Seismic Input and SoilStructure Interaction

(Ch. 5 of TBI report, PEER 2010/05)

TBI Committee Members

Y. Bozorgnia
C.B. Crouse
J.P. Stewart
October 8, 2010

Outline
1. Seismic Hazard Analysis

Probabilistic
Deterministic
Site-Response Analysis

2. Soil-Foundation-Structure Interaction

Kinematic
Inertial
Input Motion Specification

5. Ground Motion Selection and Scaling

Identification of Controlling Seismic Sources

Ground Motion Selection
Accelerogram Modification

Two SHA Approaches

Recommendation
Use General Procedure if geotechnical
engineer is inexperienced or unqualified
to perform site-specific probabilistic and
deterministic SHA.

Two SHA Approaches (cont.)

2. Site-Specific (Preferred)

Probabilistic
Deterministic

Probabilistic Seismic Hazard Analysis

(PSHA)

Source models
Eqk locations
M range
Recurrence

Probabilistic Seismic Hazard Analysis

(PSHA)

Source models
Ground motion prediction
equations (GMPEs):

Sa, Sa | (M, r, S, )

Probabilistic Seismic Hazard Analysis

(PSHA)

Source models
Ground motion prediction equations
(GMPEs)
Log
Accel.

a
am

P depends on no.
of standard deviations
A is above/below am

PSHA Output: Ground-Motion Hazard

Curves

Uniform Hazard Spectrum

Deaggregation Plot for L.A.

Recommendations for PSHA

For experienced PSHA users only

Use QA-checked software

Account for alternate seismic source

parameters and GMPEs (epistemic
uncertainty)

Logic Tree

GMPEs Recommended for

Shallow Crustal Western U.S. Earthquakes
NGA GMPEs (2008)
Abrahamson & Sliva
Boore & Atkinson
Campbell & Bozorgnia
Chiou & Youngs
Idriss

See EERI Spectra Journal

(Feb. 2008, v. 24, no. 1)

Empirical GMPEs Recommended for

Subduction Earthquakes

Atkinson & Boore (2003) Site Class B,

C, D
Crouse (1991) Soil
Youngs et al. (1997) Soil and Rock
Zhao et al. (2006) Soil Classes I IV
and Hard Rock

Ratio of 1.0 s SA 2007 and 2002 National Seismic

Hazard Maps for WUS 2% Probability in 50 Years

Deterministic MCE Calculation

Reqd per ASCE 7 Ch 21

Provides cap near major faults
Arbitrary decisions regarding:

Ruptured fault segment (closest)

Magnitude (use average of Mmax from logic
tree)

Use same GMPEs & wts from PSHA

Different sources may be most critical at
short and long periods

1992 Landers, CA M 7.3 Earthquake

Sommerville et al. (1997)

Lucerne Record, 1992 Landers Earthquake

Fault Normal (FN)

Sommerville et al. (1997)

Fault Parallel (FP)

Lucerne Response Spectra

FN

FP

Sommerville et al. (1997)

Deterministic SHA

Required per Ch 21 of ASCE 7

Usually governs MCE for sites near
active faults
Recommendations

Use same GMPEs, weights and parameter

values used in PSHA
Maximum Magnitude

May want use higher Mmax than weighted average

Mmax from logic tree

Site-Specific Deterministic Method

ASCE 7, Sect. 21.2.2

Find Fault largest median Sa

Compute 1.5 x median Sa (ASCE 7-05)

Compute Sa84th >1.5Samedian (ASCE 7-10)

Fault Map

Median Deterministic Response Spectra

H Comp.
Elysian Park Thrust
Newport Inglewood

San Andreas

Sa
0.5

0.5

1.5

T (sec)

2.5

3.5

M 7.8 San Andreas Earthquake Simulations

Graves et al. (2008)

Site Response Analysis

ASCE 7-05; Ch.21
Site-Specific Ground Motion
PSHA/DSHA Vs30
PSHA/DSHA Ref. Vs30

Recommendations

SRA not needed in absence of pronounced

impedance contrast (often the case for stiff
soil sites)
Site effect can be accounted for in such cases
through GMPE site terms
SRA advisable/required for:

Recommendations

SRA produces amplification factors, AF(T)=

Sa,soil/Sa,rock
Typically applied as deterministic modification
of UHS (Hybrid proc.): Sa,soil=AF(Sa,rock)UHS
Can avoid with modification of site term in
hazard integral (OpenSHA)
Unconservative
bias

SRA (cont.)

(may not be necessary)

Site Class F

input motion

SRA (cont.)

input
motion

Building Input Motion

Seismic Wavelengths

2. Soil-Foundation-Structure Interaction (SFSI)

SFSI for MCE

Linear springs and

dashpots model soil
-foundation interaction
Input motion same at all
points along foundation
Input can be reduced for
kinematic effects
See FEMA 440 & ASCE
41-06 for details

Basement Wall Soil Interaction

flexible
wall
pt. x
floor

Wall Reponses at Pt. x

Pt. x

3. Ground Motion Selection and

Modification

Identify controlling earthquakes

Select representative ground motions

Modify accelerograms to match target

spectrum

Identify Controlling Earthquakes

Specify natural period band SE decision

Deaggregation Plots

T = 1 sec

M1 R1

T = 5 sec

M2 R2

Issues with Ground Motion Selection

Number of ground motion sets

Multiple controlling earthquakes

Near-fault effects

Effects poorly represented in ground

motion database:

Basin Effects

M > ~ 8, long-duration motion

Use of simulations

Number of Accelerograms - N

No less than three (use maximum

responses)
Use average responses if 7 or more
motions used
More needed if multiple controlling
earthquakes

Near Fault Effects

Select a(t) for both cases

Transform FN & FP a(t) into X & Y a(t)

Fault

Simulated Ground Motions (e.g., ShakeOut)

Sa (T = 3 sec, 5 = 5%)

g
Graves et al. (2008)

Simulated Ground Motions (e.g., ShakeOut)

Can produce realistic-appearing wave

forms
Need for calibration
Most broadband methods are
inadequately validated or have biases

Basin Effects

Amplify long period motions

Increase duration

Issues with Ground Motion

Modification

Target Sa

Site-specific Sa
Conditional mean Sa (CMS)

Modification procedures

constant scaling
spectral matching

Target Sa

UHS encompasses many events

Not achievable in a given event
Scenerio spectra (CMS) more realistic; need > 1

CMS

e Parameter

M > ~ 8, Long Duration Motion

San Andreas fault M ~ 8

Cascadia and S. Alaska subduction zone
M 9+

Velocity Records for M 7.8 San Andreas Event

km

Graves et al. (2008)

Accelerogram Modification

Constant Scaling

Spectral Matching

Accelerogram Modification

Constant Scaling

Spectral Matching

Constant Scaling Method

Sa (g)

EQ-IV x 1.3
1940 Imperial Valley, El Centro (2.00)

1971 San Fernando, 8244 Orion Blvd. (1.74)

1979 Imperial Valley, El Centro Diff Array (1.74)
1989 Loma Prieta, Saratoga Aloha Ave. (1.88)
1992 Landers, Yermo Fire Station (2.00)

1994 Northridge, Sylmar Hospital (1.10)

1999 Duzce, Turkey, Duzce Station (1.36)

T (sec)

Spectral Matching

Selection and Scaling Recommendations

N > 7 (N limited by $ and time)

Use hazard deaggregations controlling EQs

CMS use several different Sa shapes

Scaling (constant or spectral matching)
SEs decision
Simulated accelerograms (M > ~ 8)
- ADV: long duration and basin effects
- DISADV: verification issues, access to quality
simulations
Peer Review Important