EGM6365: Structural Optimization EGM6365: Structural Optimization
EGM6365: Structural Optimization EGM6365: Structural Optimization
The course can be tailored to some extent The course can be tailored to some extent
for the audience: Take advantage!
Class heckler(s) needed: Reward is Class heckler(s) needed: Reward is
negotiable
Tour of web page Tour of web page
www.mae.ufl.edu/haftka/stropt and E-
learning http://lss.at.ufl.edu/ g p
Structural optimization problems Structural optimization problems
Topology optimization: Topology optimization:
Where are the holes?
Finite element topology possibly not defined Finite element topology possibly not defined
Shape optimization:
What are the boundaries? What are the boundaries?
Element shapes change during optimization
Sizing optimization: Sizing optimization:
How thick it is? In what order do I arrange
plies? p
Finite element model is fixed
Function vs. parameter
i i i optimization
.
Standard formulation
Design variables
1
x
x
2
{ , , , }
2
T
1 n
x
x x x
= =
`
)
x
Objective and constraints
n
x
)
Minimize ( )
Subjectedto ( ) 0
f
h j 1 n
x
x Subjected to ( ) 0 , ,
( ) 0 , ,
j e
k g
h j 1 n
g k 1 n
= =
s =
x
x
Numerical solution tools Numerical solution tools
Matlab fmincon: Uses an algorithm called g
sequential quadratic programming, suitable for
local search of continuously differentiable
problems problems
Microsoft Excels Solver utility. Uses
Generalized Reduced Gradient for similar Generalized Reduced Gradient for similar
problems, but can deal also with integer
variables
Learn to use them! Do first homework
assignment!
What has driven the teaching of
l i i i ? structural optimization?
Computation has become faster allowing Computation has become faster, allowing
us to solve more complicated problems
Non differentiable functions Non-differentiable functions
Global, not local optimization
More expensive structural analysis More expensive structural analysis
Software has become more widely
il bl d d d bl available and more dependable
Less concern with optimization algorithms
What has Moores lawdone for us What has Moores law done for us
Recent Past:
30 000 DOF
15-20 years ago:
200 DOF
Current:
800 000 DOF
Recent Past:
30 000 DOF
15-20 years ago:
200 DOF
Current:
800 000 DOF
Recent Past:
30 000 DOF
15-20 years ago:
200 DOF
Current:
800 000 DOF
30,000 DOF
200 DOF
800,000 DOF
30,000 DOF
200 DOF
800,000 DOF
30,000 DOF
200 DOF
800,000 DOF
The three axes of complexity The three axes of complexity
P
F
Probabilistic discrete/ global optimization
Probabilistic Continous Optimization
P
F
Probabilistic discrete/ global optimization
Probabilistic Continous Optimization
O
N
Y
C
E
Deterministic Continuous optimization
Deterministic Discrete / Global optimization
O
N
Y
C
E
Deterministic Continuous optimization
Deterministic Discrete / Global optimization
L X
I
T
Y
O
P
T
I
M
I
Z
A
T
I
O
C
O
M
P
L
E
X
I
T
Y
B
D
L X
I
T
Y
O
P
T
I
M
I
Z
A
T
I
O
C
O
M
P
L
E
X
I
T
Y
B
D
M
O
D
E
L
C
O
M
P
L
E
X
I
ANALYSIS
COMPLEXITY
OC
O
A M
O
D
E
L
C
O
M
P
L
E
X
I
ANALYSIS
COMPLEXITY
OC
O
A
Examples Examples
(A) Model and Analysis Complexity (A) Model and Analysis Complexity
Z
C
(A) Model and Analysis Complexity
e.g., Complex 3-D structure requiring
non-linear analysis
(B) Model and Optimization
Complexity
Z
C
Z
C
(A) Model and Analysis Complexity
e.g., Complex 3-D structure requiring
non-linear analysis
(B) Model and Optimization
Complexity
T
I
O
N
B
P
Complexity
e.g., Probabilistic design of a complex
3-D structure
(C) Analysis and Optimization
Complexity
T
I
O
N
B
P
T
I
O
N
B
P
Complexity
e.g., Probabilistic design of a complex
3-D structure
(C) Analysis and Optimization
Complexity
O
P
T
I
M
I
Z
A
T
O
D
E
L
Y
O
Complexity
e.g., Non-deterministic optimization with
non-linear analysis (or combined macro
and micro mechanical failure models
(P) Model Analysis and Optimization
O
P
T
I
M
I
Z
A
T
O
D
E
L
Y
O
O
P
T
I
M
I
Z
A
T
O
D
E
L
Y
O
Complexity
e.g., Non-deterministic optimization with
non-linear analysis (or combined macro
and micro mechanical failure models
(P) Model Analysis and Optimization
ANALYSIS
M
O
D
A
X
(P) Model, Analysis and Optimization
Complexity
e.g., Non-deterministic optimization of
Complex 3-D structure requiring
nonlinear analysis
ANALYSIS
M
O
D
A
X
ANALYSIS
M
O
D
A
X
(P) Model, Analysis and Optimization
Complexity
e.g., Non-deterministic optimization of
Complex 3-D structure requiring
nonlinear analysis yy