Direct Formulation and Minimum Total PE Formulation

Review: Basic FEM Steps

Preprocessing Phase
Step 1: Discretization subdivide the problem into nodes and elements
Step 2: Shape Function assume a function to represent the physical behavior of the
element
Step 3: Element Equations develop the mathematical equation for each element
Step 4: Assembly element equations throughout the FEM mesh are assembled into a
global matrix, modeling the system properties
Step 5: System Constraints apply boundary conditions, initial conditions, and loading
Solution Phase
Step 6: Solve for Primary Unknowns the global equation matrix is solved for the
results at each node
Postprocessing Phase
Step 7: Calculate Derived Variables determine other information using the nodal
values of the primary variables

Approaches to Formulating FE Problems

Direct Formulation
Minimum Total Potential Energy Formulation
Weighted Residual Formulation
Collocation Method
Subdomain Method
Galerkins Method
Least Squares Method
Variational Method
Now which to use? It depends on the application. Remember these are being used on much more
complicated models that may have 1000s or 10000s (or more!) of nodes. The best approach is the
easiest, the fastest, and most accurate (i.e., forces the error to zero at one point of interest, average
error, etc.) but usually it is a trade-off between all these criteria. This determination goes beyond
this course, but it is a valuable consideration to keep in mind.

1
1 Direct Formulation: Ex. 1.1

Bar with variable cross section

Fixed at upper end
Supports load P = 1000 lb
E = 10.4 106 lb/in2
Dimensions
w1 = 2 in
w2 = 1 in
t = 0.125 in
L = 10 in
Desired information (Weight is negligible)
Deflection (u)
Stress ()

Step 1: Discretize the Solution Domain

Subdivide the problem

Four elements
Five nodes
Accuracy increases with more nodes and elements
Element area
Constant cross section
Based on average area of nodes defining given element
ui : displacement distance of node i

Step 2: Develop a Shape Function

Model the Physical Behavior of each Element
For each element:

Average Normal Stress

F
=
A
Average Normal Strain

=

2
1.1 Elastic Equation Developement
If linearly elastic, then Hookes Law applies: = E

Substituting:  
F
=E
A
 
AE
F =

Since, for springs... F = kx
...we can see that elements are elastic springs with an equivalent stiffness or constant of...
AE
keq =

Element: Spring Model
Modeling each element as a spring: f = keq (ui+1 ui )
f Element Force
ui Displacement of node i
Aavg E
f= (ui+1 ui )

 
Ai+1 + Ai E
f= (ui+1 ui )
2
finally we get...

(Ai+1 + Ai )E
keq =
2
where:
keq Element stiffness
Length of each element
Ai , Ai+1 Cross-sectional areas at nodes i and i + 1

Step 3: Develop Element Equations

Free Body Diagram of each Element:

where... fi , fi+1 Forces on elements at nodes i and i + 1

Initial Matrix Assembly

Knowing:

fi = keq (ui ui+1 ) fi = keq ui keq ui+1

fi+1 = keq (ui+1 ui ) fi+1 = keq ui + keq ui+1

3
 Assemble in matrix form for each element:
    
fi keq keq ui
=
fi+1 keq keq ui+1

Note the form:

{Load Matrix} = [Stiffness Matrix]{Displacement Matrix}

Step 4: Assemble Global Stiffness Matrix

Element Stiffness Matrix:
For Element 1:  
1 k1 k1 u1
[K] =
k1 k1 u2
For Element 2:  
2 k2 k2 u2
[K] =
k2 k2 u3
...etc.
We can now assemble our Global Matrix Equation in either of two ways:

{Load Matrix} = [Stiffness Matrix]{Displacement Matrix}

OR

[Stiffness Matrix]{Displacement Matrix} = {Load Matrix}

Global Form of Element Matrix

u1

u2



{Displacement Matrix} = u3
u4




u5

Assemble pieces of Global Stiffness Matrix for each element:

k1 k1 0 0 0 u1
k1 k1 0 0 0 u2
1G

[K] = 0 0 0 0 0 u3
0 0 0 0 0 u4
0 0 0 0 0 u5

0 0 0 0 0 u1
0 k2 k2 0 0
u2
[K]2G

0 k2 k2
= 0 0
u3
0 0 0 0 0 u4
0 0 0 0 0 u5
...etc.

The Global Stiffness Matrix

Now combine ( i.e., [K] = [K]1G + [K]2G + [K]3G + [K]4G )

k1 k1 0 0 0
k1 k1 + k2 k2 0 0
G

[K] = 0
k2 k2 + k3 k3 0
0 0 k3 k3 + k4 k4
0 0 0 k4 k4
4
Step 5: Apply B.C. and Loads
Note that: u1 = 0, P = 1000. Therefore:

1 0 0 0 0 u1
0

k1 k1 + k2 k 2 0 0
u 2





0



0 k 2 k 2 + k 3 k 3 0 u = 0
3


0 0 k3 k3 + k4 k4 u 0
4


3
0 0 0 k4 k4 u5 10

(Note: Top row of Stiffness Matrix was changed so that u1 = 0)

Step 6: Solve Algebraic Equations

(Ai+1 + Ai )E
keq =
2
(0.21875 in2 + 0.25 in2 )(10.4 106 lb/in2 )
k1 = = 975 103 lb/in
2(2.5 in)
(0.1875 in2 + 0.21875 in2 )(10.4 106 lb/in2 )
k2 = = 845 103 lb/in
2(2.5 in)
(0.15625 in2 + 0.1875 in2 )(10.4 106 lb/in2 )
k3 = = 715 103 lb/in
2(2.5 in)
(0.125 in2 + 0.15625 in2 )(10.4 106 lb/in2 )
k4 = = 585 103 lb/in
2(2.5 in)

Solving for ui
Therefore [K]G :

1 0 0 0 0
975 975 + 845 845 0 0
G 3
lb
[K] = 10 0 845 845 + 715 715 0 in
0 0 715 715 + 585 585
0 0 0 585 585

Final matrix equation:

u1
0
u2

0



G
[K] u3 = 0
u 0

4





u5 1000

Note: UNITS must be consistent!!!

Solving...

u1

0

u2 0.001026



u3 = 0.002210 in
u4



0.003608




u5 0.005317

5
Step 7: Obtain Other Information
Stress in each element:
f
=
Aavg
keq (ui+1 ui )
=
Aavg
 
Aavg E
(ui+1 ui )
=
Aavg
 
ui+1 ui
=E

where:
f Element Force
ui Displacement of node
Note relationship to Hookes Law:
 
ui+1 ui
= E =

Calculating Normal Stress

(10.4 106 lb/in2 )(0.001026 in 0 in)
(1) = = 4268 lb/in2
2.5 in
(10.4 106 lb/in2 )(0.002210 in 0.001026 in)
(2) = = 4925 lb/in2
2.5 in
(10.4 106 lb/in2 )(0.003608 in 0.002210 in)
(3) = = 5816 lb/in2
2.5 in
(10.4 106 lb/in2 )(0.005317 in 0.003608 in)
(4) = = 7109 lb/in2
2.5 in
Reaction Force: Method 1
FBD Element 1:

F = k1 (u2 u1 )
X
FY = 0
X
FY = F + R = 0
X
FY = k1 (u2 u1 ) + R = 0
X
FY = k1 (u2 u1 ) + R = 0
R = -1000 lb

6
Reaction Force: Method 2
Draw FBD of each node
P
Write equilibrium equations for each node (i.e., F = 0)

Assemble into matrix equation of form...

{React. Matrix} = [Stiff. Matrix]{Displace. Matrix} {Ext. Load Matrix}

{R} = [K]{u} {F }

Solve for External Reactions:

R1
R2



{R} = R3
...





Ri

Method 2 Example: FBD

FBDs of nodes:

Method 2 Example: Equilibrium Equ.

Node 1: R1 k1 (u2 u1 ) = 0
Node 2: k1 (u2 u1 ) k2 (u3 u2 ) = 0
Node 3: k2 (u3 u2 ) k3 (u4 u3 ) = 0
Node 4: k3 (u4 u3 ) k4 (u5 u4 ) = 0
Node 5: k4 (u5 u4 ) P = 0

Method 2 Example: Matrix Form

Rearranging:
k1 u1 k1 u2 = R
k1 u1 + k1 u2 + k2 u2 k2 u3 = 0
k2 u2 + k2 u3 + k3 u3 k3 u4 = 0
k3 u3 + k3 u4 + k4 u4 k4 u5 = 0
k4 u4 + k4 u5 = 1000

7
In matrix form:

k1 k1
u1

R
k1 k1 + k2 k2 u2 0





k2 k2 + k3 k3 0
u3 =
k3 k3 + k4 k4 u 0
4



k4 k4 u5 103

Method 2 Example: Separate Forces

Note the form:

[Stiffness Matrix]{Displace. Matrix} = {Loads & Reactions}

Important to distinguish loads from reactionsloads are known,reactions are not!

Therefore, separate loads and reactions:

[K]{u} = {F } + {R}

OR

{R} = [K]{u} {F }

In this case solve for R:

P

0



{R} = 0
0





0

2 Total Potential Energy Formulation

Potential Energy
Xn m
X
(e)
= Fi ui
e=1 i=1

where:
Total Potential Energy for a body or system with n elements and m nodes
(e) Strain energy in each element (Lambda)
Fi External Force @ Node i
ui Displacement @ Node i
Therefore, the Potential Energy is the difference between the Total Strain Energy and the Work
done by the External Forces.

Work = F u
Not 1/2Fu, since force is a constantlike a weight moving over a distance

Strain Energy
1
Z
(e)
= dV
V 2
AND
1 2
Z
(e) = E dV
V 2
where V Volume of member
Therefore, the Strain Energy is the area under the curve multiplied by the material Volume

8
Min. Total Potential Energy
n m
X (e) X
= Fi ui = 0 i = 1, 2, . . . n
ui ui ui
e=1 i=1

For a stable system, the displacement at the equilibrium position occurs such that the value
of the systems total potential energy is a minimum.

This equation finds the displacement when the potential energy is at its minimum.

Min. Tot. Pot. Energy: Example

Find: Deflection of spring using:

(a) FBD and Statics
(b) Minimum Total Potential Energy

Example: Solution to (a)

FBD

X
MA = 0
X
MA = (6 lb)(5 in) (35 lb)(6 in) + Fs (8 in) = 0
Fs = 30 lb
Fs = kx
Fs 30 lb
x= = = 0.500 in
k 60 lb/in

9
Example: (b) Strain Energy
Elastic Energy for the spring is:

1 1
= kx2 = (60 lb/in)x2
2 2
= (30 lb/in)x2
Beam displacement:

Example: (b) Work

Similar Triangles:
x xG
=
8 5
5
xG = x
8
x xB
=
8 6
3
xB = x
4
Work of External Forces:
   
X 5 3
Fi ui = (6 lb)xG + (35 lb)xB = (6 lb) x + (35 lb) x
8 4
X 15 lb 105 lb 120 lb
Fi ui = x+ x= x
4 4 4
X
Fi ui = (30 lb)x

Example: Solution to (b)

Total Potential Energy of the System
X X
= Fi ui

= (30 lb/in)x2 (30 lb)x

Minimize

= (30x2 30x) = 0
x x
60x 30 = 0
x = 0.500 in

10
Example: Tapered Bar

Find:
Displacement (using the Minimum Total Potential Energy Formulation)

Tapered Bar: Strain Energy

Strain Energy:
1 2
Z
(e)
= E dV
V 2
1 2
Z
(e) = E dV
2 V
1
(e) = E2 V
2
From before, strain for one element:
(ui+1 ui )
=

And
V = Aavg

 2
(e) 1 ui+1 ui
= E (Aavg )
2
Aavg E 2
(e) = (ui+1 + u2i 2ui+1 ui )
2
Tapered Bar: Min. Strain Energy
Minimizing Strain Energy:

(e)
 
Aavg E 2
= (ui+1 + u2i 2ui+1 ui )
ui ui 2

(e) Aavg E
= (0 + 2ui 2ui+1 )
ui 2
(e) Aavg E
= (ui ui+1 )
ui
Also...
(e) Aavg E
= (ui+1 ui )
ui+1

11
Tapered Bar: Matrix Form and Forces
In Matrix Form ( (e) ) 
 
ui keq keq ui
(e)
=
ui+1
keq keq ui+1

Where:
Aavg E
keq =

Work of External Forces: Fi ui Fi+1 ui+1
Minimizing External Work:

(Fi ui ) = Fi
ui

(Fi+1 ui+1 ) = Fi+1
ui+1

Tapered Bar: Min. Tot. Pot. Energy

Minimum Total Potential Energy:
n m
X (e) X
= Fi ui = 0 i = 1, 2, 3, . . . n
ui ui ui
e=1 i=1

[K](G) {u} {F } = 0

k1 k1 0 0 0
k1 k1 + k2 k2 0 0
G

[K] = 0
k2 k2 + k3 k3 0
0 0 k3 k3 + k4 k4
0 0 0 k4 k4

Tapered Bar: Solution

u1

u2


{u} = u3
u4




u5



0

0



{F } = 0
0




103

Row 1 of [K] matrix will change to

1 0 0 0 0 u1
0

u 0




2


u3 =
0


u 0
4


3
u5 10

...due to B.C., u1 = 0 Same solution as Direct Formulation!!

12