You are on page 1of 18

A SPACE-TIME FINITE

ELEMENT MODEL
FOR DESIGN AND
CONTROL OPTIMIZATION
OF NONLINEAR DYNAMIC
RESPONSE


J. E. Barradas Cardoso
1
, P. Pires Moita
2
, and Anbal J. J. Valido
2


1
Instituto Superior Tcnico, Departamento de Engenharia Mecnica
Av. Rovisco Pais, 1049-001 Lisboa , Portugal

2
Escola Superior de Tecnologia, Instituto Politcnico de Setbal
Campus do IPS, Estefanilha, 2914-508, Setbal, Portugal

Introduction

Integrated methodology
structures and flexible mechanical systems
optimal design and optimal control
Bound formulation is used to solve both the
multicriterium and unicriterium problems


Response Analysis
Space discretization
Virtual work dynamic equilibrium equation



After Finite element discretization
0 0
( )
t t t t t t t t
W dV d o o o o o I = +
} }
u u S f u Q u c
t t t t t t
S S
M U + C U + K U = P
Response Analysis
Time discretization
e e e
S
= D z P
11 12 11 12
21 22 21 22
,
t t t
S S nxn nxn nxn
e t t t t t t e
S S S nxn S S nxn 1
t t t t t t
S S nxn S S nxn 2
A A
A A
+ +
+ +
(
(
= =
`
(

(
)
0 0 0
0 0
0 0
K C M P
D D D D D P P
D D D D P
( , ), ( , )
e t t t t t t t A +
= = z z z z U, U U
After the boundary conditions are applied it becames
Response Analysis

Solved at-once
Time assembly

S
= D z P

,
S c c
= = K U P P P DU

Performance functional


Total variation



Design Sensitivity Analysis Problem
( , , )
t t
G t dt + =
}
z b
o+ o+ o+ = +
Design Sensitivity Analysis Problem
Adjoint system method
adjoint fields replace the arbitrary variation of state fields by into the
virtual work equation as


define an extended 'action' function


replace the implicit design variations of the state fields by explicit design
variations and adjoint fields, then determining these adjoint fields by
vanishing the implicit design variation of the 'action' function as

total design variation of +

( ) 0
a a
S
W = = K U - P U
a
A W + =
0 A o =
A =
Design Sensitivity Analysis Problem



The sensitivities are firstly performed at the element
level and then the sensitivity equations are
assembled and time boundary equations imposed







Optimum Design Problem
A bound formulation is used to solve the
multicriterium and unicriterium problems



The diferent objectives are normalized
accordingly to


0
min , s.t.
, ( 1, 2,..., );
0, ( 1, 2,..., ); 0
0k
T
j
k m
j m
|
+ |
+ u
s =
s = =
( ) ( ), 1
min max
0i 0i 0i 0i
0i i i
w w = =

Examples
Nonlinear Impact
Absorber
v(t=0)=1;
M=1;
v=q=2
Decision variables:







M
K=K
0
| u |
v1
C=C
0
| u
.
|
q1
u
P(t)

0 0
; ;[ / ), 0 12] K C P t t s s
Case I



Starting design:



Optimal design:

Examples
0 0
0,5; 0,5;[ / ) 0, 0 12] K C P t t = = = s s
max
min . .
1
a s t
u s
0 0
0, 00667; 6,96 K C = =

-0,60
-0,40
-0,20
0,00
0,20
0,40
0 2 4 6 8 10 12
t[s ]
P
[
N
]
P0,5/0,5
P0,6/0,4
P0,7/0,3
P0,8/0,2
P0,9/0,1
-0,2
0,0
0,2
0,4
0,6
0,8
1,0
0 2 4 6 8 10 12
t[s ]
P

[
N
]
Figure 4. Optimal control for impact absorber (Case I). Figure 5. Optimal control for impact absorber (Case II)
.
Examples

-0,8
-0,6
-0,4
-0,2
0,0
0,2
0,4
0 2 4 6 8 10 12
t[s]
a
[
m
/
s
2
]
starting
Ko, Co
Ko, Co+control
Figure 2. Impact absorber max. acceleration (Case I).
-0,8
-0,6
-0,4
-0,2
0,0
0,2
0,4
0 2 4 6 8 10 12
t[s]
a
[
m
/
s
2
]
a start
a0,5/0,5
a0,6/0,4
a0,7/0,3
a0,8/0,2
a0,9/0,1
Figure 3. Impact absorber max. acceleration (Case II).
Acceleration response
Case II



Starting design:


Normalized weighted bound formulation

Optimal design

Examples
0 0
0,597; 0,597;[ / ) 0, 0 12] K C P t t = = = s s
2 2 2
0
min{ , (100 100 ) } . .
1
C E u u P dt s t
u
= + +
s
}
w
E
/w
Co
K
Ooptimu
m

C
Ooptimu
m

0.5/0.5 0.458 0.015
0.6/0.4 0.453 0.022
0.7/0.3 0.450 0.032
0.8/0.2 0.446 0.050
0.9/0.1 0.442 0.091

-0,60
-0,40
-0,20
0,00
0,20
0,40
0 2 4 6 8 10 12
t[s ]
P
[
N
]
P0,5/0,5
P0,6/0,4
P0,7/0,3
P0,8/0,2
P0,9/0,1
-0,2
0,0
0,2
0,4
0,6
0,8
1,0
0 2 4 6 8 10 12
t[s ]
P

[
N
]
Figure 4. Optimal control for impact absorber (Case I). Figure 5. Optimal control for impact absorber (Case II)
.
Examples
0.00
0.02
0.04
0.06
0.08
0.10
53 55 57 59 61
E
C
o
Criterium Space
Examples
Vehicle Suspension
Sinusoidal obstacle of height
0,1016m and wavelenght
of 24,384m

minimize the maximum
acceleration magnitude
such that the relative
vertical displacement
between masses is not
larger than 0.05 m

c3
c5
1
m4 m3
m1
k1 c1
k2 c2 k3
k4 c4 k5
2 6
4
9
11
3 7
5
8
10
Beam element 1 Beam element 2
800kg 400kg
Examples

Starting design
A = 0.1 m2, k1 = 1.75E4, k2 = k3 = 5.26E4 N/m,
c1 = 1.75E3, c2 = c3 = 4.38E3 Ns/m

Optimal design
A = 0.35579 m
2
, K1=6.94E4 N/m, K2=6.31E4 N/m, K3=5.02E4
N/m, C1=1.E6 Ns/m, C2=1.34E4 Ns/m, C3=1.E2 Ns/m
-1.4E-03
-1.2E-03
-1.0E-03
-8.0E-04
-6.0E-04
-4.0E-04
-2.0E-04
0.0E+00
2.0E-04
0.0 0.2 0.4 0.6 0.8 1.0
t[s]
P
[
N
]
| |
( ) 0., 0 1 P t t s = s s
Examples
Driver seat max. acceleration [m/s
2
]

-6.0E-05
-5.0E-05
-4.0E-05
-3.0E-05
-2.0E-05
-1.0E-05
0.0E+00
1.0E-05
2.0E-05
0.0 0.2 0.4 0.6 0.8 1.0
t[s]
a[m/s
2
]
Initial Design
Optimal design wrt area
Optimal design wrt area, Ks and Cs
Optimal design wrt area, Ks and Cs + control
CONCLUDING REMARKS
Current formulation unifies in one single formulation the problems of
optimum design and optimal control, treating the design variables as
control variables that do not change with time.

Using at-once integration of the equations of motion and its
sensitivities, the adjoint method of design sensitivity analysis can be
used with all of its advantages and without its main disadvantage in
dynamics: the necessity of backwards integration of the adjoint
system. So, the needing of backwards integration with its major
drawback - the memorization of the response history - is not
associated with path-dependent problems, as it is frequently
assumed, but it is associated with the fact of using step-by-step
integration.

Temporal boundary conditions can be applied at any point in time,
not necessarily at the start of the integration, as it is usual for the
step-by-step integration method.