Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

33
Determining NearFault FaultNormal And FaultParallel SiteSpecific Pulse Response Spectra And Selecting/Scaling SiteSpecific FaultNormal And FaultParallel Motions Including Simulated Motions & Spectral Matching Programs That Acceptably Include Pulse Motions

Transcript of Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Page 1: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Determining Near‐Fault Fault‐Normal And Fault‐Parallel Site‐Specific Pulse 

Response Spectra And Selecting/Scaling Site‐Specific Fault‐Normal And Fault‐Parallel Motions Including Simulated 

Motions & Spectral Matching Programs That Acceptably Include Pulse Motions 

Page 2: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Comments on Near‐Fault Ground Motions

N. AbrahamsonPG&E 

COSMOS Annual Meeting, Nov 14, 2014

Page 3: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Disclaimer

• Topic of the meeting is new Chapter 16 of ASCE 7‐2014

• My comments may not fit into the new Chapter 16 framework

Page 4: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Topics• Near‐Fault Effects – Velocity Pulses– Directivity– Fling

• Deterministic near‐fault response spectra– Accounting for directivity and fling

• Selecting near‐fault time histories– How many have pulses?– What is the appropriate pulse period?– FN & FP directions or principal axes?– Use of finite‐fault simulations

• Spectral matching for pulse ground motions

Page 5: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Near Fault Effects

• Directivity– Related to the direction of the rupture front• Forward directivity: rupture toward the site

(site away from the epicenter)• Backward directivity: rupture away from the site  (site near the epicenter)

• Fling– Related to the permanent tectonic deformation at the site

Page 6: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Velocity Pulses• Forward Directivity– Two‐sided velocity pulse due to constructive interference of SH waves from generated from parts of the rupture located between the site and epicenter• Constructive interference occurs if slip direction is aligned with the rupture direction

• Occurs at sites located close to the fault but away from the epicenter for strike‐slip

• Fling– One‐sided velocity pulse due to tectonic deformation– Occurs at sites located near the fault rupture independent of the epicenter location

Page 7: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Directivity Parameters for Strike‐Slip Faults

Page 8: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Abrahamson (2000) Directivity Factors 5% damping, Ave Horiz, Strike‐Slip

Page 9: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Directivity 2008 models(not centered on GMPE data) 

Centering addressedby Choiu and Youngs2014 GMPE

Page 10: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Directivity and Deterministic GM

• How to include directivity?– Choose a worst‐case hypocenter?– Choose an representative hypocenter?

• Based on deaggregation of hazard– Randomize hypocenters?• But doing probabilistic

Page 11: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Parameters for Simplified Directivity for NGA‐West2 

Model Parameters

Strike‐Slip:Rupture LengthRx, Ry

Dip‐Slip:Rupture WidthRx, Ry

Page 12: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Watson‐Lamprey (PEER Report 2014) Simplified Directivity for NGA‐West2 • Model Results– Correction to median GMPE (ln units)– Correction to the standard deviation (ln units

• Use of Results– Compute new median and standard deviation– Compute new 84th percentile ground motion

MedianwithDirectivityGMPEmed exp Med _Fac(Rx,Ry,L,W ,Dip)

SigmawithDirectivity GMPE2 Sig_Fac Rx,Ry,L,W ,Dip

Page 13: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Example Directivity FactorsSS, M7, Rx=3 km, L=80 km

Page 14: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Example 84th Percentile with Directivity Correction: SS, M7, Rx=3 km

Page 15: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Example 84th Percentile with Directivity Correction: SS, M8, Rx=3 km

Page 16: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Fling Effects

• Typically not addressed• Fling effects on the response spectral amplitudes are captured by NGA‐W2 GMPEs for periods up to 5 sec– Discussed at COSMOS 2012

• We know fling effects will happen for sites located close to the rupture of larger earthquakes• There will be a permanent displacement across the fault that will take 

2‐10 seconds to slip for M>7– Methods for adding fling effects

• Kamai et al (2014) – BBSA• If fling effects are added, check the amplitude at periods < 5 seconds 

to avoid double counting of the effects

Page 17: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Example of Fling and Processing

Red=fling Retained

Blue = standard processing

Page 18: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Example of Adding Fling into Time Histories

Page 19: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Selecting Pulse Time Histories

• How many should have a pulse?• Model probability of pulse based on source/site rupture geometry– Shahi and Baker (2011), use Rrup, s, theta

• Model probability of pulse based on the PGV epsilon– Hayden et al (2014) use idea that the above average PGV are more likely to be from pulse records

Page 20: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Probability of Pulse Motions

From Hayden et al (2014). J. Geotechnical and Geoenvironmental Engineering

Page 21: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Model of Fraction of Pulse Motions 

ExampleR=5 km and epsilon=17 time histories total

Use 65% pulse motions:5 out of 7 should be pulse

From Hayden et al (2014)

Page 22: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Selecting Pulse Time Histories

• Classified pulse motions– Lists by Shahi and Baker & Hayden et al

• Pulse period–What range pulse periods to use?• Shahi and Baker provide Tpulse(M) model• Hayden et al uses two period ranges ( break at 2.2 sec)

• Finite‐Fault Simulations– Are they ready to be used for pulse motions?

Page 23: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Pulse Period

Page 24: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Orientation of Pulse (Max PGV)

• Since 1990s, directivity has been explained as being polarized on the FN component– Over sold this concept

• Wide range of orientation of maximum PGV– Tendency for FN to be the largest PGV for forward directivity, but large range

• Consider using principal axes (max PGV direction) for time histories for the small set of selected GM– Then capture the range of angles using larger set for statistical properties

Page 25: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Orientation of Maximum PGV

From Hayden et al (2014)

Larger IDP meansForward directivity

Page 26: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

FN/FP scaling

• Somerville et al (1999) FN/Ave_H model is still reasonable, but does not address maximum direction

• Earlier, Baker already discussed directionality

Page 27: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Example Pulse RecordNorthridge: Sylmar Conv East FN

Page 28: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Scaling Pulse Records

Page 29: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

84th Percentile GM and Pulses

• Deterministic 84th percentile GM– Ground motion at the 1/6 level

• At long periods, above average ground motions (epsilon=1) are cause by pulses– 84th percentile is an envelope of the 1/6 ground motions with a range of pulse periods

Page 30: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Spectral Matching Pulse Records

• Strongly held opinions:• Method to reduce double‐counting of variability– Keeps the level of shaking consistent with the number of epsilons (deterministic) or return period (probabilistic)

– Amplitude of the pulse is what caused the above average ground motion

• Tool of the devil that must be stopped– Fitting narrow band pulse records to a broad spectrum reduces the amplitude of the pulse

– Destroying the key features of the ground motion

Page 31: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Matching Pulse Records

Page 32: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Example Matching Pulse Records

Page 33: Determning Near-Field Fault-Normal and Fault-Parallel Site-Specific ...

Evaluating Time History Methods• Using small number of recordings (scaled or spectrum compatible) 

is a simplified approach• To evaluate these simplified methods, compare risk using a full set 

of ground motions• Full Set – conditional spectra

– 500‐1000 3‐component time histories, scaled to capture the hazard over wide period range (e.g. 0.1 to 5 sec) and a wide hazard range (e.g. 1E‐2 to 1‐E5)

– These will full represent the hazard at the site– Compute hazard curve for drift (or other structural response 

parameter)• Find the probability of exceedance of the mean structural response 

from simplified methods– Is this small enough?