comparison of recorded and simulated ground motions

35
Comparison of Recorded and Simulated Ground Motions Presented by: Emel Seyhan, PhD Student University of California, Los Angeles Collaborators: Lisa M. Star, PhD Candidate, University of California, Los Angeles Robert W. Graves, PhD, USGS Jonathan P. Stewart, PhD, PE, University of California, Los Angeles

Upload: usoa

Post on 23-Feb-2016

34 views

Category:

Documents


0 download

DESCRIPTION

Comparison of Recorded and Simulated Ground Motions. Presented by: Emel Seyhan , PhD Student University of California, Los Angeles Collaborators: Li sa M. Star , PhD Candidate, University of California, Los Angeles Robert W. Graves , PhD, USGS - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Comparison of Recorded and Simulated Ground Motions

Comparison of Recorded and Simulated Ground Motions

Presented by:Emel Seyhan, PhD StudentUniversity of California, Los Angeles

Collaborators:Lisa M. Star, PhD Candidate, University of California, Los AngelesRobert W. Graves, PhD, USGSJonathan P. Stewart, PhD, PE, University of California, Los Angeles

Page 2: Comparison of Recorded and Simulated Ground Motions

OutlineMotivationHybrid Simulation ProcedureValidation Analysis & Results

Distance scalingStandard deviation

Calibration of Hybrid Simulation ProcedureDistance attenuationStandard deviation

Conclusions

Page 3: Comparison of Recorded and Simulated Ground Motions

Motivation

Broadband motions for response history analysis

Some (M, R) ranges poorly sampled by recordings

Motions needed with specific attributes, e.g.Basin effectNear fault effects

Page 4: Comparison of Recorded and Simulated Ground Motions

Motivation

Broadband motions for response history analysis

Some (M, R) ranges poorly sampled by recordings

Motions needed with specific attributes, e.g.Basin effectNear fault effects

Simulations hold potential to provide useful ground motions for engineering application in these situations

Page 5: Comparison of Recorded and Simulated Ground Motions

ShakeOut Scenario Description

Moment magnitude 7.8 earthquake 150 yr return period (last events 1857 & 1680) Evaluated for three different possible hypocenters

Hughes Lake

San Gorgonio Pass

Bombay Beach

Page 6: Comparison of Recorded and Simulated Ground Motions

Puente Hills ScenarioDirectly under down

town Los Angeles7.15 Mw EarthquakeBuried reverse fault

Page 7: Comparison of Recorded and Simulated Ground Motions

Simulation Procedure

Hybrid proceduref<1 Hz: physics based

Physics-based

Page 8: Comparison of Recorded and Simulated Ground Motions

Simulation Procedure

Hybrid proceduref<1 Hz: physics basedf>1 Hz: stochastic

Stochastic

Reference: Graves et al, 2004

Page 9: Comparison of Recorded and Simulated Ground Motions

Simulation Procedure

Hybrid proceduref<1 Hz: physics basedf>1 Hz: stochastic

Reference: Graves et al, 2004

Page 10: Comparison of Recorded and Simulated Ground Motions

Simulation ProcedureHybrid procedureSource function

Kinematically prescribed source model

Slip distributionRupture velocity

ShakeOut, Mw 7.8

Page 11: Comparison of Recorded and Simulated Ground Motions

Hybrid procedureSource functionSemi-empirical site term (fn of Vs30)

Simulation Procedure

Page 12: Comparison of Recorded and Simulated Ground Motions

Distance Attenuation

Page 13: Comparison of Recorded and Simulated Ground Motions

Calibration AnalysisApproachCalculate residuals

4 GMPEs: AS, BA, CB, CYRandom effect analysis: Separate event term (hi) from within-event residual (ei,j)

Distance-scaling evaluated from (ei,j)

, ,i j i i jR h e

i a sim,i a GMPE,iR (T)=ln(S (T)) -ln(S (T))

Page 14: Comparison of Recorded and Simulated Ground Motions

Calibration Analysis

recording "j" of eqk "i"

i

Genereal Model

ji

General Model

ei,j = Ri,j - hi

Page 15: Comparison of Recorded and Simulated Ground Motions

Intra-event Residuals

Page 16: Comparison of Recorded and Simulated Ground Motions

Intra-event Standard Deviation

s too low for T < 1.0 sLarge transition at T=1.0 s

s=stdev(e)

Page 17: Comparison of Recorded and Simulated Ground Motions

Calibration of Hybrid Simulation ProcedureFocus on high frequency stochastic modelControlling parameters

Source parameters: Stress drop, slip function, rise time, rupture velocity

Path parameters: Distance, crustal velocity & damping (Q)Site parameters: Near surface crustal velocity, shallow site

term (Vs30)Parameter selected for remove distance attenuation biasProcedure to increase intra-event standard deviation

Page 18: Comparison of Recorded and Simulated Ground Motions

ScopeDistance attenuation calibration

Strike slip fault M5, 6.5, 7.25 and 8Distributed arrays

M5 M6.5 M7.25 M8

Page 19: Comparison of Recorded and Simulated Ground Motions

Slip modelsFor M5, 6.5, 7.25 and 8Random slips

M5

M6.5

Page 20: Comparison of Recorded and Simulated Ground Motions

M7.25

M8

Page 21: Comparison of Recorded and Simulated Ground Motions

Various levels of crustal damping, QLow Qo (a=25)Mid Qo (a=41)High Qo (a=57)

Q (f) = Qo*fn

(n = 0.6)

Qo = a + b*Vs

(b = 34)

ShakeOut

Page 22: Comparison of Recorded and Simulated Ground Motions

Verification of Hybrid Trends Using Stochastic Part OnlyUsing same level of Q

(Low Qo) Original ShakeOutThis study (M8) similar

trend with previous work esp. beyond about 10 km

Page 23: Comparison of Recorded and Simulated Ground Motions

Removing Distance Attenuation Bias

Comparing different level of Q (M7.25) Using low Qo

Using high Qo

Page 24: Comparison of Recorded and Simulated Ground Motions

Removing Distance Attenuation Bias

Residuals for different level of Q (M7.25) Using low Qo

Using high Qo

Page 25: Comparison of Recorded and Simulated Ground Motions

Removing Distance Attenuation Bias

Fit semi-log line to residuals of average ground motions

For different level of Q Using low Qo

Using high Qo

Repeat for all M, GMPEs, IMs

Y = c*ln(X) + d

Page 26: Comparison of Recorded and Simulated Ground Motions

Removing Distance Attenuation Bias

Slope of residuals of average ground motionsScatter based on all

gmpesUsing low Qo

Using high Qo

PGA

Page 27: Comparison of Recorded and Simulated Ground Motions

Removing Distance Attenuation Bias

Slope of residuals of average responses

Using low Qo

Using high Qo

Page 28: Comparison of Recorded and Simulated Ground Motions

Intra-event scatter calibrationIncreasing intra-event

standard deviationRandomized velocity Randomized Fourier

AmplitudeRandomized Q

Page 29: Comparison of Recorded and Simulated Ground Motions

ApproachModify parameters e.g.

Velocity profile

Intra-event scatter calibration

Rand CaseNonRand CaseBA08

Page 30: Comparison of Recorded and Simulated Ground Motions

ApproachRandomization of

Fourier AmplitudeAdding variation

Intra-event scatter calibration

Page 31: Comparison of Recorded and Simulated Ground Motions

ApproachRandomization of

Fourier AmplitudeAdding variation

Intra-event scatter calibration

Rand CaseNonRand CaseBA08

Page 32: Comparison of Recorded and Simulated Ground Motions

Concluding RemarksCalibrated simulation procedures needed for

engineering practiceValidation process reveals:

Faster distance attenuation at shorter periodsLow intra-event standard deviation T<1s

Page 33: Comparison of Recorded and Simulated Ground Motions

Cont’dCalibration process reveals:

Possible to get slower distance attenuation by using higher Q

Randomization of Fourier Spectrum gives better results than randomization of velocity

Page 34: Comparison of Recorded and Simulated Ground Motions

More? Implementation fully hybrid simulation with revised

Q and Vs

Page 35: Comparison of Recorded and Simulated Ground Motions

Thank you