Practical Stress Analysis with
Finite Elements (2nd Edition)
Dr. Bryan
y J Mac Donald
BryanMacDonald/Glasnevin Publishing20072011
Part 1: Introduction to FEA
Pages 1 to 2 of book
WhydoWeNeedFiniteElementAnalysis?
Consider the simple beam problem shown
Are you able to determine the maximum stress and
predict where in the beam it occurs?
What if you are asked to say whether the beam had
failed due to the applied load?
H
How
would
ld you d
do thi
this?
?
Bryan Mac Donald/Glasnevin Publishing 2007-2011
WL
y
I
max
6WLmax h
bh 3
Part 1: Introduction to FEA
What happens if the problem gets more complex?
Now we have some problems:
1) There is a taper in the beam which our theory
doesnt take account of.
2) There are a number of holes in the beam which
our theory doesnt take account of.
3) There is a more complex loading system present
which our theory doesnt allow for.
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Pages 2 to 3 of book
Part 1: Introduction to FEA
Pages 2 to 3 of book
So how do we overcome these problems?
Normally we introduce a Factor
Factor of Safety
Safety (FOS)
into our calculations.
The FOS results in a safe design,
but:
1) It results in over
over-designing
designing
2) Too much material/weight
3) Too
T much
h waste
t
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 4 to 5 of book
The FOS method has been used for centuries to produce many innovative designs.
The obvious way to overcome the limitations of the FOS method is to do some
experimental tests
build prototypes and test them!
There are significant costs involved in
building and testing prototypes (known
as a build-and-break cycle)
It takes
t k a long
l
ti
time
t gett an answer to
to
t
whether the design works or not.
The designer often won
wontt get a good feel
for the behaviour of the product until
near the end of the process
Th
There
mustt be
b a better
b tt way!!
There is its called Finite Element
Analysis!
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 4 to 5 of book
What is FEA?
A simple
i l definition:
d fi iti
The finite element method (FEM) is a mathematical technique that is used to
obtain approximate answers to complex problems that cannot be solved using
basic theories
An even simpler definition:
The fundamental concept of the FEM is that it splits up a complex problem into
a greater number of simpler problems and uses complex mathematics to glue
glue
together the answers to all the simple problems to give an approximate solution
to the complex problem.
A more scientific definition:
The FEM divides the domain of interest into a finite number of simple subdomains and uses variational concepts to construct an approximation of the
solution
l ti over the
th collection
ll ti off sub-domains.
bd
i
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
What is FEA?
Pages 4 to 5 of book
A Plane Stress Problem:
Equilibrium Eqn
Eqns
s
x xy
0
y
x
yx y
x
y
x
E
y 12
xy
0
1
x
1
0
y
0 0 (1 ) / 2 zz
Linear Elastic Stress-Strain Relationship
for Plane Strain
u
v
; y ;
x
y
xy
u v
y x
Strain-displacement relationships for
Plane Strain
Great! But how do we solve it?
Well, we cant! .so we use FEA to get an approximate answer!
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
What is FEA?
Pages 6 to 7 of book
A Plane Stress Problem:
Divide into finite elements
and apply loads and
boundary conditions
Solve FE problem to obtain
displacement of each node
Further process the
displacement results
in order to obtain
strain and stress
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 6 to 7 of book
How Does FEA Work?
Divides structure into small pieces called elements.
These elements are simple shapes, so simple theory can work with them.
The collection of elements is called a finite element mesh
As the number of elements in the mesh increases their size decreases and so the
approximate solution (should) become more accurate
Analogy with Integration / Simpsons rule:
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 8 to 9 of book
How Does FEA Work?
All Finite Element Analyses follow the same procedure:
1.
2.
3
3.
4.
5.
6.
7.
8.
Step 1: Evaluate the problem and make assumptions
Step 2: Describe how the finite elements will behave
Step 3: Build the finite element model
Step 4: Form element equations
Step 5: Assemble each element equation into a global problem equation
Step 6: Specify loads and boundary conditions
Step 7: Solve the global problem
Step 8: Evaluate the results.
Lets look at an example to illustrate the above steps.
- We will cover the theory in this example later in the book.
- For now dont worry too much, just try and understand the concepts
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 8 to 9 of book
A tapered steel beam of rectangular cross section is
used to support a horizontal load of F = 10,000 N.
Use FEA to determine the deflection of the free
end of the beam and the max stress in the beam.
Step1:
Evaluatetheproblemandmakeassumptions
Thisisarelativelysimpleproblemthathasananalyticalsolutionandprobablydoesntrequire
FEAtoobtainanapproximatesolution,however,forthepurposesofexampleletsassumethat
ananalyticalsolutiondoesnotexist.
y
Wearetoldthatthebeamismadefromsteelsoletsassumethattheloadisntgreatenough
tocausepermanentplasticdeformationwhichmeanswecanassumealinearelasticmaterial
modelgovernedbyHookeslawwithusualmaterialpropertiesforsteel.
d l
db H k l
ith
l t i l
ti f t l
Assumption1:
LinearelasticmaterialmodelwithE=210x109Pa
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 8 to 9 of book
We know that the beam is embedded in a concrete
wall so it is reasonable to assume that the left hand
edge of the beam cannot move.
Assumption 2: The left hand edge of the beam is
fixed
Applied horizontal load on the right hand edge of the beam.
The
Th beams
b
own weight
i ht will,
ill however,
h
cause it to
t deflect
d fl t slightly
li htl downwards.
d
d
Lets assume selfweight to be negligible beam can only deflect in the horizontal direction.
All loads and reactions in horizontal direction = a onedimensional problem
Assumption 3:
A onedimensional analysis is sufficient to solve this problem
We are not told much about the manner that the load is applied to the right hand edge of the
b
beam.
L t assume that
Lets
th t the
th load
l d is
i applied
li d very slowly.
l l This
Thi means that
th t dynamic
d
i (vibration)
( ib ti )
effects can be ignored and that loads are applied very slowly.
Assumption
p
4:
A linear static analysis
y is sufficient to solve this p
problem
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 10 to 11 of book
We are also not told about the environment in which the beam is placed.
As a first approximation lets assume that environmental factors are not significant and hence
only a structural analysis is required.
Assumption 5: A structural analysis is sufficient to solve this problem
Step 2:
Step2:
Describe how the finite elements will behave
Describehowthefiniteelementswillbehave
Assumption3abovemeansthatwecanuseaonedimensionalelementforthisanalysis.
Thesimplestelementtypeisa1Dlineartrusselement.
Thiselementeffectivelybehaveslikeaspringthatcanonlydeforminonedirection.
2nodes Ni andNj
DuetoFnodei
D t F d i movesadistanceU
di t
Ui andnodejmovesbyU
d d j
b Uj
Sotheelementsdisplacementisafunctionofthenodaldisplacement
u((e)) =S
= SiUi +S
+ SjUj
Eqn 1.06
WhereSi andSj areshapefunctionswhichrelatedisplacementoftheelementasawholeto
thespecificdisplacementofthenodesthatmakeuptheelement.
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 10 to 11 of book
We now define a local coordinate system in the element (natural coordinate system)
given by
where is measured from the centre of the element
such that Ni has a coordinate = 1 and Nj has a coordinate = 1.
We will show in chapter 4,
4 that the shape functions are:
Si = (1 )
and
Sj = (1 + )
Eqn 1.07
Equation 1.06 is more commonly used in its matrix form:
Eqn 1.08
1 08
We can also get an expression relating element strain (general) to nodal displacement (specific)
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
From assumption 4: material obeys Hookes law
Putting equation 1.10 into equation 1.11 gives us:
Step 3: Build the Finite Element Model
Split the tapered beam into 3 regularly shaped elements:
based on average cross section at mid point of element
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Pages 10 to 11 of book
Part 1: Introduction to FEA
Pages 11 to 12 of book
The finalised model shows three elastic springs (i.e. truss elements) with equivalent stiffness
to the three uniform cross section elements shown in the previous figure:
Step 4: Form Element Equations
We will discuss this step in some detail in chapters 3 and 4. For now lets just accept that the
governing equation for this problem is:
It can be shown (see chapter 4) that for this element type:
Evaluating this equation (i.e. plugging in [B] and integrating) gives us:
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 12 to 13 of book
Calculating the cross sectional area for each element and plugging in values of E and
element length, l, gives us an equation for each element:
This allows us to form the element equations using equation 1.13
Step 5: Assemble each element equation into a global problem equaiton
Combining these three equations into a matrix equation:
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 12 to 13 of book
Step 6: Specify Loads and Boundary Conditions
Step 7: Solve the Global Problem
This example is relatively easy to solve use Gaussian elimination or similar
Solving gives:
U1 = 0,
U2 = 2.199 x 105m,
U3 = 5.374
5 374 x 105m,
m
U4 = 11.062 x 105m
So, the global nodal displacement vector is:
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Deflectionoffreeendofthe
beamis11.062x105 m
Part 1: Introduction to FEA
Pages 13 to 14 of book
Step 8: Evaluate the Results
Lets find the strain and stress in the elements:
Using the straindisplacement equations (equation 1.09)
Now that we have the strain distibution in the FE model we can easily calculate stress using
equation 1.11 (i.e. Hookes Law)
Maxstressinthebeamis
59 72 MPa atnode3
59.72MPa
at node 3
Bryan Mac Donald/Glasnevin Publishing 2007-2011
Part 1: Introduction to FEA
Pages 13 to 14 of book
Summary of Chapter 1:
After completing
p
g chapter
p 1,, yyou should:
1. Be able to explain why finite element analysis is an important tool for engineering analysis
and design
2. Be able to explain what finite element analysis is and how it works.
3. Be able to carryy out a simple
p onedimensional finite element analysis
y of an engineering
g
g
problem on a piece of paper, by following the methodology of example 1.1.
4. Understand the various steps required for any finite element analysis
Bryan Mac Donald/Glasnevin Publishing 2007-2011