CHAPTER 4

FINITE ELEMENT METHOD

The Finite Element Method is a numerical technique for finding approximate solution of
engineering problems. It is an extremely sophisticated tool for solving numerous
engineering problems and is widely used and accepted in many engineering branches.
Basic theme of FEA is to make calculations at only limited (Finite) number of
points and then interpolate the results for entire domain (surface or volume). Any
continuous object has infinite degrees of freedom and it’s not just possible to solve the
problem in this format. Finite Element Method reduces degrees of freedom from infinite
to finite with the help of discretization i.e. meshing (nodes & elements).

4.1 PRINCIPLES OF FINITE ELEMENT METHOD

Finite element analysis focuses on continuum problems; calculating the behavior and
response of a continuous surface or object. Since solving a continuum problem with an
infinite number of points is impossible, a method must be used to simplify the calculation.
The finite element method simplifies the task of solving the continuum problem by
dividing the continuum into a set of subdivisions called elements. The elements are
interconnected at nodal points into an assembly called a mesh.
Since the finite element solution is based on a set of approximate functions, the
accuracy of the method can be increased by refinement of this assumed function. For
instance, if a quadratic displacement of a beam is to be modified by a single element, then
use of higher order quadratic function for the formulation of the beam element. A simple
linear displacement function would not provide a good approximation to the actual
problem.
Another method of increasing the accuracy of solution is to increase the number
of elements in the finite element mesh. One can closely approximate the actual
displacement of the beam using a linear displacement interpolation function, provided use
of multiple elements to discretize the beam.
Any finite element analysist requiresthe following solution steps:
 Idealize the quantum as an assembly of elements.
  Establish the equilibrium requirements, in terms of field variables, for each
element.
 Establish a set of simultaneous equations for the unknown field variables based
on the element interconnection defined by the mesh.
 Solve the simultaneous equations, and calculate the element response using
element equilibrium requirements.
 Interpret the results, review the model if necessary, and perform the analysis
again.

In the finite element analysis the discretization of a continuum involves

approximation; an exact solution is never attained.

4.2 TYPES OF ELEMENTS

1-d element: It is used for geometries having one of the dimension very large in
comparison to rest of the two. Shape of 1-d element is line.

Figure 4.1 1-d Element

When the element is created by connecting two nodes, software comes to know about
only one out of three dimensions. Remaining two dimensions i.e. area of c/s must be
defined by the user as additional input data & assigned to respective elements.
Element type - rod, bar, beam, pipe etc
Practical example – long shaft, rod, beam, column, bolted joint, pin joint etc.

2-d element: It is used when two of the dimensions are very large in comparison to third
one. 2-d elements are planer just like the paper. By creating 2-d element software comes
to know 2 out of three required dimensions. The third dimension i.e. thickness has to
provided by user as an additional input data.
 Figure 4.2 2-d Element

Element type – thin shell, plate, plane stress, plane strain etc .
Practical example – sheet metal parts, skin of aircraft parts etc.

3-d element: It is used when all the three dimensions are comparable. No additional data
has to provide by user.

Figure 4.3 3-d Element

Element type – solid
Practical applications – crankshaft, engine block etc.

4.3 TWO-D MESHING

 2d element shapes

Tria Quad

L(3) P(6) L(4) P(8)

Figure 4.4 Tree diagram for 2-d elements

Where,
L = Linear element
P = Parabolic element
( ) – Indicates number of nodes/elements

Table 4.1 Different types of 2-d elements

Types of Element Displacement Function
Linear Tria 3 U = a0+a1x+a2y
(3 nodes – 3 terms in
displacement function)

Linear Quad 4 U = a0+a1x+a2y+a3xy

(one additional term in
comparison to tria3, makes it
more accurate)
Parabolic Tria 6 U = a0+a1x+a2y+a3x2+a4y2+a5xy
(6 nodes – 6 terms in
displacement function)

Parabolic Quad 8 U = a0+a1x+a2y+a3xy+a4x2+a5y2

+a6x2y+a7xy2
(two additional terms in
comparison to tria 6, makes it
more accurate)
Quad elements are better than triangular elements and Parabolic elements are better than
linear elements.

4.4 QUALITY CHECKS

Why quality checks?

Result quality is proportional to Element quality
Ideal shape of quad element is Square
Ideal shape of triangular element is Equilateral triangle
Different quality parameters like skew, aspect ratio, included angle, jacobian,
stretch etc are the measures of how far a given element deviates from ideal shape. Some
of the quality checks are based on angles (like skew, included angle) while others on side
ratio & area (like aspect stretch). Ideally all the nodes of quad element should lie in the
same plane but at curvatures and complicated geometry profiles it is not possible.
Measure of out plane angle is warp angle.
Following are general definitions of various quality checks.

Aspect ratio: maximum element edge length / minimum element edge length.
Ideal value = 1 (Acceptable < 5)

Figure 4.5 2-d Element with maximum and minimum side

Skew:
Ideal value = 0 (Acceptable < 450)
Skew for quadrilateral element = 900 minus minimum angle between two lines joining
opposite mid-sides of the element (α).
Skew for triangular element = 900 minus minimum angle between the lines from each
node to the opposing mid-side and between the two adjacent mid-sides at each node of
the element.

Figure 4.6 2-d Element with angle α

Wrap angle: warp angle is out of plane angle.

Ideal value = 00 (Acceptable < 100)
It is defined as angle between normals to two planes formed by splitting the quad element
along diagonals. Max. angle out of the two possibilities is reported as wrap angle.

Jacobian:
Ideal value = 1 (Acceptable > 0.6)
Jacobian is a scale factor arising because of transformation of co-ordinate system.
Elements are transformed from global co-ordinate to local co-ordinates, from faster
analysis point of view.

Included angles:
Included or interior angle check is applied for individual angles.
Quad ideal value = 900. (Acceptable = 450 < θ < 1350)
Tria ideal value = 600. (Acceptable = 200 < θ < 1200)

Taper:
Ideal value = 0 (Acceptable < 0.5)

4.7 2-d Element with different areas to find taper

Taper = ∑ ¿¿
A avg=( A 1+ A 2+ A 3 + A 4 )/ 4
4.5 STATIC ANALYSIS:
The static analysis problem can be stated, in the matrix form, as

[K]{u} = {F} (4.1)

Where
[K] = the global stiffness matrix
{u} = the vector of unknown displacements
{F} = the applied force vector
The matrix equation in formula 1 represents a set of simultaneous equations. The
unknown displacement vector can be obtained by inverting the stiffness matrix and
multiplying both sides of the equation by this inverse, such that

{u} = [K]-1{F} (4.2)

The global stiffness matrix in the static analysis is assembled from the stiffness
matrices for each individual element. As an example, consider a simple two-dimensional
truss element with a linear displacement function. The stiffness matrix for the element is,

(4.3)
[ K ] = AE
L

4.6 DYNAMIC ANALYSIS:

The dynamic equation of motion can be written in matrix form as:

[M]{ ẍ ¿)} + [C]{ ẋ (t)} + [K]{x(t)} = {F(t)} (4.4)

Where
[M] = the mass matrix
[C] = the damping matrix
[K] = the stiffness matrix
{x(t)} = the nodal displacement vector
{ ẋ (t)} = the nodal displacement vector
{ ẍ ¿)} = the nodal acceleration vector
{F(t)} = the applied load vector
The dynamic equilibrium equation contains two additional matrices [M] and [C], for mass
and damping, that were not included in the static equilibrium equation. All vectors for
displacement, velocity, acceleration and applied load are a function of time.
Mathematically, the equation of motion represents a set of second-order linear
differential equations. A solution can be obtained by several different methods, including:

 Modal superposition
 Frequency response
 Direct integration for transient response

The modal superposition method is normally used to solve the dynamic equation
of motion for linear elastic systems. This method is the mose popular, since the majority
of real world problems are approximately linear elastic.

The Modal Superposition Method

The Modal Superposition Method is a two-step procedure. First, obtain the mode shapes
of a structure by solving the free vibration problem. Then calculate the response to the
dynamic forcing function for each mode separately and sum them for the total response.
The mode shapes are obtained from the equation of motion for undamped free vibration,
which can be written in the matrix form as:

[M]{ ẍ ¿)} + [K]{x(t)} = 0 (4.5)

Where
[M] = the mass matrix
[K] = the stiffness matrix
{x(t)} = the nodal displacement vector
{ ẍ ¿)} = the nodal acceleration vector

Since { ẍ ¿)} = ω2 {x(t)} for free vibration, the equation can be rewritten in the following
form:

[K]{x(t)} = ω2 [M] {x(t)} (4.6)

Where

[M] = the mass matrix

[K] = the stiffness matrix
ω2 = the natural frequency
 {x(t)} = the nodal displacement vector
{ ẍ ¿)} = the nodal acceleration vector
For the non-trivial solution {x(t)} ≠ 0, the magnitude of the resonant frequencies can be
obtained by the generalized eigenvalue solution to:

[K] - ω2 [M] = 0 (4.7)

The stiffness matrix for the element is:

[ K ] = AE
L

Consider a finite element model consisting of two uniaxial truss elements of equal length,
cross sectional area and material properties, coupled in series as illusrated in figure below

Figure 4.8 A truss with two uniaxial elements

The individual stiffness matrices would be

(4.8)

(4.9)
Assembling the local stiffness matrices into a global stiffness matrix and substituting into
the static analysis yields

(4.10)

There are several different methods for approximating the mass matrix. Two of the most
common approches used in finite element analysis are the lumped mass and conisistent
mass methods.

The lumped mass method, as its name suggests, consists of evenly dividing the
total mass of the element and lumping that mass at each node. Using this technique for
the uniaxial truss element, the mass matrix would be:

(4.11)

4.7 HIGHER ORDER DISPLACEMENT FUNCTIONS:

Another method for increasing the accuracy of the solution is to use elements based on
higher order displacement functions. For instant, if the deflected shape of a structure to be
analysed is quadratic, then the finite element model comprised of elements with a
quadratic displacement function will provide accurare solution using fewer elements than
a model comprised of elements with linear displacement functions.

Shape functions
Consider a simple uniaxial truss element defined by two nodes, as shown figure 4.9.
Assuming the linear displacement function, the axial deformation anywhere along the
element can be defined as:

u(x) = a + bx (4.12)

or in matrix form as:

 (4.13)

node 1 node 2

Figure 4.9 Uniaxial Truss Element

The following conditions must hold at nodal points in order for the displacement function
to be admissible:

u(0) = x1 = 0 (4.14)

u(L) = x2 = a + bL (4.15)

such that the following matrix form of the displacement function can be written as:

(4.16)

Matrix inversion and multiplication yields:

(4.17)

Substituting the above expression into the original displacement function yields:
 (4.18)

or:

(4.19)

The matrix [H] is called shape function matrix. It contains the following shape functions:

x
H1 = 1 - L

H2 = x
L

The significance of this equation is that one can calculate the displacement
anywhere within the element from the nodal displacement, if an admissible shape
function matrix is available. The stiffness matrix and mass matrix can also be derived
from the shape function.

4.8 ANALYSIS PROCEDURES:

Procedures involved in typical finite element analysis involves:
1. Define node locations
2. Choose element type and connectivity
3. Define material and geometric properties
4. Define boundary conditions
5. Define forcing functions
6. Submit file for analysis
7. Interpret results
4.9 ADVANTAGES OF FEM :
1. It is an extremely sophisticated tool for solving numerous engineering problems
and is widely used and accepted in many branches.
2. It is applicable to any field problem, e.g. heat transfer, stress analysis, vibration
analysis, magnetic field, and etc.
3. Boundary conditions and loading are not restricted.
4. Material properties are not restricted to isotropy, and may change from one
element to another or even within an element.
5. The approximation is easily improved by refining the mesh.
6. The decrease in price and increase in availability of powerful mini and micro
computers has meant that even small companies can have access to finite
element program.