ML15154B396

Discussion of Seismic Hazard Reevaluation and Nrc'S Technical Focus Areas for Columbia Generating Station - June 4, 2015

ML15154B396
Person / Time
Site:	Columbia
Issue date:	06/04/2015
From:	Energy Northwest
To:	Office of Nuclear Reactor Regulation
DiFrancesco N, NRR/JLD, 415-1115
References
Download: ML15154B396 (111)

Text

Discussion of Seismic Hazard Reevaluation and NRCs Technical Focus Areas for Columbia Generating Station June 4, 2015

Agenda Introduction Presentation of Seismic Reevaluation Report SSHAC Activities Seismic Sources Ground Motion Model and Site Response Discussion of Interim Actions and Evaluations Flexible and Diverse Mitigation Strategies (FLEX)

Seismic Analysis (PSA)

Path Forward 2

Introduction Energy Northwest followed the approved process for development of seismic hazard reevaluation for the Columbia Generating Station site in response to Enclosure 1 of the NRCs 10 CFR 50.54(f)

Request for Information Screening determination performed in accordance with NRC-endorsed Screening, Prioritization, and Implementation Details (SPID) (EPRI 1025287)

Energy Northwest will present a detailed technical basis to demonstrate how process was followed and is prepared to discuss each of the technical focus areas 3

SSHAC Activities Kevin Coppersmith Coppersmith Consulting, Inc.

SSHAC Activities SSHAC Level 3 (SL3) conducted as Hanford Site-Wide PSHA with sponsorship from DOE and Energy Northwest Project planned and conducted to comply with NUREG-2117 and other guidance Roles and responsibilities of all project participants defined and adhered to Project-specific enhancements to SL3 process Participatory Peer Review Panel (PPRP) confirmed acceptability of both technical and process aspects of the project 5

Hanford Site-Wide SSHAC Level 3 PSHA Purpose of Study: to develop a technically defensible PSHA that can be used for design and safety evaluations at the Hanford Site, Washington, including Columbia Generating Station PSHA must enjoy high levels of regulatory assurance, as indicated by a SSHAC Level 3 process Must provide outputs that allow use at multiple facility sites within the Hanford Site, including the CGS Outputs must be compatible in format with site response analyses for site-specific facility input motions Compliant with NRC requirements, per 50.54(f) letter, and regulatory guidance Compliant with DOE Order 420.1B (later 420.1C) regarding 10-year update and expectations of DNFSB 6

SSHAC Guidelines and Guidance NUREG-2117 SSHAC Implementation Guidelines 7

Project Plan for SSHAC Level 3 PSHA Project Plan specifies:

Project organization Participant roles and responsibilities Scope Schedule Deliverables and instructions for usage Provided a basis for all project planning and contracting Ongoing information for participants and reviewers Recommended elements given in NUREG-2117 Hanford PSHA Enhancements New data collection activities PPRP participation Interfaces between seismic source characterization (SSC), ground motion characterization (GMC), and site response 8

Selection Criteria for SSHAC Participants Bommer, J.J. and Coppersmith, K.J., 2013, SMiRT-22, Lessons Learned from Application of the NUREG-2117 Guidelines for SSHAC Level 3 Probabilistic Seismic Hazard Studies for Nuclear Sites 9

Project Hanford PSHAOrganization Project Organization 10

Seismic Source Characterization Seismic Source Characterization Team Team Kevin Coppersmith - TI Lead Lorraine Wolf Kathryn Hanson

• Valentina Montaldo-Falero - Hazard Analyst Ryan Coppersmith

• Roseanne Chambers - PSHA document integrator

• Joe Lettrick - GIS data base manager Jeff Unruh 11

Ground Motion Characterization Team Julian Bommer - TI Lead Bob Youngs Linda Al Atik Gabriel Toro Adrian Rodriguez-Marek 12

Resource and Proponent Experts - WS1 Individual Affiliation Walt Silva Pacific Engineering & Analysis Carl Costantino Consultant Norm Abrahamson University of California, Berkeley Tuna Onur Onur Seemann Consulting Rob Graves U.S. Geological Survey Art Frankel U.S. Geological Survey Tom Hearns New Mexico State University Alan Rohay Pacific Northwest National Laboratory Tom Pratt U.S. Geological Survey Brian Sherrod U.S. Geological Survey Rick Blakely U.S. Geological Survey George Last Pacific Northwest National Laboratory Harvey Kelsey Humboldt State University Rex Flake Central Washington University Erick Burns U.S. Geological Survey Paul Thorne Pacific Northwest National Laboratory Bruce Bjornstad Pacific Northwest National Laboratory 13

Resource and Proponent Experts - WS2 Individual Affiliation Walt Silva Pacific Engineering & Analysis Carl Costantino Consultant Norm Abrahamson University of California-Berkeley Art Frankel U.S. Geological Survey Alan Rohay Pacific Northwest National Laboratory Suzette Payne Idaho National Laboratory Yousef Bozorgnia University of California-Berkeley Paul Spudich U.S. Geological Survey Jennifer Donahue Geosyntec Consultants Dave Boore U.S. Geological Survey Olga Ktenidou ISTerre, Université Joseph Fourier - CNRS Gail Atkinson University of Western Ontario Nick Gregor Consultant John Zhao Institute of Geological and Nuclear Sciences, New Zealand Al Rohay Pacific Northwest National Laboratory Tom Pratt U.S. Geological Survey Brian Sherrod U.S. Geological Survey Rick Blakely U.S. Geological Survey Marcia McLaren Pacific Gas and Electric Co.

Steve Reidel Washington State University Judy Zachariasen URS Corporation, Tyler Ladins Humboldt State University Craig Weaver U.S. Geological Survey 14

Participatory Peer Review Panel (PPRP)

Bill Lettis Brian Chiou Woody Savage Ken Campbell, Chair Carl Stepp 15

Goal of a SSHAC Process The fundamental goal of a SSHAC process is to properly carry out and completely document the activities of evaluation and integration, defined as:

Evaluation: The consideration of the complete set of data, models, and methods proposed by the larger technical community that are relevant to the hazard analysis.

Integration: Representing the center, body, and range of technically defensible interpretations in light of the evaluation process (i.e.,

informed by the assessment of existing data, models, and methods).

NUREG-2117 16

NUREG-2117 17

18 TI Team Working Meetings (WM)

WM1 WM2 WM3 WM4

• 3-4 days duration

• All team members

• Conference room with GIS support 19

• PPRP observers

Requirements for SSHAC Level 3 PPRP PPRP Roles and Responsibilities Technical review: ensure that the full range of data, models, and methods have been duly considered in the assessment and all technical decisions are adequately justified and documented Process review: ensure that the project conforms to the requirements of the selected SSHAC level Provide timely perspectives and advice regarding the manner in which ongoing activities can be improved or carried out more effectively Be present at all the formal workshops as observers and subsequently submit a consensus report containing comments, questions, and suggestions Provide one or more representatives of the PPRP to attend as observers the working meetings of the TI Teams Perform detailed review of all project documentation and provide written comments to ensure complete technical justification of integrated distribution Prepare PPRP Closure Letter providing final technical and process review 20

Hanford PPRP Major Activities Review of Project Plan and attendance at Kick-Off Full PPRP present at all 3 Workshops PPRP representative as observers at all 8 Working Meetings PPRP encouraged to interrogate TI Teams on their preliminary models at WS3 PPRP Briefing to review Final SSC and GMC models Review of Draft Report Preparation of PPRP 21 Closure Letter

PPRP Closure Letter November 15, 2014 22

Seismic Source Characterization Model Kevin J. Coppersmith Coppersmith Consulting, Inc.

Hanford Site-Wide SSHAC Level 3 PSHA 24

SSC-Related Activities Compilation of extensive geologic/geophysical/tectonic database Update and analysis of earthquake catalogs: crustal and subduction zone Identification of seismic source zones and future earthquake characteristics Structural geologic and Quaternary analyses of Yakima folds Assessments of behavioral characteristics of fault sources including segmentation and slip rates Incorporation of associated uncertainties, including both aleatory and epistemic components 25

Seismic Sources in SSC Model Cascadia Subduction Zone sources Plate interface Intraslab source Seismic source zones YFTB zone: serves as a background zone to fault sources Zones B, C, and D Fault sources within Yakima Fold and Thrust Belt (YFTB) 19 faults characterized More distant faults are implicitly included in source zones 26

New Data Collection and Analyses Focused studies and analyses designed to reduce uncertainties in key SSC and GMC issues, within the project schedule and budget GMC-related Velocities at recording sites Analyses of kappa Analyses of basin effects SSC-related Structural analyses of Yakima folds Quaternary geologic studies High-resolution earthquake relocation analyses 27

Seismic Source Characterization, Focus Area 1

1. Summarize the information used to constrain the slip rates on the YFTB faults, including:

a. Methodology used to evaluate fault slip from topography including associated uncertainties in the ages and offsets.

c. How potential effects of surficial erosion were accounted for in the use of an average topographic profile to represent structural relief in individual faults.

Pertains to these sections of the report Section 5.2.1 Structural Analyses Section 8.4.3 Fault Characteristics Included in the SSC Model Appendix E, Section 5 Evaluation of Long-Term Structural Relief 28

Steps in Characterizing Fault Sources Measure topographic relief along lengths Define segments for use in estimating Mchar, Mmax Alternative rupture length, area relationships Identify polygons on DEM defining fault-related deformation within seismogenic crust Compare topo relief with structural relief from boreholes: Agree.

Define fault dip and uncertainties, given mapped fault location, seismogenic thickness, and polygon Max, average, 60% of polygon width defines uncertainty in fault dip and downdip width for given thickness Include geologic indicators of style of faulting to derive net slip Alternative start times of deformation 10my, 6my for slip rate Compare with Quaternary rates where known: Agree.

29

Steps in Characterizing Fault Sources (continued)

Magnitude frequency distribution model Compare with fault-related seismicity: Characteristic earthquake model supported Incorporate recurrence interval data, if available Renewal model: elapsed times and alpha unknown Future ruptures: magnitude-dependent rupture areas, lengths can extend across segment boundaries 30

Topographic Analysis of Folds:

Defines surface evidence of fault-related deformation within seismogenic crust 31

Identification of Polygons &

Possible Segmentation Points 32

Comparison of Topographic Relief with Structural Relief 33

Fault Source Fault Source Acronym Mean Structural Relief (m)

Ahtanum Ridge AR 330 Cleman Mountain CM 650 Columbia Hills-Central-East CH-C-E 55 Mean Structural Relief Columbia Hills-East Columbia Hills-West CH-E CH-W 105 375 Columbia Hills-Central CH-C 135 Columbia Hills-Central-West CH-C-W 230 Frenchman Hills-East FH-E 155 Shape of segments Frenchman Hills-West Horn Rapids Fault FH-W HR 155 90 and slip distribution is Horse Heaven Hills-Central Horse Heaven Hills-Central-East HHH-C HHH-C-E 485 415 Horse Heaven Hills-Central-West HHH-C-W 575 likely the result of Horse Heaven Hills-East Horse Heaven Hills-West HHH-E HHH-W 205 270 repeated rupture Manastash Ridge-Central Manastash Ridge-East MR-C MR-E 300 145 Average is appropriate Manastash Ridge-West Rattles MR-W RAW 415 130 Rattlesnake Hills RH 335 for assessing slip rate Rattlesnake Mountain Saddle Mountains-East RM SM-E 619 320 when have multiple Saddle Mountains-West Selah Butte SM-W SB 335 460 Toppenish Ridge-East TR-E 300 measurements along a Toppenish Ridge-West Umtanum Ridge-Central TR-W UR-C 310 360 fault segment Umtanum Ridge-East Umtanum Ridge-Southeast Anticline UR-E UR-SA 250 90 Umtanum Ridge-West UR-W 400 Umtanum-Gable Mountain U-GM 160 Wallula Fault WF 250 Yakima Ridge-East YR-E 325 Yakima Ridge-West YR-W 250 Yakima Ridge-Southeast YR-SE 65 34

Uncertainty in Dip Included in Logic Trees 35

Logic Tree Elements Related to Slip Rate 36

Logic Tree for Rattlesnake Mtn Fault Source 37

Net Slip Rate Distributions 38

Seismic Source Characterization, Focus Area 1

1. Summarize the information used to constrain the slip rates on the YFTB faults, including:

b. Rationale for excluding thin skinned seismo-tectonic models.

d. Bases for excluding listric fault geometries or potential for backthrust structures in structural relief model.

Pertains to these sections of the report Section 4.1 Tectonic Setting Section 4.4.1.2 Instrumental Seismicity Section 6.3 Epicentral Locations Section 8.4.3.4 Fault Dip Appendix E, Section 6 Kinematic Analyses Using Earthquake Focal Mechanisms 39

Thin-skinned vs. Thick-skinned Thin-skinned model called for mechanical decoupling of faults within Columbia River Basalts with those in crystalline basement Former proponents for those models participated in WS1 and WS2 and have abandoned that model Multiple arguments for discounting thin-skinned model are given in pp. 4.5 to 4.10 Additional support was provided by high-resolution earthquake locations, which show no lack of seismicity at sub-basalt sediments Focal mechanisms in CRB and crystalline basement show comparable styles of faulting and stress orientations Geophysical properties of sub-basalt sediments suggest that they are not mechanically weak 40

Listric Fault Geometries Listric geometries were not explicitly excluded A planar fault is the simplest geometry and is consistent with the back-limb geometry of essentially all of the YFB folds Historical large-magnitude thrust earthquakes show essentially planar rupture surfaces Backthrusts would have been included if there was good mapped evidence for their presence, and that they extend to the base of the seismogenic crust (i.e., are not confined to shallow depths in the hanging wall of another fault) 41

CGS Ground Motion Characterization Robert Youngs Amec Foster Wheeler

GMC Approach Use appropriate GMPEs to develop distribution of predicted ground motions Select and appropriate GMPE shape to use as a backbone model Develop distribution of scaling factors to adjust the backbone to represent the distribution of predicted motions Develop scaling factors to adjust for unique site conditions at Hanford 43

Hanford Profile

Several 100 feet of sands and gravels, some with cementation

Sequence of basalt flows with well defined and relatively thick sedimentary interbeds - Saddle Mtns Basalts and Ellensburg Formation interbeds

Several km of more massive basalts, Wanapum and Grand Ronde

Several km of sedimentary rocks above crystalline basement rocks Waste Treatment Plant (near Site A) 44

Selection of Reference Horizon for Site-wide Study Initial concept to use top of SMB/Interbed sequence However, site response analyses indicate that treatment of the SMB/Interbed sequence as a halfspace produced different surface motions than obtained from explicitly modeling basalts and interbeds Reference horizon moved down to top of massive Wanapum basalts (top of Lolo flow with flowtop removed) 45

Implications for Downstream Site Response Analyses Assessment of baserock properties required assessment of properties of overlying SMB/Interbeds In order to maintain consistency in downstream use of baserock hazard, properties of SMB/Interbeds specified for use in site response based on properties used to develop GMC Uncertainties in damping within SMB/Interbeds incorporated into uncertainty in baserock hazard in order to minimize computation burden for subsequent analyses.

46

Crustal Earthquakes GMPEs Primary sources of hazard from shallow crustal earthquakes are reverse and reverse-oblique faulting mechanisms for sites that may be located in the hanging wall Selected candidate models that best represent these types of earthquakes - NGAW2 models, particularly those that explicitly include HW effects Select one of the candidate models to use as the backbone Compute distribution of ground motion predictions from candidate models relative to the selected backbone model Develop distribution of scaling factors to center the backbone model and represent the distribution of ground motion predictions 47

First develop footwall model using ASK14, BSSA14, CB14, and CY14 GMPEs Compute predicted ground motions for a range of magnitudes, distances, fault dips, and depths to top of rupture Compute residuals for all of the predictions as residual = ln(PSA)_i - E[ln(PSA) for 4 NGAW2]

for all 4 NGAW2 models Represent epistemic uncertainty in adjustment from CY14 by a mixed effects model ln(Y ) c1F c2 F {M 6.5} c1R , f ( M ) c2 R {M 6.5}

Fixed coefficients represent change from CY14 to average of selected 4 NGA West2 GMPEs Random coefficients represent variability in scale factors from individual GMPEs to average 48

Represent Distribution of Scale Factors Discretely Use period independent scale factors for T <= 2 seconds Scale factors nearly period independent for periods that contribute to hazard Allows use of a common Vs-kappa correction Use 9-point approximation of 2-D Gaussian distribution Account for correlation between random scale factor and random adjustment to magnitude scaling 49

Resulting Footwall Models 50

Develop Hanging Wall Adjustments Use ASK14, CB14, and CY14 HW factors Compute average scale factor for RJB = 0 sites as difference between mean HW factor [in lnPSA)] and CY14 HW factor Model HW adjustments with function form HWadjustment 1 p4 cos( ) ln p5 coshp6 max(ln(RX / p7 ),0 Compute coefficients for mean adjustment and sigma of adjustments Add the mean adjustment to the mean FW adjustment RSS the HW sigma and the sigma of fitting the mean adjustment in with the FW random C1 component 51

Crustal GMPE Logic Tree 52

Subduction Zone GMPE Selected backbone model BC Hydro model (Abrahamson et al.,

2014)

Developed as part of a SSHAC Level 3 study as a response to shortcomings of existing relationship Global dataset Includes epistemic uncertainty in magnitude scaling Includes forearc/backarc scaling 53

Modifications to BC Hydro GMPE to Address Ground Motions at Large Distances Modify dataset Re-evaluate censoring of data Additional data (KiKnet, Arango et al. for Central America, Maule EQ)

Remove Taiwan data for sites where forearc/backarc is unknown Use Arango et al. data for El Salvador Earthquake Exclude Tohoku mainshock and aftershocks Attenuation rate at high frequencies is high - low motions at large distances Still used to constrain epistemic uncertainty in large-magnitude scaling 54

Modifications to BC Hydro GMPE (2 of 2)

Modify functional form controlling anelastic attenuation Assess modified model coefficients applying higher weight to data at distances > 200 km Include forearc/backarc scaling uncertainty Nisqually earthquake data from Hanford site consistent with BC Hydro backarc attenuation predictions Regions other than Japan do not show clear difference in forearc/back arc, but data is often limited Numerical simulation based model of Atkinson and Macias (2009) derived using a Q model similar to that of Phillips et al (2014) for the Cascadia-Hanford travel path show low attenuation similar to BC Hydro Forearc model 55

Subduction Zone GMPE Logic Tree 56

Adjustments to Hanford Baserock Conditions For Crustal GMPE adjust for Vs and kappa For Subduction zone GMPE adjust for Vs only Vs and Vs-kappa adjustments made using the Inverse Random Vibration Theory Approach of Al Atik et al. (2013) 57

GMC, Focus Area 2a:

Provide additional detail on the Vs-Kappa corrections applied to the scaled backbone GMPEs - Rational for not applying a kappa correction for the subduction zone GMMs Hazard sensitivity analyses indicated that the distant, large magnitude Cascadia subduction zone interface (CSZ) earthquakes primarily contribute to the hazard at low frequencies where the effects of kappa are small Because of the large distance to the CSZ (>200 km), the effects of Q are expected to dominate over the effects of kappa. Therefore, incorporation of uncertainty in Q was the focus Because the subduction zone GMMs are defined primarily by data at large distances, separation of kappa effects from Q effects in the host GMMs is difficult.

58

GMC, Focus Area 1 (1 of 10)

Provide additional detail on the process used to define the target site kappa values and their uncertainties, including the rationale for logic tree weightings.

Utilize recordings of small earthquakes from 6 sites located on basalt or a few meters of soil over basalt to assess kappa for basalts Some sites located on SMB with interbeds, some on outcropping layers of deeper basalts Apply alternative approaches to estimate site kappa and its uncertainty Inversion of Fourier spectra of recordings Anderson and Hough (1984)

Apply kappa estimates in forward sense considering uncertainty in contribution of deeper sediments to kappa for large earthquakes 59

GMC, Focus Area 1 (2 of 10): Target site kappa Target kappa logic tree Uncertainty in Vs profile Alternative approaches Uncertainty in estimates from each approach Uncertainty in depth range contributing to kappa for larger earthquakes 60

GMC, Focus Area 1 (3 of 10): Target site kappa Alternative Vs Profiles Differ in SMB Vs Differ in subbasalt sediment Vs Favor Profile 1 2:1 over Profile 2 - prefer downhole Vs measurements over suspension logging in basalt Difference in subbasalt Vs had small impact on assessments 61

GMC, Focus Area 1 (4 of 10): Target site kappa kappa estimated by inversion Used recordings from 15 earthquakes recorded in 2005 to 2013; 10 recordings at HAWA from 2004 study Hypocenter depths 5 km to avoid double paths accelerograph recordings at HAWA; all other recordings were on BB velocity instruments Inversion process estimates kappa along with source (fC) and path (Q) parameters by nonlinear least-squares fit to FAS using point-source model Q(f) was fixed at 500(f)0.6 due to limited distance range and limited bandwidth Determined parameters: fC, kappa 62

GMC, Focus Area 1 (5 of 10): Target site kappa Assessed correlation of kappa assessed from inversion with:

Thickness of SMB Total thickness of interbeds Thickness of subbasalt sediments Conclusion that entire profile above basement contributed to kappa from the small, deeper earthquakes 63

GMC, Focus Area 1 (6 of 10): Target site kappa Anderson and Hough (1984) approach:

Select data for R < 200 km that show linear trend in FAS at high frequencies Select frequency window for linear trend Smooth FAS and noise with Konno and Omachi (1998) filter Select records with signal to noise ratio > 3 Estimate kappa by fitting a(f)=A0exp(-f)

Extrapolate trend with distance back to 0 distance 64

GMC, Focus Area 1 (7 of 10): Target site kappa Assess site kappa removing effects of shallow soils and scattering from SMB interbeds Soil kappa assessed using Campbell (2009)

Scattering kappa assessed by comparing response of an equivalent uniform Vs profile with damping to layered profile without damping Assuming Qs = Vs develop estimates of to assign kappa to SMB and to deeper layers Based on = H/Vs/Q 65

GMC, Focus Area 1 (8 of 10): Target site kappa Favored inversion over A&H 2:1 Based on fit of broader frequency range of FAS A&H produced variable assessments of Qs Application of A&H method required use of shallow earthquakes with potential multiple paths Epistemic uncertainty in A&H kappa assessed based on statistics of fit Epistemic uncertainty in Inversion kappa based on assessments of parametric variations in inversion parameters Produced an asymmetric distribution In addition, best estimate kappa based on Q(f) larger than Phillips et al. (2014), thus may be biased high. Weights adjusted to account for potential bias (lower Q would lead to lower kappa) 66

GMC, Focus Area 1 (9 of 10): Target site kappa Final component - depth extent of subbasalt sediment contribution Results for small, deep earthquakes indicates all of sediments No data available to assess contribution for shallower earthquakes or large earthquakes Considered three equally weighted alternatives All, half, or none 67

GMC, Focus Area 1 (10 of 10): Target site kappa Resulting baserock kappa distribution 68

GMC, Focus Area 2b:

Provide additional detail on the Vs-Kappa corrections applied to the scaled backbone GMPEs - Comparison of the Vs-kappa scaled median GMPEs and NGA-West 2 models M 7, Rrup 11 km, 5.00 HW Red curve - Median Crustal for VS30 760 0.50 1.00 Black curves - Median PSA Crustal scaled by Vs-kappa 0.05 0.10 corrections 0.01 0.01 0.05 0.10 0.50 1.00 5.00 10.00 Period 69

GMC, Focus Area 2c:

Provide additional detail on the Vs-Kappa corrections applied to the scaled backbone GMPEs - Whether any observational data from the region was used to assess the Vs-kappa corrected GMPEs Available ground motion data from small crustal earthquakes in the region recorded on basalts or shallow soil over basalt were used to assess kappa Recorded crustal earthquake data is generally from earthquakes too small to make meaningful comparisons with the developed GMPEs Data from the M 6.8 Nisqually earthquake was used to assess attenuation from distant CSZ earthquakes 70

Aleatory Variability Model Utilized single station sigma concept Development very similar to that of the SWUS model with similar results Specified minimum level of epistemic uncertainty in characterizing site response to address:

Lack of variability in site response at very low frequencies Possible basin effects in surficial sediments at intermediate periods 71

GMC, Focus Area 3: (1 of 5)

Provide additional discussion regarding bases for bump seen at T=0.1sec in the mean tau values for the NGA-West2 models and the decision for smoothing through this peak in developing single-station sigma.

A peak in the event-to-event variability () is commonly seen in the results of analyses of empirical strong motion data is a measure of the differences in the average motions from earthquake to earthquake May be due to source differences May be due to differences in average site conditions for each earthquake Possible mechanisms for peak explored Using point-source stochastic simulations Examination of data from limited geographical regions 72

GMC, Focus Area 3: (2 of 5) Bump in at T=0.1sec.

Performed point-source stochastic model simulations Random variation in stress parameter (stress drop)

Random variation in site Simulated motions 200 earthquakes with lognormally distributed stress drops For each earthquake, ground motions at 25 sites with lognormally distributed site Fit results with mixed effects model to compute variance components and Two cases analyzed No correlation in site between earthquakes -- produces a peak in (within event variability) but not in Half of the variance is assigned to event-to event variability in median and remaining to within-event site-to-site variability in -- produces a peak in both and 73

GMC, Focus Area 3: (3 of 5) Bump in at T=0.1sec.

Results of fitting mixed effects model to simulations 74

GMC, Focus Area 3: (4 of 5) Bump in at T=0.1sec.

Fits to CY14 residuals for only California data Inclusion of site-to-site variability term shifts peak at 10 Hz from to S2S for M < 5.5 Less conclusive for larger M, but limited data from sites recording multiple earthquakes 75

GMC, Focus Area 3: (5 of 5) Bump in at T=0.1sec.

Conclusion - peak in near 10 Hz (0.1 s) likely due to differences in average site conditions from earthquake to earthquake rather than variability in earthquake source properties Uncertainty in site effects for Hanford sites is explicitly addressed in development of GMC model Therefore, peak in is smoothed through in developing aleatory variability model 76

Provided Recommended Vertical/Horizontal Spectral Ratios for Development of Surface Vertical Spectra Gulerce and Abrahamson (2011) modified at long periods to match trends in data from the Maule earthquake 77

Baserock Hazard Results at Site C - CGS 78

Source Contributions 79

Contributions to Uncertainty (1 of 2) 80

Contributions to Uncertainty (2 of 2) 81

Development of Input Motions for Site

Response

Deaggregated mean hazard at each frequency Developed 4 scenarios to represent deaggregation M, R and weight of event varies Developed conditional mean spectrum for each scenario 82

CGS Site Response Farhang Ostadan, Bechtel Robert Youngs, Amec Foster Wheeler

Site Profile

Site profile consists of two zones:

1. Upper sands and gravels (525 ft thick)

2. Sequence of basalt flows (765 ft thick) to top of Lolo Flow

Sand and gravel properties based on FSAR data

Basalt flows characterized in PSHA study

Simulated profiles generated for each zone and then joined Labeled depths not for CGS 84

Upper Sands and Gravels Profile Velocities obtained by cross-hole and down-hole measurements at CGS (formerly WNP-2) and nearby WNP-1 and WNP-4 Considered:

one velocity profile with ln 0.15 to 0.30 EPRI and Peninsular Range (PR) nonlinear curves (D 15%)

85

Site response, Focus Area 1a Provide additional detail regarding the bases for only a single profile for the upper 525 ft.

The geology and depositional processes are also known from the nearby DOE facilities in Hanford (e.g. WTP)

The soil properties, specifically shear-wave velocity data, are consistent with the soil type and density The distinctive velocity profile in the top 525 ft is common to the 3 adjacent sites (CGS, WNP-1 and WNP-4)

Similar Vs measurements were obtained using different methods with different instruments (cross-hole & down-hole)

Seismic refraction measurements agreed well with cross-hole measurements in the top 105 ft at CGS Given the relatively small amount of variation among the measured results, there is a high degree of confidence that any additional measurements would fall close to the base case Vs profile (i.e., low epistemic uncertainty)

The variation in soil velocity is adequately covered by consideration of the aleatory uncertainty 86

Site response, Focus Area 1d Provide additional detail regarding the adequacy of EPRI and Peninsular curves for covering range of nonlinear behavior for the Pasco Gravel.

As recommended by the SPID, the EPRI and Peninsular Range curves were used These two sets of nonlinear curves are considered to span the range of nonlinearity for cohesionless soils More recent RCTS data (from other sites) often confirm the adequacy of generic curves for sands and gravels The generic curves were considered appropriate for screening analyses 87

Basalt Flows Profile Provided by PSHA study Velocities measured with PS logging and Downhole (DH) methods Two alternative profiles Weighted 2:1 (DH:PS)

Nonlinear curves for interbeds computed with Darendeli (01)

(D 15%)

Site attenuation () calibrated by recordings 88

Site response, Focus Area 1b Provide additional detail regarding the bases for two Vs profiles and their associated weights for the SMB stack.

For interbeds, DH and PS Vs values in agreement For basalts Vs from PS ~25% higher than Vs from DH Two Vs profiles created to capture epistemic uncertainty in basalt Vs Vs based on DH favored 2:1 over Vs based on PS DH Vs measured a frequencies near those of interest while PS Vs measured at 1 kHz Experts suggested that reliable PS values in stiff basalts may require use of 5-10 kHz frequencies 89

Site response, Focus Area 1c (1 of 2)

Provide additional detail regarding the thickness of the interbed deposits including their lateral extent.

Based on numerous deep borings, thickness and extent of Ellensburg Formation interbeds mapped across Hanford site Example for Mabton interbed from Rohay and Reidel (2005) 90

Site response, Focus Area 1c (2 of 2)

Provide additional detail regarding the thickness of the interbed deposits including their lateral extent Deep wells used to develop SMB stratigraphy at Site C (CGS). Details are in Last (2014)

Uncertainty/variability in thickness was included in randomization of layer thicknesses HSPSHA Fig. 7.11 91

Site response, Focus Area 1e Provide additional detail regarding the bases for randomizing the small strain damping for the basalt layers in the SMB stack in view of the uncertainty already incorporated into the determination of the site kappa.

Randomization is used to address Uncertainty in properties due to measurement error and extrapolation from other locations To characterize spatial variability within the facility footprint.

To compensate for simplifications used in conventional 1-D site response analysis methodology Randomization would be applied in SPID methodology even if median properties were known with a high degree of certainty 92

Site response, Focus Area 2 In view of the relatively high shear wave velocities (> 760 m/s) and high confining stresses in the interbed layers, provide the rationale for not considering linear behavior of these materials and instead using a single sand curve which demonstrates significant non-linear behavior.

Interbeds are relatively soft rock-like materials even at depth Vs < 1000 m/s Some degree of non-linearity expected in such materials at high loading levels, especially because of the large velocity contrasts with the basalts and resulting strain concentrations.

Selected the Darendeli (2001) model because it provided a means of incorporating the effect of depth of the interbeds on the G/Gmax and damping relationships.

Used only one set of curves with randomization because differences in average response between using alternative sets of median relationships expected to be much smaller that overall uncertainty in defining site ground motions developed for the GMC.

93

Simulated Profiles 94

Input Motions PSHA study provided CMS motions at:

20 frequencies 27 MAFEs 4 events M, R and weight of event varies Site response computed for each motion 95

Site Response Analysis Total of 2,160 input rock spectra Two velocity profiles:

1. C1 (downhole)

2. C2 (PS logging)

Two nonlinear models for the upper zone:

1. EPRI

2. PR Total of 518,400 soil column analyses 96

Site Response Performed with PSHAKE Uses random vibration theory (RVT)

Equivalent-linear wave propagation Control point defined at surface Extracted the site amplification values at the controlling frequency of the CMS motion 97

1D Site Amplification 98

1D log Mean Site Amplification 99

GMRS - Approach 3 Approach 3 (Bazzurro and Cornell, 2004):

l l Gz(z) = Horizon specific hazard curve px(xj) = probability of rock input level (i.e., slope of input hazard curve) l l l l

where 1 is the complementary standard Gaussian CDF l ) = mean amplification factor l = sigma of amplification factor 100

GMRS - Approach 3 Input Hazard Curves from PSHA Input Site Amplification Results Two Profiles:

• Profile C1 (0.67)

• Profile C2 (0.33)

Two Material Curves:

• EPRI (0.5)

• Peninsular Range (0.5)

No Minimum Amplification Factor 101

Hazard Curves: PGA 102

UHRS: Mean, AEP=10-4 103

UHRS: Mean, AEP=10-5 104

Surface UHS and GMRS 105

Site response, Focus Area 3 Provide additional detail regarding the decision not to implement a minimum site amplification value and the effect of this decision on the development of the uncertainty in the site amplification function as well as on the final hazard curves for the site.

The 0.5 limit is not used in the calculation of the surface hazard because the intended purpose of this report is to obtain the realistic and unbiased median based site amplification for SRA The basis for limitation of 0.50 is not well documented and is more relevant for additional conservatism under design applications 106

Interim Actions and Evaluation Dave Swank Michael Kennedy Energy Northwest

Interim Actions - FLEX strategy Columbia has significant flexibility for connection of portable electrical supplies following a seismic event Two portable 480 VAC diesel generators on-site Two redundant 480 VAC connection points to two safety divisions One additional 480 VAC connection point to either of two safety divisions 4160 VAC diesel generators available through SAFER Two 4160 VAC connection points to two safety divisions One additional 4160 VAC connection point capable of supplying either safety division Installed breakers and cabling allow safety divisions to be cross-tied if needed 108

Interim Actions - FLEX strategy Columbia has significant flexibility for connection of portable pumps to supply cooling water to the core and suppression pool Two portable diesel powered pumps (high head) available onsite Injection points available on all three divisions of Residual Heat Removal (RHR)

High capacity pump available through SAFER Injection point available on the Condensate system Connection points on the both divisions of Standby Service Water (SW) with piping cross-connected to RHR Inventory source is the normal plant ultimate heat sink (two SW spray ponds) 109

Interim Evaluation - Seismic Current Seismic Probabilistic Risk Assessment (SPRA) - Mean Seismic Core Damage Frequency is calculated to be 4.9 x10-6 Based upon a hazard curve developed in terms of peak ground acceleration (PGA) and fragilities in terms of PGA Current SPRA maintained in alignment with the RG 1.200-compliant Internal Events modeling and is significantly advanced beyond IPEEE requirements Reanalysis of Mean Seismic Core Damage Frequency at PGA using new ground motion response spectrum indicates margin remains Seismic walk downs completed satisfactory as required by Enclosure 3 of the NRCs CFR 50.54(f) Request for Information All results satisfactory 110

Path Forward Energy Northwest is initiating the Expedited Seismic Evaluation Process Report due Jan 2016 Energy Northwest is initiating SPRA in accordance with SPID requirements Submittal due Jun 2017 111