ML15117A069
| ML15117A069 | |
| Person / Time | |
|---|---|
| Site: | Diablo Canyon  |
| Issue date: | 04/28/2015 |
| From: | PG&E Corp |
| To: | Japan Lessons-Learned Division |
| DiFrancesco N, NRR/JLD, 415-1115 | |
| References | |
| Download: ML15117A069 (61) | |
Text
1 Diablo Canyon Seismic Hazard Reevaluation Presentation to Nuclear Regulatory Commission Public Meeting Bethesda, Maryland April 28, 2015
2 AGENDA Time Topic Speaker 1:00 - 1:15 Introductions NRC/PG&E Co.
1:15 - 1:45 Overview of R2.1 Seismic NRC
- Discussion on meeting goals and expected outcome
- General Background on 50.54(f) and NTTF 2.1 Seismic
- Introduction of seismic hazard PSHA methods and Senior Seismic Hazard Analysis Committee (SSHAC) process Presentation of Seismic Reevaluation 1:45 - 3:45 Jearl Strickland Report, Overview
- SSHAC Activities Norm Abrahamson
- Seismic Sources Norm Abrahamson
- Ground Motion Model Norm Abrahamson
- Interim Actions or Evaluations Nozar Jahangir
- Technical Focus Areas and Discussions All 3:45 - 4:05 Planned Break 4:05 - 4:20 NRC Meeting Wrap up NRC
- Technical wrap-up, review focus area, and next steps 4:20 - 5:00 Public Questions or Comments Public/NRC
3 Overview Safety is and always will be a core value for PG&E and Diablo Canyon Power Plant.
- New and extensive seismic hazard re-evaluation continues to show plant can safely withstand earthquakes.
- Seismic re-evaluation was performed with independent experts in a transparent and open public process.
- Using new regulatory guidance, the latest scientific methodologies and site-specific information, the analysis demonstrates the plants earthquake design is appropriate and safe.
- PG&E maintains a Long Term Seismic Program (LTSP) for Diablo Canyon, a unique program in the industry that continually assesses seismic safety.
- Safety commitment will continue to be reflected through ongoing seismic study.
4 SSHAC - Objectives
- Update the seismic source characterization (SSC) and ground motion characterization (GMC) models for use in an updated site-specific probabilistic seismic hazard assessment (PSHA)
- Develop a methodology for obtaining reproducible, stable estimates of probabilistic seismic hazard at a site, including explicit quantification of uncertainty.
- SSHAC guidelines are summarized in NRC documents NUREG/CR-6372 and NUREG-2117.
- DCPP incorporated new geophysical data into the SSC model, acquired as part of the State mandated AB1632 studies.
5 SSHAC - Workshops SSC/GMC Workshop 1 - Nov. 29 - Dec. 1, 2011 SWUS GMC Workshop 1 - Mar. 19 - 21, 2013
- Significant Issues, Available Data, Data Needs
- Included Resource Expert Presentations
SSC Workshop 2 - Nov. 6 - 8, 2012 SWUS GMC Workshop 2 - Oct 22 - 24, 2013
- Alternative Models and Proponent Interpretations
- Included Proponent and Resource Expert Presentations SSC Workshop 3 - Mar. 25 - 27, 2014 SWUS GMC Workshop 3 - Mar 10 - 12, 2014
- Preliminary Model and Hazard Sensitivity
- Included Proponent Expert Presentations
6 Diablo Canyon Tectonic Setting
7 Seismic Source Characterization (SSC) Model - New Data CCCSIP Study (PG&E)
- Offshore 2D/3D Seismic-Reflection Data
- Onshore 2D/3D Seismic-Reflection Data
- Updated Geologic Map Data Relocated Seismicity Catalog (J. Hardebeck, USGS)
Offshore 2D Seismic-Reflection Data (S. Johnson, USGS)
Offshore high-resolution bathymetry data (R. Kvitek, CSUMB and S. Johnson, USGS)
GPS Velocity Field (J. Murray, USGS and C. DeMets, UW)
8 SSC Models and Methods New Models
- Uniform California Earthquake Rupture Forecast, Version 3 (UCERF3) - USGS, CGS, SCEC
- Offshore Hosgri fault slip rates New Methods
- Rupture Sources: Incorporate earthquake ruptures that involve multiple faults
- Composite earthquake magnitude-frequency distributions (Wooddell et al, 2015)
- Fault geometry models that correlate geometric uncertainty
- Capture time-dependent behavior and uncertainties
- Virtual faults within the host areal source zone 8
9 Types of Seismic Sources Primary fault sources
- Contribute most of the hazard
- Hosgri, Los Osos, San Luis Bay, Shoreline faults Connected fault sources
- Can link (rupture with) Primary fault sources
- E.g., San Gregorio, San Simeon, Wilmar Avenue, Oceano faults Regional fault source
- San Andreas
- UCERF3 faults
- Additional non-UCERF3 faults Areal Source Zones
- Regional source zone
- Vicinity source zone
- Local source zone
10 Local Areal Source Zone Virtual faults capture uncertainty in location, dip, sense of slip for other known and possible faults
11 Seismic Source Characterization Focus Areas NRC SSC Topic 1:
Summarize the key data used to constrain the slip rate of the Hosgri fault, including associated uncertainties.
- Four slip rate sites
- Three new sites offshore
- Uncertainties developed for each site
- Sites are weighted for final uncertainty 11
12 Hosgri Fault Slip Rate:
Estero Bay Site
13 Hosgri Fault Slip Rate Models Documentation of Offset Uncertainty
14 Hosgri Fault Slip Rate Models Documentation of Age Uncertainty
15 Hosgri Fault Slip Rate Models Slip Rate CDF for Estero Bay Site
16 Hosgri Fault Slip Rate Models Weighting of Four Study Sites CBR of TDI (in mm/yr):
Center: 1.7 (wtd. mean)
Body: 0.8 to 2.6 (10%, 90%)
Range: 0.4 to 3.4 (1%, 99%)
17 Seismic Source Characterization Focus Areas NRC SSC Topic 2: Clarify how elements of the thrust/ reverse interpretation for the San Luis Range Thrust are incorporated into the SSC.
- San Luis Range thrust model proposes that the Irish Hills are uplifted by a northeast-dipping thrust fault
- SSC model incorporates this model as one of three alternatives of uplifting the Irish Hills
18 Fault Geometry Models 3 for Hosgri; 3 for San Luis-Pismo Block Alternative Fault Models:
- Describe the fault geometry for each tectonic model in a correlated way
- Dip variability is achieved through the differences between tectonic models
19 Seismic Source Characterization Focus Areas NRC SSC Topic 3: Clarify how the rupture models are derived from the fault source geometry models.
Fault Geometry Models (FGMs) describe the fault locations, dips, and senses of slip Rupture Models describe the alternative locations on the FGMs where maximum earthquake ruptures and smaller, floating ruptures occur.
- Represents aleatory variability in how earthquakes rupture the fault network
- Allows single- and multi-fault rupture in a standard, forward model
- Rupture models include sufficient rupture sources such that the range of alternative types of ruptures are sampled for adequate source and ground motion variability
20 New Methods, Models in DCPP SSC Model Rupture Models Allow multi-fault ruptures New composite magnitude-frequency distribution
21 Seismic Source Characterization Focus Areas NRC SSC Topic 4: Summarize the Methodology Used to Define the Equivalent Poisson Rates.
Motivation: Non-Poisson Recurrence Behavior is Likely
- In some cases, paleoseismic recurrence records are inconsistent with a Poisson process
- Renewal process includes intuitive physics (elastic strain accumulation and release)
- Simple models available that simulate renewal-type behavior
22 Methodology to Define the Equivalent Poisson Ratio (EPR)
Methodology:
- 1. Lognormal model for recurrence (also BPT, Weibull)
- 2. Requires estimates of long-term mean (LTM),
coefficient of variation (CV), and time since the most recent event (Tmre)
- CV range from global paleoseismic data (mostly California)
- Tmin constraint on Tmre from historical record (SLO Mission)
- LTM from slip rate and simple slip/event model
- 3. For each CV and slip rate, model considers joint probabilities of correct LTM and Tmre
23 EPR Methodology Use of survivor function to constrain LTM, Tmre joint probability (CV=0.6 shown)
24 EPR Methodology EPR Logic Tree, Hosgri fault 1.9
[0.25]
1.3
[0.5]
0.3
[0.25]
24
25 Ground Motion Models Reference rock ground motion model (SSHAC GMC)
- Median ground motion
- Aleatory Variability Site Amplification
- How the ground motion at the control point differs from the reference rock ground motion Capture Uncertainties in each part
- Epistemic uncertainties
26 New Data, Models, Methods for GMC Data
- NGA-W2 strong motion data set (PEER)
- European strong motion data set (RESORSE)
- Finite-fault simulations close to large earthquakes (SCEC)
Models
- Median GMPE: NGA-W2 GMPEs
- Median GMPE: European and Japanese GMPEs
- Aleatory variability: Mixture model Methods
- Sammons map approach to develop weights for GMPEs
- Additional epistemic uncertainty added to all GMPEs
- Included comparisons with empirical data and finite-fault simulations (SCEC) as part of the evaluation of the weights
- Single-station sigma approach
- Improved treatment of uncertainty for empirical DCPP site terms
27 Example Hazard Curves at Control Point with Uncertainty
28 Hazard Sensitivity to GM Model 5 Hz: 1E-4
29 Ground Motion Characterization Focus Areas NRC GMC Topic 1: Provide additional detail in the criterion used for the selection of candidate Ground Motion Prediction Equations (GMPEs) for development of the common form median ground motion models for DCPP. Specifically, please elaborate on the basis for including GMPEs based on data sets other than NGA West-2.
30 Selected Candidate GMPEs NGA-W2 Abrahamson et al. (2014), referred to as ASK14 NGA-W2 Boore et al. (2014), referred to as BSSA14 NGA-W2 Campbell and Bozorgnia (2014) , referred to as CB14 NGA-W2 Chiou and Youngs (2014), referred to as CY14 NGA-W2 Idriss (2014), referred to as Id14 Europe & Akkar et al. (2014a, 2014b), referred to as ASB14 ME Japan & CA Zhao et al. (2006), referred to as ZH06 Japan & CA TI Team implementation of Zhao and Lu (2011), referred to as ZL11
31 Selection Criteria Consider all modern GMPEs from active crustal regions
- Assumes that the magnitude and distance scaling in active crustal regions is similar around the world
- Selected GMPEs had to meet 7 criteria (SWUS, section 5.5.2)
- Most recent version
- Not an adjustment of another model
- Functional form extrapolates in a reasonable manner
- Do not combine data from active crustal and subduction earthquakes
- Not just a research tool
- Not developed for a very small region
- Peer reviewed
32 Why Select GMPEs not from NGA-W2 data
- The objective is to capture the uncertainty
- GMPEs from other regions may provide alternative credible scaling for ground motions in CA
- Early feedback from PPRP recommended that we not limit the GMPEs to NGA models because there models may not capture the full range of uncertainty
- Some of the large magnitude data in the non-NGA GMPEs are contained in the NGA data set, so there is overlap, but also different modeling approaches used
- The weights for the final models are developed considering how well the models fit the NGA data
33 Ground Motion Characterization Focus Areas NRC GMC Topic 2: Provide additional detail on the development of the common functional form used to fit the candidate GMPEs. Specifically, please discuss how model parameters such as depth to Vs equals to 1 km/s and 2.5 km/s (which are present in some of the candidate GMPEs) are accounted for in the functional form.
34 Common-Form Model The common-form models
- Developed for a single reference rock condition of VS30=760 m/s
- Footwall side only to keep the functional form simple
- Hanging-wall effects added to the common-form model The other site parameters, Z1.0 and Z2.5, are set to their default values for VS30=760 m/s
- Basin depth is not a significant issue for soft-rock sites
35 Ground Motion Characterization Focus Areas NRC GMC Topic 3: Provide additional detail on the approach for weighting the selected common form models as well as the criteria used to verify the physicality of the final model.
2000 models generated to fill in the space of possible GMPEs
- Sampled the covariance of the coefficients
- Treats the correlations of the coefficients Non-physical models
- Tails of distributions of coefficients may lead to sampled models that are unphysical
- Defined as models for which the magnitude or distance scaling is not monotonically increasing (M) or decreasing (R)
36 Mapping GMPE Differences Measure the Standard Deviation of Difference in GMPEs for a Range of M,R Magnitude
Use Sammons Maps to Describe Space of Median 37 GMPEs and Develop Weights
38 Weights for the Common-Form Models
39 Hazard Sensitivity to Median GMPE
40 Ground Motion Characterization Focus Areas NRC GMC Topic 4: Provide additional detail on how the continuous distribution for total sigma was developed by combining the between-event and within-event aleatory variabilities.
SS (M ,T ) = (M ,T ) + (M ,T )
2 SS 2
41 Example of Tau Models for 1 Hz
42 PhiSS Logic Tree
43 Combining PhiSS and Tau into Total Sigma Use Chi-Squared distribution for phiSS^2 and Tau^2 For each model (branch 2) of phiSS logic tree, sample the three PhiSS values and the three tau values Develop a cumulative distribution function (CDF)
Average the CDF using the logic tree weights for the alternative PhiSS models (branch 2) and data sets (branch 1)
Sample the total CDF at the 5th, 50th, and 95th fractile levels
44 Combining Phi and Tau into Total Sigma
45 Site Amplifications Focus Areas NRC SA Topic 1: Section 2.3.2.1 of 50.54(f) Submittal states that shear modulus and damping curves are not directly applicable to DCPP since analytical modeling is not used and that non-linear site effects are implicitly included in the empirical GM,PEs for Vs30 = 760 m/sec. However the NGA West 2 data base has a limited amount of data for sites with Vs30 near 760 m/sec and for earthquake with magnitude and source to site distance similar to those dominating the hazard for DCPP.
Please provide additional information on how these limitations in the NGA West 2 data base are accounted for in site response model for DCPP?
Limitations of the Data Base for 46 Hard Rock Sites Select a reference rock site NGA-West2 data set used
.condition by ASK14 (M>6, R<20 km)
- Range with data, but not too low VS to avoid strong non-linear site effects
- Selected VS30=760 m/s for the reference rock site condition Nonlinear Site Response in NGA-W2 GMPEs Some GMPEs used analytical modeling for noninearity.
47 Site Amplifications Focus Areas NRC SA Topic 2: Section 2.3.6 of the 50.54(f) Submittal describes the development of the site terms for DCPP.
For the calculations of between-event residuals, provide additional information on the criteria used to determine the appropriate distance range (+ and - Rrup) to the sample station. Please discuss the sensitivity of this distance range on between-event residual values. Please provide an example calculation that uses site specific values to determine the values for Phi s2s including the epistemic uncertainty in the site term.
48 Estimation of DCPP Site Terms Non-ergodic ground-motion model ln SAobs (M i , Loci , Site j ) = ln GMPE(M i , Rij ,VS30 j )
+ L2Ll + S2S j + P2Plj + Bi0 + Wij0 Estimate the combined source and path terms for each earthquake using observations from other sites (not DCPP)
L2Ll + B + P2Plj i
0
49 Estimation of DCPP Site Terms S2S j = ln SAobs (M i , Loci ,Site j ) lnGMPE(M i , Rij ,VS30 j )
( )
L2Ll + Bi0 + P2Plj + Wij0 Sources of Uncertainty of Site Term
- Uncertainty in the Event-specific source and path term:
- Randomness of the remaining aleatory variability of the within-event term
50 Selection of Distance Range for Residuals Two factors
- Distance range that includes the DCPP distance
- Distance range for which the residuals do not have a strong distance slope
Example: BSSA14 model, 5 Hz 51 San Simeon
Example: BSSA14 model, 5 Hz 52 Parkfield
53 Uncertainty in the Event-specific Source and Path Term Computed from the standard deviation and number of recordings used to estimate the term
- San Simeon: 8 recordings, Sigma = 0.68 ln units
- Parkfield: 16 recordings, Sigma = 0.55 ln units Epistemic uncertainty is the standard error of the mean
- Sigma / sqrt(N) 0.25 for San Simeon 0.14 for Parkfield NEQK SE(Source + Path) i 2
+ 02 Std Error of S2SDCPP = i=1
= 0.22 NEQK
54 Interim Evaluations and Actions DCPP Seismic Design/Licensing Basis History
55 DCPP Design / Licensing Basis DCPP was licensed prior to 2.5 LTSP App. A to 10 CFR Part 100 DE HE (Introduced OBE& SSE 2.0 DDE terms)
HE (UFSAR) LTSP DE: OBE Equivalent = 0.2 g 1.5 HE (Extrapolated)
Acceleration (g)
DDE: SSE equivalent = 0.4 g DDE 1.0 HE: Largest design ground 0.83 0.75 motion = 0.75 g 0.5 LTSP: Seismic Margin /LTSP DE 0.40 Spectra = 0.83 g 0.20 0.0 0.1 1.0 10.0 100.0 Frequency (Hz)
56 GMRS Comparisons GMRS Vs. 1977 HE Design GMRS Vs. LTSP Seismic Margin
57 GMRS Comparisons (continued) 3.0 LTSP Seismic Margin 2.5 1977 HE 2.0 Acceleration (g)
GMRS 1.5 1.0 DDE 0.5 0.0 0.1 1.0 10.0 100.0 Frequency (Hz.)
58 LTSP Seismic Margin LTSP Licensing Background
- DCPP License condition No. 2.C.(7) required in part PG&E shall develop and implement a program to reevaluate the seismic design bases used for the DCPP
- Seismic reevaluation effort was titled Long Term Seismic Program (LTSP); issued in 1988 with 1991 addendums to address the LC and committed to maintain the program going forward.
- LTSP deliverables were Seismic Probabilistic Risk Assessment (SPRA) and Seismic Margin Assessment (SMA)
- NRCs comprehensive assessment and acceptance are documented in Supplement 34 to Safety Evaluation Report (SSER-34)
59 LTSP Seismic Margin (continued)
- Key Points
- From SPRA; Mean Seismic Core Damage Frequency (SCDF) was calculated to be 3.7x10-5
- SCDF sensitivity review considering updated Hazard (with the original Fragilities) is ~ 2.06x10-5 . The fragilities will be revised to get updated risk values.
- The fragilities and HCLPF capacities are based on 5% damped horizontal spectral acceleration values, averaged over 3.0-8.5 Hz. (~ 1.94gs)
- From Seismic Margin evaluation; the Lowest, High Confidence Of Low Probability of Failure (HCLPF) of SSCs was determined to be 2.62g resulting in a minimum seismic margin of 1.35.
- NRC reviewed and acknowledged the significant seismic margin in SSER-34 .
60 ESEP, SFP Evaluation Expedited Seismic Evaluation Program (ESEP):
- The GMRS is recognized as beyond design basis. However there needs to be reasonable assurance of plant safety while new/updated risk evaluations are in-progress
- Developed to address where significant exceedance beyond design basis are identified in the 1-10Hz. frequency range.
- The GMRS is effectively bounded by the 1977 HE design spectra in 1-10 Hz. Minor high frequency exceedance is well within the LTSP seismic margins and adequately considered in the SPRA analysis.
Therefore there is reasonable assurance of plant safety.
Spent Fuel Pool Evaluation:
- SFP structure is an integral part of the Auxiliary building, which has been designed and evaluated as a seismic Design Class I structure in accordance with the DE, DDE, HE design criteria, and considered in the SPRA (Building fragility). Therefore, there is reasonable assurance of structural integrity. DCPP will perform rapid drain down evaluation activities, as required per SPID and will reevaluate the fragilities for the Auxiliary building as part of SPRA update.
61 Current and Next Actions Proceeding with SPRA Update/Upgrade
- Updating building models (3D FEA)
- Updating SSI models
- Developing Building FIRS
- Fragility evaluation preparation
- Updating/upgrading the SPRA model Next Actions
- Determine Risk evaluation Prioritization (NRC)
- Obtain agreed upon Hazard (GMRS) to proceed with the SPRA (NRC)
- Complete Seismic Risk Assessment (PG&E)