0% found this document useful (0 votes)
148 views

3D Finite Element Method

The document outlines the finite element formulation and method. It discusses the basic equations of mechanics, including the strain-displacement relation, constitutive relations, and equilibrium equations. It then describes the finite element formulation, including discretization of the domain, derivation of elemental equations, assembly, imposition of boundary conditions, and solution procedures. The key steps in finite element analysis and weighted residual methods for determining unknown variables are also summarized.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
148 views

3D Finite Element Method

The document outlines the finite element formulation and method. It discusses the basic equations of mechanics, including the strain-displacement relation, constitutive relations, and equilibrium equations. It then describes the finite element formulation, including discretization of the domain, derivation of elemental equations, assembly, imposition of boundary conditions, and solution procedures. The key steps in finite element analysis and weighted residual methods for determining unknown variables are also summarized.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 209

Finite element method

U. Saravanan
Associate Professor
Department of Civil Engineering
IIT Madras
April 2014
U. Saravanan (IIT Madras) Finite element method April 2014 1 / 81
Outline
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 2 / 81
Outline
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 2 / 81
Finite element formulation
Solution Procedure

Applied forces
Motion / displacement
Stress
Deformation gradient
or strain
Constitutive
relation
Equilibrium
equations
Strain
displacement
relation
U. Saravanan (IIT Madras) Finite element method April 2014 3 / 81
Finite element formulation
Basic equations in mechanics
Strain-displacement relation
h = grad(u), =
1
2
[h +h
t
] (1)
Constitutive relation - Isotropic Hookes law:
= tr ()1 + 2 = D. (2)
where
D =
_
_
_
_
_
_
_
_
+ 2 0 0 0
+ 2 0 0 0
+ 2 0 0 0
0 0 0 2 0 0
0 0 0 0 2 0
0 0 0 0 0 2
_
_
_
_
_
_
_
_
, (3)
U. Saravanan (IIT Madras) Finite element method April 2014 4 / 81
Finite element formulation
Basic equations in mechanics
Strain-displacement relation
h = grad(u), =
1
2
[h +h
t
] (1)
Constitutive relation - Isotropic Hookes law:
= tr ()1 + 2 = D. (2)
where
D =
_
_
_
_
_
_
_
_
+ 2 0 0 0
+ 2 0 0 0
+ 2 0 0 0
0 0 0 2 0 0
0 0 0 0 2 0
0 0 0 0 0 2
_
_
_
_
_
_
_
_
, (3)
U. Saravanan (IIT Madras) Finite element method April 2014 4 / 81
Finite element formulation
Basic equations in mechanics
Equilibrium Equations - Spatial form
div() + b = a. (4)
U. Saravanan (IIT Madras) Finite element method April 2014 5 / 81
Finite element formulation
Initial-Boundary value problem - Spatial formulation
To nd a dierentiable displacement eld, u so that:
div() + b = a,
u(x, t) |
t=0
= u
o
,
Du
Dt
|
t=0
= v
o
,
_
_
_
, on A
t
(5)
u = u, on A
u
t
, (6)
t
(n)
= n =

t, on A

t
. (7)
where A
u
t
and A

t
denote the boundary of the body where displacement
and traction are specied respectively and is such that
A
u
t
A

t
= , and A
u
t
A

t
= A
t
(8)
U. Saravanan (IIT Madras) Finite element method April 2014 6 / 81
Finite element formulation
Displacement eld
u =
0
(x) +
n

i =1
c
i
(t)
i
(x) (9)
where,
i
s are specied a priori and c
i
s are to be determined.
Then, the displacement gradient is:
h = grad(
0
) +
n

i =1
c
i
grad(
i
) (10)
U. Saravanan (IIT Madras) Finite element method April 2014 7 / 81
Finite element formulation
Desirable properties of
0

0
=
0
(x)
Is a dierentiable function
Ensures that the displacement boundary condition is met, i.e.,

0
(x) = u, x A
u
t
. (11)
U. Saravanan (IIT Madras) Finite element method April 2014 8 / 81
Finite element formulation
Desirable properties of
0

0
=
0
(x)
Is a dierentiable function
Ensures that the displacement boundary condition is met, i.e.,

0
(x) = u, x A
u
t
. (11)
U. Saravanan (IIT Madras) Finite element method April 2014 8 / 81
Finite element formulation
Desirable properties of
i
, i = 0
Basis Functions:

i
=
i
(x)
Satisfy homogeneous displacement boundary condition

i
(x) = o, x A
u
t
, (12)

i
s to be dierentiable functions
The set {
i
} to be complete.
The set {
i
} to be linearly independent.
U. Saravanan (IIT Madras) Finite element method April 2014 9 / 81
Finite element formulation
Desirable properties of
i
, i = 0
Basis Functions:

i
=
i
(x)
Satisfy homogeneous displacement boundary condition

i
(x) = o, x A
u
t
, (12)

i
s to be dierentiable functions
The set {
i
} to be complete.
The set {
i
} to be linearly independent.
U. Saravanan (IIT Madras) Finite element method April 2014 9 / 81
Finite element formulation
Desirable properties of
i
, i = 0
Basis Functions:

i
=
i
(x)
Satisfy homogeneous displacement boundary condition

i
(x) = o, x A
u
t
, (12)

i
s to be dierentiable functions
The set {
i
} to be complete.
The set {
i
} to be linearly independent.
U. Saravanan (IIT Madras) Finite element method April 2014 9 / 81
Finite element formulation
Desirable properties of
i
, i = 0
Basis Functions:

i
=
i
(x)
Satisfy homogeneous displacement boundary condition

i
(x) = o, x A
u
t
, (12)

i
s to be dierentiable functions
The set {
i
} to be complete.
The set {
i
} to be linearly independent.
U. Saravanan (IIT Madras) Finite element method April 2014 9 / 81
Finite element formulation
Desirable properties of
i
, i = 0
Basis Functions:

i
=
i
(x)
Satisfy homogeneous displacement boundary condition

i
(x) = o, x A
u
t
, (12)

i
s to be dierentiable functions
The set {
i
} to be complete.
The set {
i
} to be linearly independent.
U. Saravanan (IIT Madras) Finite element method April 2014 9 / 81
Finite element formulation
Weakened Equilibrium equation - Spatial form
Let the weight function, =

(x), then
_
A
t
[div() + b a] dv = 0 (13)
Recollecting:
div(A
t
u) = div(A) u +A grad(u), and
_
A
t
div(u)dv =
_
A
t
u nds
(14)
_
A
t
_
div(
t
) grad() + b a

dv = 0 (15)
U. Saravanan (IIT Madras) Finite element method April 2014 10 / 81
Finite element formulation
Weakened Equilibrium equation
Hence weakened equilibrium equation in spatial form
_
A
t
[ grad() + a ] dv =
_
A
t
b dv +
_
A
t
t
(n)
ds (16)
_
A
t
_

+ a
_
dv =
_
A
t
b dv +
_
A
t
t
(n)
ds (17)
U. Saravanan (IIT Madras) Finite element method April 2014 11 / 81
Finite element formulation
Weighted residual methods
How do we nd the unknown c
i
s?
Such that the weakened equilibrium equations are satised
Galerkin method:
=
i
, for i = {1, . . . , n}
Petrov-Galerkin method:
=
i
, for i = {1, . . . , n},
i
=
i
Properties of
i
Be a dierentiable function
Satisfy homogeneous boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 12 / 81
Finite element formulation
Weighted residual methods
How do we nd the unknown c
i
s?
Such that the weakened equilibrium equations are satised
Galerkin method:
=
i
, for i = {1, . . . , n}
Petrov-Galerkin method:
=
i
, for i = {1, . . . , n},
i
=
i
Properties of
i
Be a dierentiable function
Satisfy homogeneous boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 12 / 81
Finite element formulation
Weighted residual methods
How do we nd the unknown c
i
s?
Such that the weakened equilibrium equations are satised
Galerkin method:
=
i
, for i = {1, . . . , n}
Petrov-Galerkin method:
=
i
, for i = {1, . . . , n},
i
=
i
Properties of
i
Be a dierentiable function
Satisfy homogeneous boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 12 / 81
Finite element formulation
Weighted residual methods
How do we nd the unknown c
i
s?
Such that the weakened equilibrium equations are satised
Galerkin method:
=
i
, for i = {1, . . . , n}
Petrov-Galerkin method:
=
i
, for i = {1, . . . , n},
i
=
i
Properties of
i
Be a dierentiable function
Satisfy homogeneous boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 12 / 81
Finite element formulation
Weighted residual methods
How do we nd the unknown c
i
s?
Such that the weakened equilibrium equations are satised
Galerkin method:
=
i
, for i = {1, . . . , n}
Petrov-Galerkin method:
=
i
, for i = {1, . . . , n},
i
=
i
Properties of
i
Be a dierentiable function
Satisfy homogeneous boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 12 / 81
Finite element formulation
Weighted residual methods
How do we nd the unknown c
i
s?
Such that the weakened equilibrium equations are satised
Galerkin method:
=
i
, for i = {1, . . . , n}
Petrov-Galerkin method:
=
i
, for i = {1, . . . , n},
i
=
i
Properties of
i
Be a dierentiable function
Satisfy homogeneous boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 12 / 81
Finite element formulation
Weighted residual methods
How do we nd the unknown c
i
s?
Such that the weakened equilibrium equations are satised
Galerkin method:
=
i
, for i = {1, . . . , n}
Petrov-Galerkin method:
=
i
, for i = {1, . . . , n},
i
=
i
Properties of
i
Be a dierentiable function
Satisfy homogeneous boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 12 / 81
Finite element formulation
Why these restrictions on the basis and weight functions?
We require
0
to satisfy the prescribed boundary conditions and
i
i
{1, . . . , n} to satisfy the homogeneous boundary conditions so that
the value of c
i
s are independent of the boundary condition
The restriction on weight functions is so that the following term can be
computed independent of c
i
s
_
A
t
t
(n)
ds =
_
A
u
t
t
(n)
ds +
_
A

t
t
(n)
ds =
_
A

t ds, (18)
by virtue of the required form of the boundary condition and the
requirement on the weight function,
U. Saravanan (IIT Madras) Finite element method April 2014 13 / 81
Finite element formulation
Why these restrictions on the basis and weight functions?
We require
0
to satisfy the prescribed boundary conditions and
i
i
{1, . . . , n} to satisfy the homogeneous boundary conditions so that
the value of c
i
s are independent of the boundary condition
The restriction on weight functions is so that the following term can be
computed independent of c
i
s
_
A
t
t
(n)
ds =
_
A
u
t
t
(n)
ds +
_
A

t
t
(n)
ds =
_
A

t ds, (18)
by virtue of the required form of the boundary condition and the
requirement on the weight function,
U. Saravanan (IIT Madras) Finite element method April 2014 13 / 81
Finite element formulation Six steps in nite element analysis
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 14 / 81
Finite element formulation Six steps in nite element analysis
Introduction
What does the nite element method oer?
The nite element method provides a systematic way of generating
the basis functions
The nite element analysis of a typical problem involves six steps
which we shall list next
U. Saravanan (IIT Madras) Finite element method April 2014 15 / 81
Finite element formulation Six steps in nite element analysis
Introduction
What does the nite element method oer?
The nite element method provides a systematic way of generating
the basis functions
The nite element analysis of a typical problem involves six steps
which we shall list next
U. Saravanan (IIT Madras) Finite element method April 2014 15 / 81
Finite element formulation Six steps in nite element analysis
Introduction
What does the nite element method oer?
The nite element method provides a systematic way of generating
the basis functions
The nite element analysis of a typical problem involves six steps
which we shall list next
U. Saravanan (IIT Madras) Finite element method April 2014 15 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 1: Representation of the given domain into a collection of
preselected nite elements
a. Construct the nite element mesh of preselected elements
b. Number the nodes and elements
c. Generate the geometric properties needed for the problem
U. Saravanan (IIT Madras) Finite element method April 2014 16 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 1: Representation of the given domain into a collection of
preselected nite elements
a. Construct the nite element mesh of preselected elements
b. Number the nodes and elements
c. Generate the geometric properties needed for the problem
U. Saravanan (IIT Madras) Finite element method April 2014 16 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 1: Representation of the given domain into a collection of
preselected nite elements
a. Construct the nite element mesh of preselected elements
b. Number the nodes and elements
c. Generate the geometric properties needed for the problem
U. Saravanan (IIT Madras) Finite element method April 2014 16 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 1: Representation of the given domain into a collection of
preselected nite elements
a. Construct the nite element mesh of preselected elements
b. Number the nodes and elements
c. Generate the geometric properties needed for the problem
U. Saravanan (IIT Madras) Finite element method April 2014 16 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 2: Derivation of element equations for all typical elements in the
mesh
a. Consider the weighted residual formulation of the balance of linear
momentum equation over the typical element
b. Assume that the displacement is of the form,
u =
n

i =1
u
i

i
,
and substitute it into step 2a to obtain element equation of the form
[K
e
]{u
e
} = {F
e
} (19)
c. Derive element interpolation functions,
i
and compute the element
matrices.
U. Saravanan (IIT Madras) Finite element method April 2014 17 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 2: Derivation of element equations for all typical elements in the
mesh
a. Consider the weighted residual formulation of the balance of linear
momentum equation over the typical element
b. Assume that the displacement is of the form,
u =
n

i =1
u
i

i
,
and substitute it into step 2a to obtain element equation of the form
[K
e
]{u
e
} = {F
e
} (19)
c. Derive element interpolation functions,
i
and compute the element
matrices.
U. Saravanan (IIT Madras) Finite element method April 2014 17 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 2: Derivation of element equations for all typical elements in the
mesh
a. Consider the weighted residual formulation of the balance of linear
momentum equation over the typical element
b. Assume that the displacement is of the form,
u =
n

i =1
u
i

i
,
and substitute it into step 2a to obtain element equation of the form
[K
e
]{u
e
} = {F
e
} (19)
c. Derive element interpolation functions,
i
and compute the element
matrices.
U. Saravanan (IIT Madras) Finite element method April 2014 17 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 2: Derivation of element equations for all typical elements in the
mesh
a. Consider the weighted residual formulation of the balance of linear
momentum equation over the typical element
b. Assume that the displacement is of the form,
u =
n

i =1
u
i

i
,
and substitute it into step 2a to obtain element equation of the form
[K
e
]{u
e
} = {F
e
} (19)
c. Derive element interpolation functions,
i
and compute the element
matrices.
U. Saravanan (IIT Madras) Finite element method April 2014 17 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 3: Assembly of element equations to obtain the equations of the
whole problem
a. Identify the inter-element continuity conditions on the displacement
by relating element nodes to global nodes
b. Assemble element equations using 3a. and equilibrium equations.
Step - 4: Imposition of the boundary conditions of the problem
a. Identify the specied global displacement degrees of freedom
b. Identify the specied traction
U. Saravanan (IIT Madras) Finite element method April 2014 18 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 3: Assembly of element equations to obtain the equations of the
whole problem
a. Identify the inter-element continuity conditions on the displacement
by relating element nodes to global nodes
b. Assemble element equations using 3a. and equilibrium equations.
Step - 4: Imposition of the boundary conditions of the problem
a. Identify the specied global displacement degrees of freedom
b. Identify the specied traction
U. Saravanan (IIT Madras) Finite element method April 2014 18 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 3: Assembly of element equations to obtain the equations of the
whole problem
a. Identify the inter-element continuity conditions on the displacement
by relating element nodes to global nodes
b. Assemble element equations using 3a. and equilibrium equations.
Step - 4: Imposition of the boundary conditions of the problem
a. Identify the specied global displacement degrees of freedom
b. Identify the specied traction
U. Saravanan (IIT Madras) Finite element method April 2014 18 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 3: Assembly of element equations to obtain the equations of the
whole problem
a. Identify the inter-element continuity conditions on the displacement
by relating element nodes to global nodes
b. Assemble element equations using 3a. and equilibrium equations.
Step - 4: Imposition of the boundary conditions of the problem
a. Identify the specied global displacement degrees of freedom
b. Identify the specied traction
U. Saravanan (IIT Madras) Finite element method April 2014 18 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 3: Assembly of element equations to obtain the equations of the
whole problem
a. Identify the inter-element continuity conditions on the displacement
by relating element nodes to global nodes
b. Assemble element equations using 3a. and equilibrium equations.
Step - 4: Imposition of the boundary conditions of the problem
a. Identify the specied global displacement degrees of freedom
b. Identify the specied traction
U. Saravanan (IIT Madras) Finite element method April 2014 18 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 3: Assembly of element equations to obtain the equations of the
whole problem
a. Identify the inter-element continuity conditions on the displacement
by relating element nodes to global nodes
b. Assemble element equations using 3a. and equilibrium equations.
Step - 4: Imposition of the boundary conditions of the problem
a. Identify the specied global displacement degrees of freedom
b. Identify the specied traction
U. Saravanan (IIT Madras) Finite element method April 2014 18 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 5: Solution of the assembled equations
Step - 6: Postprocessing of the results
a. Compute the gradient of the determined displacement eld so that
the stresses can be computed
b. Represent the results in a tabular and/or graphical form.
U. Saravanan (IIT Madras) Finite element method April 2014 19 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 5: Solution of the assembled equations
Step - 6: Postprocessing of the results
a. Compute the gradient of the determined displacement eld so that
the stresses can be computed
b. Represent the results in a tabular and/or graphical form.
U. Saravanan (IIT Madras) Finite element method April 2014 19 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 5: Solution of the assembled equations
Step - 6: Postprocessing of the results
a. Compute the gradient of the determined displacement eld so that
the stresses can be computed
b. Represent the results in a tabular and/or graphical form.
U. Saravanan (IIT Madras) Finite element method April 2014 19 / 81
Finite element formulation Six steps in nite element analysis
Steps involved in the nite element analysis
Step - 5: Solution of the assembled equations
Step - 6: Postprocessing of the results
a. Compute the gradient of the determined displacement eld so that
the stresses can be computed
b. Represent the results in a tabular and/or graphical form.
U. Saravanan (IIT Madras) Finite element method April 2014 19 / 81
Finite element formulation for linearized elasticity
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 20 / 81
Finite element formulation for linearized elasticity Decomposition of the domain - Selection of element
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 21 / 81
Finite element formulation for linearized elasticity Decomposition of the domain - Selection of element
Elements

e
x
e
y
e
z
4
1
2
3
Figure: Tetrahedral solid element
At the end of this
step, the (x, y, z)
coordinates of
the respective
nodes as well as
the connectivity
matrix containing
the element
number and the
nodes that
constitute this
element are
assumed to be
available.
U. Saravanan (IIT Madras) Finite element method April 2014 22 / 81
Finite element formulation for linearized elasticity Decomposition of the domain - Selection of element
Elements

e
x
e
y
e
z
4
1
2
3
Figure: Tetrahedral solid element
At the end of this
step, the (x, y, z)
coordinates of
the respective
nodes as well as
the connectivity
matrix containing
the element
number and the
nodes that
constitute this
element are
assumed to be
available.
U. Saravanan (IIT Madras) Finite element method April 2014 22 / 81
Finite element formulation for linearized elasticity Decomposition of the domain - Selection of element
Elements

e
x
e
y
e
z
4
1
2
3
Figure: Tetrahedral solid element
At the end of this
step, the (x, y, z)
coordinates of
the respective
nodes as well as
the connectivity
matrix containing
the element
number and the
nodes that
constitute this
element are
assumed to be
available.
U. Saravanan (IIT Madras) Finite element method April 2014 22 / 81
Finite element formulation for linearized elasticity Decomposition of the domain - Selection of element
Elements

e
x
e
y
e
z
4
1
2
3
Figure: Tetrahedral solid element
At the end of this
step, the (x, y, z)
coordinates of
the respective
nodes as well as
the connectivity
matrix containing
the element
number and the
nodes that
constitute this
element are
assumed to be
available.
U. Saravanan (IIT Madras) Finite element method April 2014 22 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 23 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Degrees of freedom

e
x
e
y
e
z
4
1
2
3
Figure: Tetrahedral solid element
{d} =
_

_
u
1
v
1
w
1
.
.
.
u
4
v
4
w
4
_

_
(20)
U. Saravanan (IIT Madras) Finite element method April 2014 24 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Degrees of freedom
e
x
e
y
e
z
4
1
2
3
Figure: Tetrahedral solid element
{d} =
_

_
u
1
v
1
w
1
.
.
.
u
4
v
4
w
4
_

_
(20)
U. Saravanan (IIT Madras) Finite element method April 2014 24 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Degrees of freedom
e
x
e
y
e
z
4
1
2
3
Figure: Tetrahedral solid element
{d} =
_

_
u
1
v
1
w
1
.
.
.
u
4
v
4
w
4
_

_
(20)
U. Saravanan (IIT Madras) Finite element method April 2014 24 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Interpolation functions
Since there are 12 degrees of freedom, there should be 12 interpolation
functions,
i
, one corresponding to each degree of freedom, so that
u = ue
x
+ ve
y
+ we
z
=
12

i =1
d
i

i
(21)
U. Saravanan (IIT Madras) Finite element method April 2014 25 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Interpolation functions
To maintain the meaning of the degrees of freedom, we choose the
interpolation functions as

1
= N
1
(x, y, z)e
x
,
2
= N
1
(x, y, z)e
y
,
3
= N
1
(x, y, z)e
z

4
= N
2
(x, y, z)e
x
,
5
= N
2
(x, y, z)e
y
,
6
= N
2
(x, y, z)e
z

7
= N
3
(x, y, z)e
x
,
8
= N
3
(x, y, z)e
y
,
9
= N
3
(x, y, z)e
z

10
= N
4
(x, y, z)e
x
,
11
= N
4
(x, y, z)e
y
,
12
= N
4
(x, y, z)e
z
where N
i
s are the Lagrange interpolation functions and are such that
N
i
(x
j
, y
j
, z
j
) =
ij
, (22)
where (x
j
, y
j
, z
j
) are the coordinates of the four node points of the
tetrahedron.
U. Saravanan (IIT Madras) Finite element method April 2014 26 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Interpolation functions
To maintain the meaning of the degrees of freedom, we choose the
interpolation functions as

1
= N
1
(x, y, z)e
x
,
2
= N
1
(x, y, z)e
y
,
3
= N
1
(x, y, z)e
z

4
= N
2
(x, y, z)e
x
,
5
= N
2
(x, y, z)e
y
,
6
= N
2
(x, y, z)e
z

7
= N
3
(x, y, z)e
x
,
8
= N
3
(x, y, z)e
y
,
9
= N
3
(x, y, z)e
z

10
= N
4
(x, y, z)e
x
,
11
= N
4
(x, y, z)e
y
,
12
= N
4
(x, y, z)e
z
where N
i
s are the Lagrange interpolation functions and are such that
N
i
(x
j
, y
j
, z
j
) =
ij
, (22)
where (x
j
, y
j
, z
j
) are the coordinates of the four node points of the
tetrahedron.
U. Saravanan (IIT Madras) Finite element method April 2014 26 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of linear interpolation function
Assuming, the interpolation function is linear along each edge and on each
plane of the tetrahedron,
N
i
= b
i
+ c
i
x + d
i
y + e
i
z (23)
where b
i
, c
i
, d
i
and e
i
are constants determined from the condition that
N
i
(x
j
, y
j
, z
j
) =
ij
, where (x
j
, y
j
, z
j
) are the coordinates of the four node
points of the tetrahedron.
Then, the requirement: N
1
(x
j
, y
j
, z
j
) =
1j
, can be written as:
_
_
_
_
1 x
1
y
1
z
1
1 x
2
y
2
z
2
1 x
3
y
3
z
3
1 x
4
y
4
z
4
_
_
_
_
_

_
b
1
c
1
d
1
e
1
_

_
=
_

_
1
0
0
0
_

_
(24)
U. Saravanan (IIT Madras) Finite element method April 2014 27 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of linear interpolation function
Assuming, the interpolation function is linear along each edge and on each
plane of the tetrahedron,
N
i
= b
i
+ c
i
x + d
i
y + e
i
z (23)
where b
i
, c
i
, d
i
and e
i
are constants determined from the condition that
N
i
(x
j
, y
j
, z
j
) =
ij
, where (x
j
, y
j
, z
j
) are the coordinates of the four node
points of the tetrahedron.
Then, the requirement: N
1
(x
j
, y
j
, z
j
) =
1j
, can be written as:
_
_
_
_
1 x
1
y
1
z
1
1 x
2
y
2
z
2
1 x
3
y
3
z
3
1 x
4
y
4
z
4
_
_
_
_
_

_
b
1
c
1
d
1
e
1
_

_
=
_

_
1
0
0
0
_

_
(24)
U. Saravanan (IIT Madras) Finite element method April 2014 27 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of linear interpolation function
Solving the above equation we obtain:
b
1
=
1

det
_
_
_
_
x
2
y
2
z
2
x
3
y
3
z
3
x
4
y
4
z
4
_
_
_
_
, c
1
=
1

det
_
_
_
_
1 y
2
z
2
1 y
3
z
3
1 y
4
z
4
_
_
_
_
d
1
=
1

det
_
_
_
_
1 x
2
z
2
1 x
3
z
3
1 x
4
z
4
_
_
_
_
, e
1
=
1

det
_
_
_
_
1 x
2
y
2
1 x
3
y
3
1 x
4
y
4
_
_
_
_
Where
= det
_
_
_
_
_

_
1 x
1
y
1
z
1
1 x
2
y
2
z
2
1 x
3
y
3
z
3
1 x
4
y
4
z
4
_

_
_
_
_
_
(25)
U. Saravanan (IIT Madras) Finite element method April 2014 28 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of linear interpolation function
Solving the above equation we obtain:
b
1
=
1

det
_
_
_
_
x
2
y
2
z
2
x
3
y
3
z
3
x
4
y
4
z
4
_
_
_
_
, c
1
=
1

det
_
_
_
_
1 y
2
z
2
1 y
3
z
3
1 y
4
z
4
_
_
_
_
d
1
=
1

det
_
_
_
_
1 x
2
z
2
1 x
3
z
3
1 x
4
z
4
_
_
_
_
, e
1
=
1

det
_
_
_
_
1 x
2
y
2
1 x
3
y
3
1 x
4
y
4
_
_
_
_
Where
= det
_
_
_
_
_

_
1 x
1
y
1
z
1
1 x
2
y
2
z
2
1 x
3
y
3
z
3
1 x
4
y
4
z
4
_

_
_
_
_
_
(25)
U. Saravanan (IIT Madras) Finite element method April 2014 28 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Representation for the displacement eld
Hence, the displacement eld over the element can be represented in
matrix form as:
u =
_
_
_
u
v
w
_
_
_
= (N
FEM
){d} (26)
Where
(N
FEM
) =
_
_
N
1
0 0 N
2
0 0 N
3
0 0 N
4
0 0
0 N
1
0 0 N
2
0 0 N
3
0 0 N
4
0
0 0 N
1
0 0 N
2
0 0 N
3
0 0 N
4
_
_
(27)
U. Saravanan (IIT Madras) Finite element method April 2014 29 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Representation for the displacement eld
Hence, the displacement eld over the element can be represented in
matrix form as:
u =
_
_
_
u
v
w
_
_
_
= (N
FEM
){d} (26)
Where
(N
FEM
) =
_
_
N
1
0 0 N
2
0 0 N
3
0 0 N
4
0 0
0 N
1
0 0 N
2
0 0 N
3
0 0 N
4
0
0 0 N
1
0 0 N
2
0 0 N
3
0 0 N
4
_
_
(27)
U. Saravanan (IIT Madras) Finite element method April 2014 29 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Innitesimal strain tensor
A representation of the innitesimal strain tensor
=
_

_
()
xx
()
yy
()
zz
2()
xy
2()
yz
2()
xz
_

_
=
_

_
u
x
v
y
w
z
u
y
+
v
x
v
z
+
w
y
w
x
+
u
z
_

_
(28)
The strain in the element can be written as
{} = (B
FEM
){d} (29)
U. Saravanan (IIT Madras) Finite element method April 2014 30 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Innitesimal strain tensor
A representation of the innitesimal strain tensor
=
_

_
()
xx
()
yy
()
zz
2()
xy
2()
yz
2()
xz
_

_
=
_

_
u
x
v
y
w
z
u
y
+
v
x
v
z
+
w
y
w
x
+
u
z
_

_
(28)
The strain in the element can be written as
{} = (B
FEM
){d} (29)
U. Saravanan (IIT Madras) Finite element method April 2014 30 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Denition of (B
FEM
)
(B
FEM
) =
_
_
_
_
_
_
_
_
N
1,x
0 0 N
2,x
0 0 N
3,x
0 0 N
4,x
0 0
0 N
1,y
0 0 N
2,y
0 0 N
3,y
0 0 N
4,y
0
0 0 N
1,z
0 0 N
2,z
0 0 N
3,z
0 0 N
4,z
N
1,y
N
1,x
0 N
2,y
N
2,x
0 N
3,y
N
3,x
0 N
4,y
N
4,x
0
0 N
1,z
N
1,y
0 N
2,z
N
2,y
0 N
3,z
N
3,y
0 N
4,z
N
4,y
N
1,z
0 N
1,x
N
2,z
0 N
2,x
N
3,z
0 N
3,x
N
4,z
0 N
4,x
_
_
_
_
_
_
_
_
(30)
U. Saravanan (IIT Madras) Finite element method April 2014 31 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Solution Procedure

Applied forces
Motion / displacement
Stress
Deformation gradient
or strain
Constitutive
relation
Equilibrium
equations
Strain
displacement
relation
U. Saravanan (IIT Madras) Finite element method April 2014 32 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Evaluation of Cauchy stress tensor
Constitutive relation: Hookes Law
= D
A representation for Cauchy stress tensor
{} =
_

xx

yy

zz

xy

yz

xz
_

_
(31)
Now, the Hookes law can be written as
{} = (D
FEM
){} (32)
U. Saravanan (IIT Madras) Finite element method April 2014 33 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Evaluation of Cauchy stress tensor
Constitutive relation: Hookes Law
= D
A representation for Cauchy stress tensor
{} =
_

xx

yy

zz

xy

yz

xz
_

_
(31)
Now, the Hookes law can be written as
{} = (D
FEM
){} (32)
U. Saravanan (IIT Madras) Finite element method April 2014 33 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Evaluation of Cauchy stress tensor
Constitutive relation: Hookes Law
= D
A representation for Cauchy stress tensor
{} =
_

xx

yy

zz

xy

yz

xz
_

_
(31)
Now, the Hookes law can be written as
{} = (D
FEM
){} (32)
U. Saravanan (IIT Madras) Finite element method April 2014 33 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Evaluation of Cauchy stress tensor
For isotropic materials
(D
FEM
) =
E
(1 + )(1 2)
_
_
_
_
_
_
_
_
1 0 0 0
1 0 0 0
1 0 0 0
0 0 0 1 2 0 0
0 0 0 0 1 2 0
0 0 0 0 0 1 2
_
_
_
_
_
_
_
_
(33)
Substituting for the strain in the constitutive relation we obtain
{} = (D
FEM
)(B
FEM
){d} (34)
U. Saravanan (IIT Madras) Finite element method April 2014 34 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Evaluation of Cauchy stress tensor
For isotropic materials
(D
FEM
) =
E
(1 + )(1 2)
_
_
_
_
_
_
_
_
1 0 0 0
1 0 0 0
1 0 0 0
0 0 0 1 2 0 0
0 0 0 0 1 2 0
0 0 0 0 0 1 2
_
_
_
_
_
_
_
_
(33)
Substituting for the strain in the constitutive relation we obtain
{} = (D
FEM
)(B
FEM
){d} (34)
U. Saravanan (IIT Madras) Finite element method April 2014 34 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Solution Procedure

Applied forces
Motion / displacement
Stress
Deformation gradient
or strain
Constitutive
relation
Equilibrium
equations
Strain
displacement
relation
U. Saravanan (IIT Madras) Finite element method April 2014 35 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of elemental equations
Weakened equilibrium equation
_
A
t
_

+ a
_
dv =
_
A
t
b dv +
_
A
t
t
(n)
ds (35)
Dening

i
=
1
2
_
grad(
i
) + grad
t
(
i
)

(36)
Identifying the weight function, with
i
_
A
t
[

i
+ a
i
] dv =
_
A
t
b
i
dv +
_
A
t
t
(n)

i
ds (37)
for i {1, 2, . . . , 12}
U. Saravanan (IIT Madras) Finite element method April 2014 36 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of elemental equations
Weakened equilibrium equation
_
A
t
_

+ a
_
dv =
_
A
t
b dv +
_
A
t
t
(n)
ds (35)
Dening

i
=
1
2
_
grad(
i
) + grad
t
(
i
)

(36)
Identifying the weight function, with
i
_
A
t
[

i
+ a
i
] dv =
_
A
t
b
i
dv +
_
A
t
t
(n)

i
ds (37)
for i {1, 2, . . . , 12}
U. Saravanan (IIT Madras) Finite element method April 2014 36 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of elemental equations
Weakened equilibrium equation
_
A
t
_

+ a
_
dv =
_
A
t
b dv +
_
A
t
t
(n)
ds (35)
Dening

i
=
1
2
_
grad(
i
) + grad
t
(
i
)

(36)
Identifying the weight function, with
i
_
A
t
[

i
+ a
i
] dv =
_
A
t
b
i
dv +
_
A
t
t
(n)

i
ds (37)
for i {1, 2, . . . , 12}
U. Saravanan (IIT Madras) Finite element method April 2014 36 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Computation of acceleration
Acceleration:
a =

2
u
t
2
, (38)
When the developed displacement gradient, spatial velocities and its
gradient are very small
U. Saravanan (IIT Madras) Finite element method April 2014 37 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Computation of acceleration
Assuming small deformations
a

2
u
t
2
=
_
4

i =1
N
i
d
2
u
i
dt
2
_
e
x
+
_
4

i =1
N
i
d
2
v
i
dt
2
_
e
y
+
_
4

i =1
N
i
d
2
w
i
dt
2
_
e
z
= (N
FEM
){

d} (39)
Where
{

d} =
_

_
d
2
u
1
dt
2
d
2
v
1
dt
2
.
.
.
d
2
v
4
dt
2
d
2
w
4
dt
2
_

_
(40)
U. Saravanan (IIT Madras) Finite element method April 2014 38 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Computation of acceleration
Assuming small deformations
a

2
u
t
2
=
_
4

i =1
N
i
d
2
u
i
dt
2
_
e
x
+
_
4

i =1
N
i
d
2
v
i
dt
2
_
e
y
+
_
4

i =1
N
i
d
2
w
i
dt
2
_
e
z
= (N
FEM
){

d} (39)
Where
{

d} =
_

_
d
2
u
1
dt
2
d
2
v
1
dt
2
.
.
.
d
2
v
4
dt
2
d
2
w
4
dt
2
_

_
(40)
U. Saravanan (IIT Madras) Finite element method April 2014 38 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Notation and Identities
Dening
{d
i
} as a vector with only the i
th
element of the array in d
i
being 1 and
all the other elements are 0
Then

i
= (N
FEM
){d
i
} (41)

i
= (B
FEM
){d
i
} (42)
U. Saravanan (IIT Madras) Finite element method April 2014 39 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Notation and Identities
Dening
{d
i
} as a vector with only the i
th
element of the array in d
i
being 1 and
all the other elements are 0
Then

i
= (N
FEM
){d
i
} (41)

i
= (B
FEM
){d
i
} (42)
U. Saravanan (IIT Madras) Finite element method April 2014 39 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of elemental equations
Substituting the expressions we obtained for the Cauchy stress and
acceleration in the weakened spatial form of the equilibrium equations and
using the above identities we obtain
_
A
t
_
(D
FEM
)(B
FEM
){d} (B
FEM
){d
i
} + (N
FEM
){

d} (N
FEM
){d
i
}
_
dv
=
_
A
t
b (N
FEM
){d
i
}dv +
_
A
t
t
(n)
(N
FEM
){d
i
}ds
where the integration is performed over the volume or the surface of the
tetrahedron under investigation respectively
The above equation holds for d
i
, i {1, 2, . . . , 12}
Hence, it is equivalent to requiring that this hold for an arbitrary
vector d
a
(Why?)
U. Saravanan (IIT Madras) Finite element method April 2014 40 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of elemental equations
Substituting the expressions we obtained for the Cauchy stress and
acceleration in the weakened spatial form of the equilibrium equations and
using the above identities we obtain
_
A
t
_
(D
FEM
)(B
FEM
){d} (B
FEM
){d
i
} + (N
FEM
){

d} (N
FEM
){d
i
}
_
dv
=
_
A
t
b (N
FEM
){d
i
}dv +
_
A
t
t
(n)
(N
FEM
){d
i
}ds
where the integration is performed over the volume or the surface of the
tetrahedron under investigation respectively
The above equation holds for d
i
, i {1, 2, . . . , 12}
Hence, it is equivalent to requiring that this hold for an arbitrary
vector d
a
(Why?)
U. Saravanan (IIT Madras) Finite element method April 2014 40 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Formulation of elemental equations
Substituting the expressions we obtained for the Cauchy stress and
acceleration in the weakened spatial form of the equilibrium equations and
using the above identities we obtain
_
A
t
_
(D
FEM
)(B
FEM
){d} (B
FEM
){d
i
} + (N
FEM
){

d} (N
FEM
){d
i
}
_
dv
=
_
A
t
b (N
FEM
){d
i
}dv +
_
A
t
t
(n)
(N
FEM
){d
i
}ds
where the integration is performed over the volume or the surface of the
tetrahedron under investigation respectively
The above equation holds for d
i
, i {1, 2, . . . , 12}
Hence, it is equivalent to requiring that this hold for an arbitrary
vector d
a
(Why?)
U. Saravanan (IIT Madras) Finite element method April 2014 40 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Elemental equation
The above equation can be written as
(K
e
){d} + (M
e
){

d} = {f
e
b
} +{f
e
s
} (43)
Where
(K
e
) =
_
A
t
(B
FEM
)
t
(D
FEM
)(B
FEM
)dv, Element stiness matrix
(M
e
) =
_
A
t
(N
FEM
)
t
(N
FEM
)dv, Element mass matrix
{f
e
b
} =
_
A
t
(N
FEM
)
t
bdv, Element body force vector
{f
e
s
} =
_
A
t
(N
FEM
)
t
t
(n)
ds, Element surface traction vector
U. Saravanan (IIT Madras) Finite element method April 2014 41 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Elemental equation
The above equation can be written as
(K
e
){d} + (M
e
){

d} = {f
e
b
} +{f
e
s
} (43)
Where
(K
e
) =
_
A
t
(B
FEM
)
t
(D
FEM
)(B
FEM
)dv, Element stiness matrix
(M
e
) =
_
A
t
(N
FEM
)
t
(N
FEM
)dv, Element mass matrix
{f
e
b
} =
_
A
t
(N
FEM
)
t
bdv, Element body force vector
{f
e
s
} =
_
A
t
(N
FEM
)
t
t
(n)
ds, Element surface traction vector
U. Saravanan (IIT Madras) Finite element method April 2014 41 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Points to be understood
While the stiness, mass matrices and body force vector could be
computed, surface traction could not be computed at this stage
However, if part or the entire boundary of the element under
consideration happened to coincide with the boundary of the body
where the traction is prescribed, then this term can be computed
But to maintain generality, the surface traction vector is assumed to
be unknown and hence at this stage there are 2n unknowns - n
displacement degrees of freedom and n components of the surface
traction vector, where n is 12 for the tetrahedral element under
consideration
U. Saravanan (IIT Madras) Finite element method April 2014 42 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Points to be understood
While the stiness, mass matrices and body force vector could be
computed, surface traction could not be computed at this stage
However, if part or the entire boundary of the element under
consideration happened to coincide with the boundary of the body
where the traction is prescribed, then this term can be computed
But to maintain generality, the surface traction vector is assumed to
be unknown and hence at this stage there are 2n unknowns - n
displacement degrees of freedom and n components of the surface
traction vector, where n is 12 for the tetrahedral element under
consideration
U. Saravanan (IIT Madras) Finite element method April 2014 42 / 81
Finite element formulation for linearized elasticity Derivation of elemental equations
Points to be understood
While the stiness, mass matrices and body force vector could be
computed, surface traction could not be computed at this stage
However, if part or the entire boundary of the element under
consideration happened to coincide with the boundary of the body
where the traction is prescribed, then this term can be computed
But to maintain generality, the surface traction vector is assumed to
be unknown and hence at this stage there are 2n unknowns - n
displacement degrees of freedom and n components of the surface
traction vector, where n is 12 for the tetrahedral element under
consideration
U. Saravanan (IIT Madras) Finite element method April 2014 42 / 81
Finite element formulation for linearized elasticity Assembly of element equations
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 43 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Introduction
While deriving the element equations, we isolated a typical element
(say, the k
th
) from the mesh and developed its nite element model
To solve the full problem, we must sum the contributions from each
of the elements that make up the domain of the problem
Why sum?
Because
u =
N

i =1
U
i

i
(44)
U. Saravanan (IIT Madras) Finite element method April 2014 44 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Introduction
While deriving the element equations, we isolated a typical element
(say, the k
th
) from the mesh and developed its nite element model
To solve the full problem, we must sum the contributions from each
of the elements that make up the domain of the problem
Why sum?
Because
u =
N

i =1
U
i

i
(44)
U. Saravanan (IIT Madras) Finite element method April 2014 44 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Introduction
While deriving the element equations, we isolated a typical element
(say, the k
th
) from the mesh and developed its nite element model
To solve the full problem, we must sum the contributions from each
of the elements that make up the domain of the problem
Why sum?
Because
u =
N

i =1
U
i

i
(44)
U. Saravanan (IIT Madras) Finite element method April 2014 44 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Introduction
While deriving the element equations, we isolated a typical element
(say, the k
th
) from the mesh and developed its nite element model
To solve the full problem, we must sum the contributions from each
of the elements that make up the domain of the problem
Why sum?
Because
u =
N

i =1
U
i

i
(44)
U. Saravanan (IIT Madras) Finite element method April 2014 44 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Concepts used in assembling the equations
The assembly of the nite element equations is based on:
1. Continuity of the displacement eld - Related to the single valued
nature of the displacement
2. Balance of the surface force eld - Refers to the requirement that t
(n)
+ t
(n)
= o
Imposing the continuity of the displacement eld at the nodes
common to the elements amounts to the correspondence between the
local and global nodal values
U. Saravanan (IIT Madras) Finite element method April 2014 45 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Concepts used in assembling the equations
The assembly of the nite element equations is based on:
1. Continuity of the displacement eld - Related to the single valued
nature of the displacement
2. Balance of the surface force eld - Refers to the requirement that t
(n)
+ t
(n)
= o
Imposing the continuity of the displacement eld at the nodes
common to the elements amounts to the correspondence between the
local and global nodal values
U. Saravanan (IIT Madras) Finite element method April 2014 45 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Concepts used in assembling the equations
The assembly of the nite element equations is based on:
1. Continuity of the displacement eld - Related to the single valued
nature of the displacement
2. Balance of the surface force eld - Refers to the requirement that t
(n)
+ t
(n)
= o
Imposing the continuity of the displacement eld at the nodes
common to the elements amounts to the correspondence between the
local and global nodal values
U. Saravanan (IIT Madras) Finite element method April 2014 45 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Concepts used in assembling the equations
The assembly of the nite element equations is based on:
1. Continuity of the displacement eld - Related to the single valued
nature of the displacement
2. Balance of the surface force eld - Refers to the requirement that t
(n)
+ t
(n)
= o
Imposing the continuity of the displacement eld at the nodes
common to the elements amounts to the correspondence between the
local and global nodal values
U. Saravanan (IIT Madras) Finite element method April 2014 45 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Illustrative mesh
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3
I
1
3
II
2
4
4
5
1
3
(a) (b)

Global degrees of
freedom
{U
FEM
} =
_

_
U
1
V
1
W
1
.
.
.
U
5
V
5
W
5
_

_
(45)
U. Saravanan (IIT Madras) Finite element method April 2014 46 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Illustrative mesh
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3
I
1
3
II
2
4
4
5
1
3
(a) (b)

Global degrees of
freedom
{U
FEM
} =
_

_
U
1
V
1
W
1
.
.
.
U
5
V
5
W
5
_

_
(45)
U. Saravanan (IIT Madras) Finite element method April 2014 46 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Illustrative mesh
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3
I
1
3
II
2
4
4
5
1
3
(a) (b)

Global degrees of
freedom
{U
FEM
} =
_

_
U
1
V
1
W
1
.
.
.
U
5
V
5
W
5
_

_
(45)
U. Saravanan (IIT Madras) Finite element method April 2014 46 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Illustrative mesh
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3
I
1
3
II
2
4
4
5
1
3
(a) (b)

Global degrees of
freedom
{U
FEM
} =
_

_
U
1
V
1
W
1
.
.
.
U
5
V
5
W
5
_

_
(45)
U. Saravanan (IIT Madras) Finite element method April 2014 46 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Establishing the relationship between global and local
degrees of freedom
For the nite element mesh being studied the relation between the local
and global degrees of freedom is:
u
1
1
= U
1
, v
1
1
= V
1
, w
1
1
= W
1
, u
1
2
= U
2
, v
1
2
= V
2
, w
1
2
= W
2
u
1
3
= U
3
, v
1
3
= V
3
, w
1
3
= W
3
, u
1
4
= U
4
, v
1
4
= V
4
, w
1
4
= W
4
u
2
1
= U
1
, v
2
1
= V
1
, w
2
1
= W
1
, u
2
2
= U
4
, v
2
2
= V
4
, w
2
2
= W
4
u
2
3
= U
3
, v
2
3
= V
3
, w
2
3
= W
3
, u
2
4
= U
5
, v
2
4
= V
5
, w
2
4
= W
5
This transformation from the local degrees of freedom to the global is
accomplished by a boolean transformation matrix, T
FEM
, so that
{d} = (T
FEM
){U
FEM
}. (46)
U. Saravanan (IIT Madras) Finite element method April 2014 47 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Establishing the relationship between global and local
degrees of freedom
For the nite element mesh being studied the relation between the local
and global degrees of freedom is:
u
1
1
= U
1
, v
1
1
= V
1
, w
1
1
= W
1
, u
1
2
= U
2
, v
1
2
= V
2
, w
1
2
= W
2
u
1
3
= U
3
, v
1
3
= V
3
, w
1
3
= W
3
, u
1
4
= U
4
, v
1
4
= V
4
, w
1
4
= W
4
u
2
1
= U
1
, v
2
1
= V
1
, w
2
1
= W
1
, u
2
2
= U
4
, v
2
2
= V
4
, w
2
2
= W
4
u
2
3
= U
3
, v
2
3
= V
3
, w
2
3
= W
3
, u
2
4
= U
5
, v
2
4
= V
5
, w
2
4
= W
5
This transformation from the local degrees of freedom to the global is
accomplished by a boolean transformation matrix, T
FEM
, so that
{d} = (T
FEM
){U
FEM
}. (46)
U. Saravanan (IIT Madras) Finite element method April 2014 47 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Illustrative transformation matrix, (T
FEM
)
For the element 1
(T
1
FEM
) =
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 1 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 1 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 1 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 1 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 1 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 1 0 0 0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
, (47)
U. Saravanan (IIT Madras) Finite element method April 2014 48 / 81
Finite element formulation for linearized elasticity Assembly of element equations
An important identity
0 =
_
A
t
(D
FEM
)(B
FEM
)(T
e
FEM
){U
FEM
} (B
FEM
)(T
e
FEM
){U
i
FEM
}dv
+
_
A
t
(N
FEM
)(T
e
FEM
){

U
FEM
} (N
FEM
)(T
e
FEM
){U
i
FEM
}dv

_
A
t
b (N
FEM
)(T
e
FEM
){U
i
FEM
}dv

_
A
t
t
(n)
(N
FEM
)(T
e
FEM
){U
i
FEM
}ds
where {U
i
FEM
} is such that the equation
d
i
= (T
e
FEM
){U
i
FEM
}
holds.
U. Saravanan (IIT Madras) Finite element method April 2014 49 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Global matrices and force vector
The global stiness and mass matrix are obtained from
(K) =
k

e=1
(T
e
FEM
)
t
(K
e
)(T
e
FEM
), (M) =
k

e=1
(T
e
FEM
)
t
(M
e
)(T
e
FEM
),
(48)
The global force vector is obtained from
{F
FEM
} =
k

e=1
(T
e
FEM
)
t
[{f
e
b
} +{f
e
s
}], (49)
U. Saravanan (IIT Madras) Finite element method April 2014 50 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Global matrices and force vector
The global stiness and mass matrix are obtained from
(K) =
k

e=1
(T
e
FEM
)
t
(K
e
)(T
e
FEM
), (M) =
k

e=1
(T
e
FEM
)
t
(M
e
)(T
e
FEM
),
(48)
The global force vector is obtained from
{F
FEM
} =
k

e=1
(T
e
FEM
)
t
[{f
e
b
} +{f
e
s
}], (49)
U. Saravanan (IIT Madras) Finite element method April 2014 50 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand
1. (T
e
FEM
) could be obtained from the connectivity matrix
2. Even though the above formulation provides an algorithm for adding
the dierent element matrices and vectors, it is computationally
expensive
3. In practice, the global matrices are directly updated without
assembling the element matrices.
4. The continuity of the displacement at the inter element nodes
guarantees the continuity of the displacement along the entire inter
element boundary
U. Saravanan (IIT Madras) Finite element method April 2014 51 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand
1. (T
e
FEM
) could be obtained from the connectivity matrix
2. Even though the above formulation provides an algorithm for adding
the dierent element matrices and vectors, it is computationally
expensive
3. In practice, the global matrices are directly updated without
assembling the element matrices.
4. The continuity of the displacement at the inter element nodes
guarantees the continuity of the displacement along the entire inter
element boundary
U. Saravanan (IIT Madras) Finite element method April 2014 51 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand
1. (T
e
FEM
) could be obtained from the connectivity matrix
2. Even though the above formulation provides an algorithm for adding
the dierent element matrices and vectors, it is computationally
expensive
3. In practice, the global matrices are directly updated without
assembling the element matrices.
4. The continuity of the displacement at the inter element nodes
guarantees the continuity of the displacement along the entire inter
element boundary
U. Saravanan (IIT Madras) Finite element method April 2014 51 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand
1. (T
e
FEM
) could be obtained from the connectivity matrix
2. Even though the above formulation provides an algorithm for adding
the dierent element matrices and vectors, it is computationally
expensive
3. In practice, the global matrices are directly updated without
assembling the element matrices.
4. The continuity of the displacement at the inter element nodes
guarantees the continuity of the displacement along the entire inter
element boundary
U. Saravanan (IIT Madras) Finite element method April 2014 51 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Balance of surface forces
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

We require
(t
1
(n)
)
(1,3,4)
= (t
2
(n)
)
(1,3,2)
, (50)
where the subscript (1, 3, 4) (or (1, 3, 2))
denotes the local node numbers that
form the vertices of the surface being
considered
In the nite element method, we impose the above relation in a weighted-integral
sense:
_
A
1
134
(t
1
(n)
)
1
i
ds +
_
A
2
132
(t
2
(n)
)
2
j
ds = 0, (51)
for (i , j ) {(1, 1), (2, 2), (3, 3), (7, 7), (8, 8), (9, 9), (10, 4), (11, 5), (12, 6)}, where
the choice of i and j correspond to the basis functions associated with each
common node in elements 1 and 2
U. Saravanan (IIT Madras) Finite element method April 2014 52 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Balance of surface forces
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

We require
(t
1
(n)
)
(1,3,4)
= (t
2
(n)
)
(1,3,2)
, (50)
where the subscript (1, 3, 4) (or (1, 3, 2))
denotes the local node numbers that
form the vertices of the surface being
considered
In the nite element method, we impose the above relation in a weighted-integral
sense:
_
A
1
134
(t
1
(n)
)
1
i
ds +
_
A
2
132
(t
2
(n)
)
2
j
ds = 0, (51)
for (i , j ) {(1, 1), (2, 2), (3, 3), (7, 7), (8, 8), (9, 9), (10, 4), (11, 5), (12, 6)}, where
the choice of i and j correspond to the basis functions associated with each
common node in elements 1 and 2
U. Saravanan (IIT Madras) Finite element method April 2014 52 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Balance of surface forces
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

We require
(t
1
(n)
)
(1,3,4)
= (t
2
(n)
)
(1,3,2)
, (50)
where the subscript (1, 3, 4) (or (1, 3, 2))
denotes the local node numbers that
form the vertices of the surface being
considered
In the nite element method, we impose the above relation in a weighted-integral
sense:
_
A
1
134
(t
1
(n)
)
1
i
ds +
_
A
2
132
(t
2
(n)
)
2
j
ds = 0, (51)
for (i , j ) {(1, 1), (2, 2), (3, 3), (7, 7), (8, 8), (9, 9), (10, 4), (11, 5), (12, 6)}, where
the choice of i and j correspond to the basis functions associated with each
common node in elements 1 and 2
U. Saravanan (IIT Madras) Finite element method April 2014 52 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Balance of surface forces
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

We require
(t
1
(n)
)
(1,3,4)
= (t
2
(n)
)
(1,3,2)
, (50)
where the subscript (1, 3, 4) (or (1, 3, 2))
denotes the local node numbers that
form the vertices of the surface being
considered
In the nite element method, we impose the above relation in a weighted-integral
sense:
_
A
1
134
(t
1
(n)
)
1
i
ds +
_
A
2
132
(t
2
(n)
)
2
j
ds = 0, (51)
for (i , j ) {(1, 1), (2, 2), (3, 3), (7, 7), (8, 8), (9, 9), (10, 4), (11, 5), (12, 6)}, where
the choice of i and j correspond to the basis functions associated with each
common node in elements 1 and 2
U. Saravanan (IIT Madras) Finite element method April 2014 52 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Balance of surface forces
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

We require
(t
1
(n)
)
(1,3,4)
= (t
2
(n)
)
(1,3,2)
, (50)
where the subscript (1, 3, 4) (or (1, 3, 2))
denotes the local node numbers that
form the vertices of the surface being
considered
In the nite element method, we impose the above relation in a weighted-integral
sense:
_
A
1
134
(t
1
(n)
)
1
i
ds +
_
A
2
132
(t
2
(n)
)
2
j
ds = 0, (51)
for (i , j ) {(1, 1), (2, 2), (3, 3), (7, 7), (8, 8), (9, 9), (10, 4), (11, 5), (12, 6)}, where
the choice of i and j correspond to the basis functions associated with each
common node in elements 1 and 2
U. Saravanan (IIT Madras) Finite element method April 2014 52 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Why balance of surface forces?
Required from the physics of the problem - But what then is the
equilibrium equation?
Required to reduce the number of unknowns in surface force vector
Required to enforce the traction boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 53 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Why balance of surface forces?
Required from the physics of the problem - But what then is the
equilibrium equation?
Required to reduce the number of unknowns in surface force vector
Required to enforce the traction boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 53 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Why balance of surface forces?
Required from the physics of the problem - But what then is the
equilibrium equation?
Required to reduce the number of unknowns in surface force vector
Required to enforce the traction boundary condition
U. Saravanan (IIT Madras) Finite element method April 2014 53 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Alternate notation
Let the surface with vertices 1 2 3 be called surface 1, vertices
1 2 4 surface 2, vertices 1 3 4 surface 3 and vertices 2 3 4
surface 4 and then let
F
e
iJ
=
_
surfaceJ
(t
e
(n)
)
e
i
ds (52)
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Now previous equation can be
written succinctly as
F
1
i 3
+ F
2
j 1
= 0 (53)
for (i , j )
{(1, 1), (2, 2), (3, 3), (7, 7), (8, 8),
(9, 9), (10, 4), (11, 5), (12, 6)}
U. Saravanan (IIT Madras) Finite element method April 2014 54 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Alternate notation
Let the surface with vertices 1 2 3 be called surface 1, vertices
1 2 4 surface 2, vertices 1 3 4 surface 3 and vertices 2 3 4
surface 4 and then let
F
e
iJ
=
_
surfaceJ
(t
e
(n)
)
e
i
ds (52)
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Now previous equation can be
written succinctly as
F
1
i 3
+ F
2
j 1
= 0 (53)
for (i , j )
{(1, 1), (2, 2), (3, 3), (7, 7), (8, 8),
(9, 9), (10, 4), (11, 5), (12, 6)}
U. Saravanan (IIT Madras) Finite element method April 2014 54 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Alternate notation
Let the surface with vertices 1 2 3 be called surface 1, vertices
1 2 4 surface 2, vertices 1 3 4 surface 3 and vertices 2 3 4
surface 4 and then let
F
e
iJ
=
_
surfaceJ
(t
e
(n)
)
e
i
ds (52)
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Now previous equation can be
written succinctly as
F
1
i 3
+ F
2
j 1
= 0 (53)
for (i , j )
{(1, 1), (2, 2), (3, 3), (7, 7), (8, 8),
(9, 9), (10, 4), (11, 5), (12, 6)}
U. Saravanan (IIT Madras) Finite element method April 2014 54 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Alternate notation
Let the surface with vertices 1 2 3 be called surface 1, vertices
1 2 4 surface 2, vertices 1 3 4 surface 3 and vertices 2 3 4
surface 4 and then let
F
e
iJ
=
_
surfaceJ
(t
e
(n)
)
e
i
ds (52)
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Now previous equation can be
written succinctly as
F
1
i 3
+ F
2
j 1
= 0 (53)
for (i , j )
{(1, 1), (2, 2), (3, 3), (7, 7), (8, 8),
(9, 9), (10, 4), (11, 5), (12, 6)}
U. Saravanan (IIT Madras) Finite element method April 2014 54 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Alternate notation
Let the surface with vertices 1 2 3 be called surface 1, vertices
1 2 4 surface 2, vertices 1 3 4 surface 3 and vertices 2 3 4
surface 4 and then let
F
e
iJ
=
_
surfaceJ
(t
e
(n)
)
e
i
ds (52)
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Now previous equation can be
written succinctly as
F
1
i 3
+ F
2
j 1
= 0 (53)
for (i , j )
{(1, 1), (2, 2), (3, 3), (7, 7), (8, 8),
(9, 9), (10, 4), (11, 5), (12, 6)}
U. Saravanan (IIT Madras) Finite element method April 2014 54 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand
For the tetrahedron element being studied
{f
1
s
}
1
=
_
surface1
(t
1
(n
1
)
)
x
N
1
ds +
_
surface2
(t
1
(n
2
)
)
x
N
1
ds
+
_
surface3
(t
1
(n
3
)
)
x
N
1
ds +
_
surface4
(t
1
(n
4
)
)
x
N
1
ds, (54)
where n
i
denotes the outward normal to the i
th
surface
Hence
{f
e
s
}
i
= F
e
i 1
+ F
e
i 2
+ F
e
i 3
+ F
e
i 4
(55)
where {f
e
s
}
i
denotes the i
th
component in the surface traction vector
U. Saravanan (IIT Madras) Finite element method April 2014 55 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand
For the tetrahedron element being studied
{f
1
s
}
1
=
_
surface1
(t
1
(n
1
)
)
x
N
1
ds +
_
surface2
(t
1
(n
2
)
)
x
N
1
ds
+
_
surface3
(t
1
(n
3
)
)
x
N
1
ds +
_
surface4
(t
1
(n
4
)
)
x
N
1
ds, (54)
where n
i
denotes the outward normal to the i
th
surface
Hence
{f
e
s
}
i
= F
e
i 1
+ F
e
i 2
+ F
e
i 3
+ F
e
i 4
(55)
where {f
e
s
}
i
denotes the i
th
component in the surface traction vector
U. Saravanan (IIT Madras) Finite element method April 2014 55 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand

e
x
e
y
e
z
4
1
2
3
{f
1
s
}
1
is not the net force acting
on the surface of the
tetrahedron along the e
x
direction
The 12 components of {f
1
s
}
i
correspond to the weighted
average of 3 components of the
traction vector with the 4 shape
functions over the surface of the
tetrahedron
U. Saravanan (IIT Madras) Finite element method April 2014 56 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand

e
x
e
y
e
z
4
1
2
3
{f
1
s
}
1
is not the net force acting
on the surface of the
tetrahedron along the e
x
direction
The 12 components of {f
1
s
}
i
correspond to the weighted
average of 3 components of the
traction vector with the 4 shape
functions over the surface of the
tetrahedron
U. Saravanan (IIT Madras) Finite element method April 2014 56 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand

e
x
e
y
e
z
4
1
2
3
{f
1
s
}
1
is not the net force acting
on the surface of the
tetrahedron along the e
x
direction
The 12 components of {f
1
s
}
i
correspond to the weighted
average of 3 components of the
traction vector with the 4 shape
functions over the surface of the
tetrahedron
U. Saravanan (IIT Madras) Finite element method April 2014 56 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand

e
x
e
y
e
z
4
1
2
3
{f
1
s
}
1
is not the net force acting
on the surface of the
tetrahedron along the e
x
direction
The 12 components of {f
1
s
}
i
correspond to the weighted
average of 3 components of the
traction vector with the 4 shape
functions over the surface of the
tetrahedron
U. Saravanan (IIT Madras) Finite element method April 2014 56 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Points to understand

e
x
e
y
e
z
4
1
2
3
{f
1
s
}
1
is not the net force acting
on the surface of the
tetrahedron along the e
x
direction
The 12 components of {f
1
s
}
i
correspond to the weighted
average of 3 components of the
traction vector with the 4 shape
functions over the surface of the
tetrahedron
U. Saravanan (IIT Madras) Finite element method April 2014 56 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Illustrative mesh
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3
I
1
3
II
2
4
4
5
1
3
(a) (b)

U. Saravanan (IIT Madras) Finite element method April 2014 57 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Surface traction vector for the mesh being studied
{F
s
} =
_

_
(f
1
s
)
1
+ (f
2
s
)
1
(f
1
s
)
2
+ (f
2
s
)
2
(f
1
s
)
3
+ (f
2
s
)
3
(f
1
s
)
4
(f
1
s
)
5
(f
1
s
)
6
(f
1
s
)
7
+ (f
2
s
)
7
(f
1
s
)
8
+ (f
2
s
)
8
(f
1
s
)
9
+ (f
2
s
)
9
(f
1
s
)
10
+ (f
2
s
)
4
(f
1
s
)
11
+ (f
2
s
)
5
(f
1
s
)
12
+ (f
2
s
)
6
(f
2
s
)
10
(f
2
s
)
11
(f
2
s
)
12
_

_
=
_

_
F
1
11
+ F
1
12
+ F
1
13
+ F
2
11
+ F
1
14
+ F
2
12
+ F
2
13
+ F
2
14
F
1
21
+ F
1
22
+ F
1
23
+ F
2
21
+ F
1
24
+ F
2
22
+ F
2
23
+ F
2
24
F
1
31
+ F
1
32
+ F
1
33
+ F
2
31
+ F
1
34
+ F
2
32
+ F
2
33
+ F
2
34
F
1
41
+ F
1
42
+ F
1
43
+ F
1
44
F
1
51
+ F
1
52
+ F
1
53
+ F
1
54
F
1
61
+ F
1
62
+ F
1
63
+ F
1
64
F
1
71
+ F
1
72
+ F
1
73
+ F
2
71
+ F
1
74
+ F
2
72
+ F
2
73
+ F
2
74
F
1
81
+ F
1
82
+ F
1
83
+ F
2
81
+ F
1
84
+ F
2
82
+ F
2
83
+ F
2
84
F
1
91
+ F
1
92
+ F
1
93
+ F
2
91
+ F
1
94
+ F
2
92
+ F
2
93
+ F
2
94
F
1
101
+ F
1
102
+ F
1
103
+ F
2
41
+ F
1
104
+ F
2
42
+ F
2
43
+ F
2
44
F
1
111
+ F
1
112
+ F
1
113
+ F
2
51
+ F
1
114
+ F
2
52
+ F
2
53
+ F
2
54
F
1
121
+ F
1
122
+ F
1
123
+ F
2
61
+ F
1
124
+ F
2
62
+ F
2
63
+ F
2
64
F
2
101
+ F
2
102
+ F
2
103
+ F
2
104
F
2
111
+ F
2
112
+ F
2
113
+ F
2
114
F
2
121
+ F
2
122
+ F
2
123
+ F
2
124
_

_
U. Saravanan (IIT Madras) Finite element method April 2014 58 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Surface traction vector for the mesh being studied
It then follows from the balance of surface traction that only the
underlined terms are zero
The remaining terms of each equation will either be known because
t
(n)
is known on the boundary or will remain unknown because the
displacement is specied on the boundary
U. Saravanan (IIT Madras) Finite element method April 2014 59 / 81
Finite element formulation for linearized elasticity Assembly of element equations
Surface traction vector for the mesh being studied
It then follows from the balance of surface traction that only the
underlined terms are zero
The remaining terms of each equation will either be known because
t
(n)
is known on the boundary or will remain unknown because the
displacement is specied on the boundary
U. Saravanan (IIT Madras) Finite element method April 2014 59 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 60 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Illustrative boundary conditions
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Let us assume that a uniform
pressure, p
x
acts along the e
x
direction on the surface with
vertices 1 3 5
The surface with vertices
1 2 3 is xed in all three
directions for all times
All the remaining surfaces are
free of traction
U. Saravanan (IIT Madras) Finite element method April 2014 61 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Illustrative boundary conditions
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Let us assume that a uniform
pressure, p
x
acts along the e
x
direction on the surface with
vertices 1 3 5
The surface with vertices
1 2 3 is xed in all three
directions for all times
All the remaining surfaces are
free of traction
U. Saravanan (IIT Madras) Finite element method April 2014 61 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Illustrative boundary conditions
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Let us assume that a uniform
pressure, p
x
acts along the e
x
direction on the surface with
vertices 1 3 5
The surface with vertices
1 2 3 is xed in all three
directions for all times
All the remaining surfaces are
free of traction
U. Saravanan (IIT Madras) Finite element method April 2014 61 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Illustrative boundary conditions
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Let us assume that a uniform
pressure, p
x
acts along the e
x
direction on the surface with
vertices 1 3 5
The surface with vertices
1 2 3 is xed in all three
directions for all times
All the remaining surfaces are
free of traction
U. Saravanan (IIT Madras) Finite element method April 2014 61 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Illustrative boundary conditions
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Let us assume that a uniform
pressure, p
x
acts along the e
x
direction on the surface with
vertices 1 3 5
The surface with vertices
1 2 3 is xed in all three
directions for all times
All the remaining surfaces are
free of traction
U. Saravanan (IIT Madras) Finite element method April 2014 61 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Illustrative boundary conditions
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

Let us assume that a uniform
pressure, p
x
acts along the e
x
direction on the surface with
vertices 1 3 5
The surface with vertices
1 2 3 is xed in all three
directions for all times
All the remaining surfaces are
free of traction
U. Saravanan (IIT Madras) Finite element method April 2014 61 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition
Surface traction free condition implies
F
1
i 2
= F
1
i 4
= F
2
i 2
= F
2
i 4
= 0, i {1, 2, . . . , 12} (56)
Traction on surface 1 3 5 yields
F
2
i 3
=
_
surface3
p
x
e
x

2
i
ds (57)
Traction on surface 1 3 5 acts only along e
x
, hence
F
2
i 3
= 0, i {2, 3, 4, 5, 6, 8, 9, 11, 12} (58)
F
2
i 3
= p
x
_
surface3
N
i
ds = p
x
S
135
3
, i {1, 7, 10} (59)
where S
135
denotes the surface area of the triangular surface with vertices
1 3 5
U. Saravanan (IIT Madras) Finite element method April 2014 62 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition
Surface traction free condition implies
F
1
i 2
= F
1
i 4
= F
2
i 2
= F
2
i 4
= 0, i {1, 2, . . . , 12} (56)
Traction on surface 1 3 5 yields
F
2
i 3
=
_
surface3
p
x
e
x

2
i
ds (57)
Traction on surface 1 3 5 acts only along e
x
, hence
F
2
i 3
= 0, i {2, 3, 4, 5, 6, 8, 9, 11, 12} (58)
F
2
i 3
= p
x
_
surface3
N
i
ds = p
x
S
135
3
, i {1, 7, 10} (59)
where S
135
denotes the surface area of the triangular surface with vertices
1 3 5
U. Saravanan (IIT Madras) Finite element method April 2014 62 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition
Surface traction free condition implies
F
1
i 2
= F
1
i 4
= F
2
i 2
= F
2
i 4
= 0, i {1, 2, . . . , 12} (56)
Traction on surface 1 3 5 yields
F
2
i 3
=
_
surface3
p
x
e
x

2
i
ds (57)
Traction on surface 1 3 5 acts only along e
x
, hence
F
2
i 3
= 0, i {2, 3, 4, 5, 6, 8, 9, 11, 12} (58)
F
2
i 3
= p
x
_
surface3
N
i
ds = p
x
S
135
3
, i {1, 7, 10} (59)
where S
135
denotes the surface area of the triangular surface with vertices
1 3 5
U. Saravanan (IIT Madras) Finite element method April 2014 62 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition

e
x
e
y
e
z
4
1
2
3
For the assumed form for N
i
, the value of
N
i
on the plane opposite to the i
th
node is
zero
Hence, F
2
43
= 0 because N
2
is zero on the
surface with vertices 1 3 4.
For similar reasons
F
1
i 1
= F
2
i 1
= 0, i {10, 11, 12}
F
1
i 3
= 0, i {4, 5, 6}
U. Saravanan (IIT Madras) Finite element method April 2014 63 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition

e
x
e
y
e
z
4
1
2
3
For the assumed form for N
i
, the value of
N
i
on the plane opposite to the i
th
node is
zero
Hence, F
2
43
= 0 because N
2
is zero on the
surface with vertices 1 3 4.
For similar reasons
F
1
i 1
= F
2
i 1
= 0, i {10, 11, 12}
F
1
i 3
= 0, i {4, 5, 6}
U. Saravanan (IIT Madras) Finite element method April 2014 63 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition

e
x
e
y
e
z
4
1
2
3
For the assumed form for N
i
, the value of
N
i
on the plane opposite to the i
th
node is
zero
Hence, F
2
43
= 0 because N
2
is zero on the
surface with vertices 1 3 4.
For similar reasons
F
1
i 1
= F
2
i 1
= 0, i {10, 11, 12}
F
1
i 3
= 0, i {4, 5, 6}
U. Saravanan (IIT Madras) Finite element method April 2014 63 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition

e
x
e
y
e
z
4
1
2
3
For the assumed form for N
i
, the value of
N
i
on the plane opposite to the i
th
node is
zero
Hence, F
2
43
= 0 because N
2
is zero on the
surface with vertices 1 3 4.
For similar reasons
F
1
i 1
= F
2
i 1
= 0, i {10, 11, 12}
F
1
i 3
= 0, i {4, 5, 6}
U. Saravanan (IIT Madras) Finite element method April 2014 63 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition

e
x
e
y
e
z
4
1
2
3
For the assumed form for N
i
, the value of
N
i
on the plane opposite to the i
th
node is
zero
Hence, F
2
43
= 0 because N
2
is zero on the
surface with vertices 1 3 4.
For similar reasons
F
1
i 1
= F
2
i 1
= 0, i {10, 11, 12}
F
1
i 3
= 0, i {4, 5, 6}
U. Saravanan (IIT Madras) Finite element method April 2014 63 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Inference from the traction boundary condition

e
x
e
y
e
z
4
1
2
3
For the assumed form for N
i
, the value of
N
i
on the plane opposite to the i
th
node is
zero
Hence, F
2
43
= 0 because N
2
is zero on the
surface with vertices 1 3 4.
For similar reasons
F
1
i 1
= F
2
i 1
= 0, i {10, 11, 12}
F
1
i 3
= 0, i {4, 5, 6}
U. Saravanan (IIT Madras) Finite element method April 2014 63 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Surface traction vector for the mesh being studied
{F
s
} =
_

_
(f
1
s
)
1
+ (f
2
s
)
1
(f
1
s
)
2
+ (f
2
s
)
2
(f
1
s
)
3
+ (f
2
s
)
3
(f
1
s
)
4
(f
1
s
)
5
(f
1
s
)
6
(f
1
s
)
7
+ (f
2
s
)
7
(f
1
s
)
8
+ (f
2
s
)
8
(f
1
s
)
9
+ (f
2
s
)
9
(f
1
s
)
10
+ (f
2
s
)
4
(f
1
s
)
11
+ (f
2
s
)
5
(f
1
s
)
12
+ (f
2
s
)
6
(f
2
s
)
10
(f
2
s
)
11
(f
2
s
)
12
_

_
=
_

_
F
1
11
+ F
1
12
+ F
1
13
+ F
2
11
+ F
1
14
+ F
2
12
+ F
2
13
+ F
2
14
F
1
21
+ F
1
22
+ F
1
23
+ F
2
21
+ F
1
24
+ F
2
22
+ F
2
23
+ F
2
24
F
1
31
+ F
1
32
+ F
1
33
+ F
2
31
+ F
1
34
+ F
2
32
+ F
2
33
+ F
2
34
F
1
41
+ F
1
42
+ F
1
43
+ F
1
44
F
1
51
+ F
1
52
+ F
1
53
+ F
1
54
F
1
61
+ F
1
62
+ F
1
63
+ F
1
64
F
1
71
+ F
1
72
+ F
1
73
+ F
2
71
+ F
1
74
+ F
2
72
+ F
2
73
+ F
2
74
F
1
81
+ F
1
82
+ F
1
83
+ F
2
81
+ F
1
84
+ F
2
82
+ F
2
83
+ F
2
84
F
1
91
+ F
1
92
+ F
1
93
+ F
2
91
+ F
1
94
+ F
2
92
+ F
2
93
+ F
2
94
F
1
101
+ F
1
102
+ F
1
103
+ F
2
41
+ F
1
104
+ F
2
42
+ F
2
43
+ F
2
44
F
1
111
+ F
1
112
+ F
1
113
+ F
2
51
+ F
1
114
+ F
2
52
+ F
2
53
+ F
2
54
F
1
121
+ F
1
122
+ F
1
123
+ F
2
61
+ F
1
124
+ F
2
62
+ F
2
63
+ F
2
64
F
2
101
+ F
2
102
+ F
2
103
+ F
2
104
F
2
111
+ F
2
112
+ F
2
113
+ F
2
114
F
2
121
+ F
2
122
+ F
2
123
+ F
2
124
_

_
U. Saravanan (IIT Madras) Finite element method April 2014 64 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Surface traction vector for the mesh being studied after
application of boundary condition
{F
s
} =
_

_
F
1
11
+ p
x
S
135
3
F
1
21
F
1
31
F
1
41
F
1
51
F
1
61
F
1
71
+ p
x
S
135
3
F
1
81
F
1
91
0
0
0
p
x
S
135
3
0
0
_

_
There are 9 unknowns in the
surface traction vector
U. Saravanan (IIT Madras) Finite element method April 2014 65 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Surface traction vector for the mesh being studied after
application of boundary condition
{F
s
} =
_

_
F
1
11
+ p
x
S
135
3
F
1
21
F
1
31
F
1
41
F
1
51
F
1
61
F
1
71
+ p
x
S
135
3
F
1
81
F
1
91
0
0
0
p
x
S
135
3
0
0
_

_
There are 9 unknowns in the
surface traction vector
U. Saravanan (IIT Madras) Finite element method April 2014 65 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Surface traction vector for the mesh being studied after
application of boundary condition
{F
s
} =
_

_
F
1
11
+ p
x
S
135
3
F
1
21
F
1
31
F
1
41
F
1
51
F
1
61
F
1
71
+ p
x
S
135
3
F
1
81
F
1
91
0
0
0
p
x
S
135
3
0
0
_

_
There are 9 unknowns in the
surface traction vector
U. Saravanan (IIT Madras) Finite element method April 2014 65 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Surface traction vector for the mesh being studied after
application of boundary condition
{F
s
} =
_

_
F
1
11
+ p
x
S
135
3
F
1
21
F
1
31
F
1
41
F
1
51
F
1
61
F
1
71
+ p
x
S
135
3
F
1
81
F
1
91
0
0
0
p
x
S
135
3
0
0
_

_
There are 9 unknowns in the
surface traction vector
U. Saravanan (IIT Madras) Finite element method April 2014 65 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Displacement boundary condition
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

The surface with vertices
1 2 3 is xed in all three
directions for all times
The non-zero components of
{

U
FEM
} is also same as that of
{U
FEM
}
There are 6 unknown
displacement dof
{U
FEM
} =
_

_
U
1
V
1
W
1
U
2
V
2
W
2
U
3
V
3
W
3
U
4
V
4
W
4
U
5
V
5
W
5
_

_
=
_

_
0
0
0
0
0
0
0
0
0
U
4
V
4
W
4
U
5
V
5
W
5
_

_
U. Saravanan (IIT Madras) Finite element method April 2014 66 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Displacement boundary condition
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

The surface with vertices
1 2 3 is xed in all three
directions for all times
The non-zero components of
{

U
FEM
} is also same as that of
{U
FEM
}
There are 6 unknown
displacement dof
{U
FEM
} =
_

_
U
1
V
1
W
1
U
2
V
2
W
2
U
3
V
3
W
3
U
4
V
4
W
4
U
5
V
5
W
5
_

_
=
_

_
0
0
0
0
0
0
0
0
0
U
4
V
4
W
4
U
5
V
5
W
5
_

_
U. Saravanan (IIT Madras) Finite element method April 2014 66 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Displacement boundary condition
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

The surface with vertices
1 2 3 is xed in all three
directions for all times
The non-zero components of
{

U
FEM
} is also same as that of
{U
FEM
}
There are 6 unknown
displacement dof
{U
FEM
} =
_

_
U
1
V
1
W
1
U
2
V
2
W
2
U
3
V
3
W
3
U
4
V
4
W
4
U
5
V
5
W
5
_

_
=
_

_
0
0
0
0
0
0
0
0
0
U
4
V
4
W
4
U
5
V
5
W
5
_

_
U. Saravanan (IIT Madras) Finite element method April 2014 66 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Displacement boundary condition
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

The surface with vertices
1 2 3 is xed in all three
directions for all times
The non-zero components of
{

U
FEM
} is also same as that of
{U
FEM
}
There are 6 unknown
displacement dof
{U
FEM
} =
_

_
U
1
V
1
W
1
U
2
V
2
W
2
U
3
V
3
W
3
U
4
V
4
W
4
U
5
V
5
W
5
_

_
=
_

_
0
0
0
0
0
0
0
0
0
U
4
V
4
W
4
U
5
V
5
W
5
_

_
U. Saravanan (IIT Madras) Finite element method April 2014 66 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Displacement boundary condition
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

The surface with vertices
1 2 3 is xed in all three
directions for all times
The non-zero components of
{

U
FEM
} is also same as that of
{U
FEM
}
There are 6 unknown
displacement dof
{U
FEM
} =
_

_
U
1
V
1
W
1
U
2
V
2
W
2
U
3
V
3
W
3
U
4
V
4
W
4
U
5
V
5
W
5
_

_
=
_

_
0
0
0
0
0
0
0
0
0
U
4
V
4
W
4
U
5
V
5
W
5
_

_
U. Saravanan (IIT Madras) Finite element method April 2014 66 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Displacement boundary condition
e
x
e
y
e
z
4
1
2
3
5
I
II
2
4
1
3 I
1
3
II
2
4
4
5
1
3
(a) (b)

The surface with vertices
1 2 3 is xed in all three
directions for all times
The non-zero components of
{

U
FEM
} is also same as that of
{U
FEM
}
There are 6 unknown
displacement dof
{U
FEM
} =
_

_
U
1
V
1
W
1
U
2
V
2
W
2
U
3
V
3
W
3
U
4
V
4
W
4
U
5
V
5
W
5
_

_
=
_

_
0
0
0
0
0
0
0
0
0
U
4
V
4
W
4
U
5
V
5
W
5
_

_
U. Saravanan (IIT Madras) Finite element method April 2014 66 / 81
Finite element formulation for linearized elasticity Imposition of the boundary condition
Finite element governing equation
The 15 unknowns - 9 in surface traction vector and 6 in displacement
vector has to be determined from 15 linear equations
(K){U
FEM
} + (M){

U
FEM
} = {F
s
} +{F
b
} = {F
FEM
} (60)
U. Saravanan (IIT Madras) Finite element method April 2014 67 / 81
Finite element formulation for linearized elasticity Solution procedure
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 68 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Reformulation of the nite element governing equation
_
(K
11
) (K
12
)
(K
21
) (K
22
)
__
{U
1
}
{U
2
}
_
=
_
{F
1
}
{F
2
}
_
, (61)
Where
{U
1
} is the column of known
displacement variables
{U
2
} is the column of unknown
displacement variables
{F
1
} is the column of unknown
surface traction vector
{F
2
} is the column of known
surface traction vector
(K
ij
) are matrices obtained from partitioning the global stiness
matrix, depending on the number of known and unknown
displacement variables
U. Saravanan (IIT Madras) Finite element method April 2014 69 / 81
Finite element formulation for linearized elasticity Solution procedure
Determination of unknown displacement and traction
vector
Writing the above equation as two matrix equations
(K
11
){U
1
} + (K
12
){U
2
} = {F
1
} (62)
(K
21
){U
1
} + (K
22
){U
2
} = {F
2
} (63)
Solving for {U
2
} we obtain
{U
2
} = (K
22
)
1
[{F
2
} (K
21
){U
1
}] (64)
Then {F
1
} is computed from
{F
1
} = (K
11
){U
1
} + (K
12
)(K
22
)
1
[{F
2
} (K
21
){U
1
}] (65)
U. Saravanan (IIT Madras) Finite element method April 2014 70 / 81
Finite element formulation for linearized elasticity Solution procedure
Determination of unknown displacement and traction
vector
Writing the above equation as two matrix equations
(K
11
){U
1
} + (K
12
){U
2
} = {F
1
} (62)
(K
21
){U
1
} + (K
22
){U
2
} = {F
2
} (63)
Solving for {U
2
} we obtain
{U
2
} = (K
22
)
1
[{F
2
} (K
21
){U
1
}] (64)
Then {F
1
} is computed from
{F
1
} = (K
11
){U
1
} + (K
12
)(K
22
)
1
[{F
2
} (K
21
){U
1
}] (65)
U. Saravanan (IIT Madras) Finite element method April 2014 70 / 81
Finite element formulation for linearized elasticity Solution procedure
Determination of unknown displacement and traction
vector
Writing the above equation as two matrix equations
(K
11
){U
1
} + (K
12
){U
2
} = {F
1
} (62)
(K
21
){U
1
} + (K
22
){U
2
} = {F
2
} (63)
Solving for {U
2
} we obtain
{U
2
} = (K
22
)
1
[{F
2
} (K
21
){U
1
}] (64)
Then {F
1
} is computed from
{F
1
} = (K
11
){U
1
} + (K
12
)(K
22
)
1
[{F
2
} (K
21
){U
1
}] (65)
U. Saravanan (IIT Madras) Finite element method April 2014 70 / 81
Finite element formulation for linearized elasticity Post processing of the results
1
Finite element formulation
Six steps in nite element analysis
2
Finite element formulation for linearized elasticity
Decomposition of the domain - Selection of element
Derivation of elemental equations
Assembly of element equations
Imposition of the boundary condition
Solution procedure
Post processing of the results
Remarks
U. Saravanan (IIT Madras) Finite element method April 2014 71 / 81
Finite element formulation for linearized elasticity Post processing of the results
Computing the displacement eld
u =
_

_
U
1
=

n
i =1
d
1
i

1
i
U
2
=

n
i =1
d
2
i

2
i
.
.
.
U
e
=

n
i =1
d
e
i

e
i
, (66)
where e is the number of elements in the mesh and n is the number of
degrees of freedom. Depending on the value of x, the corresponding
element in equation (66) is used
U. Saravanan (IIT Madras) Finite element method April 2014 72 / 81
Finite element formulation for linearized elasticity Post processing of the results
Computing the displacement gradient
grad(u) =
_

_
grad(U
1
) =

n
i =1
d
1
i
grad(
1
i
)
grad(U
2
) =

n
i =1
d
2
i
grad(
2
i
)
.
.
.
grad(U
e
) =

n
i =1
d
e
i
grad(
e
i
)
(67)
Note that the derivative calculated from dierent elements meeting at a
node is always discontinuous in all approximations in which only the
function values are interpolated, unless the approximate solution coincides
with the actual solution
U. Saravanan (IIT Madras) Finite element method April 2014 73 / 81
Finite element formulation for linearized elasticity Post processing of the results
Computing the surface traction
Two ways to compute surface traction
1. Determined from elemental equation:
{f
s
}
equil
= (K
e
){d} + (M
e
){

d} {F
b
} (68)
2. Determined from the denition:
{f
e
s
}
def
=
_
A
t
t
(n)

i
ds (69)
where t
(n)
is computed from its denition, t
(n)
= n, where the
Cauchy stress is determined from Hookes law and strain deduced
from the computed gradient of the displacement
U. Saravanan (IIT Madras) Finite element method April 2014 74 / 81
Finite element formulation for linearized elasticity Post processing of the results
Computing the surface traction
Two ways to compute surface traction
1. Determined from elemental equation:
{f
s
}
equil
= (K
e
){d} + (M
e
){

d} {F
b
} (68)
2. Determined from the denition:
{f
e
s
}
def
=
_
A
t
t
(n)

i
ds (69)
where t
(n)
is computed from its denition, t
(n)
= n, where the
Cauchy stress is determined from Hookes law and strain deduced
from the computed gradient of the displacement
U. Saravanan (IIT Madras) Finite element method April 2014 74 / 81
Finite element formulation for linearized elasticity Post processing of the results
Points to understand
Since {f
e
s
}
def
i
are calculated using the approximate displacement eld,
they are not as accurate as {f
e
s
}
equil
i
{f
e
s
}
def
i
= {f
e
s
}
equil
i
Because of computational aspects, in nite element computer codes
{f
e
s
}
def
i
are calculated instead of {f
e
s
}
equil
i
However, the dierence between {f
e
s
}
def
i
and {f
e
s
}
equil
i
decreases as
the number of elements is increased or the degree of the interpolation
is increased, hopefully!
U. Saravanan (IIT Madras) Finite element method April 2014 75 / 81
Finite element formulation for linearized elasticity Post processing of the results
Points to understand
Since {f
e
s
}
def
i
are calculated using the approximate displacement eld,
they are not as accurate as {f
e
s
}
equil
i
{f
e
s
}
def
i
= {f
e
s
}
equil
i
Because of computational aspects, in nite element computer codes
{f
e
s
}
def
i
are calculated instead of {f
e
s
}
equil
i
However, the dierence between {f
e
s
}
def
i
and {f
e
s
}
equil
i
decreases as
the number of elements is increased or the degree of the interpolation
is increased, hopefully!
U. Saravanan (IIT Madras) Finite element method April 2014 75 / 81
Finite element formulation for linearized elasticity Post processing of the results
Points to understand
Since {f
e
s
}
def
i
are calculated using the approximate displacement eld,
they are not as accurate as {f
e
s
}
equil
i
{f
e
s
}
def
i
= {f
e
s
}
equil
i
Because of computational aspects, in nite element computer codes
{f
e
s
}
def
i
are calculated instead of {f
e
s
}
equil
i
However, the dierence between {f
e
s
}
def
i
and {f
e
s
}
equil
i
decreases as
the number of elements is increased or the degree of the interpolation
is increased, hopefully!
U. Saravanan (IIT Madras) Finite element method April 2014 75 / 81
Finite element formulation for linearized elasticity Post processing of the results
Points to understand
Since {f
e
s
}
def
i
are calculated using the approximate displacement eld,
they are not as accurate as {f
e
s
}
equil
i
{f
e
s
}
def
i
= {f
e
s
}
equil
i
Because of computational aspects, in nite element computer codes
{f
e
s
}
def
i
are calculated instead of {f
e
s
}
equil
i
However, the dierence between {f
e
s
}
def
i
and {f
e
s
}
equil
i
decreases as
the number of elements is increased or the degree of the interpolation
is increased, hopefully!
U. Saravanan (IIT Madras) Finite element method April 2014 75 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 1: Determining {f
e
s
}
i
for point forces
If a point force, F
o
acts at a point, say x
o
, then
t
(n)
(x) = F
o
(x x
o
), (70)
where the Dirac delta function () is dened by
_

F(x)(x x
o
)dv = F(x
o
). (71)
Hence, when traction vector is given by (70)
{f
e
s
}
i
=
_
A
t
t
(n)

i
ds = F
o

i
(x
o
). (72)
U. Saravanan (IIT Madras) Finite element method April 2014 76 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 1: Determining {f
e
s
}
i
for point forces
If a point force, F
o
acts at a point, say x
o
, then
t
(n)
(x) = F
o
(x x
o
), (70)
where the Dirac delta function () is dened by
_

F(x)(x x
o
)dv = F(x
o
). (71)
Hence, when traction vector is given by (70)
{f
e
s
}
i
=
_
A
t
t
(n)

i
ds = F
o

i
(x
o
). (72)
U. Saravanan (IIT Madras) Finite element method April 2014 76 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 2: Sources of error in the nite element solution
1. Domain approximation error, which is due to the approximation of the
domain
2. Computational errors, which are due to inexact evaluation of the
coecients of the element stiness matrix and force vector or are
introduced owing to the nite arithmetic in a computer.
3. Approximation errors, which is due to approximation of the solutions
by piecewise polynomials.
The main source of error in a boundary value problem is the third
error.
U. Saravanan (IIT Madras) Finite element method April 2014 77 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 2: Sources of error in the nite element solution
1. Domain approximation error, which is due to the approximation of the
domain
2. Computational errors, which are due to inexact evaluation of the
coecients of the element stiness matrix and force vector or are
introduced owing to the nite arithmetic in a computer.
3. Approximation errors, which is due to approximation of the solutions
by piecewise polynomials.
The main source of error in a boundary value problem is the third
error.
U. Saravanan (IIT Madras) Finite element method April 2014 77 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 2: Sources of error in the nite element solution
1. Domain approximation error, which is due to the approximation of the
domain
2. Computational errors, which are due to inexact evaluation of the
coecients of the element stiness matrix and force vector or are
introduced owing to the nite arithmetic in a computer.
3. Approximation errors, which is due to approximation of the solutions
by piecewise polynomials.
The main source of error in a boundary value problem is the third
error.
U. Saravanan (IIT Madras) Finite element method April 2014 77 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 2: Sources of error in the nite element solution
1. Domain approximation error, which is due to the approximation of the
domain
2. Computational errors, which are due to inexact evaluation of the
coecients of the element stiness matrix and force vector or are
introduced owing to the nite arithmetic in a computer.
3. Approximation errors, which is due to approximation of the solutions
by piecewise polynomials.
The main source of error in a boundary value problem is the third
error.
U. Saravanan (IIT Madras) Finite element method April 2014 77 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 3: Symmetry of the stiness and mass matrix
The element matrices - stiness and mass matrix - derived from
Galerkin method is symmetric, irrespective of the type of element
used or the order of the polynomial used to approximate the solution
Because of the symmetry of the element matrices, the assembled
global matrix will also be symmetric
The symmetry of the global and elemental matrices depends on the
dierential equation, the weighted residual formulation and
numbering of the nite element equations.
U. Saravanan (IIT Madras) Finite element method April 2014 78 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 3: Symmetry of the stiness and mass matrix
The element matrices - stiness and mass matrix - derived from
Galerkin method is symmetric, irrespective of the type of element
used or the order of the polynomial used to approximate the solution
Because of the symmetry of the element matrices, the assembled
global matrix will also be symmetric
The symmetry of the global and elemental matrices depends on the
dierential equation, the weighted residual formulation and
numbering of the nite element equations.
U. Saravanan (IIT Madras) Finite element method April 2014 78 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 3: Symmetry of the stiness and mass matrix
The element matrices - stiness and mass matrix - derived from
Galerkin method is symmetric, irrespective of the type of element
used or the order of the polynomial used to approximate the solution
Because of the symmetry of the element matrices, the assembled
global matrix will also be symmetric
The symmetry of the global and elemental matrices depends on the
dierential equation, the weighted residual formulation and
numbering of the nite element equations.
U. Saravanan (IIT Madras) Finite element method April 2014 78 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 4: Sparseness of the stiness and mass matrix
Since, K
ij
= M
ij
= 0 if global nodes i and j do not belong to the
same element, the global matrices are banded, i.e., all components
beyond a certain distance from the diagonal are zero
A banded matrix is a sparse matrix
Thus, the sparseness of the matrix is a result of the nite element
interpolation functions, which have non-zero values only over an
element of the domain.
U. Saravanan (IIT Madras) Finite element method April 2014 79 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 4: Sparseness of the stiness and mass matrix
Since, K
ij
= M
ij
= 0 if global nodes i and j do not belong to the
same element, the global matrices are banded, i.e., all components
beyond a certain distance from the diagonal are zero
A banded matrix is a sparse matrix
Thus, the sparseness of the matrix is a result of the nite element
interpolation functions, which have non-zero values only over an
element of the domain.
U. Saravanan (IIT Madras) Finite element method April 2014 79 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 4: Sparseness of the stiness and mass matrix
Since, K
ij
= M
ij
= 0 if global nodes i and j do not belong to the
same element, the global matrices are banded, i.e., all components
beyond a certain distance from the diagonal are zero
A banded matrix is a sparse matrix
Thus, the sparseness of the matrix is a result of the nite element
interpolation functions, which have non-zero values only over an
element of the domain.
U. Saravanan (IIT Madras) Finite element method April 2014 79 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 5: Estimate of stress
The balance of the surface forces at the inter element surfaces is
expressed in a weighted integral sense, is an approximation over the
actual requirement
Even this condition will not be satised by the nite element solution
Since, {f
e
s
}
def
i
= {f
e
s
}
equil
i
, nite element solution also only ensures
that (F
e
iJ
)
equil
+ (F
e+1
kL
)
equil
= 0 and not (F
e
iJ
)
def
+ (F
e+1
kL
)
def
= 0.
Thus, the conditions on the traction are only satised approximately
and consequently the estimate of the stresses by the nite element
method is also poor in comparison to the estimate of the displacement
U. Saravanan (IIT Madras) Finite element method April 2014 80 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 5: Estimate of stress
The balance of the surface forces at the inter element surfaces is
expressed in a weighted integral sense, is an approximation over the
actual requirement
Even this condition will not be satised by the nite element solution
Since, {f
e
s
}
def
i
= {f
e
s
}
equil
i
, nite element solution also only ensures
that (F
e
iJ
)
equil
+ (F
e+1
kL
)
equil
= 0 and not (F
e
iJ
)
def
+ (F
e+1
kL
)
def
= 0.
Thus, the conditions on the traction are only satised approximately
and consequently the estimate of the stresses by the nite element
method is also poor in comparison to the estimate of the displacement
U. Saravanan (IIT Madras) Finite element method April 2014 80 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 5: Estimate of stress
The balance of the surface forces at the inter element surfaces is
expressed in a weighted integral sense, is an approximation over the
actual requirement
Even this condition will not be satised by the nite element solution
Since, {f
e
s
}
def
i
= {f
e
s
}
equil
i
, nite element solution also only ensures
that (F
e
iJ
)
equil
+ (F
e+1
kL
)
equil
= 0 and not (F
e
iJ
)
def
+ (F
e+1
kL
)
def
= 0.
Thus, the conditions on the traction are only satised approximately
and consequently the estimate of the stresses by the nite element
method is also poor in comparison to the estimate of the displacement
U. Saravanan (IIT Madras) Finite element method April 2014 80 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 5: Estimate of stress
The balance of the surface forces at the inter element surfaces is
expressed in a weighted integral sense, is an approximation over the
actual requirement
Even this condition will not be satised by the nite element solution
Since, {f
e
s
}
def
i
= {f
e
s
}
equil
i
, nite element solution also only ensures
that (F
e
iJ
)
equil
+ (F
e+1
kL
)
equil
= 0 and not (F
e
iJ
)
def
+ (F
e+1
kL
)
def
= 0.
Thus, the conditions on the traction are only satised approximately
and consequently the estimate of the stresses by the nite element
method is also poor in comparison to the estimate of the displacement
U. Saravanan (IIT Madras) Finite element method April 2014 80 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 6: Estimate of the displacement eld
The displacement eld obtained from nite element method satises
the equilibrium equation only in an approximate sense, since the
procedure is based on the weak form of the dierential equation.
The assumed displacement eld, in many cases, does not contain the
actual displacement eld and hence there will be an error in the
estimated displacements
U. Saravanan (IIT Madras) Finite element method April 2014 81 / 81
Finite element formulation for linearized elasticity Remarks
Remark - 6: Estimate of the displacement eld
The displacement eld obtained from nite element method satises
the equilibrium equation only in an approximate sense, since the
procedure is based on the weak form of the dierential equation.
The assumed displacement eld, in many cases, does not contain the
actual displacement eld and hence there will be an error in the
estimated displacements
U. Saravanan (IIT Madras) Finite element method April 2014 81 / 81

You might also like