SCHOOL OF ENGINEERING SCIENCE AND TECHNOLOGY

DEPARTMENT OF MECHATRONIC ENGINEERING

___________________________________________________________
NAME

MPUMELELO NCUBE

STUDENT NUMBER

C121372Q

LEVEL

4.2

COURSE

FINITE ELEMENT ANALYSIS

Assignment: Explain how finite element analysis is linked with Lagranges

equations, Jacobian equations, Hookes law and Prandtl Stress Function.
[100marks].

1 INTRODUCTION
1.1 What is the finite element method
The finite element method (FEM) is a numerical technique for solving problems which are described
by partial differential equations or can be formulated as functional minimization. A domain of interest
is represented as an assembly of finite elements. Approximating functions in finite elements are
determined in terms of nodal values of a physical field which is sought. A continuous physical
problem is transformed into a discretized finite element problem with unknown nodal values. For a

linear problem a system of linear algebraic equations should be solved. Values inside finite elements
can be recovered using nodal values.
Two features of the FEM are worth to be mentioned:
1. Piece-wise approximation of physical fields on finite elements provides good precision even
with simple approximating functions (increasing the number of elements we can achieve any
precision).
2.

Locality of approximation leads to sparse equation systems for a discretized problem. This
helps to solve problems with very large number of nodal unknowns.

1.2 How the FEM works

To summarize in general terms how the finite element method works we list main steps of the finite
element solution procedure below.
1. Discretize the continuum. The first step is to divide a solution region into finite elements. The
finite element mesh is typically generated by a pre-processor program. The description of mesh
consists of several arrays main of which are nodal coordinates and element connectivity.
2. Select interpolation functions. Interpolation functions are used to interpolate the field variables over
the element. Often, polynomials are selected as interpolation functions. The degree of the polynomial
depends on the number of nodes assigned to the element.
3. Find the element properties. The matrix equation for the finite element should be established which
relates the nodal values of the unknown function to other parameters. For this task different
approaches can be used; the most convenient are: the variation approach and the Galerkin method.
4. Assemble the element equations. To find the global equation system for the whole solution region
we must assemble all the element equations. In other words we must combine local element equations
for all elements used for discretization. Element connectivity is used for the assembly process. Before
solution, boundary conditions (which are not accounted in element equations) should be imposed.
5. Solve the global equation system. The finite element global equation system is typically sparse,
symmetric and positive definite. Direct and iterative methods can be used for solution. The nodal
values of the sought function are produced as a result of the solution.
6. Compute additional results. In many cases we need to calculate additional parameters. For
example, in mechanical problems strains and stresses are of interest in addition to displacements,
which are obtained after solution of the global equation system.

LAGRANGE EQUATIONS

2.1 Definition
L = T V (Kinetic Potential energies)
Lagrange equations
Used if a body undergoes
a. large displacements
b. large rotations
c. large strains
As a result the body is considered subjected to arbitrary large motions
2.2 Lagrange theory
Lagrange interpolation, a variety of finite element families has been proposed for 2D and 3D
elements, i.e. for squares and cuboids. For example, there is the regular Lagrange family according to
Argyris et al.with nodes regularly spaced everywhere on the grid, and the mid-edge Lagrange family
of Taylor and others. composed of only vertex-type and side (or mid-edge) nodes. We note that the
trial function spaces of their all members are fully known.
For simplex finite elements, on the other hand, only the complete Lagrange family with regularly
placed nodes is fully recognized. However, it has been suggested that there may exist the mid-edge
Lagrange family for a simplex, composed of only vertices and arbitrarily placed side nodes. In
addition, rather general finite elements have also been presented in the adaptive finite element
procedures utilizing not the conventional interpolation bases but the peculiar mixed bases with both
nodal and node less basis functions. It seems that potential exists to develop at least as many families
for Lagrange interpolations over simplex finite elements simplexes as have already been found for
cubes. In order to clarify such general simplex families, it is required that the interpolation element
ology be reconstructed in a unified manner. The native simplex interpolations are thus easily and
systematically applicable also to fracture mechanics, with only minor modification of the Jacobian
transformation routine.
2.3 Applicability to FEA
For conservative systems

Results in the differential equations that describe the equations of motion of the system

Key points:
i)

ii)

Newton approach requires that you find accelerations in all 3 directions, equate F=ma,
solve for the constraint forces, and then eliminate these to reduce the problem to
characteristic size
Lagrangian approach enables us to immediately reduce the problem to this characteristic
size. We only have to solve for that many equations in the first place.

The ease of handling external constraints really differentiates the two approaches.

An Example
Spring mass system

Lagrangian L = T V

Lagranges Equation

Do the derivatives

Put it all together

Consider the MGR problem with the mass oscillating between the two springs. Only 1 degree of
freedom of interest here so, take qi=R

JACOBIAN

3.1 Jacobian theory

In the Finite Element Method, an element's Jacobian Matrix relates the quantities wrote in the natural
coordinate space and the real space. The bigger the element is distorted in comparison with a ideal
shape element, the worse will be the transformation of the quantities from the natural space to the real
space. In fact, it means that the distortion on the element shape will introduce error in the
mathematical trasformation from natural space to the real space.
It is important to mention that the two definitions proposed here rest upon the Jacobian matrix of the
mapping function F relating the parent or reference element frame to its actual coordinates within the
mesh. In the remainder of the article this matrix is denoted as J. Note that other geometrical functions
can be used to define the quality of an element.
3.2 Applicability to FEA
As mentioned by the definition, the FEM computes for each node of the underlying mesh an
approximated solution to the system of PDEs describing the physical phenomenon under study. For
this purpose, a perfect element is defined in a reference parent system ( 1 , 2 , 3

where each

reference element point is related to its actual counterpart within the element in the modeled domain
referential

(x 1 , x 2 , x 3)

through the mapping F, as shown in Fig. 2. The Jacobian matrix of

mapping F considered at parent frame

point is defined as:

J ()=

F
( )

The FE analysis can be carried out on a mesh as long as the mapping F, considered for each mesh
element, remains a onetoone application. This property can be assessed by considering the
determinant of Jacobian matrix of F,

J ( ) , at a number of element control points, such as the

element nodes or Gaussian quadrature (integration) points, for all the elements forming the mesh.

Figure 2: The referencetoactual frame mapping from which the Jacobian matrix
Mesh relaxation also relies on the iterative computation of the Jacobian matrix of the mapping

( 1 , 2 , 3)

between the element parent (or reference) coordinates system

element coordinates

( x 1 , x 2 , x 3)

and the actual

as shown on Figure 2. The Jacobian matrix is computed at a

number of control points which can be either mesh nodes or element integration points. Let

()=( 1 , 2 , 3 ,)be

an element point in parent coordinates and

x=( x 1 , x 2, x 3)

the same

point in actual element coordinates.

x () :=F ( )= ( )x
n

n=1

x n1 , x n2 , x n3
x n= ) are the coordinates of
Where N is the number of nodes of the considered element and

the

nth element node. The Jacobian matrix J is defined at parent frame point as:

J 11 J 12 J 13
F
J ( ) :=
( )= J 21 J 22 J 23

J 31 J 32 J 33
Where

J ij =( x i / j )

Finally, the Jacobian matrix of mapping F computed at point

precomputed shape function derivatives

J ( ) =

Pj

is assembled using the

3N

as follows:

( x i / j)

Jacobian equations
After mapping is performed for the coordinates, we can now evaluate the strain matrix. To do this, it
is necessary to express the differentials in terms of the natural coordinates, since the relationship
between the x and y coordinates and the natural coordinates is no longer as straightforward as in the
case for rectangular elements. Utilizing the chain rule in application to partial differentiation, we have

 Ni xi
=
j

The basic Jacobi method has been developed over a century ago for the solution of standard Eigen
problems. A major advantage of the procedure is its simplicity and stability.

4 HOOKES LAW
4.1 Definition

It states that within the limit of proportionality, force applied is directly proportional to
extension.

4.2 Applicability to FEA

A complete description of the general state of stress at a point consists of:

normal stresses in three directions, x (or 11), y (or 22) and z (or 33),

shear stresses on three planes, x (or 12 ...), y (or 23 ..), and z (or 31 ...).

The stress, x in the x-direction produces 3 strains:

longitudinal strain (extension) along the x-axis of:

y z x

x
E

transverse strains (contraction) along the y and z

-axes, which are related to the Poissons ratio:

normal stresses in three directions, x (or 11), y (or 22) and z (or 33),
shear stresses on three planes, x (or 12 ...), y (or 23 ..), and z (or 31
...).

The stress, x in the x-direction produces 3 strains:

longitudinal strain (extension) along the x-axis of:

transverse strains (contraction) along the y and z -axes, which are related to the Poissons ratio

Properties of
Absolute values of are used in calculations.

The value of is about:

0.25 for a perfectly isotropic elastic materials.
0.33 for most metals

In order to determine the total strain produced along a particular direction, we can apply the
principle of superposition.
For Example, the resultant strain along the x-axis, comes from the strain contribution due to
the application of x, y and z.
x causes:
in the x-direction

y causes:

in the x-direction

z causes:
in the x-direction
Applying the principle of superposition (x-axis):

1
x x ( y z )
E

By superposition of the components of strain in the x, y, and z directions, the strain along each axis
can be written as:

y
x

1
y v z x
E

1
x v y z
E

z v x y
E

The shearing stresses acting on the unit cube produce shearing strains.

xy G xy
yz G yz

 xz G xz
The proportionality constant G is the modulus of elasticity in shear, or the modulus of rigidity. Values
of G are usually determined from a torsion test.

5 PRANDTL STRESS FUNCTION

5.1 Prandtl stress function theory

The Prandtl stress function is a special case of the Morera stress functions and it is useful
when torsion is present.
Prandlt Stress state conditions
Lets take an example of a rod with a constant cross-section. Its axis lies on the z-axis. We
can apply a torsion T to both its sides. This torsion T is said to be positive when it is directed
counterclockwise about the z-axis (according to the right-hand rule). Since we only apply
torsion, we can assume there are no normal (direct) stresses, so x=y =z=0. The same
goes for the shear stress xy, so xy = 0. From this follows that also x =y =z =xy

= 0.

We also assume no body forces are present. So most of the stresses are zero. We only have
two non-zero stresses left, being zyzx .
5.2 Prandtl stress function applicability to FEA
In Saint Venants theory of torsion for non-circular sections, the displacements are given by
uz = y(x)
uy = z (x)

(1)

u x=

d
( y , z)
dx

If d / dx = is a constant, then the only non-zero stresses are

(2)

and all the equilibrium equations are satisfied if

( 3)

This equation can be satisfied automatically by writing the stresses in terms of a function, ,
called the Prandtl stress function, where

(4)
However, from Eq. (2), we have

Which implies

However, these mixed derivatives of the displacement ux must be equal, if we are to be able
to integrate the strains to find this displacement, and this compatibility condition requires that
the stress function satisfy
(5)

or equivalently

(6)

which is called Poissons equation. We know that on the outer boundary of the bar we have
no applied tractions so that

and y and z components of the traction vector are identically zero, so that in terms of the
Prandtl stress function we have

(7)

But by examining a small element near the surface (Fig.1), we see that Eq. (8) also implies
that

(9)

which says that must be a constant on the boundary. For a cross-section with no holes, we
can take the constant to be zero in general.

To find the torque, T, in terms of the stress function, we start from the relation between the
torque and stresses and write that relation in terms of the stress function

where the integral over the boundary C vanishes because =0 . Thus, the torque T is just
twice the area under the (y, z) surface. It can be shown that on the boundary C, the total
shear stress takes on the largest value anywhere in the cross section. Since at the boundary
this total shear stress must be tangent to the boundary

We have

so this maximum total shear stress is just the negative of the largest slope of the (y, z)
surface at the boundary. We can make these results appear similar to the familiar formulas for
the torsion of a circular cross section if we define a modified stress function = G ,
where
2 = 2 in the cross section
= 0 on the boundary
If we let T = GJeff then

and for the maximum shearing stress we have

The prandtl stress function deals with the analysis of torsion of non-circular sections. It can be applied
to find the torsion of these sections. For the torsion of non-circular forces, displacements are given by:

u z= y ( x)

Equation [1]

u y =z ( x)
u x=

If

d
= '
dx

Equation [2]

d
( x)
dx

Equation [3]

is a constant then the only non-zero stresses are

xz =(G ( u x / z ) +( u z / x))

Equation [4]

(G ( ux / z ) + ' y)

xy = (G ( ux / z ) +(u z / x))
ux
(G
' z )
=
y

( )

equation [5]
equation [6]

equation [7]

And all the equilibrium equations are satisfied by

( xy / y )

( xz / z )

=0

equation [8]

The equation can be written using stresses in terms of a function , called the Prandtl Stress Function.

xy = ( / z ) , xz = ( / y )

equation [9]

From equation [7]

( uz )

ux
y

( )

= ( / y ) - G ' y

= ( / y ) + G ' z

equation [9]

equation [10]

It implies that

ux
G
z y

2
= ( / y ) - G '

equation [11]

2 u x
G
y z

2
= ( / z ) + G '

equation [12]

However these denominations of the displacement

u x must be equal if we are to be able to find

the strains to find displacement. This compatibility condition requires that the Prandtl stress function
satisfy

2
2
G '
2
2
y + z = -2

equation [13]

We know that at the end of the bar we have no applied

(n)
T x = xy n y + xz n z =0

equation [14]

y and z are identical so that the equation in terms of the Prandtl stress function becomes

( / z)n y -

( z ) n =0
z

Equation [15]

The equation implies that

z
y
z
y
+
=0
s
zs
Which states that

Equation [16]

must be a constant on the boundary. To find the torque in terms of the

Prandtl stress function we start from the relationship between the torque and stresses.

T = ( xz y xy z ) dA

Equation [17]

( y y z z)dA

=-

equation [18]

dA

equation [19]

[ ( y n y ) + ( z n z ) ] ds+2 dA
A

equation [20]

2 dA

The Prandtl Stress Function

equation [21]

=0,

therefore for any material the torque T

is twice the area

( y , z ) . the total shear stresses take the largest value anywhere in the cross

und the surface

section of the material. Since at the boundary this total shear stress must be tangent to the boundary
therefore for maximum shear stress

max =T /J eff (

n max onboundary

equation [22]

Where

J eff =2 dA
A

equation [23]

Guided by Observations from Mechanics of Materials

projection of each section on x,y-plane rotates as rigid-body about central axis
amount of projected section rotation is linear function of axial coordinate
plane cross-sections will not remain plane after deformation thus leading to a warping
displacement
Stress Function Formulation

Equilibrium Equations

Compatibility Relation

Prandtl Stress Function = (x,y)

Equilibrium will be identically satisfied and compatibility relation gives

A Poisson equation that is amenable to several analytical solution techniques.

CONCLUSION
As highlighted above, the Jacobian, lagrange equations, hookes law and the prandlt stress function
are all important tools in the finite element analysis of different substances.

REFERENCES

1. A. Weinstein, The center of shear and the center of twist, Quart. of Appl. Math. 5(1), 97
99 (1947).
2. E. Reissner and W. T. Tsai, On the determination of the centers of twist and of shear for
cylindrical shell beams, J. Appl. Mechanics 39 10981102 (1972).
3. S. P. Timoshenko and J. N. Goodier, Theory of Elasticity, 3rd edn, McGrawHill
International Book Company, 1984.
4. I. S. Sokolniko, Mathematical Theory of Elasticity, McGrawHill, New York, 1956.
5. W. E. Mason and L. R. Herrmann, Elastic shear analysis of general prismatic beams, J.
Eng. Mech. Div. ASCE 94, EM4, 965983 (1968).
6. L. R. Herrmann, Elastic torsional analysis of irregular shapes, J. Eng. Mech. Div. ASCE
91, EM6, 1119 (1965).
7. J. L. Krahula and G. L. Lauterbach, A nite element solution for SaintVenant Torsion,
AIAA Journal 7(12), 22002203 (1969).
8. G. Haberl and F. Och, Eine FiniteElementLosung fur die Torsionssteigkeit und den
Schubmittelpunkt beliebiger Querschnitte, Z. f. Flugwiss. 22(4), 115119 (1974).
9. C. Zeller, Querschnittsverformungen von Staben, IngenieurArchiv 52, 1737 (1982).