A Genetic Algorithm Solution for the Problem of Selection and Scaling of Ground Motion Records...
-
date post
22-Dec-2015 -
Category
Documents
-
view
214 -
download
0
Transcript of A Genetic Algorithm Solution for the Problem of Selection and Scaling of Ground Motion Records...
A Genetic Algorithm Solution for the Problem of Selection and
Scaling of Ground Motion Records
Arzhang Alimoradi and Farzad NaeimJohn A. Martin & Associates
First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
2
Agenda
• Why Genetic Algorithms?– Formulation of an optimal search and scaling
method (minimal or no scaling)• Example I: SAC-20 LA• Example II: PEER-GMSM• Concluding remarks
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
3
Why GA?
Different philosophies with the analysis of complex systems:
1. Stochastic analysis, non-stationary indeterministic processes, applied with ‘reductionism’
2. Analysis of complex systems (critical, chaotic) Evidence for deterministic process (Olson and Allen, Nature,
Vol. 438, pp. 212-215, November 2005) Physically critical systems (Corral, Physical Review Letters, Vol.
92, No. 10, March 2004)
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
4
1. Fault location, type, size and activity, strain rate
2. Rupture dynamics (slip rate, seismic moment, stress drop,
directivity, mechanism,…) 3. Path (Q, topology, distances,
refraction and reflection,…)4. Local site effects (amplification,
filtering, nonlinear soil deformations,…)
5. Basin effects6. Soil-foundation-structure interaction 7. … ?!?!?!
6 DOF Time Series of Point Displacement
Earthquake Ground Motion Process (Complex System)
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
6
Chromosomes
Nucleus
Optimization through adaptation!
Genetic algorithms: what does biology have to do with ground
motion selection?
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
7
(Genotype) (Phenotype)
(Structure) (Solution Alternative)
(Environment)
(Problem)
[0110100101][1110101101][0000110011] {Optimal Variables}
Adaptation through generations
(Genotype) (Phenotype)
(Structure) (Solution Alternative)
(Environment)
(Problem)
[0110100101][1110101101][0000110011] {Optimal Variables}
Adaptation through generations
GA BasicsN
atur
al S
yste
mA
rtif
icia
l Sys
tem
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
8
A population of five chromosomes:
10111001100010100110111010011000101001110010101110
01100111011101001000111100101111010111000111000101
00101001100010100110001010011000101001100000100110
00101001110010110111001010011100101101101010110110
10111101100010100110001111111000101001100110100010
00100001110010100000001110011000101001100000010110
11101001100010100110001010011001101001100010110000
00100111011101000010110001110101000111001110100111
00101001110010100111001010101000101001100110101111
00100111011101001000111001110101010111010010100100
Gene 1: Gene 1: record 1 record 1 and the and the scalingscaling
Gene 2: scaling set 2Gene 2: scaling set 2Gene 3-5: …Gene 3-5: …
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
9
211080.3
Total number of different combinations =
8010
The number of stable particles in the universe =
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
10
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
12
SAC-20 LA, Post-Northridge
• 3 Dimensional inelastic model in CSi-Perform
• Prof. Krawinkler’s M2 Model (with panel zone distortions are modeled)
• Gravity beam-to-column connections are pinned
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
13
Comparison of Modal Properties
Krawinkler’s M2 Perform-3D
T1, sec. 3.65 3.64
T2, sec. 1.26 1.28
T3, sec. 0.72 0.74
Effective Modal Mass 1 80.1 78.2
Effective Modal Mass 2 11.8 11.3
Effective Modal Mass 3 3.1 3.9
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
14
Denali Pump Station #10 -051 Denali Pump Station #10 -321
Imperial Valley El Centro Array Station 6 -140
Imperial Valley El Centro Array Station 6 -230
Landers Yermo EW Landers Yermo NS
Loma Prieta Anderson Dam 270 Loma Prieta 360
Loma Prieta Hollister EW Loma Prieta Hollister EW
Northridge Newhall EW Northridge Newhall NS
Northridge Sylmar County Hospital Parking Lot 360
Northridge Sylmar County Hospital Parking Lot 90
Selected Bin
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
15
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
16
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
18
Estimate I
• Cumulative distribution function of maximum interstory drift ratio conditioned on a given earthquake event on a site with a given 30m soil shear-wave velocity. (To be used for distribution and mean evaluations)
n
i
t
trecord etT
record dttSa
tSaZ
27
1 arg
1min
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
19
Estimate I
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
20
Estimate II
• Median of maximum interstory drift ratio (To be used only for the evaluation of the arithmetic mean of response)
n
i
t
tetT
recordrecord
dttSa
tSa
Z 17
minarg
7
1
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
21
Estimate II
Oct. 27, 2006 First PEER Workshop on Ground Motion Selection and Modification for Nonlinear Analysis
22
Conclusions
• Need for a consistent theory with regards to selection and scaling,
• Scaling scheme shall not distort the spatial distribution of response when compared to ‘conforming records,’
• A computational tool was presented for efficient search and scaling of ground motion records in a database.