Beruflich Dokumente
Kultur Dokumente
OpenSees-SNOPT Framework
SNOPT (Sparse Nonlinear OPTimization)
Specific
optimization tool
Specific
application/purpose
of optimization
General purpose
Design point search
optimization
Example 1: Structural Optimization of an Electrical
Transmission Tower
Serviceability: F = 25 kN
Ultimate: F = 100 kN
¾ Design variables
(1) Cross-section Area A: in range [8.0e-4, 1.6e-2] m2 , initial 8e-3 m2
(2) Cross-section Area B: in range [3.0e-4, 6.0e-3] m2 , initial 3e-3 m2
(3) Cross-section Area C: in range [2.0e-4, 4.0e-3] m2 , initial 2e-3 m2
¾ Minimize the total cost (or volume) of the tower such that
(1) when F = 25 kN, umax < 1.50 cm (at the top of the tower)
(2) when F = 100 kN, umax < 15.0 cm (at the top of the tower)
¾ Optimal design
(1) A = 3.17e-3 m2, B = 3.51e-4 m2, C = 2.00e-4 m2
(2) Total volume = 0.274 m3 (compared with initial volume
of 1.20m3 ).
Optimization Results
(Iteration # 40)
1.2
0.8
0.6
0.4
0.2
0 10 20 30 40
Example 2: Nonlinear FE Model Updating of Soil
Column
¾ 2D soil column modeled by three layers of pressure-
independent multi-yield surface J2 soil plasticity models
-6m
Surface
z -11m
Accel. [m/s2]
y -6 m
x -17m
-11 m
σ xz
Time [sec]
τ xz
Objective & Constraint Functions and Results
¾ Objective Function: ¾ Method 1: Gradient by FFD
# stations ⎛ # time steps 2⎞
1 FE exp ⎟ ∂F ΔF
F = ∑ ⎜⎜ ∑ ⎜ ( u j,ti ) ⎟⎟
u j,ti −
≈
∂θ Δθ
j=1 ⎝ i=1 2 ⎠⎟
20, 000 < G i < ∞ 20 < τi < ∞ ∂F # stations ⎛⎜ # time steps u j,t i ⎞⎟
∂
= ∑ ⎜ ∑ ( u j,t i )
u j,t i − exp
⎟⎟
∂θ j=1 ⎝
⎜ i=1 ∂θ ⎠⎟
¾ Results obtained by OpenSees-SNOPT:
Param. G1 (kPa) G2 (kPa) G3 (kPa) τmax,1 (kPa) τmax,2 (kPa) τmax,3 (kPa)
True/Ref
28,800 39,200 57,800 31 33 34
Value
Initial
30,000 30,000 30,000 30 30 30
value
SNOPT
28,797.40 38,941.76 57,977.04 31.145 33.660 34.156
(FFD)
SNOPT
28,800.03 39,199.89 57,800.08 31.0 33.0 34.0
(DDM)
Comparison of Ground Acceleration after FE
model Updating (DDM)
¾ Comparison between ‘experimental’ (reference) and FE predicted ground surface
accelerations: (a) before FE model updating, (b) after FE model updating
(a)
ug [m / s 2 ]
(b)
ug [m / s 2 ]
Time [sec]
F= F = 2.8E-4
398.10
Initial value of Final value of
objective function objective function
Convergence Process (DDM versus FFD)
⎛
# stations # time steps 2⎞
u j,ti ) ⎟⎟⎟
1
F= ∑ ⎜⎜
⎜⎝ ∑ ( u j,ti −
exp
F j=1 i=1 2 ⎠⎟
Iteration #
Note: FE model updating converges MUCH FASTER using DDM versus FFD.
Example 3: FE Reliability Analysis of R/C Frame
Structure
7.0 7.0 (unit: m)
P C C
3.6
B A B
P/2 C C
7.2
B A B
z
z z (unit: cm)
30 25 20 25 20 20
A B C
25
25
25
y y
20
y
20
20
utop = 0.144m
Example 4: Probabilistic Performance-based Optimum
Seismic Design
k0 = 100, 000 kN/m Objective function : f k0 , F y ( )
Fy = 14, 000 kN Starting Point
= ∑| ν
i
LT (k0 , F y ) - ν LObj
T
|2
F SDOF Model
(Menegotto-Pinto) PBEE
Fy 1 b ⋅ k0 Analysis
ν LObj ν LT ( k0 , Fy )
m = 6,150 tons T
b = 0.10
k0
1 ξ = 0.02
U y = 0.075 m U
A priori selected optimum
design parameters
•Objective Function
Zoom
Optimization Results
v EDP=μD ( k 0* , Fy * )
Xstart =
[100,000kN/m,
X* = 14,000 kN]
[137,200kN/m,
10,290 kN]
Xend =
[135,774kN/m, Demand Hazard Curve
10,038 kN]
Objective Function
20
Current research based on application
of OpenSees-SNOPT
Investigation of Seismic Isolation for CHSR Prototype Bridge
¾ Probabilistic Performance-based
Optimization of SIS
Schematic View of 110ft-Span Prototype Bridge
Expansion
Joint
23
Transversal Section Around Expansion Joint Continuous Joint
Gu, Q., Barbato, M., Conte, J. P., Gill, P. E., and McKenna, F., “OpenSees-
SNOPT Framework for Finite-Element-Based Optimization of Structural and
Geotechnical Systems,” Journal of Structural Engineering, ASCE, 138(6), 822-
834, 2012.
Thank you !