Scribd
Upload
528 views

FEM of Spring

University of Washington Mechanical Engineering ME 478, Summer 13 Finite Element Analysis Topic 1: FEM of spring system

Uploaded by

Lei Chi Hou
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
528 views

FEM of Spring

University of Washington Mechanical Engineering ME 478, Summer 13 Finite Element Analysis Topic 1: FEM of spring system

Uploaded by

Lei Chi Hou
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

ME 478 FEA

Chi Hou Lei @ UW 13

Formulation of one-dimensional spring element


In this section, we introduce the two approaches for finite element analysis: The classical method (which uses force
equilibrium) and the energy method.

This structure consists of three spring elements. Our approach is to formulate the matrix equation for one of the springs,
apply these formulations to other spring elements and then assembly these matrix equations to the global matrix
equation.
First we develop the matrix equation for 1 element.

u1 and u2 are the nodal displacements. F1 and F2 are the nodal force. The convention is positive outwards.
Now recall the governing equation (force-deformation relation) of spring:

T
where T is the tension and

u2

k(u2

u1 ) ,

u1 is the deformation of the spring.

By the sign convention for nodal forces and equilibrium, we have:

f1
f2

T
T
1

ME 478 FEA
Substitute these two equations into the governing equation, we obtain:

f1
f2

k(u2

k(u2

u1 )

u1 )

Chi Hou Lei @ UW 13

k(u1
k(u1

u2 )
u2 )

Expressed in matrix form,

f1

f2

k u1
k

u2

This is the building block (for one single element).

Now going back to the system of springs, we apply the matrix equation to each spring, that is;

f1

k1

f2

f2

k1 u1

k1

k1

k2

f3

u2

k2

f3

k2 u2
k2

f4

u3

k3

k3 u3

k3

k3

u4

There are in total 4 nodes, thus 4 nodal forces and 4 nodal displacements. We may expand each matrix to include all
these 4 nodal forces and displacements.

u1

f1

u2

f2

f3

u3

f3

f4

u4

f4

f1
f2

k1
k1

k1
k1

k2
k2

k2
k2

u1

f1

u1

u2

f2

u2

u3

f3

u4

f4

k3
k3

k3 u3
k3

u4

The three equations are now added to form the global matrix equation:

f1
f2

k1
k1

k1
k1

f3

u1

k2
k2

k2
k2

f4

k3
k3

u2
k3 u3
k3

u4

To deal with the boundary conditions, we update some of the nodal forces and displacements.

f1
0
0
F

k1
k1

k1
k1

k2
k2

k2
k2

k3
k3

u2
k3 u3
k3

u4

ME 478 FEA
Chi Hou Lei @ UW 13
Due to zero displacement, the above 4 X 4 matrix equation can be reduced to a 3 X 3 matrix equation. Here we simply
delete the first row and first column because the first nodal displacement is zero.

k1

k2

k2

k2

k2

u2

k3

k3 u3

k3

k3

u4

The unknown displacements are evaluated by solving this matrix equation, or simply, taking inverse matrix.

u2

k1

u3

k2

k2

k2

k2

u4

k3

k3

k3
1
k1
1
k1

1
k1
k1 k2
k1k2

1
k1

k1 k2
k1k2

0
0

k3

F
1
k1

0
k1 k2
0
k1k2
F
k1k2 k2k3 k3k1
k1k2k3

1
F
k1
k1 k2
F
k1k2
k1k2 k2k3 k3k1
F
k1k2k3

We still have one unknown, the nodal force, f1 , which can be found from the original matrix equation.
..
From the matrix equation, we know that f1 is in fact the dot product of the first row of the matrix and the nodal
displacement vector. Thus,

f1

k1u1

k1u2

k1 0 k1

1
F
k1

F
This makes sense, why? Remember cut set (connectivity equation) from ME 373.
Physically, f1 represents the reaction force at the wall.
3

ME 478 FEA

Chi Hou Lei @ UW 13

We may also the deformation (elongation or contraction) and force (tension or compression) in any spring of the
system. Here as an example, we consider spring 2. Using the governing equation for spring, the deformation of element
2 is:
2

element 2

u3

k1 k2
F
k1k2
1
F
k2

u2

1
F
k1

The force in element is:

Telement

T2

k2

k(u3
k2

u2 )

1
F
k2

ME 478 FEA
SUMMARY: All finite element problems are solved with the following steps
Step #
1

Chi Hou Lei @ UW 13

Discrete and select element type

The problem has done the discretization for you.


2

Displacement function

Two DOFs, one at each end of the element.


The problem has extracted the modal of displacement for you.
Here we use linear function to describe displacement.
3

Governing equation

stiffness matrix and matrix


equation for single element

f1

f
5

KU

k(u2
k1

f2

ku

Assembly to obtain (global)


stiffness matrix and (global) matrix
equation

f1
f2

k1

k1 u1

k1

k1

u1 )

k1

k1
k1

f3

u1

k2

k2

k2

k2

Incorporate BC and solve for the


unknown degrees of freedom

U K 1F
7

u2

k3

f4
6

u2

k3 u3

k3

u2
u3

k1

k2
k2

k3
1

k2
k2

u4

k3

k3

k3

Reaction force at wall: f1

Deformation in spring 2:

u3

Force in spring 2: T2

k(u3

KU

k2

k1u1

0
0

k3

Post-processing
Interpret results

u4

k1u2

1
F
k2
u2 ) F

u2

ME 478 FEA

Chi Hou Lei @ UW 13

Energy method

potential

strain energy work done


U W
1 2
ei
i:elements 2

Fju j
j:nodes

The system is stable (giving you the state of equilibrium) when all partial derivatives of the total potential energy
become zero, that is,

uj

Now we solve the same problem using energy method.

U W
1 2
ei
i:elements 2

j:nodes

1 2
e1
2
1
(u2
2

1 2
e3 F1u1 F2u2 F3u3 F4u4
2
1
1
(u3 u2 )2
(u4 u3 )2 F1u1 F2u2
2
2

1 2
e2
2
u1 )2

Fju j

F3u3

F4u4

ME 478 FEA
Chi Hou Lei @ UW 13
Taking partial derivatives with respect to each nodal displacement yield the same set of equations.

1
k1 (u2
2

1
k2 (u3
2

u1 )2

1
k3 (u4
2

u2 )2

u3 )2

F1u1

u1
0

1
k1 2(u2 u1 )( 1)
2
k1 (u1 u2 ) F1

u2
0

1
1
k1 2(u2 u1 )(1)
k2 2(u3 u2 )( 1) 0 F2
2
2
k1 (u2 u1 ) k2 (u2 u3 ) F2

u3

1
k2 2(u3
2

u3
0

k2 (u3

u2 )

1
2k3 (u4
2
k3 (u4

0 F1

1
2k3 (u4
2

u2 )(1)
k3 (u3

F2u2

F3u3

F4u4

0 0 0

0 0

u3 )( 1) 0 0 F3

u4 ) F3

u3 )(1) 0 0 0 F4

u3 ) F4

We may rearrange these 4 equations so that all nodal displacements are on one side and all nodal forces are on the
other side.

k1 (u1

u2 ) F1

k1 (u2

u1 )

k2 (u2

u3 ) F2

k2 (u3

u2 )

k3 (u3

u4 ) F3

k3 (u4

u3 ) F4

k1u1

k1u2

F1

k1u1

(k1

k2 )u2

k2u3

F2

k2u2

(k2

k3 )u3

k3u4

F3

k3u3

k3u4

F4

We obtain the same matrix equation as in the classical approach, that is,

ME 478 FEA

Chi Hou Lei @ UW 13

f1
f2
f3

k1
k1

k1
k1

u1

k2
k2

f4

k2
k2

k3
k3

u2
k3 u3
k3

u4

The remained steps are identical and are not repeated here.

You might also like