ML20258A002
| ML20258A002 | |
| Person / Time | |
|---|---|
| Site: | Vogtle  |
| Issue date: | 12/10/2020 |
| From: | John Lamb Plant Licensing Branch II |
| To: | Gayheart C Southern Nuclear Operating Co |
| Lamb J | |
| References | |
| EPID L-2019-LLR-0109 | |
| Download: ML20258A002 (16) | |
Text
December 10, 2020 Ms. Cheryl A. Gayheart Regulatory Affairs Director Southern Nuclear Operating Co., Inc.
3535 Colonnade Parkway Birmingham, AL 35243
SUBJECT:
VOGTLE ELECTRIC GENERATING PLANT, UNITS 1 AND 2 - AUDIT REPORT FOR THE PROMISE VERSION 1.0 PROBABILISTIC FRACTURE MECHANICS SOFTWARE USED IN RELIEF REQUEST VEGP-ISI-ALT-04-04 (EPID L-2019-LLR-0109)
Dear Ms. Gayheart:
By letter dated December 11, 2019, as supplemented by letter dated September 9, 2020, Southern Nuclear Operating Company (SNC) submitted to the U.S. Nuclear Regulatory Commission (NRC), a proposed alternative to the inservice inspection (ISI) requirements of the American Society of Mechanical Engineers Boiler and Pressure Vessel Code for the steam generator main steam outlet nozzle-to-vessel welds and feedwater nozzle-to-vessel welds and nozzle inside radius sections of the Vogtle Electric Generating Plant (Vogtle) Units 1 and 2.
SNC included in its submittal the non-proprietary Electric Power Research Institute (EPRI)
Report No. 3002014590, Technical Bases for Inspection Requirements for [Pressurized-Water Reactor] PWR Steam Generator Feedwater and Main Steam Nozzle-to-Shell Welds and Nozzle Inside Radius Sections, April 2019 (ADAMS Accession No. ML19347B107), as the technical basis for the proposed alternative.
The NRC staff review is for the Vogtle, Units 1 and 2, plant-specific alternative proposed by SNC. The NRC staff is not approving the EPRI Report generically.
The EPRI technical basis report contains probabilistic fracture mechanics (PFM) analyses that SNC is citing, in part, as the technical basis for extending the ISI interval for the subject plant-specific components from 10 to 30 years. The PFM analyses were performed using the software, PRobabilistic OptiMization of InSpEction (PROMISE) Version 1.0, developed by Structural Integrity Associates.
On June 29, July 1, and July 27, 2020, the NRC staff conducted an audit of PROMISE Version 1.0 to verify that it properly implements PFM principles and has undergone adequate verification and validation, following the audit plan issued by letter dated May 14, 2020 (ADAMS Accession No. ML20128J311).
Pursuant to Section 2.390 of Title 10 of the Code of Federal Regulations (10 CFR), we have determined that the Enclosure does not contain proprietary information.
C. Gayheart If you have any questions, please contact me at 301-415-3100.
Sincerely, Digitally signed by John G. Lamb John G. Lamb Date: 2020.12.10 09:01:30 -05'00' John G. Lamb, Senior Project Manager Plant Licensing Branch II-1 Division of Operating Reactor Licensing Office of Nuclear Reactor Regulation Docket Nos. 50-424 and 50-425
Enclosure:
Audit Report cc: Listserv
AUDIT REPORT BY THE OFFICE OF NUCLEAR REACTOR REGULATION FOR THE PROMISE VERSION 1.0 PROBABILISTIC FRACTURE MECHANICS SOFTWARE REGARDING RELIEF REQUEST VEGP-ISI-ALT-04-04 VOGTLE ELECTRIC GENERATING PLANT, UNITS 1 AND 2 DOCKET NOS. 50-424 AND 50-425
1.0 BACKGROUND
By letter dated December 11, 2019 (Agencywide Documents and Access Management System (ADAMS) Accession No. ML19347B105), as supplemented by letter dated September 9, 2020 (ADAMS Accession No. ML20253A311), Southern Nuclear Operating Company (SNC, the licensee), submitted to the U.S. Nuclear Regulatory Commission (NRC), a proposed alternative to the inservice inspection (ISI) requirements of the American Society of Mechanical Engineers Boiler and Pressure Vessel Code (ASME Code) for the steam generator (SG) main steam outlet nozzle-to-vessel welds (NVWs) and SG feedwater NVWs and nozzle inside radius (NIR) sections of the Vogtle Electric Generating Plant (Vogtle), Units 1 and 2. SNC requested the alternative for the remainder of the 4th ISI Interval through to the end of the 6th ISI Interval.
Specifically, pursuant to Title 10 of the Code of Federal Regulations (10 CFR), Part 50, Paragraph 50.55a(z)(1), the licensee proposed to increase the ISI interval for the subject components to 30 years, from the current ASME Code Section,Section XI requirement of 10 years. The regulation 10 CFR 50.55a(z)(1) requires the licensee to demonstrate that the proposed alternative provides an acceptable level of quality and safety. The licensee included in its submittal non-proprietary Electric Power Research Institute (EPRI) Report No.
3002014590, Technical Bases for Inspection Requirements for PWR Steam Generator Feedwater and Main Steam Nozzle-to-Shell Welds and Nozzle Inside Radius Sections, April 2019 (ADAMS Accession No. ML19347B107), as the technical basis for the proposed alternative. In its request, the licensee included an applicability evaluation of the EPRI technical basis report to Vogtle, Units 1 and 2.
The NRC staff review is for the Vogtle, Units 1 and 2, plant-specific alternative proposed by SNC. The NRC staff is not approving the EPRI Report generically.
The EPRI technical basis report contains probabilistic fracture mechanics (PFM) analyses that the licensee is citing, in part, as technical basis for extending the ISI interval for the subject components from 10 to 30 years. The PFM analyses were performed using the software, PRobabilistic OptiMization of InSpEction (PROMISE) Version 1.0, developed by Structural Integrity Associates, Inc. (SIA).
2.0 SCOPE AND PURPOSE By letter dated May 14, 2020 (ADAMS Accession No. ML20128J311), the NRC staff issued an audit plan for the PROMISE Version 1.0 software. The original audit plan was for June 29, 2020, and July 1, 2020. The audit was not completed in 2 days and a third day was added on Enclosure
July 27, 2020. On June 29, July 1, and July 27, 2020, the NRC staff conducted the audit with SNC and its contractors, EPRI and SIA, via Microsoft Skype. The NRC staff conducted the audit of PROMISE Version 1.0 to verify that it properly implements PFM principles and has undergone adequate Verification and Validation (V&V).
3.0 AUDIT TEAM The audit team consists of Mr. John Lamb, Senior Project Manager for Vogtle, Units 1 and 2, in the Division of Operating Reactor Licensing; Dr. David Rudland, Senior Technical Advisor in the Division of New and Renewed Licenses; and Mr. David Dijamco, Technical Reviewer in the Vessels and Internals Branch, Division of New and Renewed Licenses.
4.0 AUDIT REPORT During the audit, the licensee presented information on the audit items listed in Section 4.0 of the audit plan. The licensee also established an online portal and placed in the portal information on each audit item. Section 5.0 of this audit report lists the documents and files the NRC staff reviewed. The NRC staff discussed with the licensee each audit item and made the following relevant observations:
o Item 1.a - Show the listing of V&V cases/tests that were conducted in the software V&V plan (SVVP) and V&V reports (References 73 and 74 of the EPRI technical basis report).
What were the validation activities? Just benchmarking?
o The NRC staff reviewed the SVVP, proprietary report DEV1806.401, and the accompanying calculation packages, proprietary documents DEV1806.301, DEV1806.302, and DEV1806.303, for PROMISE Version 1.0 provided in the audit portal.
The NRC staff noted the descriptions of the test plans in Section 7.0 of the SVVP but could not locate in the calculation packages the documentation of the tests. The licensee clarified the exact locations within DEV1806.301, DEV1806.302, and DEV1806.303 where the tests were documented. The NRC staff understood the licensees clarifications, but noted that the SVVP could have provided a better explanation of the alignment between the test plan and test execution.
o The observations in Item 3.a apply to the validation and benchmark questions for this audit item.
o Item 1.b - Show the V&V documents for the nozzle corner crack stress intensity factors (SIFs) from TIFFANY software and how these SIFs are input into PROMISE Version 1.0.
o The NRC staff reviewed proprietary documents DEV1801.302 and DEV1806.301 on the NIR SIFs provided in the portal.
o The licensee stated that stresses due to transients are calculated at incremental time steps; PROMISE Version 1.0 then interacts internally with TIFFANY to determine SIFs for a range of crack sizes from the resulting stresses for each time step of applicable transients; then the maximum and minimum SIFs are calculated and input into PROMISE Version 1.0 as a table.
o The NRC staff asked what version of TIFFANY interacts with PROMISE Version 1.0, was V&V performed on the version, and what were the differences between this version
and the TIFFANY version that interacts with the NRC-sponsored PFM software, xLPR.
The licensee stated that the version of TIFFANY that interacts with PROMISE Version 1.0 is called SI-TIFFANY, Version 3.1 (Reference 66 of the EPRI technical basis report) and that V&V has been performed on this version. The licensee also summarized the differences with the version that interacts with xLPR. Among the differences were the capability to import stresses determined from a separate source and addition of the nozzle inner radius (NIR) crack model. The NRC staff noted that proprietary document DEV1806.301 included a summary of the verification performed for the NIR crack model and the interaction of PROMISE Version 1.0 and TIFFANY.
The NRC staff also noted that the licensee documented the development SIFs for the NIR based on the Universal Weight Function method in proprietary document DEV1801.302.
o During the discussion of Figure 7 in the SMiRT-25 Conference paper (see table of documents and files reviewed), the NRC staff asked how SIFs at different aspect ratios for the NIR crack model are addressed. The licensee explained that unlike the NVW crack model, the aspect ratio for the NIR crack model is set to a single depth-to-length aspect ratio of 0.5.
o The NRC staff asked how the SIFs for cylinders used in the EPRI technical basis report were appropriate for the NVWs. The licensee stated that the cylindrical SIFs were demonstrated to be conservative with respect to the geometric configuration of the NVW.
o Item 1.c - Show users manual (typically called for by software quality assurance guidance).
o The NRC staff reviewed the users manual for PROMISE Version 1.0, proprietary report DEV1806.402, provided in the portal and noted the discussion of percent error in Section 2.8 of the manual. This prompted the NRC staff to ask how the accuracy of the probability values were considered in the analyses in the EPRI technical basis report.
As an example, the NRC staff discussed one of the probability values in the EPRI technical basis report that gave a large percent error using the percent error equation in the users manual. The licensee explained that overall, the resulting probability values were so low that a large percent error is expected.
o Item 2.a.i - Clarify if one realization is equivalent to one nozzle.
o The licensee clarified that one realization in PROMISE Version 1.0 is equivalent to one nozzle.
o Item 2.a.ii - Table 8-7 of the EPRI technical basis report indicates that there were 10 million aleatory realizations and 1 epistemic realization. How would results change if a combination of aleatory and epistemic realizations were implemented?
o The licensee explained that PROMISE 1.0 has an outermost Epistemic loop. Within the outer loop, there is an aleatory loop. All variables identified as epistemic are sampled within the epistemic loop. The variables identified as aleatory are sampled within the aleatory loop. The time loop is the innermost loop. Within the time loop, stress intensity factors (KI) are calculated for the current crack. Based on the KI, crack growth rate is calculated. The crack size is updated using the crack growth rate during this time step.
This process is repeated until either the end of life is reached or failure occurs.
o The licensee explained that if a combination of aleatory and epistemic realizations were implemented, the resulting probabilities will have percentile bands. The licensee showed a probability plot for case FEW-P1N as an example, ran in PROMISE Version 1.0 with a combination of aleatory and epistemic realizations and a fracture toughness (KIC) value of 80 ksiin (standard deviation = 5 ksiin) specified as an epistemic variable. The plot included probabilities of rupture at percentiles of 0.05, 0.10, 0.50, 0.90, and 0.95. The NRC staff noted that the probability of rupture per year at 80 years at the 50th percentile (0.50) is about (7x10-6)/80 = 8.75x10-8 per year, which, as expected, is close to the value of 8.95x10-8 per year in Table 8-13 of the EPRI report for the same case (the case in the EPRI report was performed with 1 epistemic realization and 10 million aleatory realizations).
o Item 2.a.iii - Show an input deck to clarify which parameters are random and which parameters are constant.
o The licensee showed a table of input parameters that specified whether each parameter is random or constant. The observations on the input deck in Item 2.h.i apply to this audit item.
o Items 2.a.iv.1 and 2.a.iv.2 - Explain the sampling and time loops and how crack depth is updated in the crack growth calculations.
o The licensee explained that for each realization or trial, PROMISE Version 1.0 determines a time to failure to develop a histogram of failure times. The software estimates the probability of failure in a given time as the fraction of the total number of trials that the computed failure time is less than the given time.
o The licensee explained that PROMISE 1.0 has an outermost Epistemic loop. Within the outer loop there is an aleatory loop. All variables identified as epistemic are sampled within the epistemic loop. The variables identified as aleatory are sampled within the aleatory loop. The time loop is the innermost loop. Within the time loop, stress intensity factors (KI) are calculated for the current crack. Based on the KI, crack growth rate is calculated. The crack size is updated using the crack growth rate during this time step.
This process is repeated until either the end of life is reached or failure occurs.
o Item 2.b.i - To understand the flaw distribution model used in PROMISE Version 1.0, show the values of depth and length (and density? or is density a single density instead of a distribution of densities?) for several realizations.
o The licensee showed sampled values of depth and length for several realizations. Also, file FEW-3PA.smp.csv provided for Item 2.h.i shows more sampled values of depth and crack length.
o The NRC staff asked whether it is conservative to treat the flaws in the NIR and NVW separately. The licensee cited Figures 4-8 and 4-10 of the EPRI technical basis report to explain that these two locations are far enough from each other in the same nozzle that flaws in either location may be treated independently.
o Item 2.b.ii - Section 8.2.4.3.5 of EPRI technical basis report: explain single flaw in each case.
o The licensee clarified that in the sensitivity study on the number of flaws in Section 8.2.4.3.5 of the EPRI report, two cases were analyzed. Each realization in the first case had exactly one flaw in the NVW location. In the second case, some realizations may not have had any flaws, while other realizations may have had more than one flaw. The licensee stated that the number of flaws in the second case represented the Poisson distribution, with a mean value of one flaw.
o The licensee explained that when there was more than one flaw in a realization, flaws do not interact. The NRC staff asked whether it was conservative to not consider flaw interaction. The licensee stated that flaw interaction is not significant and that the assumption of no flaw interaction is consistent with the analyses in report BWRVIP-108, Technical Basis for the Reduction of Inspection Requirements for the Boiling Water Reactor Nozzle-to-Shell Welds and Nozzle Blend Radii (Reference 2 of the EPRI technical basis report).
o Item 2.b.iii - Section 8.2.4.3.6 of EPRI technical basis report: why did the PVRUF [Pressure Vessel Research Users Facility] distribution end up with slightly lower probability of failure value? It was expected to result in higher probability of failure value. Show a comparison of the PVRUF and Marshall flaw distribution inputs. (Section 8.2.2.2 of the report states that PVRUF was slightly more conservative than the Marshall.)
o The licensee showed a plot of cumulative probability of the PVRUF and Marshall distributions and explained that at a certain cumulative probability value (the y-axis of the plot), the Marshall curve gives a deeper flaw than PVRUF, which can lead to a higher probability result. The licensee stated, however, that overall the PVRUF flaw distribution is slightly more conservative and that the difference in probability results shown in Table 8-22 of the EPRI technical basis report between Marshall and PVRUF is insignificant.
o Item 2.b.iv - Show SIA calculation package 1700313.301 for derivation of depth and length.
o The NRC staff reviewed proprietary document 1700313.301 provided in the portal and asked what the three curves on flaw length in Figure 7 of the document represented.
The licensee stated that those curves represent three separate flaw data from NUREG-6817, Generalized Procedure for Generating Flaw-Related Inputs to the FAVOR Code, August 2013 (Reference 60 of the EPRI technical basis report). The licensee also stated that one of the three curves was selected for the flaw length distribution in the PFM analysis because it had the most data and was the most conservative of the three.
o Item 2.b.v - Clarify: Page 8-31 of the EPRI technical basis report mentions a distribution derived by SI[A] based on data from NUREG/CR-6817. Is this distribution the same as Equation 8-1 of the report? Can you change the depth parameter in Equation 8-1?
o The licensee clarified that the SIA-derived flaw distribution mentioned in Section 8.2.4.3.6 of the EPRI technical basis report is not the same as the flaw distribution represented by Equation 8-1 of the report. The licensee also stated that the crack depth parameter in Equation 8-1 was not changed.
o Item 2.c.i - The EPRI technical basis report does not discuss the interaction between the probability of detection (POD) curve and ISI. Show how POD and ISI are implemented into the PROMISE Version 1.0 code by showing the software requirements for this feature and running some example problems. Refer to Sections 8.2.2.3 and 8.2.2.8 of the EPRI technical basis report.
o The licensee explained how the POD curve is implemented into PROMISE Version 1.0.
At the times inspection (i.e., ISI) is applied, flaws are either detected or not detected with the chance of detection/non-detection given by the POD curve. If detected, a flaw is assumed to be repaired or properly dispositioned, and thus cannot cause failure; if not detected, the flaw continues to grow and thus can lead to failure. No adjustment factor is applied to the failure probabilities as a result of the POD curve.
o The licensee showed in the input deck the fields for the POD curve, number of ISIs and ISI frequency, and examination coverage (expressed as a number between 0 to 1, inclusive).
o The licensee stated that POD and ISI are independent, and that the POD curve is the same, for example, if ISI is performed every 10 years or every 30 years. The number of ISIs and its frequency are input into PROMISE Version 1.0.
o The licensee explained that each ISI does not depend on the results of the previous ISI and stated that this independence is conservative.
o The licensee benchmarked the application of ISI as discussed in Item 3.a.
o The NRC staff asked about one of the inputs related to POD and ISI: examination coverage, which refers to the amount of weld volume examined during ISI. The licensee analyzed the effect of examination coverage and presented its results in SIA letter report 1900064.406.R0, which was uploaded into the portal. The licensee confirmed that PROMISE Version 1.0 was used for the PFM analyses in the letter report. The licensee explained how examination coverage is implemented in the software through a hypothetical case with 50 percent examination coverage (meaning only 50 percent of the weld volume is examined) for 10 realizations. The licensee stated that for approximately 50 percent of the realizations (5 realizations for this example) at the specified times of ISI, the POD curve will not be applied for the postulated flaws; thus, the postulated flaws will continue to grow for these 5 realizations. Given the importance of the results of the PFM analyses in 1900064.406.R0, the NRC staff requested the letter report to be submitted to the docket.
o Item 2.d.i - Discuss fracture toughness and the standard deviation used in the base and sensitivity cases discussed in the EPRI technical basis report. Show the runs for FEW-P3A for the base case and those in the sensitivity study for fracture toughness.
o The licensee stated that because the median value of 200 ksiin is close to the lower bound of the fracture toughness data shown in Figure 8-8 of the EPRI technical basis report, a nominal standard deviation of 5 ksiin was used and that this approach is consistent with the value assumed in BWRVIP-108.
o The NRC staff observed the run files (see table of documents and files reviewed) provided in the portal for this item and a related run file in Item 3.b. These runs are for
the FEW-P3A case in Tables 8-13 and 8-14 of the EPRI technical basis report. The NRC staff noted in the output files that the runs include the impact of preservice inspection (PSI) with ISI at 20, 40, and 60 years.
o The licensee explained how the results with low probabilities were calculated using the FEW-P3A case as an example. The probabilities of rupture and leakage reported for this case in Tables 8-13 and 8-14 of the EPRI technical basis report are 1.25x10-9 and 2.5x10-9 per year, respectively; these values are calculated by adding 1 failure or 1 leak in 10 million realizations to the corresponding values in the output file for this case. The NRC staff observed that the probabilities of rupture and leakage at 80 years in output file FEW-P3A.rpt are 0.0E+00 and 1E-7, respectively. Adding 1/10 million to these values and dividing by 80 years gives 1.25x10-9 and 2.5x10-9 per year, which match the values in Tables 8-13 and 8-14 for FEW-P3A.
o Items 2.e.i and 2.e.ii - Show how stresses are input into PROMISE Version 1.0 and explain how the software handles thermal transients.
o The licensee deferred to the response in Item 1.b and reiterated that the maximum and minimum SIFs for each crack size are calculated and input to PROMISE Version 1.0 as a table. During the discussion, the NRC staff summarized its understanding of how stresses and SIFs are handled by PROMISE Version 1.0: stresses are determined outside of the software (for example, through finite element analysis); PROMISE Version 1.0 then interacts internally with TIFFANY to determine SIFs from the stresses.
The licensee confirmed that the NRC staffs understanding is correct.
o The NRC staff asked for confirmation if all the analysis runs include crack face pressure, and the licensee confirmed that crack face pressure is included.
o The NRC staff reviewed output files FEW-P3A.rpt and FEW-P3A-ISI.rpt (provided for Item 3.b) to confirm if the transients shown in Table 5-5 of the EPRI technical basis report were included in the analysis runs. The NRC staff noted that all transients in Table 5-5, except the Loss of Power transient (see related observation on Loss of Power transient in the discussion of Item 2.e.iii), were clearly included. The NRC staff could not locate in the output files evidence that the Loss of Power transient was included. The NRC staff also reviewed the corresponding input files, FEW-P3A.vnz3 and FEW-P3A-ISI.vnz3, and noted the following input block for an LTOP [low temperature overpressure protection] Event.
RS.mxn 0.1 The NRC staff consulted the PROMISE 1.0 Theory and Users Manual (see table of documents and files reviewed) provided in Item 1.c, and noted that, compared to the input fields for the other four transients in Table 5-5 of the EPRI report, this LTOP Event only has load RS.mxn with a multiplier of 0.1 (and 1 cycle per year). The Heat-up/Cooldown transient, for example, has these other loads and multipliers: HUCD.mxn
with a multiplier of 1.0, Pressure.mxn with a multiplier of 1.0, CrFPr.mxn with a multiplier of 1.0, and HUCD.mnn with a multiplier of 1.0.
Note that even though the LTOP event is programmed in PROMISE 1.0, it can be ignored in PROMISE runs. LTOP was not used in the EPRI report for the SG components. Therefore, although the LTOP event appears in these input files, it was not considered in the PFM runs.
o Item 2.e.iii - Show the SIF or stress history with the temperature history for the assumed transients to ensure maximum SIF or stress does not occur at low temperatures.
o In the EPRI technical basis report, the licensee assumed a KIC value of 200 ksiin, the value of the ASME Code KIC curve at high temperatures, and explained that this is valid since the temperatures for all applicable transients remains high. The NRC staff noted in Table 5-5 of the EPRI technical basis report that some temperatures in the listed transients are low enough to invalidate the assumption of a KIC value of 200 ksiin.
Therefore, the NRC staff asked for histories of applied SIF and temperature at the locations of interest for Heat-Up/Cooldown, Plant Loading, and Plant Unloading transients. The licensee showed applied SIF and temperature history plots for these transients at both the NVW and NIR locations of the feedwater nozzle (no plots were shown at the NVW location for the main steam nozzle). The licensee confirmed that the applied SIF shown were the total applied SIF due to the appropriate loads.
o For the ramp periods in the beginning and end of the Heat-Up/Cooldown transient at the NVW location, the licensee showed that the applied SIF intersects the KIC curve at approximately 130 ksiin at about 175 degrees Fahrenheit (°F). The applied SIF was for a depth of 80 percent of the thickness and for the maximum length-to-depth ratio. Since 130 ksiin is lower than the assumed KIC value of 200 ksiin, the licensee explained that the fracture toughness sensitivity studies in Section 8.2.4.3.1 of the EPRI technical basis report (results shown in Tables 8-13 and 8-14 of the report), in which a KIC value of as low as 80 ksiin was investigated, address the low-temperature periods at the beginning and end of the Heat-Up/Cooldown transient when the KIC value can drop below 200 ksiin. For the case of KIC = 80 ksiin for case FEW-3PA (the limiting NVW location), the probability of rupture is 3.75x10-8 per year at 80 years as shown in Table 8-13 of the EPRI report, which is below the criterion of 1x10-6 per year. The NRC staff confirmed the probability of failure and probability of leakage values in Table 8-13 and 8-4 with those in the output files for FEW-3PA. For the case of KIC = 80 ksiin, the NRC staff noted in output file FEW-3PA-080.rpt that there were 29 instances in 10 million realizations when the rupture times were less than 80 years (see observation in Items 2.a.iv.1 and 2.a.iv.2), which, adding 1/10 million (see observation in Item 2.d.i), is equivalent to the value of 3.75x10-8 per year in the EPRI report.
o The NRC staff asked if the fracture toughness sensitivities studies in Tables 8-13 and 8-14 of the EPRI technical basis report were run with the same number of realizations, 10 million aleatory realization, and the licensee responded, yes. The NRC staff confirmed the number of realizations in the output files for case FEW-3PA in Tables 8-13 and 8-14 provided in the portal for Items 2.d.i and 3.b. The NRC staff also confirmed the KIC values shown in the column headings of Tables 8-13 and 8-14 in the input and output files provided for FEW-3PA.
o For the Plant Loading and Plant Unloading transients, the licensee presented plots that showed that at the NVW and NIR locations of the feedwater nozzle, the temperatures remained high (~ 500°F or greater) even though the water temperature entering the feedwater nozzle ranged from 70°F to 430°F as shown in Table 5-5 of the EPRI report.
o The NRC staff also asked about the Loss of Power transient in which the feedwater temperature is at a low value of 32°F and the pressure is at a maximum value of 1,120 pounds per square inch gauge (psig) as shown in Table 5-5 of the EPRI report.
Even though the licensee did not show a temperature history plot for this transient, the licensee stated that the temperature comparison between the NVW and NIR locations of the feedwater nozzle and piping end location of the feedwater nozzle would be similar to the comparison for the Plant Loading and Plant Unloading transients.
o Item 2.f.i - Show a plot of failure probability versus number of realizations for a low probability case and a high probability case.
o The licensee showed probability of leakage and rupture plots for 107 and 108 realizations for high and low probability cases. For the low probability case, the probability of leakage for 107 realizations showed slightly larger probabilities than for 108 realizations.
All other cases provided for this item showed essentially the same probabilities for 107 and 108 realizations.
o Item 2.g.i - Show model dependencies between the inputs, if any, considered in the sensitivity studies in Section 8.2.4.3 of the EPRI technical basis report (example: interaction between POD and ISI).
o The licensee stated that the only modeling dependency is the correlation of crack length with crack depth, and that the crack length is a log-normal distribution as explained in Section 8.2.2.2 of the EPRI technical basis report. See also related observation on SIA proprietary document 1700313.301 in Item 2.b.iv.
o Item 2.h.i - Show code execution and input deck for one or two cases presented in the EPRI technical basis report. Clarify which parameters are random and which are constant.
o At the beginning of the audit the licensee demonstrated the use of PROMISE Version 1.0 by running case FEW-P3A in Table 8-9 of the EPRI technical basis report.
Running the case with 10 million realizations took about 20 minutes. When the run was completed, the licensee showed the probability of leakage value of 9.528x10-4 at 80 years; this is equivalent to (9.528x10-4)/80 years = 1.19x10-5 per year, which matches the value in Table 8-9 of the EPRI technical basis report.
o During the demonstration, the licensee showed an input deck and showed the input fields for the flaw distribution, the POD curve, ISI, and percent examination coverage.
o The NRC staff noted that some input parameters can be set as random or constant (i.e.,
not random). For example, transient stresses are set as random. The NRC staff reviewed input file FEW-3PA.vnz3 provided for this item for the Loss of Load transient.
This transient is set as random with a normal distribution as shown in the input block below:
LOL STR_LOL_MAP_P3.CSV 1 4 3 8 0.001 Aleatory Normal 1,0 1,0,2 This is consistent with what the licensee stated during the discussion of the table provided for Item 2.a.iii that transient and weld residual stresses are random. However, the licensee stated during the discussion of Item 2.h.i that stresses are constant.
o Item 3.a - In addition to the benchmark cases shown in Table 8-5 of the EPRI technical basis report, show a benchmark run with VIPERNOZ for the FEW-P3A base case in Table 8-9.
o In SIA letter report 1900064.407.R1, the licensee analyzed the effect of ISI in the benchmark runs with VIPERNOZ. The licensee confirmed that PROMISE Version 1.0 was used for the PFM analyses in the letter report. The NRC staff observed that 1900064.407.R1 showed the results only for probability of leakage and explained that the results for probability of rupture are needed to get a complete picture of the analysis.
The licensee updated the letter report and uploaded it in the portal. The updated letter report, 1900064.407.R2, showed the results for probability of rupture in addition to the results for probability of leakage. Given the importance of the results of the PFM analyses in 1900064.407.R2, the NRC staff requested the letter report to be submitted to the docket.
o Item 3.b - Show underlying calculations and/or calculation packages for the base cases in Tables 8-8 and 8-9 of the EPRI technical basis report.
o The licensee stated that there are no underlying calculations and/or calculation packages for the cases in Tables 8-8 and 8-9, but provided input and output files (see table of documents and files reviewed). The NRC staff observed these files as needed.
o Item 3.c - Show calculation of values in Table 8-26 of the EPRI technical basis report (probability of leak for 80 years) from the plot in Figure 8-12 (probability of leak versus time).
o The NRC staff reviewed the information provided for this item, which showed how the values in Table 8-26 of the EPRI technical basis report for the 0, 10, and 20 ISI case were calculated from the plot in Figure 8-12 of the report.
o Item 3.d - Show underlying calculations and/or calculation packages for the cases in Table 8-28 or Table 8-29 of the EPRI technical basis report.
o The licensee stated that there are no underlying calculations and/or calculation packages for the cases in Tables 8-28 and 8-29, but provided some output files. The NRC staff reviewed output files FEW-P1N-ISI-Table 8-28.rpt and FEW-P1N-ISI-Table 8-29.rpt, calculated the probability of rupture values per year at
80 years based on the values in the files, and compared them with those reported in Tables 8-28 and 8-29 of the EPRI technical basis report for case FEW-P1N. The NRC staffs calculations are one order of magnitude higher than those in Tables 8-28 and 8-29. As a result of this observation, the NRC staff calculated the probability of leakage per year with PSI only from output file FEW-P1N.rpt, and noted that the resulting value is three orders of magnitude higher than the value reported in Table 8-9 for FEW-P1N. A related observation is that the probability of rupture per year for FEW-P3A (1.25x10-9) in Tables 8-28 and 8-29 is three orders of magnitude higher than the value for FEW-P1N (1.25x10-12). The NRC staff asked during the audit whether the differences in values are due to the assumed number of flaws (i.e., flaw density) in the NIR. The licensee responded that yes, the differences are due to the assumed flaw density in the NIR. The assumed flaw density for the results shown in Tables 8-28 and 8-29 is 0.1, while the assumed flaw density in the NIR for the base cases in Tables 8-8 and 8-9 is 0.001.
Thus, the NRC staff observed that the probability values in the output files for the NIR need to be multiplied by the flaw density assumed in the NIR.
The NRC staff review is for the Vogtle, Units 1 and 2, plant-specific alternative proposed by SNC. The NRC staff is not approving the EPRI Report generically.
5.0 DOCUMENTS AND FILES REVIEWED In addition to the discussions of each audit item, the staff reviewed the following documents and files:
Revision Need in Document or File Document Title or File Description and/or Date Docket?
Responses to each audit item in Each response contains the requested information N/A No Section 4.0 of the for the audit item audit plan.
NRC Audit of PROMISE 1.0 Software Relief Request VEGP-ISI-ALT-04-04 Inspection Interval June 29 and Slides for Day 1 Extension for Vogtle Steam Generator Feedwater July 1, 2020 No and Main Steam Nozzle-to-Shell Welds and Nozzle Inside Radius Areas Day 2 - NRC Audit of PROMISE 1.0 Software Relief Request VEGP-ISI-ALT-04-04 (Inspection Slides for Day 2 Interval Extension for Vogtle Steam Generator July 1, 2020 No Feedwater and Main Steam Nozzle-to-Shell Welds and Nozzle Inside Radius Areas)
Continuation of NRC Audit of PROMISE 1.0 Software Relief Request VEGP-ISI-ALT-04-04 Slides for Day 3 Inspection Interval Extension for Vogtle Steam July 27, 2020 No Generator Feedwater and Main Steam Nozzle-to-Shell Welds and Nozzle Inside Radius Areas SIA report Software Project Plan, Software Requirements, Revision 0 DEV1806.401 Functional Specification and Software Verification November No (PROPRIETARY) and Validation Plan 2018
Revision Need in Document or File Document Title or File Description and/or Date Docket?
SIA report Revision 0 DEV1806.402 PROMISE 1.0 Theory and Users Manual November No (PROPRIETARY) 2018 SIA report Revision 0 Software Verification and Validation Report for DEV1806.403 November No PROMISE 1.0 (PROPRIETARY) 2018 SIA calculation Crack Models and Crack Growth Module Revision 0 DEV1806.301 No Verifications 11/14/2018 (PROPRIETARY)
SIA calculation Revision 0 DEV1806.302 Sampling Verification No 11/14/2018 (PROPRIETARY)
SIA calculation Revision 0 DEV1806.303 Probabilistic Verification No 11/14/2018 (PROPRIETARY)
Transactions of Weight Function Based Stress Intensity Factor August 4 - 9, No SMiRT-25 Conference Solution for Nozzle Corner Cracks 2019 SIA calculation Stress Intensity Factors (SIFs) for Nozzle Corner Revision 0 DEV1801.302 No Cracks 3/12/2018 (PROPRIETARY)
SIA calculation 1700313.301 Fabrication Flaw Size Distribution (excerpts only) Revision 0 No (PROPRIETARY)
Subject:
Evaluations to Address Limited SIA letter report Examination Coverage of Vogtle Electric 1900064.406.R0 to Revision 0 Generating Plant Units 1 and 2 Steam Generator Yes address follow-up 7/16/2020 Main Steam and Feedwater Nozzle-to-Shell action for Item 2.c.i Welds and Nozzle Inside Radius Sections FEW-3PA.rpt/.vnz3 FEW-3PA-30.rpt/.vnz3 FEW-3PA- PROMISE Version 1.0 run files provided for audit 12/18/2018 No 080.rpt/.vnz3 Item 2.d.i FEW-3PA-100.rpt/.vnz3 FEW- PROMISE Version 1.0 run files provided for audit 12/18/2018 No 3PA.rpt/.vnz3/.smp.csv Item 2.h.i SIA letter reports 1900064.407.R1 and Revision 1
Subject:
Evaluations to Address Benchmarking of R2 7/19/2020 Yes the PROMISE Software to Include the Effects of to address follow-up Revision 2 (Rev 2)
Inspection action for Item 3.a 8/10/2020 Various files names for the cases shown in 10/27/2018 PROMISE Version 1.0 run files provided for audit Tables 8-8 and 8-9 of to No Item 3.b the EPRI technical 12/18/2028 basis report
Revision Need in Document or File Document Title or File Description and/or Date Docket?
FEW-P1N-ISI-Table 8-12/18/2018 28.rpt PROMISE Version 1.0 run files provided for audit and No FEW-P1N-ISI-Table 8- Item 3.d 11/7/2018 29.rpt
6.0 CONCLUSION
As a result of the audit, the NRC staff has gained a better understanding of how PFM principles were implemented in PROMISE Version 1.0 and of the V&V performed on the software. The audit also assisted the NRC staff in identifying information that needed to be docketed (as identified in Section 5.0 of this report) either through voluntary submittal by SNC or through requests for additional information.
The NRC staff review is for the Vogtle, Units 1 and 2, plant-specific alternative proposed by SNC. The NRC staff is not approving the EPRI Report generically.
ML20258A002 OFFICE NRR/DORL/LPL2-1/PM NRR/DORL/LPL2-1/LA NRR/DNRL/D NAME JLamb KGoldstein ABradford DATE 10/15/2020 10/16/2020 9/10/2020 OFFICE NRR/DORL/LPL2-1/BC NRR/DORL/LPL2-1/PM NAME MMarkley (SWilliams for) JLamb DATE 12/9/2020 12/10/2020