Equipment Selection Methodology of Seismic Probability Safety Assessment for Nuclear Power Plant

1. Introduction

A typical seismic probability safety assessment (SPSA) requires geological, structural, and systems analysts. In the case of probabilistic seismic hazard analysis (PSHA), geological experts can independently perform the analysis, but structural experts performing seismic fragility analysis (SFA). Probability safety assessment (PSA) model analysts performing system analysis, cannot perform the analysis independently. Only when the analysis is performed through mutual co-operation between the two fields can a reasonable result be achieved. This is a unique characteristic of SPSA.

For SPSA, a seismic equipment list (SEL) is first prepared, and reviews of the design, construction installation data, walkdown and system are conducted to analyze structures, systems and components (SSCs) that have little or no effect on the safety of the entire power plant. The final selected SSCs are reflected in the PSA model to perform SPSA. The following are the considerations in selecting a general SEL [1]:

• Identify SSCs that are important to safe shutdown from full-power PRA models.
• Identify SSCs from a review of seismic evaluation performed for the IPEEE (Individual Plant Examination for External Events).
• Identify structures and passive components that are important to the seismic response.
• Identify additional SSCs from a plant walkdown.

As can be seen from the above SEL selection criteria, equipment is required systematically for the safe shutdown of a nuclear power plant, and critical equipment vulnerable to earthquakes is selected through seismic response analysis and site surveys. In this paper, we propose the equipment selection methodology that should be essential to SPSA using the characteristics of individual plants and internal events PSA for existing qualitative and ambiguous selection criteria.

2. Practices for seismic equipment selection and screening criteria

In this paper, we review the practice of SEL selection and characteristics based on cases applied in Korea and the United States. In Korea, when an SPSA is performed, the SEL is selected in consideration of the general points mentioned in Chapter 1. After preparing the SEL, special screening criteria are set, which will be reviewed in detail here. Quantitative screening means that SSCs, which generally establish and meet appropriate screening criteria that are expected not to affect overall plant safety, are excluded from the SPSA model, and reflect and analyze only equipment and buildings that do not meet the screening criteria of the SPSA model.

For example, there are two screening criteria: the first is the specific median acceleration capacity (A_m), and the second is the high confidence of a low probability of failure (HCLPF). First, if A_m is set as the screening criteria, it should not contribute to plant safety risk regardless of the results of the site’s PSHA, so a relatively high value should be set as the screening criteria, which can be an overly conservative analysis. Moreover, there is a weakness in that there is no difference in core damage frequency (CDF), which is a risk metric for SPSA between a power plant with a high probability of seismic occurrence on the site and a power plant with a low probability of seismic occurrence.

Second, when HCLPF is set as the screening criteria, its value is the result of the convolution of the hazard curve derived by PSHA and fragility of a single piece of equipment. As an example, if 1.00E-05/yr, the convolution value computed in between the fragility of specific equipment and the site hazard curve is set as the screening criterion HCLPF, even if all of this equipment directly causes core damage, its CDF is 5.00E-07/yr, meaning less impact. In other words, HCLPF corresponds to a probability of equipment failure of 0.05 based on 95 % reliability; so conservatively, if the corresponding earthquake occurrence frequency is 1.00E-05/yr, the final CDF corresponds to 5.00E-07/yr. This will be analyzed in detail in the following sections; however, the disadvantage is that the SPSA results will be optimistic if seismic-induced failures are excluded on a single screening criterion because equipment that is critical to plant safety shutdown can be used in various seismic-induced initiating events and contributes to the CDF of each initiating event.

In the case of the United States, ASME/ANS RA-S–2008, which can be called a PRA standard, describes as follows [2]:

• DEVELOP seismic fragilities for all those structures, systems, or components, or combination thereof, identified by the systems analysis.
• If screening of high seismic capacity components is performed, DESCRIBE fully the basis for screening and supporting documents. For example, it is acceptable to apply the guidance given in EPRI NP-6041-SL, Rev. 1, and NUREG/CR-4334 to screen out components with high seismic capacity. However, CHOOSE the screening level high enough that the contribution to core damage frequency and large early release frequency from the screened-out components is not significant.

This provides a vague criterion that fragility analysis should be performed to reflect all SSCs in the SPSA model and screening can be applied, and this case should not have a serious impact on both CDF and large early release frequency (LERF). Next, guidance suggesting the US SPSA methodology recommends determining the detailed fragility analysis SSCs through the steps below [1].

• Step 1: Screen Inherently Rugged Structures, Systems, and Components.
• Step 2: Assign Initial, Representative Fragility Values for Structures, Systems, and Components.
• Step 3: Create and Quantify an Initial Seismic Probability Risk Assessment Model.
• Step 4: Rank the Systems, Structures, and Components by Importance Measure.
• Step 5: Perform Detailed Fragility Calculations for the Risk-Important Structures, Systems, and Components.
• Step 6: Document the Seismic Equipment List Screening Process.

All equipment except “inherently rugged structures, systems, and components” is considered in the SPSA model, so the evaluation result is quite conservative and the SPSA model must be established in advance to perform this analysis. In addition, since there are no specific screening criteria for detailed fragility analysis, there is a weakness in that the results may differ depending on the judgment criteria of the performing expert. As mentioned above, it is recommended that all SSCs should be included in the SEL and qualitatively selected and removed within the range that does not affect the entire CDF or LERF; however, no methodology can find a systematic method for this.

3. Sensitivity analysis for single HCLPF screening criteria

The Nuclear Energy Institute (NEI) presents the screening criteria for performing fragility calculation [3], which is similar to the HCLPF screening criteria applied in Korea. To confirm the validity of the screening criteria presented in this report, we performed a sensitivity analysis using the top 50 cutsets of SPSA based on the data contained in this report. The sensitivity analysis consisted of SPSA models assuming 4th seismic acceleration intervals for convenience. As a quantification tool, software using minimal cutsets upper bound (MCUB) methodology was used, but over-estimated CDF was derived from high seismic acceleration interval and quantified using FTeMC [4] software with Monte Carlo simulation. The top 50 cutsets presented in the report consist of 22 seismic-induced failures, ten random failure events, and four human errors events, as shown in Tables 1 and 2. These cutsets are also considered in the success sequence.

(Table 1)

(Table 2)

In this paper, sensitivity analysis is carried out on two cases. First, we analyzed the changes in CDF according to the application of the screening criteria using the results of the convolution with the single piece of equipment; and second, we analyzed the changes in CDF when each piece of equipment with similar HCLPF size is not considered in the model. To utilize the results of the first, single piece of equipment convolute with seismic hazard curve. The convolution value was derived using PRASSE [5], which is used in the quantification analysis of seismic-induced initiating events with binary decision diagram method and Monte Carlo simulation. The results are shown in Table 3.

(Table 3)

As Table 2 shows, the higher the HCLPF value, the smaller the convolution result, but it can be seen that there is a slight difference, depending on β_R and β_U. Table 4 shows the sensitivity analysis according to two screening criteria.

(Table 4)

When screening criteria 5.00E-5/yr is applied, four pieces of equipment are excluded from the model, and the CDF reduction is 0.5 % based on the base model, so there is no significant effect. With screening criteria 1.00E-4/yr, the change in the overall CDF is around 4 %, indicating that applying higher screening criteria can have a significant impact on the overall results. Therefore, the appropriate screening criteria was found to be effective because they might not have a significant impact on the entire CDF.

Second, we examine the amount of change in CDF for equipment with similar HCLPF. Even with similar HCLPF value, the A_m value may show large difference according to the β value, and the effect may be very different depending on the seismic acceleration intervals determined. In addition, the accident sequence varies depending on the plant characteristics, so the target equipment contributes to plant safety which can have a very different effect on the entire CDF. Therefore, screening based on single HCLPF criteria will not be able to understand all the effects without performing various sensitivity analyses and may have an optimistic effect on the overall results. Table 5 shows the analysis of the effect on CDF that is reduced when four pieces of equipment with similar HCLPF (battery; emergency diesel generator fuel oil day tank; 4kV-480V transformer; and bearing cooling heat exchanger) are excluded from the model.

(Table 5)

As a result, it can be seen that in the case of the battery, the reduction rate of CDF is very large, corresponding to 2.25 %, but in the case of the transformer, it is very small, corresponding to 0.2 %. Therefore, it can be seen that if a single HCLPF criterion is applied and screened-out, the effect on the entire CDF is different for each piece of equipment, and thus further analysis is required.

4. Equipment selection methodology

As discussed in the previous section, equipment selection for SPSA is not being carried out using a rational approach, such as by single screening criteria or conservatively considering all equipment. In this paper, we propose a methodology for an equipment selection methodology based on all three analysis steps of SPSA.

4.1. Determination of site seismic hazard region using PSHA

PSHA is the step of evaluating the site-specific seismic probability that is the input to the SPSA. Basically, the probability of an earthquake is the most important step because it directly affects the probability of a seismic-induced initiating event. In this paper, we reviewed the correlation between various SPSA cases and PSHA of each site. Table 6 shows the specific values of the site-specific seismic hazard curve and CDF values for each nuclear power plant [6, 7].

(Table 6)

 Although an attempt was made to review more cases of PSHA and seismic core damage frequency (SCDF), the published data are limited and insufficient to infer an accurate correlation; however, the trend can be confirmed using published data. The values that represent the correlation between site SPHA and SCDF are the probability of exceeding 1.0g and the probability of exceeding the safety shutdown earthquake (SSE). First, the probability of seismic acceleration exceeding 1.0g corresponds to a very large earthquake magnitude, so most power plants assume that direct core damage occurs. However, as a result of reviewing the case of power plants based on this value, it is difficult to represent the characteristics of seismic hazards based on Pr(a>1.0g), since the core damage frequency for earthquakes exceeding 1.0g accounts for a range of approximately 2 % – 14 %.

Next, we reviewed the probability of exceeding SSE. Since SSE is the reference value for seismic design of safety-related SSC, it can be assumed that in the event of an earthquake below SSE, most equipment can maintain its function, which has a significant impact on plant safety before and after the seismic acceleration corresponding to this value. When plotting the correlation between the probability of occurrence of SSE and CDF, it can be seen that the correlation between the two variables is high, as shown in Figure 1.

Figure 1. Relationship of Seismic Core Damage Frequency (SCDF) and probability for SSE.

Therefore, if this value is broadly divided into four types, as shown in Table 7, it is possible to classify the possibility of earthquake occurrence on the site by reoccurrence period.

(Table 7)

In the methodology proposed in this paper, the probability of SSE occurrence is basically divided into four major sections, and we intend to use this as the basis for determining the equipment groups that need to be subjected to fragility analysis. In the case of Region A, since earthquakes are very likely to occur, it is suggested that the number of equipment groups that need to be analyzed for fragility is most; and in the case of Region D, the equipment target for fragility analysis is least because of the lowest probability of earthquakes. To examine the adequacy of the four distinct values, the data on the seismic hazard values of power plants operating in the US [8] are examined as shown in Figure 2.

Figure 2. Hazard curve for US nuclear power plant [8].

Based on these values, it can be seen that the 61 power plants comprise 7 % in Region A, 21 % in Region B, and 26 % in Region C. Lastly, Region D occupies 46 %. It can be seen that the criteria are valid for classifying the overall seismic hazard trend of a specific site.

4.2. Determination of equipment group for fragility analysis

The second step of the SPSA is seismic fragility analysis. Seismic fragility analysis is performed by a structural expert, and it is a step to evaluate the seismic resistance for the weakest failure mode of each piece of equipment and to determine the median acceleration capacity (A_m); with β_R and β_U representing the variability. Seismic fragility analysis requires the greatest consumption of time and budget among the three steps, as structural experts with the particular expertise need to perform the analysis, as well as review and conduct on-site verification of many data such as design, construction, and installation works. Therefore, the methodology proposed in this paper is a method that suggests a way to minimize such analysis. First, through a survey of sufficient data on the fragility of general equipment, we examine the characteristics of the fragility of equipment generally installed in power plants. A great deal of research has been conducted on the fragility of general equipment. Among this research, according to the SPID report [9], the fragility of SSCs can be largely classified into three, and the first inherently rugged SSCs generally have very high seismic resistance, so there is no need to include inherently rugged SSCs in the SPSA model. Second, SSCs with somewhat high seismic resistance are reflected in the SPSA model if the impact on SPSA is significant after reviewing the magnitude. In the last case, SSCs must be considered in the SPSA model. In this paper, equipment groups are classified into four types based on generic fragility data, and the results are shown in Table 8.

(Table 8)

 Table 8 refers to three reports [1, 10, 11], and the equipment is categorized conservatively based on the lowest HCLPF value. The equipment group according to HCLPF value has the following characteristics:

• Group I (HCLPF: 0.2g or less): Generally, the seismic acceleration corresponding to plant SSE is taken to 0.2g, so when an earthquake occurs, SSCs will fail with a probability of around 5 %. This equipment group includes offsite power sources, which are generally considered the most vulnerable of plant installations, fixed electrical panels without anchor bolts, non-safety class buildings, and yard tanks. In this case, the equipment corresponding to the most vulnerable group should always be considered regardless of the magnitude of the site seismic hazard curve. Typical yard tanks include condensate storage tank and refueling water storage tank.
• Group II (HCLPF: 0.2g to 0.4g): Typically, this equipment group comprises active components which are operated by power source, including fixed electrical panels using anchor bolts, relays, batteries, small tanks, and non-safety grade diesel generators.
• Group III (HCLPF: 0.4g to 0.6g): Generally, this equipment group corresponds to the safety-related class component; mainly the equipment and power sources considered for accident mitigation.
• Group IV (HCLPF: 0.6g or higher): Generally, this equipment group includes various valves, pipes that make up the inherently rugged SSCs, including pressurizer, steam generators, and equipment installed in the main system, such as safety injection tanks. The main component of this group is equipment that directly causes core damage.

After applying these groups to individual power plants, there may be cases different from the general fragility due to a plant’s unique design characteristics or constructability. Therefore, equipment different from the general fragility should be included in the analysis through a walkdown that is essential for SPSA.

4.3 Seismic plant response analysis (SPRA)

The final step is the plant seismic response analysis, which is divided into an analysis of equipment that can cause an initial event and equipment that should be used to mitigate safety shutdown such as engineering safety feature when a seismic-induced initial event occurs. Equipment that may cause an initial event should be included conservatively without a screening process, as all of the initial events may cause direct core damage. However, this equipment may be excluded if the frequency of occurrence is calculated to be below 1.00E-07/yr, which corresponds to the typical initial event screening criteria [12]. Table 9 shows equipment that can initiate seismic event in general light-water reactor nuclear power plant.

(Table 9)

In general, seismic events can occur in seismic-induced loss of coolant accident, loss of power, loss of control, loss of ultimate heat sink, main steam line break, and anticipated transient without scram.

Next, when the seismic-induced initiating event occurs, an analysis of the equipment considered to mitigate the event is performed. Like all external event analysis, SPSA basically uses the event tree and fault tree used in the internal event PSA. Here, non-safety and non-seismic equipment that cannot be used conservatively is excluded from the model. However, although seismic-induced initiating events are different from internal events, the primary heat removal, which is essential for mitigation of the accident, must be performed by the same equipment and procedure. Therefore, in the ASME Standard, which is considered the standard of the PSA, equipment corresponding to the standard can be listed based on FV (Fussell-Vesely) importance 0.005 higher and RAW (Risk Achievement Worth) value of two or higher, which are the criteria that can be used to classify significant basic events. To confirm the application of importance value, a sensitivity analysis was performed on the reference nuclear power plant. In Table 10, we show the difference between the results of the existing SPSA and CDF when only the equipment that was evaluated as important in internal events was modeled.

(Table 10)

As a result, it is confirmed that the results of the sensitivity analysis showed no significant difference between the case of modeling all equipment and the case of modeling only equipment that was evaluated as important in internal events. When 242 items of equipment, 13 % of the total of 1798, were modeled in the SPSA model, the CDF was 97.6 % of the baseline CDF, and even when 49 pieces of equipment classified as important in basic events based on FV are modeled, a result equivalent to 97 % of the baseline CDF was obtained. In addition, even when only 24 pieces of equipment were considered, a value corresponding to 95.5 % of the CDF value of the base model was derived, indicating that the importance of other equipment other than this was much lower than expected. Therefore, it is judged that the equipment selection methodology based on the values of the internal event FV and RAW is much more rational than the existing HCLPF-based method, and is an efficient method that can reflect the characteristics of the SPSA model.

4.4 Summary of the proposed equipment selection methodology

The procedure is shown in Figure 3, a flow chart of the method of selecting equipment for seismic events over the three steps described above.

(Figure 3)

First, based on the site seismic hazard curve obtained as a result of PSHA, the SSE value is checked, which is the design standard of the power plant, and probability value exceeding SSE value is checked. After that, the region of ​​the site is determined by checking the SSE reoccurrence period according to the SSE excess probability value. Through this value, the group of equipment that should perform fragility analysis is determined. Next, based on the results of the PSA importance of internal events, basic events with an FV value of 0.005 or more or RAW value of two or more, are listed and described in terms of the function of the system. However, basic events related to non-seismic equipment and human error are excluded among the results of the PSA importance of internal events. Finally, equipment is derived through cross-examination between the equipment required for the function of the mitigating equipment derived from an internal event and the equipment group subject to fragility analysis.

5. Case study for the equipment selection methodology

In this paper, we compare and analyze the seismic PSA results of the reference nuclear power plant using the proposed methodology. Existing PSA results could not be considered, as detailed data on the walkdown could not be collected. However, even if the walkdown is excluded, the results are sufficient to demonstrate the effectiveness of the methodology proposed in this paper. The first step is the analysis of site-specific PSHA results. The review of the corresponding hazard curves indicates that the probability of earthquake occurrence is relatively low, and the probability of exceeding 0.2g on the SSE basis corresponds to 1.06E-04/yr, corresponding to the Region C medium risk region proposed in this paper. In terms of the likelihood of earthquake occurrence, it is a value that is approximately 54 % of the total power plants in terms of seismic hazard considered by Figure 2. Based on this, the plant needs to review equipment groups I and II subject to fragility analysis. Fragility analysis groups I and II include general vulnerable equipment including offsite power sources and yard tanks, as well as general active equipment. The next analysis is the plant seismic response analysis using internal events PSA results; 74 basic events were selected with an FV value of 0.005 or more and 253 basic events with a RAW value of two or more. Among them, 30 pieces of equipment that satisfy both FV and RAW are considered, along with 297 basic events. Excluding 20 human error basic events and 26 non-seismic basic events, 251 basic events are considered. In addition, except for valves, flow elements, flow transmitters, radiation transmitters, dampers and filters, which are inherently rugged SSCs, the total 136 basic events are derived.

The 136 basic events derived can be divided into related systems and equipment types to achieve the following 14 critical system functions. Here, if we review the 14 important functions, we can confirm easily the unique operating characteristics of the reference plant.

• Auxiliary feedwater (AF) supply
• AF pump room cooling
• Emergency power supply
• Component cooling water supply
• Diesel generator fuel supply
• Diesel generator room cooling
• Essential chilled water supply
• High-pressure injection
• Low-pressure injection
• Plant control
• Ultimate heat sink
• Reactor containment cooling
• Safety actuation signal
• Ultimate heat sink pump room cooling

In the end, cross-examining 14 critical systems and functions with fragility equipment groups I and II, the results shown in Table 11 are obtained.

(Table 11)

The number of equipment items derived through cross-examination is a total of 23, which is a very small result considering the overall equipment in nuclear power plants. However, when reviewing the previously analyzed SPSA results, it can be confirmed that all devices are considered important in the existing SPSA model except for the three pieces of equipment that initiate seismic-induced initiating events, so the methodology proposed in this paper is very efficient. It can be confirmed that it is reasonable.

6. Conclusion

In this study, a methodology of the equipment selection for SPSA is proposed. The single HCLPF screening criterion which has been applied for SPSA reflects some of the site-specific PSHA results but does not reflect the plant design characteristics, so if the single HCLPF screening criterion is applied to the model based on this, an optimistic evaluation can be made. The methodology proposed in this paper has the following advantages:

• Safety aspects: Single HCLPF screening criteria is not applied, and all equipment is not reflected in the model using general fragility data, so realistic and reasonable SPSA results are expected.
• Economics aspects: As the largest portion of the required manpower for SPSA is the detailed fragility analysis work, the methodology proposed in this paper is economically beneficial as the number of pieces of equipment subject to SPSA decreases.

The equipment selection methodology proposed in this paper requires analysis of all three parts of SPSA, unlike the previously proposed methodology. This means that selecting equipment through consideration of only one part, such as PSHA, may lead to incorrect results. SPSA has its own uncertainty, so if one factor affects several steps, the uncertainty becomes very large. Therefore, the analysis of equipment not essential for SPSA can increase this uncertainty, so it can be said that it is necessary for a much more realistic analysis that is not considered in advance.

In recent years, SPSA has evolved from a single unit criterion evaluation to an evaluation of multiple units operating at one site. In this case, the number of equipment items handled by the SPSA increases, resulting in unnecessary model enlargement. Therefore, even in such case, it is necessary to select equipment that has a significant influence on seismic events, and applying the equipment selection methodology proposed in this paper can contribute to the simplification of the SPSA model and the accuracy of the analysis of the quantitative results, and minimize unnecessary analysis. Therefore, it is expected that uncertainty errors will be minimized.

References

[1] Seismic Probabilistic Risk Assessment Implementation Guide. USA: Electric Power Research Institute; 2013, 3002000709.

[2] Standard for Level 1/Large Early Release Frequency Probabilistic Risk Assessment for Nuclear Power Plant Applications. USA: The American Society of Mechanical Engineers; 2009, ASME/ANS Ra-SA-2009.

[3] White Paper: Criterion for Capacity-based Selection of SSCs for Performing Fragility Analysis in a Seismic Risk-based Evaluation. USA: Nuclear Energy Institute; 2012.

[4] FTeMC Quick Guide Fault Tree Top Event Probability Evaluation Using Monte Carlo Simulation. Korea: Korea Atomic Energy Research Institute; 2017, KAERI-ISA-memo-FTeMC-01, Rev. 1.

[5] Development and Validation of the Seismic Probabilistic Safety Assessment Software PRASSE. Korea: Korea Atomic Energy Research Institute; 2012, KAERI/TR-4649.

[6] Probabilistic safety assessment for seismic events. IAEA; 1993, TECDOC-724.

[7] M Vermaut, Ph Monette P Shah R.D Campbell, Methodology and results of the seismic probabilistic safety assessment of Krško Nuclear Power Plant, Nuclear Engineering and Design, 2 May 1998, Volume 182, Issue 1, Pages 59-72

[8] Risk Assessment of Operational Events Handbook Volume 2 – External Events. USA: Nuclear Regulatory Commission; 2008, Revision 1.01

[9] Seismic Evaluation Guidance Screening, Prioritization and Implementation Details (SPID) for the Resolution of Fukushima Near-Term Task Force Recommendation 2.1: Seismic. USA: Electric Power Research Institute; 2013, 1025287

[10] A Methodology for Analyzing Precursors to Earthquake-Initiated and Fire-Initiated Accident Sequences. USA: Nuclear Regulatory Commission; 1998, NUREG/CR-6544.

[11] Surry Seismic Probabilistic Risk Assessment Pilot Plant Review. USA: Electric Power Research Institute; 2010, 1020756.

[12] Identification of External Hazards for Analysis in Probabilistic Risk Assessment. USA: Electric Power Research Institute; 2011, 1022997.