ML18109A507
| ML18109A507 | |
| Person / Time | |
|---|---|
| Issue date: | 05/08/2019 |
| From: | Sara Lyons, Shilp Vasavada NRC/NRR/DRA, NRC/RES/DE |
| To: | |
| Wayne Davis | |
| References | |
| IMECE2018-87677 | |
| Download: ML18109A507 (8) | |
Text
Proceedings of the ASME 2018 International Mechanical Engineering Congress and Exposition IMECE2018 November 9-15, 2018, Pittsburgh, PA, USA IMECE2018-87677 SEISMIC PROBABILISTIC RISK ASSESSMENT OF NUCLEAR POWER PLANTS:
10 CFR 50.69 ASSUMPTIONS AND SOURCES OF UNCERTAINTY Sara Lyons, Shilp Vasavada U.S. Nuclear Regulatory Commission1 Rockville, MD, USA ABSTRACT BACKGROUND The U.S. Nuclear Regulatory Commission (NRC) Nuclear power plants (NPPs) in the United States have been promulgated Part 50.69 to Title 10 of the Code of Federal designed to withstand credible natural and manmade hazards, Regulations (CFR), Risk-informed categorization and including earthquakes. However, since the original design of treatment of structures, systems and components for nuclear many NPPs, the technical communitys understanding of the power reactors, in November 2004 (hereafter referred to as 10 seismic hazard in areas across the U.S. has continued to evolve CFR 50.69). The rule provides a voluntary alternative to such that seismic events in some regions of the country are now compliance with many regulations which require special thought to be more likely. As a result, various efforts to assess treatment, or regulatory requirements which go beyond the seismic risk of these plants have been undertaken, including:
industrial controls, including: specific inspection, testing, qualification, and reporting requirements. The voluntary Unresolved Safety Issue A-46, Seismic Qualification of alternative includes a process for categorization of structures, Equipment in Operating Plants, 1980 [1],
systems, and components (SSCs) as having either low safety Individual Plant Examination of External Events (IPEEE),
significance (LSS) or high safety significance (HSS). The 1991 [2],
categorization process can result in increased requirements for Generic Issue 199, Implications of Updated Probabilistic HSS SSCs which were previously treated as non-safety-related, Seismic Hazard Estimates in Central and Eastern United and reduced requirements for LSS SSCs which were previously States, 2005 [ 3], and treated as safety-related. Fukushima Dai-ichi Response, Near Term Task Force The categorization process includes plant-specific risk Recommendation 2.1, 2012 [ 4]
analyses which are used in combination with an integrated decision-making panel (IDP) to determine whether the SSC has These efforts have resulted in an improved understanding of a low or high safety significance. Seismic probabilistic risk seismic risk at U.S. NPPs and a number of safety improvements.
assessment (SPRA) is one of the risk analyses options to account However, the seismic risk assessment methodologies used in for the seismic risk contribution. Because the 10 CFR 50.69 rule response to these programs have varied and the assessments were has currently not been implemented widely, the significance of not necessarily maintained beyond initial completion.
various SPRA assumptions and sources of uncertainty to the Some U.S. NPPs have developed high quality SPRAs and categorization process has had limited evaluation for a broad have chosen to use them to support the implementation of 10 spectrum of U.S. nuclear power plants. CFR 50.69. This rule provides a voluntary alternative to This paper will assess the importance of certain aspects of compliance with many regulations which require special the seismic risk contribution to the categorization process. NRC treatment, or regulatory requirements which go beyond Standardized Plant Analysis Risk (SPAR) models will be used to industrial controls, including specific inspection, testing, perform sensitivity studies to quantify the impact of various qualification, and reporting requirements. The voluntary assumptions and sources of uncertainty on the outcome of the alternative under 10 CFR 50.69 includes a process for categorization process. categorization of SSCs as having either LSS or HSS. The categorization process can result in increased requirements for HSS SSCs where such requirements did not exist previously 1
The views expressed herein are those of the authors and do not represent an official position of the U.S. NRC.
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.
Approved for public release. Distribution is unlimited.
1
(termed non-safety related SSCs) and reduced requirements for conditional on the seismic acceleration experienced by the SSC.
LSS SSCs where such requirements existed previously (termed The fragility of any SSC is usually expressed in the form of a as safety-related SSCs). family of fragility curves. The fragility curves are often The categorization process includes plant-specific risk described by a lognormal-lognormal distribution with analyses which are used in combination with an IDP to determine parameters Am (median capacity), R (randomness), and U whether the SSC has a low or high safety significance. The IDP, (uncertainty). The lognormal standard deviation associated in its determination of the appropriate categorization of a with randomness, R, is included to account for the aleatory particular SSC, considers the risk analyses in conjunction with uncertainty which is present due to the random nature of the non-risk related attributes such as the impact of the proposed seismic hazard and is described by the peak and valley categorization on the defense-in-depth and safety margins at the variation of an actual earthquake. The lognormal standard plant. SPRA is one of the risk analyses options to account for deviation associated with uncertainty, U, is included to account the seismic risk contribution. SPRAs have evolved over the last for the epistemic uncertainty associated with the actual spectral few decades and are garnering increased safety and regulatory shape of the earthquake as compared to the reference use as the technology matures. Because the 10 CFR 50.69 rule earthquake. Oftentimes, R and U are combined into a single has currently not been implemented widely, the significance of composite lognormal standard deviation, C. In an SPRA, the various SPRA assumptions and sources of uncertainty to the plant response logic model is used to identify combinations of categorization process has had limited evaluation for a broad failures that may result in core damage and subsequent large spectrum of U.S. nuclear power plants. early release of radionuclides. The risk of core damage and This paper will assess the importance of certain aspects of large early release is calculated by combining the appropriate the seismic risk contribution to the categorization process. NRC fragility curves with the site hazard curves and plant response Standardized Plant Analysis Risk (SPAR) models will be used to model. Both the fragility curves and site hazard curves must perform sensitivity studies to quantify the impact of various use the same reference earthquake intensity parameter, most assumptions and sources of uncertainty on the categorization commonly the peak ground acceleration (PGA) [5]. The process. physics of failure modeling approach is based on a stress-strength model where the SSC fails if the stress exceeds its IMPORTANCE OF SEISMIC RISK TO 10 CFR 50.69 capacity [6]. In SPRA fragility analysis, several variables are considered in estimating the seismic demand (i.e., stress) and The importance of the seismic risk to the 10 CFR 50.69 capacity (i.e., strength) for various earthquakes at the PGA.
categorization process can vary depending on several factors, In this paper, nine NRC Standardized Plant Analysis Risk including the geographic location, seismic hazard, design, (SPAR) models will be used to explore the importance of seismic construction codes, and vintage of the plant as well as the seismic risk to the 10 CFR 50.69 categorization process. These models risk relative to the risk from other hazards. In order to assess the were developed to represent the as-built, as-operated plant, but impact of the seismic risk on the categorization process, 10 CFR have limitations with respect to plant representation and level of 50.69 requires a systematic evaluation process and reasonable detail primarily due to the model update frequency.
confidence that any potential increases in core damage frequency Furthermore, the SPAR models use generic seismic fragility (CDF) or large early release frequency (LERF) are small. values for SSCs and are not expected to be representative of Small increases are discussed in Regulatory Guide (RG) 1.174 refined plant-specific fragility calculations. Thus, the results of as resulting in a change in CDF (termed CDF and read as delta the SPAR models are used in this paper as examples to explore CDF) between approximately 1E-6/year and 1E-5/year and the importance of certain aspects of the seismic risk in the change in LERF (LERF) of between approximately 1E-7/year context of the categorization process. The results presented in and 1E-6/year for NPPs that have a total CDF and LERF of less this paper are not used and should not be used to draw definitive than approximately 1E-4/year and 1E-5/year, respectively. conclusions related to particular plants. For the purposes of this Based on these requirements, it may be inferred that seismic risk paper the risk contribution is considered from only internal will be important to the categorization process for plants with a events, internal fire, and seismic events as these hazards are relatively high seismic CDF or seismic LERF. Such plants will generally expected to have the more significant effect on the likely need a high quality SPRA to support implementation of 10 categorization results. Figure 1 shows the relative risk CFR 50.69. Conversely, the seismic risk contribution may be contribution from each of these hazards for the nine example negligible for those plants that are robust relative to the seismic plants. These SPAR models were selected to represent a variety hazard at their site, provided there is reasonable confidence that of NPPs, including pressurized and boiling water reactors, aspects that were used to determine that the as-built, as-operated various containment designs, and some NPPs which are located plant is seismically robust are maintained. in geographic regions which have had a significant increase in An SPRA consists of three major parts: hazard analysis, expected seismic hazard relative to that used for the plants fragilities evaluation, and plant response analysis. It is original design.
important to provide an overview of seismic fragility to support One method for performing categorization in accordance the discussion in the remainder of this paper. The seismic with 10 CFR 50.69 is described in Nuclear Energy Institute fragility of a SSC is the probability of failure of that SSC (NEI) Report 00-04, 10 CFR 50.69 SSC Categorization 2
Guideline [8] which has been endorsed by the NRC in RG assumption of the robustness of the plant SSCs against the 1.201, Guidelines for Categorizing Structures, Systems, and seismic hazard is re-evaluated, if necessary.
Components in Nuclear Power Plants According to their Safety Due to the low seismic contribution, the results for Example Significance [9]. NEI 00-04 allows for not considering the Plant 9 are not expected to be meaningful for the evaluations in seismic risk in the categorization if that risk (expressed in this paper and will not be considered further.
seismic CDF) is less than 1% of the internal events plant risk (internal events CDF). Note that the CDF for Example Plant 9 is less than 1% of the total CDF. Some plants, such as Example Plant 9, which are expected to have a low seismic risk may not have a seismic margins analysis (SMA) or SPRA to support the categorization process. However, 10 CFR 50.69 does not restrict the evaluation approaches to SMA or SPRA. There can exist alternative systematic evaluations which may be sufficient to support 10 CFR 50.69 implementation. The systematic evaluation must recognize that the low hazard of the site does not directly translate into low seismic risk because the seismic design of the SSCs is based on the hazard. As a result, the convolution of the low hazard with SSC-specific fragilities may not result in low risk.
A possible semi-quantitative approach can be postulated by seeking the plant level fragility (i.e., a fragility curve hypothesized to represent the failure of all basic events necessary to prevent core damage) that would result in the seismic CDF being less than 1% of the internal events CDF as follows:
The target seismic CDF, 1% of the internal events CDF, is calculated and the plant-specific mean seismic hazard curve is known a priori.
A composite lognormal standard deviation for the plant level fragility curve can then be selected with supporting technical justification.
A plant level fragility can be found using the above information that, when convoluted with the plant-specific Figure 1. Example Plant Internal Event, Seismic, and Fire hazard curve, will result in the target seismic CDF. Risk Core Damage Frequency (CDF) Contribution Limited fragility calculations, possibly for SSCs that were identified in historical evaluations as necessary for safe shutdown of the plant during seismic events, can be used to USE OF IMPORTANCE MEASURES FOR 10 CFR 50.69 support the justification that the resulting plant level fragility is equal to or above the calculated value. The 50.69 categorization process described in NEI 00-04, as Sensitivity studies on the impact of the composite lognormal endorsed in RG 1.201, includes plant-specific risk analyses standard deviation can provide information about the which are used in combination with the IDP. SSCs which meet variability in that parameter as well as the calculated plant certain criteria in the plant-specific risk analyses may not be re-level fragility. categorized by the IDP, including SSCs which are identified as HSS by the integrated risk characterization portion of the Regardless of whether or not an SPRA is used, information process, the internal events PRA assessment, a non-PRA method should be provided to the IDP with respect to the seismic hazard to address external events or shutdown risk, or the defense-in-as well as the components necessary for safe shutdown under depth assessment.
seismic events to inform their decision-making. The defense-in- The integrated risk characterization portion of the process depth aspect of categorization by the IDP is done relative to the identifies SSCs as HSS if the integrated importance measures design basis of the plant. The lack of an SPRA can result in a meet the criteria to be determined HSS. That is, if the integrated potential decoupling between the current (and future) seismic Fussell-Vesely (F-V) importance measure is greater than 0.005, hazard at the plant and the design basis seismic hazard. the integrated Risk Achievement Worth (RAW) is greater than 2, Therefore, it is important for the 10 CFR 50.69 implementation or the integrated RAW associated with common cause failure is that the IDP is informed of the difference between the design greater than 20, the SSC will be categorized as HSS. NEI 00-04 basis hazard and the most recent seismic hazard so that the states that the integrated importance measures for contribution to 3
core damage frequency (CDF) are calculated per Equations (1) Eight SPAR models from the eight example plants which and (2), below, which are reproduced from NEI 00-04. have a seismic risk contribution greater than 1% (Example Plants 1 - 8 in Figure 1) were selected to further understand the
(, x ) significance of potential assumptions and sources of uncertainty
=
(1) within the seismic PRA which may impact the categorization process. The following assessments were performed for all or a where, subset of these eight SPAR models:
= Integrated F-V importance of component i over all CDF contributors Case 1: Baseline Assessment
, = F-V importance of component i for CDF contributor j Case 2: Use of truncated lognormal fragility curves
= CDF of contributor j Case 3: The probability of a loss of offsite power due to a seismic event was reduced
= 1 +
(, 1)x (2) Case 4: Select HEPs were decreased to guarantee success The baseline assessment (Case 1) was performed to gauge where, the significance of the SPRA results in the context of the overall
= Integrated RAW of component i over all CDF categorization results and to provide a control case for contributors comparison purposes. The use of truncated lognormal fragility
, = RAW of component i for CDF contributor j curves (Case 2) was evaluated as a potentially more realistic
= CDF of contributor j representation of the conditional failure of SSCs. The reduction in the probability of a loss of offsite power due to a seismic event The expressions in Equations (1) and (2) are also applicable for (Case 3) was evaluated because it is generally a dominant LERF related importance measures. contributor to seismic CDF and seismic LERF and its Under the process described in NEI 00-04, SSCs would be significance could be reduced at some NPPs in future years. The categorized as HSS based, in part, on the SPRA if (1) the decrease in select HEPs (Case 4) was evaluated because the integrated importance measure met the criteria identified above human reliability analysis can be significant to the SPRA results, or (2) based on seismic PRA specific sensitivity analysis the IDP but some currently used methods are simplified based on a determined that the SSC would be categorized as HSS. There bounding assessment. Cases 2, 3, and 4 represent areas of are other aspects of the NEI 00-04 process that would result in potential future refinement in the development and application an SSC being identified as HSS, but those aspects would not of SPRAs.
necessarily be driven by the results of the seismic PRA. The results of these sensitivity assessments are discussed below in the context of the impact of the seismic CDF on the EVALUATION APPROACH categorization results. However, further study of the impact with respect to seismic LERF would be needed to reach a definitive There are several traditional engineering methods which conclusion. It has been observed from recent, high quality SPRA provide the foundation and basis for the seismic PRA results. For results of U.S. NPPs that seismic LERF can be a significant example, the fragility curves for specific SSCs are based on the percentage of the seismic CDF due to dependencies associated expected stress (seismic loading) and strength (capacity) of the with the seismic hazard.
component during a given seismic event. There are several inputs which are needed to support this analysis, including the CASE 1: BASELINE ASSESSMENT probabilistic seismic hazard analysis, the spectral shape of the seismic event, the soil-structure interaction analysis, the The eight SPAR models were run to identify the SSCs that propagation of the seismic load through any relevant structures, would likely be categorized as HSS based, in part, on the SPRA the analysis used to model the response of the SSC itself, and any results using a methodology which is similar to the endorsed correlation of the SSC with similar SSCs which are expected to process described in NEI 00-04. This evaluation included be subject to similar loading during a given seismic event. As a solving the respective SPAR models, calculating the integrated result, many assumptions and sources of uncertainty can affect importance measures for each basic event which was the importance measures associated with a specific basic event consistently defined across the hazards of interest, and or SSC which may potentially impact the 10 CFR 50.69 evaluating the results. Basic events which did not meet the categorization results. Similarly, the associated human candidate HSS criteria based on SPRA results and those that reliability analyses, which considers the impact of the seismic could be determined not to meet the integrated importance event on NPP operator actions and responses, contain several measure results were excluded from further evaluation. The assumptions and sources of uncertainty which may impact the 10 remaining results were divided into the following three groups:
CFR 50.69 categorization results.
4
Result 1: Basic events that met the HSS criteria based on the seismically-induced failure of RCS, vessel internals or integrated importance measures, and met the HSS criteria secondary side SSCs based on the internal events PRA model. seismically-induced loss of electrical equipment not modeled Result 2: Basic events that met the HSS criteria based on the to a similar level of detail as in the associated internal events integrated importance measures, but did not meet the HSS or fire PRA models criteria based on the internal events PRA model. failure of SSCs or failure modes not directly modeled in the Result 3: Basic events that met the candidate HSS criteria associated internal events or fire PRA models: seismically-based on the SPRA results, but did not directly map to the fire induced failure of heat exchangers, tanks, battery racks, walls, and internal events PRA models for calculation of the compressors, instrumentation, valves integrated importance measures. seismically-induced external events (e.g., dam failure)
Of the SPAR models that were evaluated, many of the SPRA Many of these basic events would have met the integrated basic events which met the HSS criteria based on the integrated importance measure HSS criteria, even if the contribution from importance measure calculations also met the HSS criteria based other hazards was negligible.
on the internal events (IE) PRA model (Result 1). This result is not surprising because many of the same SSCs are relied upon to protect the plant, regardless of the hazard of concern. In the CASE 2: TRUNCATED FRAGILITY CURVE cases where the SSC would be found to be HSS based on the IE PRA results alone, the additional consideration of the SPRA The fragility evaluation is a crucial aspect of SPRAs and can results would offer minimal benefit if the process described in significantly influence the insights and metrics derived from an NEI 00-04 was used as the basis for 10 CFR 50.69 SPRA. As a result, the categorization under 10 CFR 50.69 using categorization, provided the seismic failure modes were also SPRAs will be influenced by such evaluations. Refinements in addressed by the internal events PRA categorization. the fragility evaluation are sought in the development of SPRAs Four of the eight SPAR models had SPRA basic events through the use of more detailed approaches and such which met the HSS criteria based on the integrated importance refinements usually result in a decrease in seismic CDF.
measure calculations but did not meet the HSS criteria based on In the context of the refinement in the fragility evaluations the IE PRA model (Result 2). In many cases, the basic events to support SPRA development and the corresponding impact on that fell into this category appeared to align to SSCs which would 10 CFR 50.69 categorization, it is conceivable that a truncated have been categorized as HSS based on the IE PRA model had a fragility curve for determining the plant-specific seismic CDF is full functional assessment and SSC mapping effort been used. The possibility of a fragility cutoff has been recognized in completed. However, it is likely that the seismic failure modes past studies, including NUREG/CR-4334, An Approach to the and associated components (e.g., supports) would not have been Quantification of Seismic Margins in Nuclear Power Plants [7].
appropriately considered in the absence of the SPRA results as In this study, the panel recognized that the conservative would be necessary for accurate categorization results. The capacities are close to the lower-bound cutoff values below SPRA results appeared to influence the HSS determination for which there is no significant likelihood of failure. That study several component types, including: batteries, swing diesel further stated that although lower-bound capacity values have generators, valves, tanks, and turbine driven pumps. This not been rigorously established, it is the belief of many engineers determination is expected to vary from plant to plant. that lower bound capacity values do exist in the absence of major The standardized nature of the SPAR models allowed for design and construction errors, and that earthquakes below or simplified comparisons across hazards and between plants. near the [safe shutdown earthquake] SSE are found not to Nonetheless, in some cases, components could not be directly contribute significantly, which is not surprising in light of the mapped across hazards. All eight SPAR models had SPRA basic generally conservative design practices used [7]. Such an events that met the candidate HSS criteria, but did not directly approach is customary in SMAs which do not require further map to the fire and internal events PRA models for calculation consideration if the High Confidence of Low Probability of of the integrated importance measures (Result 3). The basic Failure (HCLPF3) capacity exceeds the seismic margin events which did not directly map to the fire and internal events earthquake (SME; also known as the review level earthquake PRA models were highly dependent on how the specific model [RLE]) [7]. A similar screening approach was also used during was developed, and included: the recent Expedited Seismic Evaluation Process (ESEP) which was initiated following the accident at Fukushima Dai-ichi NPP seismically-induced failure of buildings, containment, the [11]. Therefore, the plant SSE, plant HCLPF value, or a fraction polar crane, or cooling towers thereof can be considered to be a reasonable candidate for the 3
HCLPF values represent the peak ground acceleration that corresponds to HCLPF values were originally developed using an expert elicitation approach a 95% confidence level of a 5% or less probability of failure, which can be shown and are currently more objectively defined in accordance with existing to correspond to a failure probability of 1% or less on the mean fragility curve. calculation procedures [7, 10].
5
truncation of the fragility curve for use in SPRAs if appropriately justified. where,
= mean probability of failure (l) = conditional probability of failure from fragility curve with respect to PGA
= derivative of hazard curve with respect to PGA The ASME/ANS PRA Standard [14] that is currently endorsed by NRC allows screening of sequences that contribute less than 1% to the hazard-specific CDF. Therefore, the SSE can represent a possible truncation level for the SPRA for a plant where an evaluation similar to that presented here shows a contribution that the PRA community considers as small. In such a case, the evaluation can also be extended to the plant HCLPF (or an intermediate value between the SSE and HCLPF) and a similar determination can be made. As noted previously, the numerical values presented in Table 1 and the above discussion are for illustration purposes only.
Figure 2. Notional Hazard and Fragility Curve During a relatively low intensity seismic event, the dominant failure mode may be different than it was for the In Figure 2, example mean hazard curves for two sites and failures which were used to develop and validate the fragility an example fragility curve is plotted. The hazard curve is plotted curves. The introduction of new failure modes can be on a logarithmic scale because the annual exceedance frequency problematic for defining representative failure models. Since decreases significantly as the intensity of the earthquake there is very limited test and experience data for failures which increases. As a result of this relationship, the left tail of the occur at PGA below the HCLPF value, the current fragility curve fragility curve can contribute a significant portion of the may not be representative of the true conditional probability of resulting risk. According to a study by Shiu, up to 50% of the failure in this region [15].
core damage frequency can be attributed to peak ground accelerations below the design level - or safe shutdown - Table 1. Illustration of Fragility Truncation earthquake [12]. This analytical result is contrary to the expected Example SSE Contribution of acceleration levels up outcome because the SSCs are not expected to fail when exposed Plant # PGA (g) to SSE to total seismic CDF (%)
to earthquake events that are within their design capacity. The 1 0.25 7 HCLPF is also indicated in Figure 2. 2 0.12 5 In order to gauge the potential of truncating the left tail of 3 0.15 1 the fragility curve for use in an SPRA, an assessment of the 4 0.2 38 potential risk contribution from the truncated portion has been 5 0.12 <<0.1 performed. The results from the SPAR models discussed 6 0.15 2 previously were used for the assessment. The intent of the 7 0.2 <<0.1 assessment performed here is to illustrate the concept. Table 1 provides information on the SSE and the contribution of the It is possible that the left tail of the fragility curve may be seismic CDF at the SSE to the total calculated seismic CDF using dominated by low cycle fatigue or even random failures. It has the plant-specific SPAR model results. As noted from Table 1, been hypothesized that SSCs do not fail at earthquake intensities certain plants, such as Plants 3, 5, and 7, have a relatively small that are low relative to the component capacities [7, 10]. If this contribution from accelerations at or below the SSE. However, is the case, removing this failure contribution may result in a the rest of the plants, notably Plant 4, have an appreciable more realistic seismic risk profile.
contribution from acceleration at or below SSE to the total seismic CDF. Such an evaluation can be performed before Certain caveats on the above evaluation are in order:
developing an SPRA by using the available plant level fragility curve, the plant-specific hazard curve, and the plant-specific SSE. The plant level fragility curve and the plant-specific It needs to be stressed that any truncation, including the hazard curve can be combined using the convolution method to assessment described above, will represent an important develop the seismic CDF as indicated in Equation 3 [13]. assumption in the development of the SPRA model. It is normal and expected practice to confirm the validity of such assumptions prior to using the SPRA results.
= 0 (l) ( ) (3) 6
Because the technical communitys understanding of the Similar to the results of Case 3, reducing the probability of seismic hazard at a particular NPP may increase in the a select class of HEPs resulted in variations in the importance future, it may be worthwhile for plants with even moderate measures of many other basic events. However, this sensitivity hazards to continue to refine and develop more realistic did not result in changes which appeared to significantly alter the modeling approaches. categorization results. Again, it did not appear that the changes would result in significantly different categorization results if a CASE 3: DECREASE FRAGILITY DRIVING LOSS OF full functional assessment and SSC mapping effort been OFFSITE POWER GIVEN SEISMIC EVENT completed. However, it was noted that a change in the categorization result would be possible for SSCs that were near Three SPAR models (representing Example Plants 1, 2, and the threshold values and may be likely in cases where the refined
- 7) were run with an improved LOOP fragility to identify whether HRA results in different failure sequences.
the categorization results would likely change based in part on the SPRA results using a process which is similar to the endorsed CONCLUSIONS process described in NEI 00-04. The improved LOOP fragility was developed by shifting the fragility curve such that the 10 CFR 50.69 requires a systematic evaluation process for conditional failure probability associated with the Bin 1 hazard categorization of SSCs based on their risk significance from was assigned to Bin 2, the conditional failure probability different hazards including the seismic hazard. The importance associated with the Bin 2 hazard was assigned to Bin 3, etc. The of seismic risk to the 10 CFR 50.69 categorization process can new Bin 1 fragility was reduced by an order of magnitude from vary based on various plant-specific attributes. A possible semi-the original value. The results were evaluated using a quantitative approach was postulated and presented for plants methodology similar to that discussed in the baseline evaluation which do not have a SMA or SPRA due to historical reasons.
and compared to the baseline results. The postulated approach was formulated based on the endorsed Because importance measures are relative, shifting the guidance that allows screening of the seismic risk if the seismic LOOP fragility resulted in variations in the importance measures CDF is less than 1% of the internal events CDF.
of many basic events. However, this sensitivity did not result in NRC SPAR models were used to explore the importance of changes which appeared to significantly alter the categorization seismic risk to the 10 CFR 50.69 categorization process. In order results. While there were some changes, as compared to the to further understand the significance of potential assumptions baseline evaluation with respect to the basic events which met and sources of uncertainty within the SPRA which may impact the criteria of Result 1, 2, or 3, it did not appear the changes the outcome of the categorization process, the following would result in significantly different categorization results if a assessments were performed:
full functional assessment and SSC mapping effort been completed. However, it was noted that a change in the Case 1: Baseline Assessment categorization result would be possible for SSCs that were near Case 2: Use of truncated lognormal fragility curves the threshold values. Interestingly, the resulting seismic CDF Case 3: The probability of a loss of offsite power due to a was at least 95% of the baseline seismic CDF for all three seismic event was reduced models. Case 4: Select HEPs were decreased to guarantee success The baseline assessment (Case 1) was performed to gauge CASE 4: DECREASE OPERATOR RECOVERY the significance of the SPRA results in the context of the overall ACTIONS GIVEN SEISMIC EVENT categorization results and to provide a control case for comparison purposes. The use of truncated lognormal fragility The same three SPAR models that were used to evaluate curves (Case 2) was evaluated as a potentially more realistic Case 3 (representing Example Plants 1, 2, and 7) were run and representation of the conditional failure of SSCs. The reduction all HEPs which represented failure to manually align and actuate in the probability of a loss of offsite power due to a seismic event were set to zero implying guaranteed success of those actions. (Case 3) was evaluated because it is generally a dominant The goal of this evaluation was to identify whether a more contributor to seismic CDF and seismic LERF and its refined seismic human reliability analysis would likely change significance could be reduced at some NPPs in future years. The the categorization results based in part on the SPRA results using decrease in select HEPs (Case 4) was evaluated because the a process which is similar to the endorsed process described in human reliability analysis can be significant to the SPRA results, NEI 00-04. This simplified assessment was intended to gauge but some currently used methods are simplified based on a the potential significance of the HEPs to the categorization bounding assessment. Cases 2, 3, and 4 represent areas of results. It does not address the significant changes to the potential future refinement in the development and application modeling logic which may occur as the HRA methodology is of SPRAs.
refined and additional SSCs are credited. The results were The baseline assessment showed that, for the SPAR models evaluated using a methodology similar to that discussed in the that were evaluated, many of the SPRA basic events which met baseline evaluation and compared to the baseline results. the HSS criteria based on the integrated importance measure 7
calculations also met the HSS criteria based on the IE PRA 2. Generic Letter 88-20, Supplement 4, Individual Plant model. This result is not surprising because many of the same Examination of External Events (IPEEE) for Severe SSCs are relied upon to protect the plant, regardless of the hazard Accident Vulnerabilities10 CFR 50.54(f), NRC, June of concern. However, four of the eight SPAR models had SPRA 1991.
basic events which met the HSS criteria based on the integrated importance measure calculations but did not meet the HSS 3. Information Notice 2010-18, Generic Issue 199, criteria based on the IE PRA model. In many cases, the basic Implications of Updated Probabilistic Seismic Hazard events that fell into this category appeared to align with SSCs Estimates in Central and Eastern United States on Existing which would have been categorized as HSS based on the IE PRA Plants, NRC, September 2010.
model if a full functional assessment and SSC mapping effort 4. Letter from the NRC, Request for Information Pursuant to been completed. However, in such circumstances, it is important Title 10 of the Code of Federal Regulations 50.54(f) to ensure that seismic failure modes and associated components Regarding Recommendations 2.1, 2.3, and 9.3, of the (e.g., supports) are appropriately considered in the absence of the Near-Term Task Force Review of Insights from the seismic PRA results. Fukushima Dai-ichi Accident, March 2012.
The truncated fragility curve assessment showed that truncating the fragility curve (i.e., not considering the structural 5. EPRI Report 1019200, Seismic Fragility Applications failures below a certain threshold), may be technically defensible Guide Update, 2009.
and may result in a reduction in calculated risk. Further, the 6. Modarres, Kaminskiy, Krivtsov, Reliability Engineering comparison of the likelihood of structural failure as compared to and Risk Analysis, Second Edition, 2010.
random failures can also be used to support such an argument.
Such an assumption would need to be adequately validated prior 7. NUREG/CR-4334, An Approach to the Quantification of to use. Seismic Margins in Nuclear Power Plants, U.S. Nuclear The assessments that varied the LOOP fragility and select Regulatory Commission, Washington, DC. August 1985.
HEPs demonstrated that, because importance measures are relative, variations in the importance measures of many other 8. Nuclear Energy Institute Report 00-04, 10 CFR 50.69 basic events were noted. However, the sensitivities did not result SSC Categorization Guideline, Revision 0, July 2005.
in changes which appeared to significantly alter the 9. U.S. Nuclear Regulatory Commission Regulatory Guide categorization results. It was noted that a change in the 1.201, Guidelines for Categorizing Structures, Systems, categorization result would be possible for SSCs that were near and Components in Nuclear Power Plants According to the threshold values. their Safety Significance, Revision 1, 2006.
The assessments in this paper were focused on the impacts on seismic CDF and the categorization therefrom. However, 10. EPRI NP-6041SLR1, Revision 1, A Methodology for further study of the impact with respect to seismic LERF is Assessment of Nuclear Power Plant Seismic Margin, needed. It has been observed from recent, high quality SPRA 1991.
results of U.S. NPPs that seismic LERF can be a significant due 11. EPRI Report 3002000704, Seismic Evaluation Guidance:
to dependencies associated with the seismic hazard. Augmented Approach for the Resolution of Fukushima Near-Term Task Force Recommendation 2.1 - Seismic, 2013.
ACKNOWLEDGMENTS
- 12. NUREG/CP-0070, Proceedings of the Workshop on The authors acknowledge the staff at the Idaho National Seismic and Dynamic Fragility of Nuclear Power Plant Laboratory (INL) and in NRCs Office of Nuclear Regulatory Components, August 1985.
Research (RES) for their contributions to the development and maintenance of the SPAR models. The authors also thank 13. Kennedy, R.P, Risk-Based Seismic Design Criteria, various stakeholders for informative discussions, via public Nuclear Engineering and Design, Vol. 192, 1999, pp. 117-meetings, on the topics presented in this paper. 135.
- 14. ASME/ANS RA-Sa-2009, Addenda to ASME/ANS RA-REFERENCES S-2008: Standard for Level 1/Large Early Release Frequency Probabilistic Risk Assessment for Nuclear
- 1. Generic Letter 87-03, Verification of Seismic Adequacy Power Plant Applications, an American National of Mechanical and Electrical Equipment in Operating Standard, American Society of Mechanical Engineers, Reactors, Unresolved Safety Issue (USI) A-46, NRC, 2009.
February 1987. 15. Modarres, Amiri, Jackson, Probabilistic Physics of Failure Approach to Reliability, Second Edition, 2017.
8