ML18306A499

Technical Basis for Updating Fcse Seismic Hazard Analysis Guidance Final

ML18306A499
Person / Time
Issue date:	11/20/2018
From:	Patrick Koch NRC/NMSS/DFCSE/LOB
To:	Margie Kotzalas NRC/NMSS/DFCSE/LOB
Koch P
Shared Package
ML18306A479	List: ML18306A483 ML18306A499
References
NRC-50-14-E-0001
Download: ML18306A499 (35)

Text

TECHNICAL BASIS FOR UPDATING FCSE SEISMIC HAZARD ANALYSIS GUIDANCE Prepared for U.S. Nuclear Regulatory Commission Contract NRC-50-14-E-0001 Prepared by John Stamatakos and Miriam Juckett Center for Nuclear Waste Regulatory Analyses San Antonio, Texas September 2018

ABSTRACT In this technical report, we evaluated the currently acceptable approaches employed by licensees and license applicants to develop the seismic hazard analyses that are reviewed by the U.S. Nuclear Regulatory Commission (NRC) as part of licensing and safety evaluations of fuel fabrication facilities regulated under Title 10 of the Code of Federal Regulation (10 CFR)

Part 70. As described in NUREG-1520, acceptable seismic hazard assessment methods for fuel cycle facilities include a range of options, such as U.S. Geological Survey (USGS) seismic hazard maps, the USGS seismic hazard mapping tool, or an approximated hazard by using the Regulatory Guide (RG) 1.60 spectra anchored at the Safe Shutdown Earthquake of the nearest nuclear power plant. In this report, we compare the results of these methods to site-specific fully probabilistic seismic hazard analyses at four sites in the central and eastern United States.

These four sites were selected to represent a range in site conditions, from hard bedrock to thick soft soil, and relative hazard potential, from a high hazard area near the center of the New Madrid seismic zone to a low hazard site in eastern Pennsylvania. The results of the evaluation and associated recommendations described in this report are not intended to substitute for other important aspects of fuel cycle safety, nor are the results intended to challenge existing technical bases for seismic safety at NRC-regulated fuel cycle facilities.

However, our results demonstrate the benefits of updating NRCs seismic hazard guidance for fuel cycle facilities to include fully probabilistic and site-specific seismic hazard analyses as a method for assessing seismic hazard. These updates will harmonize NRCs guidance on seismic hazard analyses with the current state of practice and across the agency, increase technical confidence and risk consistency, and support the NRCs goal of risk-informed, performance-based regulations.

ii

CONTENTS ABSTRACT ................................................................................................................................. ii FIGURES ................................................................................................................................... iv TABLES ...................................................................................................................................... v ABBREVIATIONS/ACRONYMS ................................................................................................. vi ACKNOWLEDGMENTS............................................................................................................ vii 1 INTRODUCTION ..............................................................................................................1-1 1.1 Purpose ...................................................................................................................1-1 1.2 Scope ......................................................................................................................1-2 1.3 Approach .................................................................................................................1-2 2 REGULATORY FRAMEWORK AND EXISTING APPROACHES FOR SEISMIC HAZARD ANALYSES .......................................................................................................2-1 2.1 Nuclear Power Plants ..............................................................................................2-1 2.2 NRC Fuel Cycle Facilities ........................................................................................2-2 2.3 Department of Energy Facilities ...............................................................................2-3 2.4 Strengths and Weaknesses of the Different Approaches .........................................2-4 3 METHODOLOGY ..............................................................................................................3-1 4 RESULTS .........................................................................................................................4-1 4.1 Columbia Airport ......................................................................................................4-1 4.2 Clarks Hill ................................................................................................................4-3 4.3 Lehigh University .....................................................................................................4-5 4.4 St. Francis Bridge ....................................................................................................4-7 5 DISCUSSION AND RECOMMENDATIONS......................................................................5-1 5.1 Discussion ...............................................................................................................5-1 5.2 Specific Recommendations......................................................................................5-2 6 REFERENCES .................................................................................................................6-1 iii

FIGURES Figure 3-1. Map of the eastern United States showing the location of the four sites analyzed in this report ........................................................................................ 3-1 Figure 3-2. Method used to estimate the equivalent AEP and the ratio between the equivalent AEP and the fully-probabilistic hazard value. Sa is the spectral acceleration. ....................................................................................................... 3-4 Figure 4-1. Geologic map of the state of Missouri showing the location of the Columbia and St. Francis Bridge sites relative to the Callaway Nuclear Generating Station ............................................................................................. 4-1 Figure 4-2. Fully probabilistic and USGS seismic hazard results for the Columbia, Missouri site ....................................................................................................... 4-2 Figure 4-3. Uniform hazard response spectra (UHRS) for the Columbia, Missouri site ......... 4-2 Figure 4-4. Geologic map of the southeastern U.S. showing the location of the Clarks Hill site relative to the MOX site at Savannah River and the Vogtle Electric Generating Station ...................................................................... 4-3 Figure 4-5. Fully probabilistic and USGS seismic hazard results for the Clarks Hill South Carolina site ............................................................................................. 4-4 Figure 4-6. Uniform hazard response spectra (UHRS) for the Clarks Hill, South Carolina site ............................................................................................. 4-4 Figure 4-7. Location map for the Lehigh University site in eastern Pennsylvania, relative to the Limerick Generating Station ......................................................... 4-5 Figure 4-8. Fully probabilistic and USGS seismic hazard results for the Lehigh University site ......................................................................................... 4-6 Figure 4-9. Uniform hazard response spectra (UHRS) for the Lehigh University site ............ 4-6 Figure 4-10. Geologic cross section beneath the Rte. 60 St. Francis Bridge. The cross section depths are shown here in ft. To convert to meters, divide the values in feet by 3.28. Adapted from Figure 7.2 of Anderson et al. (2001)......... 4-7 Figure 4-11. Example VS profile beneath the St. Francis Bridge from the MODOT SASW analysis................................................................................................... 4-8 Figure 4-12. Fully probabilistic and USGS seismic hazard results for St. Francis Bridge, Missouri site ....................................................................................................... 4-9 Figure 4-13. Uniform hazard response spectra (UHRS) for the St. Francis Bridge site ........... 4-9 iv

TABLES Table 3-1. Location and description of the four evaluated sites ........................................... 3-1 Table 4-1. Comparison of the NRC, USGS, and RG 1.60 Spectrum results for Columbia, Missouri ............................................................................................. 4-1 Table 4-2. Comparison of the NRC, USGS, and RG 1.60 spectrum results for Clarks Hill, South Carolina.................................................................................. 4-5 Table 4-3. Comparison of the NRC, USGS, and RG 1.60 spectrum results for Lehigh University, Pennsylvania ......................................................................... 4-7 Table 4-4. Comparison of the NRC, USGS, and RG 1.60 Spectrum results for St. Francis Bridge site, Missouri ....................................................................... 4-10 Table 5-1. Summary of USGS results ................................................................................. 5-1 Table 5-2. Summary of RG 1.60 results .............................................................................. 5-1 v

ABBREVIATIONS/ACRONYMS 1D one-dimensional AEP annual exceedance probabilities ASCE American Society of Civil Engineers CEUS-SSC central and eastern United States seismic source characterization CFR Code of Federal Regulations CNWRA Center for Nuclear Waste Regulatory Analyses DOE U.S. Department of Energy EPRI Electric Power Research Institute GMM ground motion models ISA integrated safety analysis MODOT Missouri Department of Transportation MOX Mixed Oxide NEHRP National Earthquake Hazard Reduction Program NGA next generation attenuation NPP nuclear power plant NRC U.S. Nuclear Regulatory Commission NSF National Science Foundation NTTF near term task force PEER Pacific Earthquake Engineering Research PGA peak ground acceleration PSHA probabilistic seismic hazard analysis RG Regulatory Guide RLME repeated large magnitude earthquake RVT random vitiation theory SASW spectral analysis of surface waves SDC seismic design categories SPRA seismic probabilistic risk assessment SRA site response analysis SSC seismic source characterization SSE safe shutdown earthquake SSHAC Senior Seismic Hazard Analysis Committee SwRI Southwest Research Institute UHRS uniform hazard response spectra U.S. United States USGS U.S. Geological Survey vi

ACKNOWLEDGMENTS This report was prepared to document work performed by the Center for Nuclear Waste Regulatory Analyses (CNWRA) for the U.S. Nuclear Regulatory Commission (NRC) under Contract No. NRC-50-14-E-0001. The activities reported here were performed on behalf of the NRC Office of Nuclear Material Safety and Safeguards. The report is an independent product of CNWRA and does not necessarily reflect the views or regulatory position of the NRC.

We thank Cliff Munson, Jon Ake, Patrick Koch, and Jonathan Marcano of the U.S. NRC for their support on this project. Appreciation is extended to Lora Neill and Nora Naukam for assistance in preparation of the report and Biswajit Dasgupta for technical review.

QUALITY OF DATA, ANALYSES, AND CODE DEVELOPMENT DATA: There is no original CNWRA-generated data in this report. Sources of other data should be consulted for determining the level of quality of those data. CNWRA staff developed seismic hazard inputs using the U.S. Geological Survey (USGS) seismic hazard tool, (https://earthquake.usgs.gov/hazards/interactive/). These data were analyzed and plotted using standard functions in Excel.

ANALYSES AND CODES: NRC codes PROBHAZ_NRC (Rev 2.0) and ELSRAP were used by the NRC staff to develop the full probabilistic seismic hazard analysis (PSHA) results in this report. NRC authors supplied the resulting seismic data, which also were plotted using standard functions in Excel.

vii

1 INTRODUCTION 1.1 Purpose The purpose of this technical report is to evaluate and compare the currently accepted methods used to assess seismic hazards at sites regulated under Title 10 of the Code of Federal Regulations (10 CFR) Part 70 with site-specific and fully probabilistic methods of determining seismic hazards and to draw conclusions regarding the differences in these approaches. In addition, this report demonstrates the potential improvements in evaluating uncertainty and in quantifying performance metrics or risk analyses that can be achieved when fully probabilistic and site-specific seismic hazard analyses are used. The results of the evaluation and associated recommendations described in this report are not intended to substitute for other important aspects of fuel cycle safety. We acknowledge that there are many additional safety considerations that provide the regulatory assurance, either by prevention of accidents or through defense in depth, that existing fuel facilities can continue to operate safely and without incident, should a seismic event occur. We also acknowledge that this report does not challenge existing technical bases for seismic safety at U.S. Nuclear Regulatory Commission (NRC) regulated fuel cycle facilities. Nonetheless, our results demonstrate the benefits of updating NRCs seismic hazard guidance to include fully probabilistic and site-specific seismic hazard analyses as a method for assessing seismic hazard. These updates will harmonize NRCs guidance on seismic hazard analyses with current state of practice and across the agency, increase technical confidence and risk consistency, and support the NRCs goal of risk-informed, performance-based regulations.

Seismic hazard analyses have evolved considerably over the past two decades, applying new models and methods developed by government agencies, industry, and academia, and incorporating substantial new earthquake data from world-wide seismic events. These recent advances have enabled seismic hazard evaluations to become fully mature probabilistic assessments, such that they now provide a more reliable and defensible basis for quantitative analyses capable of supporting decision making. In addition, these analyses also quantify the uncertainty associated with these assessments. Accurate estimates of the hazard, coupled with a quantitative assessment of uncertainty, are important elements of performance-based and risk-informed decision making. In fact, NRC has recognized the value that a fully probabilistic and site-specific hazard analysis adds to implementing performance-based and risk-informed regulations. The purpose of a site-specific probabilistic seismic hazard analysis (PSHA) is to quantify the probability of exceeding various ground-motion levels at a site, given all possible future earthquakes that could cause vibratory ground motion at the site, including the uncertainties associated with this estimate of the ground shaking. In doing so, PSHA results provide reliable quantitative inputs to performance and risk assessments.

The NRC has relied on PSHA studies since the late 1970s and early 1980s, first as a research tool to better understand the ambiguities associated with deterministically-derived design ground motion parameters and the resultant seismic safety margins of nuclear power plants (NPPs). In the 1990s, as the NRC moved toward the use of risk-informed and performance-based principles, PSHA was introduced into regulatory guidance. Specifically, 10 CFR 100.23 states that because uncertainties are inherent in seismic hazard estimates, these uncertainties must be addressed through an appropriate analysis, such as a PSHA or suitable sensitivity analyses.

In recent years, PSHA, including studies that incorporate a site response analysis based on the specific geomechanical properties of soil and bedrock beneath the sites, have become much 1-1

more prevalent (although not universal) for many types of critical facilities. In addition, and as described in greater detail in Section 5.2 of this report, the availability of seismic training, computer codes, other software or web-based applications, seismic input databases, and earthquake data has increased substantially in the past few years. These increases in accessibility now allow industry consultants and engineers to relatively easily develop the kind of site-specific PSHA results that we discuss in this report.

1.2 Scope The scope of this report is limited to a comparison of methods used to develop the seismic hazard inputs to quantitative risk and performance assessments. The evaluation presented in this report demonstrates, by practical examples, the differences in seismic hazard that result from use of the currently accepted methods described in NUREG-1520 versus site-specific evaluations that are based on the most recent application tools available to develop PSHA at nuclear facilities. These tools include seismic source characterization and ground motion models based on existing Senior Seismic Hazard Analysis Committee (SSHAC) studies; specifically NUREG-2115 (NRC, 2012a) and Electric Power Research Institute (EPRI) (2013).

These tools also include site-specific and fully probabilistic site response models, following Approach 3 from NUREG/CR-6728 (Risk Engineering, 2001). We recognize that seismic safety of fuel cycle facilities includes other important design and operational elements to withstand or mitigate the potential consequences from external hazards, including conservatisms inherent in seismic designs that result from the application of existing codes and standards, such as American Society of Civil Engineers (ASCE) 43 (ASCE, 2005) and ASCE 7 (2010). The scope of this report does not include an assessment or evaluation of those important additional factors.

1.3 Approach We conducted the evaluation described in this report by comparing the seismic hazard results at four sites in the eastern United States (U.S.) using three different methods, which are discussed in greater detail in Section 3. In brief, we first determined the site-specific seismic hazard at each of the four sites based on the most recent seismic hazard codes and analyses used to develop site-specific PSHA results at all U.S. NPPs sites in response to the NRCs Near Term Task Force (NTTF) Recommendation 2.1 (NRC, 2011a). Second, we determined the hazard at each site using the most recent version of the U.S. Geological Survey (USGS) seismic hazard tool (https://earthquake.usgs.gov/hazards/interactive/). For this analysis, we assumed a site class as hard rock or firm soil, depending on the site soil conditions, per the National Earthquake Hazard Reduction Program (NEHRP) recommended provisions (Building Seismic Safety Council, 2003). Third, we approximated the hazard at each site by using the Regulatory Guide (RG) 1.60 spectra (NRC, 2014) anchored at the Peak Ground Acceleration (PGA) corresponding to the Safe Shutdown Earthquake (SSE) of the nearest NPP. Finally, we compared the results of these three methods in terms of the range in predicted annual exceedance probabilities (AEPs) using the fully probabilistic results as the reference hazard curve. The comparison of AEPs provides insights into the likelihoods for seismic initiating events that are important to evaluating seismic performance and seismic risk at a given facility or site.

1-2

2 REGULATORY FRAMEWORK AND EXISTING APPROACHES FOR SEISMIC HAZARD ANALYSES 2.1 Nuclear Power Plants Although all of the original seismic hazard analyses used at U.S. NPPs were based on deterministic criteria defined in 10 CFR Part 100, Appendix A, site-specific PSHA studies have been used at NRC regulated NPPs since the early 1980s to address emerging knowledge. For example, when USGS asserted that a Charleston-type source cannot be ruled out for the Eastern seaboard), both the NRC and industry conducted PSHAs for all central and eastern United States (CEUS) sites. These hazard results were subsequently used to address the Charleston issue. New reactor siting, which began in the 2003-2004 timeframe, is subject to the updated regulations in 10 CFR 100.23(d)(1), which require that uncertainty inherent in estimates of the SSE be addressed through an appropriate analysis, such as a PSHA.

The PSHA results were used in the 1990s to support the NRCs request to all operating plants to examine their plants for plant-specific vulnerabilities to external hazards. More recently, the most comprehensive set of PSHA results followed the seismic hazard re-evaluations at all U.S. NPPs following the 2011 accident at the Fukushima Dai-ichi NPP in Japan. In response to that accident, the NRC issued an information request (NRC, 2012b) and other orders that ultimately resulted in improvements in data, models, and methods used to develop seismic hazard analyses at U.S. NPPs. The information request and basis for the updated seismic hazard analyses are documented in the Near-Term Task Force Review of Insights from the Fukushima Dai-ichi Accident (NRC, 2011b). In particular, the NRC NTTF Recommendation 2.1 and subsequent Staff Requirements Memoranda associated with Commission Papers SECY-11-0124 (NRC, 2011c) and SECY-11-0137 (NRC, 2011d) instructed the licensees and NRC to perform a re-evaluation of the seismic hazards at all U.S. NPP sites using present-day NRC requirements. Specifically, these seismic hazard updates relied on SSHAC studies, such as NUREG/CR-6372 (NRC, 1997), EPRI (2012), NUREG-2115 (NRC, 2012a), NUREG-2117 (NRC, 2012c), NUREG/CR-6728 (Risk Engineering, 2001), and RG 1.208 (NRC, 2007).

An important outcome from the updated seismic hazard analyses is the recognition that studies are technically challenging and complex, with variability and uncertainty in inputs and results that need to be formally evaluated and quantified. As discussed in NUREG/CR-6372, the SSHAC process was designed as a formal, structured approach for incorporating the judgments of experts that support the development of reliable and stable PSHAs. Thus, the application of the SSHAC methods (or other NRC-endorsed approaches) in developing seismic source characterizations and ground motion models could provide significant benefits to licensees, applicants and regulators in terms of developing reliable results and thereby improving regulatory assurance.

Another significant technical finding from the updates to seismic hazard analyses following NTTF Recommendation 2.1 is the recognition that the local site conditions, especially the geotechnical and geomechanical properties of the soils and bedrock, have a profound influence on the resulting hazard. These properties are integral to developing accurate site response analyses. Site response modeling accounts for changes in seismic energy (amplification or deamplification, attenuation, and damping) due to the velocity and impedance contrasts in the uppermost 100 to 200 m of strata (including soils, sedimentary layers, and bedrock) as seismic waves propagate through strata directly beneath the site. Use of site-specific data to develop 2-1

fully probabilistic site response models therefore also provides significant benefit to applicants and regulators over more simplistic models that approximate the site conditions.

The NRC has used risk-informed regulatory decision-making to address seismic issues. At the heart of this risk-informed decision-making is the ability of NRC licensees to perform a seismic probabilistic risk assessment (SPRA). In fact, to implement NTTF Recommendation 2.1, the licensees with sites that had a significantly higher seismic hazard than their SSE design level ground motions were asked to conduct an SPRA and provide information that will be used in the risk-informed decision-making. The SPRA identifies dominant accident sequences and the plant vulnerabilities that contribute most to the risk. The SPRAs need to be realistic, including uncertainties, in order to provide meaningful results and robust insights. The PSHA, as a direct input to the SPRA, should also provide a realistic site-specific spectral shape. This site-specific spectral shape is a critical factor, along with the understanding of dominant earthquake-distance relationship, in subsequent engineering analysis. The American Society of Mechanical Engineers/American Nuclear Society (ASME/ANS) PRA Standard discusses the attributes and results from a PSHA that are important to a risk study.

2.2 NRC Fuel Cycle Facilities Seismic hazard assessments at fuel cycle facilities do not necessarily rely on a site-specific PSHA. As described in NUREG-1520, acceptable seismic hazard assessment methods include a range of options such as USGS seismic hazard maps or the USGS seismic hazard mapping tool (https://earthquake.usgs.gov/hazards/interactive/), seismic design specifications in building codes (e.g., IBC, 2012; or ASCE/SEI 7 Standard, 2010), or even the seismic design values from the nearest NPP.

In particular, 10 CFR 70.62(c)(1) requires each licensee or applicant shall conduct and maintain an integrated safety analysis, that is of appropriate detail for the complexity of the process, that identifies:

• Potential accident sequences caused by process deviations or other events internal to the facility and credible external events, including natural phenomena;

• The consequence and the likelihood of occurrence of each potential accident sequence identified pursuant to paragraph (c)(1)(iv) of 10 CFR 70.62, and the methods used to determine the consequences and likelihoods.

For new facilities or new processes at existing facilities, 10 CFR 70.64(a), Baseline design criteria, requires, in part, that the design must provide for adequate protection against natural phenomena with consideration of the most severe documented historical events for the site.

For facilities processing special nuclear materials, 10 CFR 70.61, Performance Requirements, requires that individual accident sequences resulting in high consequences to workers and the public be highly unlikely and that sequences resulting in intermediate consequences to these receptors be unlikely. Definitions of what constitutes unlikely, highly unlikely, and credible accidents or event sequences are defined in the applicants Integrated Safety Analysis (ISA)

Summary, per the requirement in 10 CFR 70.65.

A critical factor in establishing the performance requirements for fuel cycle facilities regulated under 10 CFR Part 70 is the probability of an initiating event that could cause an accident sequence with credible high or intermediate consequences to the public or workers. The 2-2

initiating events in seismic event sequences can be the probability of exceeding a given level of ground motion at the site or the probability that seismic ground motion at the site causes a key system, structure, or component to fail. However, to adequately assess the risk associated with a seismic event, the resulting accident sequences need to be considered. As described in NUREG-1520, Appendix C, accident sequences are identified and quantified in the ISA Summary, where the risk of each credible consequence event (probability weighted consequence) is determined by evaluating the severity of the potential consequences of an unmitigated accident with the probability that the accident sequence could occur. These are then binned into categories of likely, unlikely, or highly unlikely.

Although NUREG-1520 identifies a variety of methods available to evaluate accident sequences, one common method is to evaluate the probability of the accident and then, if necessary, to take additional protective measures to reduce the probability of the accident to highly unlikely. This evaluation is done by first multiplying the probability of the initiating event by the probabilities of all the subordinate events that complete the accident sequence. If, for instance, highly unlikely is defined in the ISA Summary as having an annual probability of less than 1 x 105, and the composite annual probability (product of the initiating and subordinate events) of the accident sequence is less than 1 x 105, the accident sequence can be considered as highly unlikely. If, on the other hand, the resulting annual probability is greater than 1 x 105 the applicant or licensee could introduce additional safety measures or a more robust design to reduce the probability of the initiating event or a subordinate event, such that the resulting revised event sequence annual probability falls below 1 x 105.

In past practice, seismic hazard analyses for nuclear fuel facilities have relied on a variety of methods to characterize the seismic hazard at the site or to develop inputs for the seismic design. These include site-specific PSHA, the USGS seismic hazards maps or iterative seismic hazard mapping tool, seismic design values that were derived from the USGS hazard maps, or deterministic seismic design values derived from site-specific data. In some cases, the deterministic seismic design values were derived from a nearby NPP, using a specified response spectra (such as RG 1.60 spectra) anchored at the SSE of the nearest NPP. Note that for USGS hazard analysis, ground motion values are adjusted in the USGS calculation to account for site soil conditions by selecting hazard results according to one of six site soil classes (Site Soil Class A through F), where Soil Class A is a very stiff hard rock site with average shear wave velocities above 1,500 m/s, while a Soil Class D is a less stiff soil site with an average shear wave velocity between 180 and 360 m/s.

2.3 Department of Energy Facilities The value of conducting site-specific hazard analyses also is being recognized by other federal agencies whose responsibilities include seismic safety of critical facilities, particularly the U.S. Department of Energy (DOE). DOE issued an update to its standard regarding external hazard analysis (DOE-Standard-1020-2012) (DOE, 2012) to emphasize the importance of site-specific analyses, especially the site response analysis. Updates to the American National Standards Institute/American Nuclear Society Standard (ANSI/ANS-2.29, 2008) are ongoing, and these also emphasize the importance of fully probabilistic seismic hazard studies that are based on site-specific data, including probabilistic site response analyses.

The DOE approach differs from the current NRC guidance in that DOE relies on site-specific PSHA studies developed for each regulated site, but applies the hazard results differently, depending on the facilities risk profiles. As described in DOE Standard-1020-2012, DOE defines five Seismic Design Categories (SDC), SDC-1 through SDC-5. These SDC categories 2-3

are assigned to a facility (or a safety significant structure, system, or component within the facility) based on the potential severity of adverse radiological and toxicological effects that may result from any seismic-initiated failures. SDC-1 is for a conventional building whose failure may not result in any radiological or toxicological consequences, while an SDC-5 is for an NPP or a nuclear material processing facility with a large inventory of radioactive material. Each of these SDC categories have corresponding target performance goals, which is the mean AEP that a seismic source characterization (SSC) will exceed its specified damage limit state given the seismic hazard at the site. Target performance goals of 1 x 104/yr, 4 x 105/yr, and 1 x 105/yr are used in ASCE 43-05 (ASCE, 2005) for SSCs defined as SDC-3, SDC-4, and SDC-5, respectively. Under this DOE approach, the SDC category determines how the hazard study is applied in the risk or performance analysis, but the PSHA itself is developed without specific regard for the SDC category. The underlying basis is that if a facility is designed to withstand a hazard with a certain AEP, consistent with a SDC and using the corresponding design method outlined in ASCE-43, it will achieve its performance goal.

2.4 Strengths and Weaknesses of the Different Approaches There are clear strengths and weaknesses associated with each of these approaches. The generalized and even qualitative approaches currently recommended in NUREG-1520 typically provide fuel cycle facility applicants and licensees with the simplest and least expensive way to characterize the seismic hazard at the site. These can usually be developed by structural engineers without specific training and experience in seismic hazard analysis or geotechnical engineering. For facilities where the NRC-regulated radiological or chemical risks from a seismic event are well below the highly unlikely threshold, a simplified seismic hazard analysis may be the only necessary assessment. In this case, the USGS hazard results are typically used as input to the standard building code provisions.

However, for facilities where there is the potential for seismically-initiated event sequences that could lead to radiological or chemical releases above regulatory limits, a more accurate and site-specific PSHA could be significantly beneficial, especially in quantifying risk insights. As we demonstrate in the comparisons described in Section 3 of this report, uncertainties in the probability of exceeding ground motion values, and thus the probability of a seismic initiating event, can vary (too high or too low) by up to nearly two orders of magnitude, depending on which method is used to characterize the seismic hazard at the site. Should one of the more generalized methods underestimate the probability of the ground motions, then the potential risks of the seismically-initiated accident could be underestimated by an equal factor in the ISA.

If unrecognized, this has the potential to misrepresent the risk significance of seismically-initiated accidents. On the other hand, one of the more generalized methods may overestimate the actual hazard. Should this outcome go unrecognized, then licensees or applicants may unnecessarily expend resources to mitigate the potential accident sequences with a more robust design or addition defense in depth; expenditures that may not be needed if a more accurate and reliable hazard is computed for the site.

These fully probabilistic PSHA studies require some level of training in seismic hazard analysis and geotechnical engineering. However, as described more fully in Section 5.2, computer codes and web-based applications that allow users to develop site-specific PSHA results are now abundant and readily available. These can be readily accessed by applicants, licensees, and their contractors. Moreover, the number of engineering firms across the U.S. that now have applicable experience in developing PSHAs has expanded over the past decades, commensurate with increased use of PSHA methods.

2-4

Another advantage to using a fully probabilistic PSHA is that it captures the uncertainty in the analysis, which is typically represented by the 5th, 15th, 85th, and 95th percentiles plotted about the mean and median values. Although obtaining these percentile curves requires additional computations, users can generate them easily from many of the open-source seismic hazard codes (Hale et al., 2018). The distribution of the hazard shown by these plots provides engineers and risk analysts additional perspective on the reliability and robustness of the results, especially in evaluating results that are near the margins of regulatory limits. These distributions arise from the epistemic uncertainty and aleatory variability captured in the PSHA logic tree, such as the distribution of magnitudes about the calculated maximum magnitude for a seismic source or the aleatory variability in the ground motion estimated from one of the ground motion models.

As described in USGS (Petersen et al., 2014), the USGS hazard code incorporates these uncertainties in the logic trees for seismic source characterization and ground motion models based on data analyses that are derived by the ground motion modelers. Lee et al. (2008) showed that it is mathematically possible to incorporate these uncertainties for the USGS results. However, because of the complexity and scale of the USGS code, they are not derived or included in the results of the USGS seismic hazard mapping tool. Thus, results from the USGS seismic hazard mapping tool only provide the mean hazard values, making it impossible to assess the uncertainties of the mean value results from the USGS tool.

The deterministic method, using a specified response spectra (such as RG 1.60 spectra) anchored at the SSE of the nearest NPP, does not include any estimates of uncertainty. The SSE is a single value, typically the PGA from a deterministic seismic hazard analysis, engineering experience, or engineering judgment. The specified response spectra is not uniform in terms of AEP, and may be conservative or not conservative relative to the actual hazard. These specified response spectra could also distort the results and insights of a risk study.

Thus, a site-specific PSHA provides the most robust response spectra and is the only method among those evaluated in this report that explicitly provides a characterization of the uncertainty in the hazard.

2-5

3 METHODOLOGY For the evaluation in this report, three methods were used to estimate the seismic hazard at four sites (Figure 3-1). These four sites were selected to represent a range in site conditions-from hard bedrock to thick soft soil-and relative hazard potential, from a high hazard area near the center of the New Madrid seismic zone to a low hazard site in eastern Pennsylvania. Table 3-1 provides a summary of the site and hazard conditions at each of the four sites.

Figure 3-1. Map of the eastern United States showing the location of the four sites analyzed in this report Table 3-1. Location and description of the four evaluated sites Nearest Commercial Lat. Long. Earthquake Nuclear Power Site (dd) (dd) Potential Site Geology Plant Site Columbia 38.8161 92.2197 Moderate: distant Flat-lying terrane Callaway Nuclear Airport New Madrid and with moderately Generating Station, Wabash Valley fault thick soil of clay and 65 km east of the site, sources shale above a SSE = 0.20 g Paleozoic sedimentary rock and Precambrian granite and gneiss.

Clarks Hill 36.7902 90.2013 Moderate: moderate Moderately thick Vogtle, 80 km south-distances to active strata of sandy clay southeast of the site, Charleston, SC and and sandy silt above SSE = 0.20 g Eastern Tennessee porphyritic granite seismic sources (saprolite in many areas) 3-1

Table 3-1. Location and description of the four evaluated sites Nearest Commercial Lat. Long. Earthquake Nuclear Power Site (dd) (dd) Potential Site Geology Plant Site Lehigh 33.6494 82.2411 Low: no nearby Sits atop a Limerick Generating University known fault sources, prominent ridge in Station, 50 km diffuse small the Great Valley southwest of the site, magnitude regional Physiographic SSE = 0.15 g earthquakes Province, with thin glacial till above Precambrian crystalline granite and gneiss.

St. Francis 40.6055 75.3716 High: proximal to the Thick soils resting Callaway Nuclear Bridge New Madrid Fault atop flat-lying Generating Station, System Paleozoic bedrock. 260 km south-southeast of the site, SSE = 0.20 g The first method we considered uses the most recent version of the USGS seismic hazard tool (https://earthquake.usgs.gov/hazards/interactive/) with the outputs adjusted as hard rock (Site Class A) or firm rock (Site Class B/C). The USGS approach provides site-specific values of ground motion at each site, but the underlying algorithm in the USGS code extrapolates results from a one-degree latitude by one-degree longitude grid of calculated hazard values. For the eastern U.S., one degree of latitude or longitude equals approximately 110 km. The USGS also provides hazard results at only four spectral frequencies: PGA, 5 Hz, 1 Hz, and 0.5 Hz. The hazard curve data for these four spectral frequencies was downloaded from the USGS website and replotted. For each of the four spectral frequencies, the digital data output from the USGS program provides 20 data points, each data point consisting of an AEP and its corresponding ground motion value. The typical range in AEP values recorded in the USGS output was between 1 x 101 and 1 x 107. To approximate the ground motion corresponding to the 1 x 104 AEP values, we extrapolated from the ground motion values as a ratio-adjusted midpoint between the reported ground motions for AEPs just below and just above the AEP = 1 x 104 target.

The second method we considered simply uses the SSE from a nearby NPP along with the RG 1.60 seismic design response spectrum to characterize the seismic hazard. The NPP must be designed so that, if the SSE ground motion occurs, risk-significant structures, systems, and components will remain functional and within applicable stress, strain, and deformation limits.

The SSE ground motion is the free-field ground motion response spectra. In view of the limited data available on vibratory ground motions of strong earthquakes, the response spectra are typically smoothed, such as the one developed in RG 1.60, and then anchored at the sites PGA or the SSE from a nearby NPP.

The third method is a fully-probabilistic approach following the same process the NRC staff used for the NTTF Recommendation 2.1 seismic hazard re-evaluations. This method uses the Central and Eastern United States SSC (CEUS-SSC) models in NUREG-2115 (NRC, 2012b),

EPRI (2013) CEUS ground motion models (GMM), and a site-specific site response analysis Approach 3 from NUREG/CR-6728 (Risk Engineering, 2001) to develop the final hazard curves. The CEUS-SSC model, which consists of both distributed seismicity sources referred to as background zones as well as repeated large magnitude earthquake (RLME) sources based on paleoseismic evidence of repeated large-magnitude earthquakes, was implemented by selecting the seismic background source zones located within 320 km and the RLME sources within 500 km of each of the four sites. For ground motion calculations, assuming the reference 3-2

hard rock condition, we used the EPRI (2013) GMM, applied to each of the CEUS-SSC sources to develop mean baserock seismic hazard curves at seven spectral frequencies (0.5, 1, 2.5, 5, 10, 25, and 100 Hz). Subsequently, a one-dimensional (1-D) site response analysis (SRA) was performed for each of the sites using the equivalent linear approach. The steps used in performing the 1-D site response analysis include:

• Selection of the baserock (i.e., elastic half space) elevation beneath the site

• Development of a basecase site profile in terms of shear wave velocity (VS), layer thickness, density, and damping

• Characterization of the nonlinear dynamic material properties (shear modulus and damping as a function of shear strain)

• Estimate of total site kappa value (near surface attenuation factor)

• Development of input spectra using random vibration theory (RVT)

• Development of 60 random profiles about the basecase site profile

• Determination of site amplification factor in terms of median and natural log standard deviation as a function of the input rock ground motion acceleration level Appendix B of EPRI (2012) provides detailed guidance on an acceptable approach for performing the site response analysis. The final step in the development of surface seismic hazard curves and response spectra involves the combination of the site amplification factors with the baserock seismic hazard curves. Current guidance (e.g., ASCE-4, ASCE-43, ANS 2.29, and EPRI, 2012) recommends the use of probabilistic approaches in order to develop fully-probabilistic, site-specific seismic hazard curves and response spectra.

The fully probabilistic results were plotted both as hazard curves (seven spectral frequencies and PGA), as well as uniform hazard response spectra (UHRS) for the 1 x 104 AEP. These results are shown for each of the four sites in Section 4 of this report. Comparisons among the three methods were then made in terms of the differences in the predicted amplitudes of ground motions as well as the relative difference in estimated AEP for a specific ground motion.

Specifically, we used the fully-probabilistic hazard curves as baselines from which we could derive corresponding AEPs for the ground motion values obtained from the other two methods.

For example, the 5 Hz ground motion from the USGS curves at 1 x 104 AEP was computed to be 0.30 g. We then determined that the corresponding AEP for 0.30 g on the fully probabilistic 5 Hz hazard curve is 8 x 104 AEP. The resulting ratio of 8 (8 x 104 /1 x 104) indicates that the USGS result is nonconservative compared to the fully probabilistic results by a factor of eight.

In the results tables presented in Section 4, we document the 1 x 104 ground motion values for PGA, 5 Hz, and 1 Hz spectral acceleration [S(a)] for each site that were determined by the three methods (fully probabilistic, USGS, and RG 1.60), the corresponding equivalent AEP values for the USGS and RG 1.60 ground motion values based on the fully-probabilistic curves, and the ratios comparing these equivalent AEP values with 1 x 104.

The most direct comparisons were computed as (1) ratios between the USGS and the fully-probabilistic results and (2) ratios between the deterministic nearby NPP RG 1.60 spectra and fully-probabilistic results. Figure 3-2 illustrates how we derived these ratios. In making 3-3

Figure 3-2. Method used to estimate the equivalent AEP and the ratio between the equivalent AEP and the fully-probabilistic hazard value. Sa is the spectral acceleration.

these comparisons, we assert that the fully-probabilistic results most accurately reflect the seismic hazard at these sites. This is because the fully-probabilistic results were derived using site-specific geologic information and incorporated the most up-to-date methods for seismic source characterization, ground motion modeling, and site response analysis. By comparison, the SSE approach relies on a set of approximations or simplifications to assess the hazard at the fuel cycle facility.

Several examples are listed next that illustrate how differences in the accuracy of the results could come about. First, in some cases, the nearest nuclear power plant may be several hundred kilometers away and situated on a soil profile that is very different from the fuel cycle facility site, resulting in inaccurate estimates of attenuation and damping. Second, seismic sources that cause significant ground motion at the NPP may actually be too far away to produce significant ground motions at the fuel cycle facility site, or the reverse may be true.

Finally, the reference design spectra for either facility is not a uniform hazard spectrum, meaning that AEPs of all the associated spectral frequencies are not fixed. Thus, while the SSE may have an AEP of 1 x 104, the AEP of the 5 Hz or 10 Hz may be significantly higher or lower than1 x 104. This makes using these spectral frequencies to assess performance or risk analyses uncertain because these response spectra could also distort the results and insights of a risk study.

Similarly, the USGS seismic hazard results also rely on simplifications and approximations of the hazard. For example, the USGS hazard analysis tool provides uniform hazard results at only four spectral frequencies (PGA, 5, 1, and 0.5 Hz), which limits the ability of these results to 3-4

accurately capture ground motion peaks caused by possible resonance of the soil column. In addition, the USGS results are computed on a geographic grid with one degree latitude and longitude spacing (approximately 100 km), so results for sites between grid points are extrapolated from the nodes. More importantly, the USGS results rely on generalized site-response factors based on a coarse differentiation of soil types, rather than a site-specific site response that corrects the reference rock ground motions for actual site-soil conditions.

3-5

4 RESULTS 4.1 Columbia Airport The Columbia, Missouri site is located in central Missouri, 65 km east of the Callaway Nuclear Generating Station (Figure 4-1). The site rests atop mature soils that overlie bedrock composed of Paleozoic limestones and Precambrian granite and gneiss (Table 4-1). The seismic hazard here is moderate, with significant influence from the active New Madrid fault system in southeastern Missouri and the Wabash Valley seismic source in eastern Illinois and southwestern Indiana. For this site, the NRC site response was based on a stratigraphic and shear wave velocity profile similar to that at Callaway, but adjusted for depth to bedrock based on the depth to bedrock map developed by the Missouri Department of Natural Resources (Voight, 2012).

Figure 4-1. Geologic map of the state of Missouri showing the location of the Columbia and St. Francis Bridge sites relative to the Callaway Nuclear Generating Station Table 4-1. Comparison of the NRC, USGS, and RG 1.60 Spectrum results for Columbia, Missouri USGS USGS RG 1.60 RG 1.60 Ratio Ratio Fully Prob.

S(a) GM Equivalent GM Equivalent USGS/Fully RG 1.60/Fully GM (g)

(g) AEP (g) AEP Prob. Prob.

PGA 0.26 0.16 4.48e-04 0.20 2.25e-04 4.48 2.25 5.0 Hz 0.69 0.30 7.78e-04 0.57 1.75e-04 7.78 1.75 1.0 Hz 0.12 0.16 4.61e-05 0.29 8.09e-04 0.46 8.08 4-1

Figures 4-2 shows the USGS and fully-probabilistic seismic hazard results. For this site, the fully-probabilistic results for PGA and 5 Hz are considerably higher than those from the USGS or RG 1.60 ground motions (Figure 4-3). As shown in Table 4-1, the ratios for these two spectral frequencies range between 1.8 and 7.8, indicating that if the USGS method or the RG 1.60 method were used to construct a new fuel cycle facility at this site, per the guidance in NUREG-1520, the resulting assumed performance of the facility could be non-conservative by a factor of almost eight times. We note the peak of the NRC UHRS curve at 5 Hz. There may be a similar peak in the USGS data at a slightly different spectral frequency, but these data are limited because the current USGS hazard tool can only calculate four spectral frequencies.

Figure 4-2. Fully probabilistic and USGS seismic hazard results for the Columbia, Missouri site Figure 4-3. Uniform hazard response spectra (UHRS) for the Columbia, Missouri site 4-2

As shown in Figure 4-3, and depending on the structures, systems, and components at the facility, the RG 1.60 spectrum can be quite conservative or non-conservative, in this example by a factor of eight at low frequencies and by a factor of 1.5 to 2 at higher frequencies. This comparison is somewhat typical of many CEUS sites as the RG 1.60 spectra were developed in 1970s using the western U.S. earthquake data then available. In contrast, the PSHA results are based on CEUS specific ground motions and a site-specific site-response analysis.

4.2 Clarks Hill The Clarks Hill site is located adjacent to the dam at Clarks Hill Lake (still officially designated as the J. Strom Thurmond Reservoir at the federal level). The site is situated at the boundary between the Appalachian Piedmont and the Coastal Plain (Figure 4-4). The site is approximately 50 km northwest of the MOX facility at Savannah River, and 80 km north-northwest of the Vogtle Electric Generating Plant (Figure 4-4). The site sits on 100-200 m sandy clay and sandy silt above porphyritic granite.

For this site, there is relatively close agreement between the fully probabilistic and USGS results (Figures 4-5 and 4-6). In this case, the USGS results are conservative compared to the fully probabilistic results, up to a factor of two (Table 4-2). By contrast, the RG 1.60 results are non-conservative at PGA and correlate poorly with the fully probabilistic results above 5 Hz (Figure 4-6).

Figure 4-4. Geologic map of the southeastern U.S. showing the location of the Clarks Hill site relative to the MOX site at Savannah River and the Vogtle Electric Generating Station 4-3

Figure 4-5. Fully probabilistic and USGS seismic hazard results for the Clarks Hill South Carolina site Figure 4-6. Uniform hazard response spectra (UHRS) for the Clarks Hill, South Carolina site 4-4

Table 4-2. Comparison of the NRC, USGS, and RG 1.60 spectrum results for Clarks Hill, South Carolina Ratio USGS USGS RG 1.60 RG 1.60 Ratio Fully Prob. RG S(a) GM Equivalent GM Equivalent USGS/Fully GM (g) 1.60/Fully (g) AEP (g) AEP Prob.

Prob.

PGA 0.32 0.39 6.06e-05 0.20 2.94e-04 0.61 2.94 5.0 Hz 0.54 0.61 7.97e-05 0.57 9.17e-05 0.80 0.92 1.0 Hz 0.11 0.16 4.98e-05 0.48 5.58e-06 0.50 0.06 4.3 Lehigh University The Lehigh University site is located in eastern Pennsylvania. The Lehigh campus is built on South Mountain, overlooking the city of Bethlehem, Pennsylvania (Figure 4-7). South Mountain consists of crystalline bedrock with a thin veneer of glacial soil. Thus, it is a hard rock site with shear wave velocities greater than 1,000 m/s. The Limerick Generating Station is the closest NPP, located 50 km to the south-southwest, outside the town of Pottstown, Pennsylvania. The seismic hazard is low, with no known active faults or active seismic zones within 200 km of the site. For this site, there was no NRC site response analysis needed because the stiff rock beneath the site does not amplify or deamplify the upward propagating seismic ground motion.

Figure 4-7. Location map for the Lehigh University site in eastern Pennsylvania, relative to the Limerick Generating Station 4-5

For this site, there is good agreement between the NRC and USGS results (Figures 4-8 and 4-9). However, the fully probabilistic UHRS is rich in high frequency energy, and it is not clear whether the USGS results would capture this feature of the hazard because of the limited number of spectral frequencies provided by the current USGS seismic hazard tool (Figure 4-9).

The high frequency energy is not unexpected, given that this is a hard rock site. As shown in Table 4-3, the RG 1.60 spectrum does not capture the high frequency energy, but it is substantially higher at low frequencies. At 1 Hz, the RG 1.60 spectrum is about 20 times more conservative than the USGS or fully probabilistic hazard.

Figure 4-8. Fully probabilistic and USGS seismic hazard results for the Lehigh University site Figure 4-9. Uniform hazard response spectra (UHRS) for the Lehigh University site 4-6

Table 4-3. Comparison of the NRC, USGS, and RG 1.60 spectrum results for Lehigh University, Pennsylvania Fully USGS USGS RG 1.60 RG 1.60 Ratio Ratio S(a) Prob. GM Equivalent GM Equivalent USGS/Fully RG 1.60/

GM (g) (g) AEP (g) AEP Prob. Fully Prob.

PGA 0.26 0.20 1.55e-04 0.15 2.38e-04 1.55 2.38 5.0 Hz 0.27 0.20 1.17e-04 0.52 2.73e-05 1.17 0.27 1.0 Hz 0.06 0.06 1.11e-04 0.25 5.25e-06 1.11 0.05 4.4 St. Francis Bridge The St. Francis River, Missouri site is located in southeastern Missouri, about equidistant from the Callaway Generating Station (Figure 4-1) and the Arkansas Nuclear One site in northern Arkansas. The site is next to Highway 60, where the highway crosses the St. Francis River, just north of Fisk, Missouri. The seismic hazard here is high, with significant contribution from the New Madrid seismic zone. At this site, the NRC site response was based on shear wave velocity data obtained from Spectral Analysis of Surface Waves (SASW) reported in a Missouri Department of Transportation report (Anderson et al, 2003) and a stratigraphic profile from Figure 7-2 of Anderson et al., (2001), which is replotted in Figure 4-10. Example results from the Missouri Department of Transportation report SASW analysis are shown in Figure 4-11.

The results show that the upper soils have shear wave velocity (VS) values between 100 m/s and 300 m/s to a depth of approximately 45 m.

Figure 4-10. Geologic cross section beneath the Rte. 60 St. Francis Bridge. The cross section depths are shown here in ft. To convert to meters, divide the values in feet by 3.28. Adapted from Figure 7.2 of Anderson et al. (2001).

4-7

Figure 4-11. Example VS profile beneath the St. Francis Bridge from the MODOT SASW analysis Seismic hazard at this site is high, with ground motions at relatively low AEPs well above 1.0 g (Figure 4-12). For this site, the RG 1.60 and USGS results do not match the fully-probabilistic results (Figure 4-13), although the RG 1.60 results may not be meaningful at this site because the nearest commercial NPPs are so far away that using the SSE from Callaway or Arkansas Nuclear One would not be considered reliable, even given the option in NUREG-1520.

However, the comparison of the USGS and fully probabilistic results show significant differences. USGS results are conservative at high frequencies, but nonconservative, by a factor of 5.6, at 1 Hz (Table 4-4). This is most likely due to the more detailed site response in the fully-probabilistic analysis, in which the thick and soft soils beneath the bridge (Figure 4-10) amplify the low frequency ground motions and dampen the high frequency motions. The simple site classification approach used in the USGS hazard calculations does not capture the effects of the local geology on the site response.

4-8

Figure 4-12. Fully probabilistic and USGS seismic hazard results for St. Francis Bridge, Missouri site Figure 4-13. Uniform hazard response spectra (UHRS) for the St. Francis Bridge site 4-9

Table 4-4. Comparison of the NRC, USGS, and RG 1.60 Spectrum results for St. Francis Bridge site, Missouri USGS USGS RG 1.60 RG 1.60 Ratio Ratio Fully Prob. GM S(a) GM Equivalent GM Equivalent USGS/Fully RG 1.60/

(g)

(g) AEP (g) AEP Prob. Fully Prob.

PGA 0.77 1.33 3.61e-06 0.20 4.01e-03 0.04 40.12 5.0 Hz 2.14 2.29 7.88e-05 0.57 2.64e-03 0.79 26.36 1.0 Hz 1.56 0.72 5.59e-04 0.30 1.75e-03 5.59 17.45 4-10

5 DISCUSSION AND RECOMMENDATIONS 5.1 Discussion As discussed in Section 3, the most accurate representation of the hazard is from the site-specific and fully-probabilistic results, especially because the site-specific analysis includes a site response component that accounts for the geotechnical and geomechanical properties of the soil and bedrock beneath the site. In Section 4, we demonstrated the significant differences that arise among the methods by comparing the resulting hazards at the four sites. Table 5-1 and Table 5-2 summarize the AEP ratios that were derived in Section 4. Values above 1.0 indicate non-conservative cases. Values below 1.0 indicate conservative cases.

Table 5-1. Summary of USGS results Columbia Lehigh St. Francis S(a) Airport Clarks Hill University Bridge PGA 4.48 0.61 1.55 0.04 5 Hz 7.78 0.80 1.17 0.79 1 Hz 0.46 0.50 1.11 5.59 Table 5-2. Summary of RG 1.60 results Columbia Lehigh St. Francis S(a) Airport Clarks Hill University Bridge PGA 2.25 2.94 2.38 40.12 5 Hz 1.75 0.92 0.27 26.36 1 Hz 8.08 0.06 0.05 17.45 As shown from the example site comparisons, the two more generalized approaches (USGS hazard mapping tool and deterministic seismic design values from nearby commercial NPPs) produce results that, in terms of probability, could be in error by more than one order of magnitude. Examples from this study show that these outcomes are not systemically high or low, but can be either conservative or unconservative depending on the site conditions and distance from active seismic sources. In the performance-based and risk-informed approach under 10 CFR Part 70, robust and reliable estimates of the hazard may be needed in order to maintain a realistic representation of the risk. If the initial hazard probability is incorrect by an order of magnitude or more, then the resulting estimate of the performance may also be incorrect by that same amount.

In addition, results from this study and from the NTTF Recommendation 2.1 re-evaluations for all NPPs across the U.S. show that the resulting UHRS vary considerably from site to site and region to region. The rich high-frequency energy at the Lehigh University site or the peak at 5 Hz spectral frequencies at the Columbia, Missouri site demonstrate the potential variability in seismic hazards because of the site conditions. The current USGS hazard tool, which only provides results for four spectral frequencies, may also be too coarse to accurately characterize the hazard.

If one of the more generalized methods underestimates the probability of the ground motions, then the potential risks of the seismically-initiated accident could be underestimated by an equal factor in the ISA. If unrecognized, this has the potential to misrepresent the risk significance of 5-1

seismically-initiated accidents. On the other the hand, if one of the more generalized methods overestimates the actual hazard, and this outcome is unrecognized, then licensees or applicants may unnecessarily expend resources to mitigate the potential accident sequences with a more robust design or additional defense in depth; expenditures that may not be needed if a more accurate and reliable hazard is computed for the site.

The results from this study complement similar findings from NTTF Recommendation 2.1 re-evaluations, all of which have led the seismic hazard community to advocate for fully probabilistic and site specific seismic hazard studies for all NRC and DOE nuclear facilities.

Site-specific and fully-probabilistic methods are now consistently relied on in the most recent version of the DOE Standard (DOE, 2012); in other NRC seismic guidance documents, such as RG 1.208 (NRC, 2007), NUREG-2117 (NRC, 2012c) and NUREG-2213 (in press, NRC, 2018);

in recent EPRI guidance (EPRI, 2012); and the American Nuclear Standard 2.29 (ANSI/ANS, 2008). Implementing the recommendations listed in the following section of this report would provide risk-consistent results and harmonize the 10 CFR Part 70 guidance with the rest of the NRC and with the larger seismic hazard community supporting seismic safety at nuclear facilities across the U.S.

5.2 Specific Recommendations

• Based on the analysis in this report, the use of site-specific PSHA studies for future regulatory evaluations of fuel cycle facilities should be considered for addition as an acceptable approach in the updates to NUREG-1520, or to new or updated regulatory guidance for fuel cycle facilities.

• This guidance should emphasize a reliance on site-specific probabilistic seismic hazard curves, spanning an appropriate range of spectral frequencies. This can be accomplished using existing seismic sources, ground motion models, and site response codes described later in this section. The site response model should specifically account for the facility-specific soil conditions.

• The updated or new guidance should also recommend the development of an appropriate UHRS and associated seismic design inputs (e.g., response spectra or scaled time histories).

• Finally, use of the SSHAC methodology, which has now been applied in developing seismic hazard analyses across the U.S., is recommended to develop PSHA inputs or the complete PSHA. The SSHAC process provides applicants, licensees, regulators, and the public an objective, transparent, and peer-reviewed process in which there is a high degree of technical confidence in the results and regulatory stability and assurance in meeting the safety objectives in 10 CFR 70. NUREG-2213 describes how to develop a SSHAC hazard study under four increasingly more complex study levels (Levels 1 through 4), depending on the specific needs of the SSHAC study. Simplified processes described for a Level 1 or Level 2 study would be sufficient for most fuel cycle and fuel storage facility applications. The SSHAC process can also be used to update or make an existing regional analyses site-specific.

The availability of computer codes and web-based applications that allow users to develop site-specific PSHA results are now abundant and can be readily accessed and utilized by applicants, licensees, and their contractors. For example, the Pacific Earthquake Engineering Research (PEER) Center just completed a PSHA code verification project that evaluated 5-2

13 commercially-available and open-source seismic hazard codes (Hale et al., 2018). These codes can implement the seismic source characterization developed in NUREG-2115 (NRC, 2012b). PEER also is nearing completion of a SSHAC Level 3 study to develop the Next Generation Attenuation Relationships for Central & Eastern North America (NGA-East)

(PEER, 2018). The NGA-East ground motion characterization model consists of a set of new ground motion prediction equations for median and standard deviation of ground motions and their associated weights in the logic trees for use in PSHA. There also are many available codes for developing fully-probabilistic site response analyses, including Strata, RASCAL, and PSHAKE. Collectively, these tools and the associated user manuals provide the necessary tools for seismologists and seismic engineers to develop site-specific PSHA results similar to the ones developed in this report.

The approach described in the preceding paragraph is applicable to all sites in the central and eastern U.S., which is defined as the area east of 104 degrees of longitude. For a site in the western U.S., there is not a regional seismic source model comparable to the one in NUREG-2115 (NRC, 2012b). Thus, a site-specific SSC model would need to be developed.

This site-specific SSC model could rely on existing fault characterizations, such as the Quaternary fault database developed by the USGS (2006) or the California Reference Fault Parameter Database (Field et al., 2013). In addition, areal source zones in the western U.S.

could be developed using the USGS (2006) gridded magnitude recurrence rates. The Southwestern Ground Motion Characterization project (GeoPentech, 2015) developed a regional ground motion model, in which key inputs would need to be adjusted to account for site-specific conditions. However, as previously noted, the optimal process to develop assessments of these site specific hazards in the western U.S. would be to follow a SSHAC Level-1 or SSHAC Level-2 process, as described in NUREG-2213 (NRC, 2018, in process),

and consistent with current DOE practice (e.g., Payne et al., 2002).

5-3

6 REFERENCES Anderson, N.L., G. Chen, S. Kociu, R. Luna, T. Thitimakorn, A. Malovichko, D. Malovichko, and D. Shylakov. Vertical Shear-Wave Velocity Profiles Generated From Spectral Analysis of Surface Waves: Field Examples. RDT 03-006. Jefferson City, Missouri: Missouri Department of Transportation Research, Development and Technology. 2003.

Anderson, N.L., H. Baker, G. Chen, T. Hertell, D.J. Hoffman, R. Luna, Y. Munaf, S. Prakash, P.M. Santi, and R.W. Stephenson. Earthquake Hazard Assessment along Designated Emergency Vehicle Priority Access Routes. Research, Development, and Technology RDT 01-009. University of Missouri-Rolla. 2001.

ANSI/ANS. Probabilistic Seismic Hazard Analysis. ANSI/ANS-2.29-2008.

La Grange Park, Illinois: American National Standards Institute/American Nuclear Society.

2008.

ASCE. Minimum Design Loads for Buildings and Other Structures. ASCE/SEI 7 and SEI Standards. New York, New York: American Society of Civil Engineers. 2010.

ASCE/SEI. Seismic Design Criteria for Structures, Systems, and Components in Nuclear Facilities. ASCE/SEI 43-05. New York, New York: American Society of Civil Engineers and Structural Engineering Institute. 2005.

Building Seismic Safety Council. NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, Part1: Provisions (FEMA 368). Washington, DC:

Building Seismic Safety Council. 2003.

DOE. Natural Phenomena Hazards Analysis and Design Criteria for DOE Facilities.

DOE-STD-1020-2012. Washington, DC: U.S. Department of Energy. 2012.

EPRI. Seismic Evaluation Guidance, Screening, Prioritization and Implementation Details

[SPID] for the Resolution of Fukushima Near-Term Task Force Recommendation 2.1: Seismic.

EPRI Report 1025287. ML12333A170. Palo Alto, California: Electric Power Research Institute. November 2012.

_____. Ground Motion Model Review Final Report. ML13155A553. Palo Alto, California:

Electric Power Research Institute. June 2013.

Field, E.H., et al. Uniform California Earthquake Rupture Forecast, Version 3 (UCERF3)-The Time-Independent Model. U.S. Geological Survey Open-File Report 2013-1165. 2013.

(http://www.wgcep.org/data-ref_fault_db).

GeoPentech. Southwestern United States Ground Motion Characterization SSHAC Level 3-Technical Report. Rev. 2. March 2015.

<http://www.pge.com/en/safety/systemworks/dcpp/sshac/index.page>

Hale, C., N. Abrahamson, and Y. Bozorgnia. PEER PSHA Code Verification Project. 2018.

(https://www.coursehero.com/file/29515300/7-hale-2016pdf/)

IBC. International Building Code', International Code Council, ICC. Country Club Hills, Illinois: International Building Code. 2012.

6-1

Lee, Y., W. Graf, and Z. Hu. Characterizing the Epistemic Uncertainty in the USGS 2014 National Seismic Hazard Mapping Project (NSHMP). Bulletin of the Seismological Society of America. Vol. 108 (3A). pp. 1,465-1,480. doi: https://doi.org/10.1785/0120170338 NRC. NUREG-2213, Updated implementation Guidelines for SSHAC Hazard Studies.

Washington, DC: U.S. Nuclear Regulatory Commission. 2018 (in press).

_____. NUREG-1520, Standard Review Plan for Fuel Cycle Facilities License Applications.

Revision 2. Washington, DC: U.S. Nuclear Regulatory Commission. 2015.

_____. Design Response Spectra for Seismic Design of Nuclear Power Plants.

Regulatory Guide 1.60, Revision 2. Washington, DC: U.S. Nuclear Regulatory Commission.

2014.

_____. NUREG-2115, Central and Eastern United States Seismic Source Characterization for Nuclear Facilities. ADAMS stores the NUREG as multiple ADAMS documents, which are accessed through the webpage. 2012a.

http://www.nrc.gov/reading-rm/doc-collections/nuregs/staff/sr2115/.

_____. Letter from Michael R. Johnson, Director, Office of New Reactors, to All Power Reactor Licensees and Holders of Construction Permits in Active or Deferred Status, dated March 12, 2012. ML12053A340. Washington, DC: U.S. Nuclear Regulatory Commission.

2012b.

_____. NUREG-2117, Practical Implementation Guidelines for SSHAC Level 3 and 4 Hazard Studies. Rev. 1. Washington, DC: U.S. Nuclear Regulatory Commission. 2012c.

_____. Near-Term Report and Recommendations for Agency Actions Following the Events in Japan. SECY-11-0093. ML11186A950. Washington, DC: U.S. Nuclear Regulatory Commission. July 2011a.

_____. Recommendations for Enhancing Reactor Safety in the 21st Century: The Near-Term Task Force Review of Insights from the Fukushima Dai-Ichi Accident. Enclosure to SECY-11-0093. ML112510271. Washington, DC: U.S. Nuclear Regulatory Commission.

July 2011b.

_____. Recommended Actions to be Taken Without Delay from the Near-Term Task Force Report. SECY-11-0124. ML11245A158. Washington, DC: U.S. Nuclear Regulatory Commission. September 2011c.

_____. Prioritization of Recommended Actions to be Taken in Response to Fukushima Lessons Learned. SECY-11-0137. ML11272A111. Washington, DC: U.S. Nuclear Regulatory Commission. October 2011d.

_____. Regulatory Guide 1.208, A Performance-Based Approach to Define the Site-Specific Earthquake Ground Motion. Washington, DC: Nuclear Regulatory Commission. 2007.

_____. NUREG/CR-6372. Recommendations for Probabilistic Seismic Hazard Analysis:

Guidance on Uncertainty and Use of Experts. Washington, DC: U.S. Nuclear Regulatory Commission. 1997.

6-2

Payne, S.J., V.W. Gorman, S.A. Jensen, M.E. Nitzel, M.J. Russell, and R.P. Smith.

Development of Probabilistic Design Basis Earthquake (DBE) Parameters for Moderate and High Hazard Facilities at INEEL. INEEL/EXT-99-00775. Final Report. Rev. 2.

Idaho Falls, Idaho: Idaho National Engineering and Environmental Laboratory. 2002.

PEER. Next Generation Attenuation Relationships for Central & Eastern North-America (NGA-East). Berkeley, California: Pacific Earthquake Engineering Research Center. 2018.

https://peer.berkeley.edu/research/nga-east.

Petersen, M.D., M.P. Moschetti, P.M. Powers, C.S. Mueller, K.M. Haller, A.D. Frankel, Y. Zeng, S. Rezaeian, S.C. Harmsen, O.S. Boyd, N. Field, R. Chen, K.S. Rukstales, N. Luco, R.L.

Wheeler, R.A. Williams, and A.H. Olsen. Documentation for the 2014 Update of the United States National Seismic Hazard Maps. U.S. Geological Survey Open-File Report 2014-1091.

p. 243. Denver, Colorado: U.S. Geological Survey. 2014.

https://dx.doi.org/10.3133/ofr20141091.

Risk Engineering Inc. NUREG/CR-6728, Technical Basis for Revision of Regulatory Guidance on Design Ground Motions: Hazard- and Risk-consistent Ground Motion Spectra Guidelines.

Washington, DC: U.S. Nuclear Regulatory Commission. 2001.

USGS. Quaternary Fault and Fold Database for the United States. U.S. Geological Survey.

2006. http//earthquake.usgs.gov/hazards/qfaults/.

Voight, V. Map of Depth to Bedrock for the State of Missouri. Missouri Department of Natural Resources, Division of Geology and Land Survey, Missouri Geological Survey. 2012. Isopach intervals of 25 feet; scale 1:500,000.

6-3