated in Figure 3.15 and their relationship with the basic stress parameters are
as defined in Eq.3.55.

Figure 3.15. Flexure action Moments, bending and twisting

; ; (3.55)

31
 Finite Element Methodology
(3.56)

Wherein; is the matrix of elastic constant as defined in Eq.3.57.

(3.57)

3.4.2.4 Stiffness matrix

Element stiffness matrix is given by Eq.3.58.

(3.58)

The integral to be taken over the element are. Once again the three point numerical integration
is convenient for the purpose.

3.4.2.5 Element Load Vector

Element may be subjected to the pressure of intensities acting in Z direction and

corresponding turns out to be (Eq.3.59);

(3.59)

The integral is evaluated over the element surface through three point integration scheme.

3.4.3 Drilling Action

Drilling action is characterized by rotation around Z axis, as also couple around Z

axis. Consequently, the element nodal displacements and the element nodal loads are as
shown in Figure 3.16 (a) and Figure 3.16 (b) respectively. These are denoted through vectors
and as defined in Eq. 3.60

; (3.60)

Equilibrium of the element in the drilling mode deformation is defined by;

(3.61)

Wherein; is the stiffness matrix associated with the drilling mode. could be derived
by following procedure similar to the one employed for deriving or . In the present
32
 Finite Element Methodology
formulation, however in the drilling action is introduced to avoid numerical ill conditioning of
equations of equilibrium. For this is assumed to be as defined in Eq.3.62

(3.62)

Wherein; is a constant having a value of 0.01 to 0.1. Lower the value of better the result
would be. is obviously zero.

Y
Z3
2

1 Z2
X
Z1

Z
a) Element nodal displacement

Y Z3
2

1 Z2
X
Z1

Z
b) Element nodal loads

Figure 3.16 (a b). Drilling Action

3.4.4 Element Local Equilibrium

33
 Finite Element Methodology
The element local equilibrium is illustrated in Figure 3.17 (a d), the same is defined by
Eq.3.67.

(3.67)

Wherein; is the element local stiffness matrix and , are as shown in Eq.3.68.

; (3.68)

The structure of the sub vectors being as defined earlier in Eq.3.44

V3
U3
3
V2
Y W3
V1 U2
2

1 U1
W2
X
W1

Z a) Element nodal Loads local reference system

Y3 3 X3

Y2
Y
Z3
X2
Y1 2

1 X1
Z2
X
Z1

Z
b) Element nodal displacements local reference system

Once again , could be constructed from appropriate model vectors. Therefore

could be built from appropriate placement of the coefficient of the model stiffness matrices.

34
 Finite Element Methodology

FY3
FX3
3
FY2
Y FZ3
Fy1 FX2
2

1 FX1
FZ2
X
FZ1

Z c) Element nodal loads local reference system

CY3
CX3
3
CY2
Y CZ3
Cy1 CX2
2

1 CX1
CZ2
X
CZ1

d) Element nodal loads local reference system

Z

Figure 3.17 (a d). Element local equilibrium

3.4.5 Element Global Equilibrium

The element with its local (X, Y, Z) is placed arbitrarily in the global (X, Y, Z) space as
shown in Figure 3.18. The equation of equilibrium in the global system being;

(3.69)

Wherein; is the element global stiffness matrix and , as shown in Eq.3.70

; (3.70)
35
 Finite Element Methodology

The structure of the sub vectors is as defined earlier in Eq.3.47.

Y
3

1
Z X

X
Z

Figure 3.18. Element characteristics: 3 node triangular element (Placement of element in

global space

3.4.6 Transformations

Following the details of section 3.3.7 above, the transformation matrix defined therein
is once more valid and the relationships appearing in Eq.3.51 also valid. The rotation matrix
is however modified to suit the three nodded element as indicated in Eq.3.71.

[T]
[T]
(3.71)
[T]
[T]
[T]
[T]

3.5 Four Node Quadrilateral Plate Elements

A four node quadrilateral plate element (1 2 3 4) is shown in Figure 3.19 (a) is

advocated for a sufficient solution. As far possible it is recommended that the 04 - node plate

36
 Finite Element Methodology
element be used, reserving the use of 03 - node plate element for representing the regions
involving the sharp corners with its kind of philosophy the idealization details get simplified
to a considerable extent, as may be seen from Figure 3.19 (b) and Figure 3.19 (c).

4 3 4 3
3
4 5 2

1 1
1 2 2
a) Four node quadrilateral plate b) Assembly
29
28 24
27 23
26 22 19
Y 18
25 21 14
20 17 13 X
16 8 9
12
15
11 7
10
6
5
4

3
2

1
c) Idealized structure with four node plate element

Figure 3.19 (a c). Four node quadrilateral plate element

A 04 - node quadrilateral plate element could be built from the assembly of four triangular
elements generated by joining its diagonals as shown in Figure 3.19 (b).This however
introduces an internal node designated as node number 5. The degrees of freedom of the
internal node are to be eliminated so as to generate the required four node elements. For this a
process called static condensation is employed. The details are as follows.

Element nodal displacement vector and the element nodal load vector for the
four node plate element are as defined in Eq.3.72.

37
 Finite Element Methodology

; (3.72)

Equilibrium of the element is defined as;

(3.73)

Wherein; is the stiffness matrix of the element.

As pointed out, the 04 - node quadrilateral element is treated as assembly of four

triangular elements, as shown in Figure 3.19 (c). The assembly is defined by the following
element nodal connectivity (Table 3.3).

Table 3. 3. Element connectivity

Element Number Nodes

1 1 2 5
2 2 3 5
3 3 4 5
4 4 1 5

and of each of the triangular elements are derived by employing above connectivity
details. Assemblage of and is derived from the standard super positioning technique.

and of the assemblage are as defined in Eq.3.74

; (3.74)

And its equilibrium is defined by,

(3.75)
Eq.3.75 is partitioned as per the scheme shown in Figure 3.19 (c),

(3.76)

From Eq.3.76

38
 Finite Element Methodology

Therefore;
(3.77)

Again from Eq.3.76,

(3.78)

Substituting from Eq.3.73, it follows that;

(3.79)

Comparing Eq.3.79 with Eq.3.73 it follows that;

(3.80)

3.6 08 - Node Solid Hexahedral Element

In this section we shall discuss the development of [K e] and [Fe] of the solid hexahedral
element. [Fe] is derived for the body forces prevailing over the element. A typical element is
as shown in Figure 3.20. The element is referred to the Cartesian reference axes (x, y, z);
ically upward direction. At a point over
the element there prevails 3 -
directions. A force at a point over the element is expressed through the components (F x, Fy,
Fz) in (x, y, z) directions. The nodes over the bottom faces of the element are numbered (1-2-
3-4) in anticlockwise direction. Correspondingly the nodes over top element are numbered as
(5 6 7 8) in anticlockwise direction.

We shall denote the displacement at ith node through a

components at the ith node through sub vector [F]. These sub vectors are defined in Eq. 3.81

i= I= Eq. 3.81

39
 Finite Element Methodology
7
b) Reference axes
Y
6
X
8
5 w
3
v c) Displacements

u
2
4 Fz
1 Fy
d) Forces
a) Element
Fx

Figure 3.20. Basic features of solid hexahedral element

e] and the element nodal load vector [Fe] are as

shown in Eq.3.82

= = (3.82)

3.6.1 Interpolation Functions

The inter element compatibility is ensured if the straight edges of the element contributes
to remain all the straight edges, even after the deformations. Displacements (u, v, w) are

axes of reference as the parent element as shown in Figure 3.21 and treating (x, y, z, u, v, w)
as the parameters varying linearly over the edges of the parent element, we could write the
linear iso-parametric characteristics as defined in Eq.3.83 and Eq.3.84.

40
 Finite Element Methodology

2.0

2.0 b) Reference axes (

at centroid of the cube (2x2x2)

2.0

a) Parent element (2x2x2)

Figure 3.21. Basic features of parent element

Eq.3.83

u=

N1= 0.125 (1- - - 5= 0.125 (1- -

N2 - - 6 -
N3= 0.125 ( - 7

N4= 0.125 (1- - 8= 0.125 (1-

3.6.2 Strain Vectors

At appoint over the element domain, their prevails six components of strains. These
vectors are repr

x- Normal stress in x- xy - Shear stress in (x, y) plane

y - Normal stress in y- yz Shear stress in (y, z) plane
z - Normal stress in z- directi zx Shear stress in (z, x) plane

41
 Finite Element Methodology

= (3.85)

= (3.86)

Where: [B] is the element strain matrix

Following the details of [B] matrix, a sub vector [B]i is defined as below

0 0
0 0
0 0
[B]i = . (3.87)
0
0
0

[B] = [B]1 [B]2 [B]3 [B]4 [B]5 [B]6 [B]7 [B]8 (3.88)

3.6.3 Stress vectors

At the point over the element domain their prevails six components of stresses. These are
represented by stress vector [ ] as given in Eq.3.87

42
 Finite Element Methodology

x
y

[ ] = z . (3.89)
xy
yz
zx

Where; x = normal stress in x- direction

y = normal stress in y- direction

z = normal stress in z- direction

xy = shear stress over (x,y) plane
yz = shear stress over (y-z) plane
zx = shear stress over (z-x) plane

Assuming the material of the element to be homogeneous, isotropic and elastic, the
vectors [ -

[ e] (3.90)

Where [C] is the matrix of elastic coefficient as given in Eq. 3.91

C1 C2 C2 0 0 0
C2 C1 C2 0 0 0
[C] = . (3.91)
C2 C2 C1 0 0 0
0 0 0 C3 0 0
0 0 0 0 C3 0
0 0 0 0 0 C3

Where;

C1 = C2 = C3 = G= (3.92)

E- Modulus of elasticity
- Poi
G- Shear modulus

43
 Finite Element Methodology
3.6.4 Element Stiffness Matrix [Ke]
Employing the principles of virtual work, we could prove that;

[Ke] = T [C] [B] dx dy dz (3.93)

3.6.5 Element Load Vector due to Body Forces [Fe]

Let, / unit volume body force intensity in x- direction
/ unit volume body force intensity in y- direction
/ unit volume body force intensity in z- direction
The vector [P] defining as body force intensities as defined in Eq. 3.94
Px
[P] = Py
Px
(3.94)

u
v
. (3.95)
w

Where; [N] is the interpolation matrix as shown in Eq. 3.96 and the sub vector [N]i is given
in Eq. 3.97

[N] = [N]1 [N]2 [N]3 [N]4 [N]5 [N]6 [N]7 [N]8 . (3.96)

[N]i = N 0 0 (3.97)
0 N 0
0 0 N

Employing the principle of virtual work the load vector [Fe] due to the body force could be as
shown in Eq.3.98

T
[Fe] = [P] dx dy dz (3.98)

44
 Finite Element Methodology
3.7 Summary

Finite element method is used to solve physical problems in engineering analysis and
design. A physical problem comprises of an actual structure or structural components
subjected to certain loadings. Discretization of physical problem in to a mathematical model
requires certain assumption which leads to evolution of set of differential equations governing
the mathematical model. The fundamental scheme of finite element method is to divide a
structure into smaller parts. The structure divided into number of elements leads to
simplification of governing equations. For this discretized 4 node plate elements and 8
node solid hexahedral elements, primary unknowns are nodal displacements. The strains and
stresses are further obtained by post processing. The general FEM procedure followed in
developing software in Fortran for parametric analysis of concrete arch dams can be
summarized as follows;

1. Discretization: divide the structure into finite elements defining nodal coordinates and
element connectivity.
2. Formation of element stiffness matrix [Ke].
3. Formation of global stiffness matrix.
4. Formation of global load vector [F].
5. For
6. Assemblage of global nodal displacements.
7. Incorporation of specified boundary conditions.

8.
9. Computation of element strains and stresses from the known displacements for each
element.

Flow diagrams (Figure 3.22) to arrive the objectives of research work are as shown below.
Finite element program is developed for 2 D plate elements and 3 D, 8 node hexahedral
solid elements.

45
 Finite Element Methodology

Mathematical
Physical model: Geometry,
Concrete model (System
loads, material properties,
Arch Dam of equations)
boundary conditions

Results Presentation Verification of results

(Solved Problems)

Analysis Results:
Displacements, Computer Program: Analytical/Numerical
strains and stresses FORTRAN, Rhino GH solutions

Figure 3.22. Flow diagram of Fortran Programming

46