Discretization of the Problem

FEM PROCEDURE

 Step 1: Domain discretization

 Step 2: Displacement interpolation
 Step 3: Formation of FE equation in local coordinates
 Step 4: Coordinate transformation
 Step 5: Assembly of FE equations
 Step 6: Imposition of displacement constraints
 Step 7: Solving the FE equations
Introduction
 This is the first step in the finite element procedure
 The body under examination is divided into elements in
such a way that the unknown field variable is adequately
represented through the body.
 Proper care must be taken for planning and preparation
of the problem – considerable savings in time and effort
 Economics - by approximating the dimensionality of the
problem and by taking advantage of any symmetry in the
body
Geometrical Approximations
 If geometry and loads of a problem can be prescribed in
one plane, then problems can be modelled as two-
dimensional.
 Plane Strain: Bodies which are long and whose geometry
and loading do not vary significantly in the logitudinal
direction.
Example: Analysis of dams, long pipes subjected to
internal pressure
 Plane Stress: Bodies that have negligible thickness in one
direction, and are loaded in the plane of the body.
Example: Thin plates subjected to inplane load
What to model?
 Many modeling decisions must be made before building an
analysis model:
 How much detail should be included?
 Does symmetry apply?
 Will the model contain stress singularities?
GEOMETRY MODELLING
 Reduction of a complex geometry to a manageable one.
 3D? 2D? 1D? Combination?

Bulky solids 3-D solid element mesh

Neutral surface

z
z

y
x x y

h
Shell
(Using 2D or 1D makes
2-D shell element mesh
Neutral surface

meshing much easier)

fy2 Centroid
fy1



x

z 



Beam member 1-D beam element mesh
… What to model?
Details
 Small details that are unimportant to the analysis should
not be included in the analysis model.
 For some structures, however, "small" details such as fillets
or holes can be locations of maximum stress and might be
quite important, depending on your analysis objectives.
Symmetry
 Many structures are symmetric in some form and allow
only a representative portion or cross-section to be
modeled.
 The main advantages of using a symmetric model are:
 It is generally easier to create the model.
 It allows you to make a finer, more detailed model and thereby
obtain better results than would have been possible with the
full model.
…Symmetry
 To take advantage of symmetry, all of the following must
be symmetric:
 Geometry
 Material properties
 Loading conditions
 There are different types of symmetry:
 Axisymmetry
 Rotational
 Planar or reflective
 Repetitive or translational
 USE OF SYMMETRY

 Different types of symmetry:

Use of symmetry reduces number
of DOFs and hence computational
time. Also reduces numerical error.

Mirror symmetry
Axial symmetry

Cyclic symmetry Repetitive symmetry

Axisymmetry
 Symmetry about a central axis, such as in light bulbs,
straight pipes, cones, circular plates, and domes.
 Plane of symmetry is the cross-section anywhere around
the structure. Thus you are using a single 2-D “slice” to
represent 360° — a real savings in model size!

 Loading is also assumed to be

axisymmetric in most cases. However, if it
is not, and if the analysis is linear, the loads
can be separated into harmonic
components for independent solutions
that can be superimposed.
Rotational symmetry
Repeated segments arranged about a central axis, such as in
turbine rotors.
Only one segment of the structure needs to be modeled.
Loading is also assumed to
Planar or reflective symmetry
 One half of the structure is a mirror image of the
other half. The mirror is the plane of symmetry.
 Loading may be symmetric or anti-symmetric about
the plane of symmetry.

This model illustrates both

reflective and rotational
symmetry
Repetitive or translational symmetry
 Repeated segments arranged along a straight line, such
as a long pipe with evenly spaced cooling fins.
 Loading is also assumed to be “repeated” along the
length of the model.

This model illustrates both repetitive and reflective symmetry.

 In some cases, only a few minor details will disrupt a
structure's symmetry. You may be able to ignore such
details (or treat them as being symmetric) in order to gain
the benefits of using a smaller model. How much accuracy
is lost as the result of such a compromise might be difficult
to estimate.
Stress singularities
 A stress singularity is a location in a finite element
model where the stress value is unbounded (infinite).
Examples:
 A point load, such as an applied force or moment
 An isolated constraint point, where the reaction force behaves like a
point load
 A sharp re-entrant corner (with zero fillet radius)
 As the mesh density is refined at P s = P/A
As A  0, s  
a stress singularity, the stress value
increases and never converges.
…Stress singularity
 Real structures do not contain stress singularities. They
are a fiction created by the simplifying assumptions of the
model.
 So how do you deal with stress singularities?
 If they are located far away from the region of interest, you can
simply ignore them by deactivating the affected zone while
reviewing results.
 If they are located in the region of interest, you will need to
take corrective action, such as:
 adding a fillet at re-entrant corners and reruning the analysis.
 replacing a point force with an equivalent pressure load.
 “spreading out” displacement constraints over a set of nodes.
Meshing
 Meshing is the process used to “fill” the solid model
with nodes and elements, i.e, to create the FEA model.
 Remember, you need nodes and elements for the finite
element solution, not just the solid model. The solid model
does NOT participate in the finite element solution.

meshing

Solid model FEA model

Element Type
 The element type is an important choice that determines
the following element characteristics:
 Degree of Freedom (DOF) set. A thermal element type, for
example, has one dof: TEMP, whereas a structural element type
may have up to six dof: UX, UY, UZ, ROTX, ROTY, ROTZ.
 Element shape -- brick, tetrahedron, quadrilateral, triangle, etc.
 Dimensionality -- 2-D (X-Y plane only), or 3-D.
 Assumed displacement shape -- linear vs. quadratic.
Element category
 There are many different categories of elements. Some of
the commonly used ones in commercial codes are:
 Line elements
 Shells
 2-D solids
 3-D solids
Line elements
 Beam elements are used to model bolts, tubular members,
C-sections, angle irons, or any long, slender members
where only membrane and bending stresses are needed.
 Spar elements are used to model springs, bolts, preloaded
bolts, and truss members.
 Spring elements are used to model springs, bolts, or long
slender parts, or to replace complex parts by an equivalent
stiffness.
Shell elements
 Used to model thin panels or curved surfaces.
 The definition of “thin” depends on the application, but as a
general guideline, the major dimensions of the shell
structure (panel) should be at least 10 times its thickness.
2-D Solid elements:
 Used to model a cross-section of solid objects.
 Must be modeled in the global Cartesian X-Y plane.
 All loads are in the X-Y plane, and the response
(displacements) are also in the X-Y plane.
 Element behavior may be one of the following:
 plane stress
 plane strain
 axisymmetric
 axisymmetric harmonic
Y

Z X
Plane Stress
 Plane stress assumes zero stress
in the Z direction.
Y
 Valid for components in which
the Z dimension is smaller than Z X
the X and Y dimensions.
 Z-strain is non-zero.
 Optional thickness (Z direction)
allowed.
 Used for structures such as flat
plates subjected to in-plane
loading, or thin disks under
pressure or centrifugal loading.
Plane Strain
 Plane strain assumes zero strain in the Z
direction.
 Valid for components in which the Z
dimension is much larger than the X and Y
dimensions.
Z
 Z-stress is non-zero.
 Used for long, constant cross-section Y
X
structures such as structural beams.
Axisymmetry
 Axisymmetry assumes that the 3-D model
and its loading can be generated by
revolving a 2-D section 360° about the Y
axis.
 Axis of symmetry must coincide with the global
Y axis.
 Negative X coordinates are not permitted.
 Y direction is axial, X direction is radial, and Z
direction is circumferential (hoop) direction.
 Hoop displacement is zero; hoop strains and
stresses are usually very significant.
 Used for pressure vessels, straight pipes, shafts,
etc.
3-D Solid elements
 Used for structures which, because of geometry, materials,
loading, or detail of required results, cannot be modeled
with simpler elements.
 Also used when the model geometry is transferred from a
3-D CAD system, and a large amount of time and effort is
required to convert it to a 2-D or shell form.
Element Order
Element order refers to the polynomial order of the
element’s shape functions.
 What is a shape function?
 It is a mathematical function that gives the “shape” of the
results within the element. Since FEA solves for DOF
values only at nodes, we need the shape function to map
the nodal DOF values to points within the element.
 The shape function represents assumed behavior for a
given element.
 How well each assumed element shape function matches
the true behavior directly affects the accuracy of the
solution, as shown on the next slide.
 Linear approximation
(Poor Results)
Quadratic distribution of
DOF values
Actual quadratic
curve

Linear approximation
with multiple elements Quadratic approximation
(Better Results) (Best Results)
 When you choose an element type, you are implicitly
choosing and accepting the element shape function
assumed for that element type. Therefore, check the
shape function information before you choose an element
type.
 Typically, a linear element has only corner nodes, whereas
a quadratic element also has midside nodes.
 Linear elements Quadratic elements
 Can support only a linear variation  Can support a quadratic variation
of displacement and therefore of displacement and therefore a
(mostly) only a constant state of linear variation of stress within a
stress within a single element. single element.
 Highly sensitive to element  Can represent curved edges and
distortion. surfaces more accurately than
 Acceptable if you are only linear elements. Not as sensitive
interested in nominal stress results. to element distortion.
 Need to use a large number of  Recommended if you are
elements to resolve high stress interested in highly accurate
gradients. stresses.
 Give better results than linear
elements, in many cases with
fewer number of elements and
total DOF.
Mesh Density
 The fundamental premise of FEA
is that as the number of elements
(mesh density) is increased, the
solution gets closer and closer to
the true solution.
 However, solution time and
computer resources required also
increase dramatically as you
increase the number of elements.
…Mesh Density
 If you are interested in highly accurate stresses:
 A fine mesh will be needed, omitting no geometric details at
any location in the structure where such accuracy is needed.
 Stress convergence should be demonstrated.
 Any simplification anywhere in the model might introduce
significant error.
 If you are interested in deflections or nominal stresses:
 A relatively coarse mesh is sufficient.
 Small geometry details may be omitted.
 If you are interested in mode shapes (modal analysis):
 Small details can usually be omitted.
 Simple mode shapes can be captured using a relatively coarse
mesh.
 Complex mode shapes may require a uniform, moderately fine
mesh.
 Thermal Analyses:
 Small details can usually be omitted, but since many thermal
analyses are followed by a stress analysis, stress considerations
generally dictate the detail of the model.
 Mesh density is usually determined by expected thermal
gradients. A fine mesh is required for high thermal gradients,
whereas a coarse mesh may be sufficient for low gradients.
Multiple attributes
 Most FEA models have multiple attributes. For example,
the silo shown here has two element types, three real
constant sets, and two materials.

TYPE 1 = shell
TYPE 2 = beam MAT 1 = concrete
MAT 2 = steel

REAL 1 = 3/8” thickness

REAL 2 = beam properties
REAL 3 = 1/8” thickness
 There are two main meshing
methods: free and mapped.
 Free Mesh
 Has no element shape restrictions.
 The mesh does not follow any pattern.
 Suitable for complex shaped areas and
volumes.
 Mapped Mesh
 Restricts element shapes to
quadrilaterals for areas and hexahedra
(bricks) for volumes.
 Typically has a regular pattern with
obvious rows of elements.
 Suitable only for “regular” areas and
volumes such as rectangles and bricks.
 Free Mesh Mapped Mesh
+ Easy to create; no need to + Generally contains a lower
divide complex shapes into number of elements.
regular shapes. + Lower-order elements may be
– Volume meshes can contain only acceptable, so the number of
tetrahedra, resulting in a large DOF is lower.
number of elements. – Areas and volumes must be
– Only higher-order (10-node) “regular” in shape, and mesh
tetrahedral elements are divisions must meet certain
acceptable, so the number of criteria.
DOF can be very high. – Very difficult to achieve,
especially for complex shaped
volumes.
Element distortion
 Use of distorted elements in irregular and
complex geometry is common but there are some
limits to the distortion.
 The distortions are measured against the basic
shape of the element
 Square  Quadrilateral elements
 Isosceles triangle  Triangle elements
 Cube  Hexahedron elements
 Isosceles tetrahedron  Tetrahedron elements

39
Element distortion

 Aspect ratio distortion

b
b 3 Stress analysis
Rule of thumb: 
a 10 Displacement analysis

40
Element distortion

 Angular distortion

skew Taper
b a



b<5a

41
Element distortion

 Curvature distortion



42
Element distortion
 Volumetric distortion

Area outside distorted

element maps into an
internal area – negative
volume integration

43
Element distortion

 Volumetric distortion (Cont’d)






44
Element distortion

 Mid-node position distortion

Shifting of nodes beyond limits

can result in singular stress
field (see crack tip elements) a

b
>b/4

>a/4

45
MESH COMPATIBILITY

 Requirement of Hamilton’s principle – admissible

displacement

 The displacement field is continuous along all the

edges between elements

46
Different order of elements

1 1

Linear Linear
Quad 2 Quad
2

3 3

Crack like behaviour – incorrect results

47
Different order of elements
 Solution:
 Use same type of elements throughout
1 1

Linear Linear Quad 2 Quad

2 3
 

 Use transition elements

Quad Transition Linear

Vary quadratically along this edge Vary linearly along this edge

Transition
Element

48
 Use MPC equations