Probabilistic Seismic Hazard Analysis Project
Origins and Objectives
12-05-2012
BC has all four types of tectonic
regions
1. Active Crustal: WNA
2. Stable Continental: CENA
3. Subduction Zones: Cascadia
4. Volcanic: Volcanic Arc/Cascades
>80 dams
18 Extreme Consequence
6 > 10,000 PAR
2
Background
•
BC Hydro has been involved in probabilistic seismic hazard analysis (PSHA)
since the 1980s.
•
Evolution in Approach:
• 1984 – first PSHA for Lower Mainland & Vancouver Island, using GSC
seismic source model
• 1991-92 – Provincial Seismic Hazard Review; developed BCH seismic
source model
• 1997 – Introduced “multiple model” approach, with BCH and GSC “H” &
“R” source models as alternates, & alternate ground motion models
• Post-1997 – Ongoing introduction of new ground motion models
•
Results have been used for seismic performance evaluations and designs for
dams and the electric system as they have become available
Concerns About BCH PSHAs in 2000s
• Seismic source and ground motion models becoming outdated
• Cascadia subduction zone not incorporated into probabilistic model
• Large computed ground motions
• Large uncertainties
• Sensitivity of computed ground motions to some model inputs (especially
subduction zone ground motion models)
• Stability and consistency of hazard estimates, same site and site-to-site
•
•
4
Dr. C. A. Cornell retained in 2007.
Alternative approaches to update BCH’s PSHA were considered.
Project Objectives and Startup
• Develop an up-to-date and technically sound PSHA model for the BC Hydro
service area.
• Achieve technical stability of the PSHA model and its results for 10-15 years.
• Model to be used by BC Hydro for seismic hazard assessment and
characterization at all sites.
• Initial work undertaken mainly with internal resources
• Review by Dr. Cornell indicated that more rigour required to meet Project
Objectives
• Dr. Cornell recommended a SSHAC III approach
Probabilistic Seismic Hazard Analysis Project
Results and Policies for Use
+$10M and 4 years later… .
PSHA Model – ‘Shallow’ Seismicity
7
PSHA Model – ‘Deep’ Seismicity
Subduction slab
represented by 4
source zones which
become increasingly
deeper to the
northeast.
Source zone
boundary uncertainty
is included in the
model
The Nootka Fault is
represented as a
separate source
8
PSHA Model – Evaluated & Selected GMPEs
Active Crustal
Stable
Continental
Subduction Zone
Abrahamson & Silva Silva et al (2003)
(2008)
Youngs et al (1997)
Chiou & Youngs
(2008)
Campbell (2003)
Atkinson & Boore (2003)
Campbell &
Bozorgnia (2008)
Atkinson & Boore
(2006, 2011)
Garcia et al (2005)
Boore & Atkinson
(2008)
Atkinson
(2008, 2011)
Zhao et al (2006)
Idriss (2008)
Kanno et al (2006)
Macias & Atkinson (2009)
Lin and Lee (2010)
Volcanic Arcs: Evaluated McVerry et al (2011) but did not use it.
PSHA Model: BC Hydro Subduction Model
ln( Sa )   1   4 *  C 1  ( 2   14 * F event   3 * (M  7.8)) * ln R  C 4 * exp[( M  6) *  9 ] 
 6 * R   10 * F event  f Mag (M )  f depth (Z h )  f FABA (R ) 
f site (PGA
1000
, V S 30 )
1000 .0
*
V s  
 V S 30

for
V S 30  1000
for V S 30  1000
 4 * ( M  (C 1   C 1 ))   13 * (10  M ) 2
f Mag ( M )  
2
 5 * ( M  (C 1   C 1 ))   13 * (10  M )
for
M  C1   C1
for
M  C1   C1
f depth ( Z h )   11 * ( Z h  60 ) * F event


f FABA
max[ R,85 ]

[   8 * Ln (
)] * F FABA
 7
40
( R )  
max[ R,100 ]
[ 15   16 * Ln (
)] * F FABA
40


f site (PGA
1000
for F event  1
for F event  0
*

V
 12 * Ln ( s )  b * Ln (PGA 1000  c ) 
V lin

*

V
n

b * Ln (PGA 1000  c * ( s ) )
, Vs 30 m )  
V lin


*
*

Vs
V
)  b * n * Ln ( s )
  12 * Ln (
V lin
V lin

for
V S 30  V lin
for
V S 30  V lin
10
Uncertainty in PGA Hazard
Annual Exceedance Frequency
0.01
0.95
0.85
MEAN
0.50
0.15
0.001
0.05
Now what??
0.0001
0.00001
0
20
40
PGA (%g)
60
80
Now that we understand the uncertainty…
• what to use?
• Different philosophies at different sites
50
45
40
Typical Results
(actual data : Site C)
Level
Relative confidence
Probability
(%) (%)
35
30
25
20
15
10
5
Median
Mean
84th percentile
0
0
0.1
0.2
0.3
0.4
Peak Ground Acceleration (g)
0.5
0.6
0.7
12
PSHA Policy
•For High, very High and extreme Consequence dams, ground motion
hazards at a mean 10-4 AEF to be a starting point
•Best representation of the uncertainties
•Use full spectrum of results for risk analyses
•Consider the potential for more extreme events, and the risks
associated with those events
•Policy decision to embrace risk informed decision making (finally…)
•Consider combinations of events. For example, if a dam is damaged
due to an earthquake, would it be possible to subsequently pass the
mean annual flood (or larger flood) safely?
•Incorporate system thinking and reliability principles in design
13
Effects of PSHA project and policy
14
Effects of PSHA project and policy
LaJoie
15
Effect of PSHA Policy and Results
Different effects at different sites
Site C : increase in ground motion
(median to mean)
HLK: decrease in ground motion
(84th to mean)
RUS : decrease in seismic hazard
JHT : decrease in PGA, but
increase in hazard
JOR : increase in ground motion
16
Probabilistic Seismic Hazard Analysis Project
Sharing the Work
Giving away the hard-earned IP….
12-05-2012
Requests for Model/Model Components
Agency
Request
Purpose
Standing Committee on
Earthquake Design (SCED)
PSHA Model
National Building Code of Canada
USGS
Subduction GM
Model
USGS National Seismic Hazard maps
Pacific Northwest National
Laboratory, WA
Cascadia
Subduction Model
Hanford
URS Corporation
Subduction GMPE
Consulting
PEER
Berkeley, CA
Subduction GMPE
and Data
Research & Development
Shannon& Wilson
Seattle, WA
Subduction GMPE
Consulting
GeoMotions, LLC
Lacey, WA
Subduction GMPE
Incorporation into Commercial
Software
Geosyntec Consultants
Houston, TX
Source Model
Subduction GMPE
Consulting
Golder Associates
Burnaby, BC
PSHA Model
Consulting
Klohn Crippen Berger
Vancouver, BC
PSHA Model
Consulting
Miscellaneous Individuals
PSHA Model
-------------
Sharing the Model
Project documents ALMOST signed off :
- GIS based Seismic Source Model
logic trees that characterize the various source areas
- Ground Motion model (attenuation formulae)
logic trees
IP will be shared by
Publication of scientific papers
Access to hardcopy reports
Actual software code held by AMEC
With access to IP, others can develop alternative codes
19
end
20
What Has Changed Since 1997?
•
Seismic Data Base
– Another 15 years of data : increased data base of western Canada recordings
– no large, “surprise” earthquakes
– Several important worldwide earthquakes, some with strong motion records
•
Improved understanding of BC seismotectonics, e.g.
– greatly advanced knowledge of 3D structure & geological evolution of
Canada’s continental landmass and its margins.
– Improved understanding of the configuration & behaviour of the Cascadia
subduction zone, its seismic history and seismic potential
•
Ground Motion Models
– Ongoing development & publication of new attenuation relationships,
e.g.
Frequency of Exceedance
Cumulative Rate (EQ > M/yr)
Elements of PSHA – Cornell Methodology
Acceleration
Identify and model sources of aleatory (random) and epistemic (model and
Uncertainty in
parameter)
Fault uncertainty
Confidence
Attenuation
Limits
Available information
often supports multiple, credible
(scientifically sound)
Site
Magnitude
M1
interpretations Area
Sources
M2
Mean
Minputs
SSHAC : The goal is to develop
that represent the composite
x
Magnitude (M)
Acceleration
Distance
distribution of the informed scientific community
STEP 1
Seismogenic Zone Model
GSC-H
Alternate
seismogenic
zone models
STEP 2
Recurrence Model
Best
Estimate
Alternate recurrence
curves for each
seismogenic zone
STEP 3
Ground Motion
Attenuation
STEP 4
Ground Motion
Hazard
Best Estimate
Alternate Mx values
for each recurrence
curve
Alternate attenuation
relationships
Area
Sources
STEP 1
Seismogenic Zone Model
Uncertainty in
Attenuation
Magnitude
M1
M2
Mx
Magnitude (M)
STEP 2
Recurrence Model
Seismic Source Characterization Model
Ivan Wong led the SSC group!
Distance
STEP 3
Ground Motion
Attenuation
Frequency of Exceedance
Site
Acceleration
Fault
Cumulative Rate (EQ > M/yr)
Elements of PSHA – Cornell Methodology
Confidence
Limits
Mean
Acceleration
STEP 4
Ground Motion
Hazard
Ground Motion Model
PSHA Policy Review
Peer Reviewed as part of Project
April 28, 2011 at Stanford U
Marty McCann
Ivan Wong
Kevin Coppersmith
All agreed on use of mean : entire scientific process of hazard analyses revolves
around the understanding of uncertainty
Use of mean already considered best practice in the US
Confirmed that policy is defendable
Use of Risk Informed approach most welcome :
“Dam community is two decades behind the Nuclear industry”
USBR and USACE have now moved to risk-informed decision making
24
PSHA Policy : Use of the Mean
• Broadly representative of the distribution
• Means add to means mathematically for combined risk analyses
• Accounts for uncertainty
50
45
Uncertainty not
Preliminary
accounted
forResults
if
for
Site
C
Median is adopted
40
Level
Relative confidence
Probability
(%) (%)
35
30
25
20
15
10
5
Median
Mean
84th percentile
0
0
0.1
0.2
0.3
0.4
Peak Ground Acceleration (g)
0.5
0.6
0.7
25
PSHA Policy : Use of the Mean
• Broadly representative of the distribution
• Means add to means mathematically for combined risk analyses
• Accounts for uncertainty
50
45
Do we really need
Preliminary
Results
this
little
for Site C
uncertainty?
40
Level
Relative confidence
Probability
(%) (%)
35
30
25
20
15
10
5
Median
Mean
84th percentile
0
0
0.1
0.2
0.3
0.4
Peak Ground Acceleration (g)
0.5
0.6
0.7
26
PSHA Policy : Use of the Mean
• Broadly representative of the distribution
• Means add to means mathematically for combined risk analyses
50
45
Use of mean supported
by ICOLD draft bulletin
40
Level
Relative confidence
Probability
(%) (%)
35
30
25
20
15
10
5
Median
Mean
84th percentile
0
0
0.1
0.2
0.3
0.4
Peak Ground Acceleration (g)
0.5
0.6
0.7
27