MECH3300 Finite Element Methods: Lecture 1 - Introduction To The FEM

MECH3300 Finite Element Methods
Lecture 1 - Introduction to the FEM

We will initially look at beam elements, then briefly review other types.
The solution process will be briefly discussed - more in Lecture 2
Concepts
Node - a generalised joint
- connection point at which equations are written
- there are at most 6 unknowns (degrees of freedom) at a node
(3 displacements, 3 rotations)

Element - connection between a group of nodes representing
stiffness or other properties approximately
- interpolation of displacement between the nodal values is used
- this becomes more accurate as element size is reduced

Reference node - an extra point used to orient the cross-section
of a beam

Beam elements
Beam elements were developed first, as a stiffness matrix [K
e
] of a
beam can be found exactly for elastic behaviour and small deflections
if there are boundary conditions and loading at each end only.

The matrix [K
e
] links forces and moments (a vector F) to
displacements and rotations (a vector u) at each end (each node).
F = [K
e
]u

In 3D 3 force components and 3 moment components act on each
end - hence the element matrix is 12 by 12.

Individual terms are listed on the fifth slide. They depend on the
length L, the area of cross-section A, the second moments of area, an
effective area deforming in shear etc.
Euler-Bernoulli v Timoshenko beam elements
The simplest beam element uses Euler-Bernouilli beam theory.

This neglects deflection transversely due to shear strain.

The same theory is used as in hand calculation
M/(EI) = 1/R integrated twice implies deflection

However, if the length of a beam is short (eg similar to its depth)
then this assumption leads to error, so most FE packages use
beam elements obtained from Timoshenko beam theory, which
includes deflection due to shear strain, as well as that due to
bending strain and rigid body motion.

An element may revert to a Euler-Bernoulli beam if the shear
area (cross-sectional area used to find shear deformation) is
not specified by the user.
Terms in a Beam Element Matrix
With Euler-Bernoulli beam theory, the following types of terms arise
in a beam element matrix, for bending a beam of length L on a
principal axis of the cross-section, with 2
nd
moment of area I.

Transverse force/deflection relations give 12EI/L
3

Transverse force/rotation or bending moment/displacement
relations give 6EI/L
2

Bending moment/rotation relations give 4EI/L or 2EI/L

Twisting a beam gives GI
P
/L

Axial deformation gives EA/L (A = cross-sectional area0

Transverse shear deformation gives GA
S
/L
where A
S
= effective cross-sectional area for shear.
Terms in one column of a stiffness matrix
One column of [K
e
] get multiplied by a particular element in the
vector of displacements and rotations, u.

Hence one column corresponds to those forces/moments
causing unit movement of one component of displacement or
rotation, at one end of a beam only.

Eg for unit transverse displacement at the left end of the beam
below (but no rotation), the forces and moments acting as
shown below - for Euler-Bernoulli beam theory.
1
12EI/L
3
12EI/L
3

6EI/L
2

6EI/L
2

L
Unit rotation of one end of a beam element
The forces and moments acting are as follows (for small angles)
1
6EI/L
2

6EI/L
2

4EI/L
2EI/L
These terms become a column of [K
e
]

Find these in the energy methods assignment.
The full matrix for transverse deflection
For bending about one principal axis only, the transverse
deformation can be described by 4 equations:

(
(
(
(
(
(
(
(
=
(
(
(
(
2
2
1
1
2 2
2 3 2 3
2 2
2 3 2 3
2
2
1
1
/ 4 / 6 / 2 / 6
/ 6 / 12 / 6 / 12
/ 2 / 6 / 4 / 6
/ 6 / 12 / 6 / 12
z
y
z
y
z
y
z
y
u
u
L EI L EI L EI L EI
L EI L EI L EI L EI
L EI L EI L EI L EI
L EI L EI L EI L EI
M
F
M
F
u
u
Note this assumes the cross-section does not change.
F
y2
F
y1

M
z1

M
z2

L
Rotation of coordinates
A beam will not in general be aligned with the global xyz axes. To
rewrite the stiffness terms in terms of forces/displacements in the global
axis directions, a rotation matrix R is found, the individual terms of which
are direction cosines.
The rotation R applies to both forces and displacements. The inverse of
R is just is transpose. This corresponds to rotating back the other way.
Hence if in local axes aligned with a beam F = K u, then in global
coordinates RF
G
= K Ru
G
or F
G
= R
T
KR u
G

Even for a single spring, this transformation leads to a 6 by 6 matrix,
containing products of direction cosines, as each end of the spring can
move in 3 directions, giving 6 equations.
A beam has rotations at each end as
well, giving 12 equations.

Solution Process
Once they are all written in the same coordinate system, element
matrices are combined (assembled) to give a large sfiffness matrix
for the full structure, by assuming that forces and moments at ends
of beams connected (nodes) sum to zero or to the resultant external
loading. Eg F
x1
+ F
x2
+ F
x3
= 0 at the joint below.

The full set of linear equations is solved for the displacements and
rotations. This is only possible if some displacements or rotations
are restrained to prevent rigid body motion.

The forces and moments acting on any one beam can then found
from its stiffness matrix times known displacements.

These then imply the stresses within a beam.

More detail later on this ..
F
x2

F
x1

F
x3

Geometric issues
A beam element is geometrically a line connecting nodes at
which forces/moments are applied - the centroidal axis of the
real beam.
The cross-section is represented by properties of the element.
Real beams have differing stiffnesses for bending about 2
principal axes. To indicate the direction of the minor principal
axis, a third node is often used.
Node A
Node B
Reference node in
plane of beam and
principal axis 2
Beam (axis 3 is
along the beam)
Principal axis 2
Minor principal axis
Principal axis 1
Major principal axis
1
2
Plate elements
Shell structures are common in engineering (drums, vehicle
bodies, ship hulls, etc.)
A plate element has the thickness implied. Geometrically it is a
surface representing the mid-surface of the real plate.
Stiffness of a plate element can only be found approximately, by
assuming polynomial variations of displacement or of stress
related quantities such as bending moment per length over its
area.
Accuracy improves as more elements are used.
Force/moment - deflection/rotation relations are again written at
node points.
Linear polynomials lead to an element with corner nodes only,
and sides that remain straight lines.

A plate model - a container pallet
View from below of front half of pallet.
A model combining beams and plates
A dragline rope tower
Solid elements
By assuming displacement variations in all three directions,
there is no missing geometry is implied by properties.
Such solid elements typically use linear interpolation of
displacement between nodal values, giving corner nodes only,
or quadratic interpolation giving mid-edge nodes as well.
Only displacements, not rotations, are present as unknowns at
the nodes. Rotations are only needed to describe bending and
twisting, not explicit in the geometry of an element. Rotations
imply the variation of axial displacement through the depth of a
beam or the variation of in-plane displacement through the
thickness of a plate.
Linear solid elements - edges remain straight lines.

MECH3300 Finite Element Methods: Lecture 1 - Introduction To The FEM

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

MECH3300 Finite Element Methods: Lecture 1 - Introduction To The FEM

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MECH3300 Finite Element Methods: Lecture 1 - Introduction To The FEM

Uploaded by

Copyright:

Available Formats

MECH3300 Finite Element Methods

Lecture 1 - Introduction to the FEM

You might also like