ML15364A073
ML15364A073 | |
Person / Time | |
---|---|
Site: | Salem ![]() |
Issue date: | 01/08/2016 |
From: | Tekia Govan Japan Lessons-Learned Division |
To: | Braun R Public Service Enterprise Group |
Govan, Tekia NRR/JLD 415-6197 | |
References | |
CAC MF3790, CAC MF3791 | |
Download: ML15364A073 (21) | |
Text
UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, O.C. 20555-0001 January 8, 2016 Mr. Robert Braun President and CNO PSEG Nuclear LLC-N09 P. 0. Box 236 Hancocks Bridge, NJ 08038
SUBJECT:
NUCLEAR REGULATORY COMMISSION REPORT FOR THE AUDIT OF PSEG NUCLEAR LLC'S FLOOD HAZARD REEVALUATION REPORT SUBMITTALS RELATING TO THE NEAR-TERM TASK FORCE RECOMMENDATION 2.1-FLOODING FOR SALEM NUCLEAR GENERATING STATION, UNITS 1AND2 (CAC NOS. MF3790 AND MF3791)
Dear Mr. Braun:
By letter dated June 1, 2015 (Agencywide Documents Access and Management System (ADAMS) Accession No. ML15146A220), the U.S. Nuclear Regulatory Commission (NRC) informed you of the staff's plan to conduct a regulatory audit of PSEG Nuclear LLC's (PSEG, the licensee) Flood Hazard Reevaluation Report (FHRR) submittal related to the Near-Term Task Force Recommendation 2.1-Flooding for Salem Nuclear Generating Station, Units 1 and 2 (Salem). The audit was intended to support the NRC staff review of the licensee's FHRR and the subsequent issuance of a staff assessment.
The audits conducted on June 22, 2015, and July 16, 2015 were performed consistent with NRC Office of Nuclear Reactor Regulation, Office Instruction LIC-111, "Regulatory Audits,"
dated December 29, 2008, (ADAMS Accession No. ML082900195). Therefore, the purpose of this letter is to provide you with the final audit report which summarizes and documents the NRC's regulatory audit of the licensee's FHRR submittal. Based on shared site characteristics, this audit was combined with the audit of PSEG's Hope Creek Generating Station, Unit 1 (Hope Creek). The results of this audit report are applicable to both Salem and Hope Creek. The NRC staff has prepared a separate audit report for Hope Creek with the same audit results.
R. Braun If you have any questions, please contact me at (301) 415-6197 or by e-mail at Tekia.Govan@nrc.gov.
Sincerely, Tekia V. Govan, Project Manager Office of Nuclear Reactor Regulation Japan Lessons-Learned Division Hazards Management Branch Docket Nos. 50-272 and 50-311
Enclosure:
Audit Report cc w/encl: Distribution via Listserv
UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, D.C. 20555-0001 NUCLEAR REGULATORY COMMISSION AUDIT REPORT FOR THE AUDIT OF PSEG NUCLEAR LLC'S FLOOD HAZARD REEVALUATION REPORT SUBMITTALS RELATING TO THE NEAR-TERM TASK FORCE RECOMMENDATION 2.1-FLOODING FOR SALEM NUCLEAR GENERATING STATION, UNITS 1 AND 2 BACKGROUND AND AUDIT BASIS By letter dated March 12, 2012, the U.S. Nuclear Regulatory Commission (NRC) issued a request for information to all power reactor licensees and holders of construction permits in active or deferred status, pursuant to Title 10 of the Code of Federal Regulations (10 CFR),
Section 50.54(f) "Conditions of license" (hereafter referred to as the "50.54(f) letter"). The request was issued in connection with implementing lessons-learned from the 2011 accident at the Fukushima Dai-ichi nuclear power plant, as documented in The Near-Term Task Force Review of Insights from the Fukushima Dai-ichi Accident. Recommendation 2.1 in that document recommended that the NRC staff issue orders to all licensees to reevaluate seismic and flooding for their sites against current NRC requirements and guidance. Subsequent Staff Requirements Memoranda associated with Commission Papers SECY 11-0124 and SECY-11-0137, instructed the NRC staff to issue requests for information to licensees pursuant to 10 CFR 50.54(f).
By letter dated March 11, 2014, PSEG Nuclear LLC ( PSEG, the licensee) submitted its Flood Hazard Reevaluation Reports (FHRRs) for Salem Nuclear Generating Station, Units 1 and 2 (Salem) (Agencywide Documents Access and Management System (ADAMS) Accession No. ML14071A401). The NRC is in the process of reviewing the aforementioned submittal and has completed a regulatory audit of the licensee to better understand the development of the submittal, identify any similarities/differences with past work completed and ultimately aid in its review of the licensees' FHRR. This audit summary is being completed in accordance with the guidance set forth in NRC Office of Nuclear Reactor Regulation, Office Instruction LIC-111, "Regulatory Audits," dated December 29, 2008, (ADAMS Accession No. ML082900195).
AUDIT LOCATION AND DATES The audit was completed by document review via a webinar session in conjunction with the use of the licensee's established electronic reading room (ERR) and teleconference on June 22, 2015, from 9am to 3pm and July 16, 2015, from 1:OOpm to 4:00pm.
Enclosure
AUDIT TEAM Title Team Member Organization Team Leader, NRR/JLD Tekia Govan NRC Technical Monitor Michael Willingham NRC Technical Staff Mike Lee NRC Technical Staff Michelle Bensi NRC Technical Team Lead Kenneth Erwin NRC Technical Branch Chief Christopher Cook NRC NRC Contractor Philip Meyer Pacific Northwest National Lab NRC Contractor Christopher Bender Taylor Engineering A list of the Licensee's participants can be found in Attachment 2.
DOCUMENTS AUDITED of this report contains a list which details the documents that were reviewed by the NRC staff, in part or in whole, as part of this audit. The documents were located in an electronic reading room during the NRC staff's review. The documents, or portions thereof, that were used by the NRC staff as part of the technical analysis and/or as reference in the completion of the staff assessment, were submitted by the licensee and docketed for completeness of information, as necessary. These documents are identified in Table 1.
AUDIT ACTIVITIES In general, the audit activities consisted mainly of the following actions:
- Review background information on site topography and geographical characteristics of the watershed.
- Review site physical features and plant layout.
- Understand the selection of important assumptions and parameters that would be the basis for evaluating the individual flood causing mechanisms described in the 50.54(f) letter.
- Review model input/output files to computer analyses such as Delft-3D and FL0-2D to have an understanding of how modeling assumptions were programmed and executed.
- Review of data filtering and statistical approaches utilized to develop storm recurrence rate and distribution of storm parameters used in the Joint Probability Method (JPM).
- Review of key assumptions, approximations, and heuristics utilized in conjunction with the JPM.
Table 1 summarizes specific technical topics (and resolution) of important items that were discussed and clarified during the audit. The items discussed in Table 1 may be referenced/mentioned in the staff assessment in more detail.
EXIT MEETING/BRIEFING On August 14, 2015, the NRC staff closed out the discussion of the technical topics described above. There are no outstanding information needs remaining as a result of this audit.
Table 1: Salem/Hope Creek Information Needs - Audit/Post-Audit Summary INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED 1 Storm Surge - Data used for Development of Distributions In response to the NRC staff information requests:
Evaluation of the effects of flooding from storm surge on water surface a. The licensee described the approach used to geographically filter data elevations at the Salem/Hope Creek site (hereafter the "PSEG site") is using a capture zone approach. The licensee provided a Fortran requested in the 50.54 (f) letter. The flood hazard reevaluation reports computer code (file name "hdatpsegf.for") that was used by the (FHRRs) for Salem and Hope Creek (References [1] and [2]) provide licensee for investigations of tropical cyclone statistics. It was noted by Table 2.4-10 containing storms crossing a line of demarcation running the licensee that the input file for the Fortran computer code was the from a point at 36.5°N,76°N to a point at 41.5°N, 71°W (shown in Figure "latest HURDAT file that was available at the time of the work 2.4-11 ). The table lists values for central pressure, forward velocity, and conducted for PSEG in late summer and fall of 2014" (the file name track angle for each of the storms. The September 2014 request for assigned by the licensee is "newhurdat2013.txt"). The licensee stated additional information (RAI) response (Reference. [3]) contains similar that a line of demarcation was drawn from 75.5 W, 37.0 N to 71.0 W, information in Table RAl-5-2 for storms crossing a line a line extending 41.0 N. Storms that crossed this line were identified by the licensee as from a point at 37°N,76°N to a point at 41°N,?2°W. However, being close enough to the PSEG site to include in the analysis. The differences are noted between the two tables with respect to the storms licensee noted that the coordinates used were slightly different than included in the tables as well as values of parameters assigned to what was provided in the RAI response (Reference [3]); however, the storms appearing in both tables. It is understood that the information in licensee clarified that these coordinates reflected those used in the FHRR Table 2.4-10 is used to develop distributions for central pressure calculation (i.e., the discrepancy was identified as a reporting error differential. The data is Table RAl-5-2 is used to develop distributions rather than a calculation error). The licensee stated that a central for central pressure differential, forward velocity, and storm heading. pressure threshold of 980mb was established in the computer code to In connection with the 50.54 (f) request, the licensee was requested to filter any storms that were "too weak to produce a significant surge provide the following information: response." Using this information, the licensee's execution of the hdatpsegf.for code produces an output file (file name 'table.txt') that
- a. Description of the approach used for screening or filtering data.
contains the filtered storm dataset. The licensee noted that the track
- b. Clarification of the reason(s) for the differences between the dataset angle assigned to storm event Sandy was manually adjusted.
used in the FHRRs (Table 2.4-10) and that used in the September 2014 RAI response (Table RAl-5-2). Specifically, clarification is The licensee further noted that there are several discrepancies requested regarding differences in storms considered within the two between the Table RAl-5-2 (Reference [3]) and the values used in the datasets, as well as differences in parameter values assigned to FHRR analysis. The licensee stated that the values in Table RAl-5-2 storms that appear in both datasets. were incorrect due to an apparent transcription error in generating the
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED
- c. Description of the method for assigning central pressure values to tables. The licensee stated that values used in subsequent storms for which central pressure data is not included in the calculations are as shown in Table 2 of this audit summary.
HURDAT2 dataset.
- b. The licensee stated that there are two "main reasons for the different
- d. Discussion of the treatment of storm events in the analysis passing datasets" observed between FHRR Table 2.4-10 and RAI Table 5-2:
close to but not across the line of demarcation, as well as storm
- i. The "line of demarcation" was modified causing the number of events that are not included the HURDAT2 database (Reference filtered storms to be potentially altered. The licensee noted that
[4]).
the Fortran code was setup to identify the point at which the storm passes the line of demarcation. Once a storm has crossed into the area of interest, the Fortran code will identify the minimum central pressure and associated forward speed and heading (based on the specific point that is chosen as the minimum central pressure). As a result, changes to the line of demarcation would potentially change the number of filtered storms and the specific data values used in the analysis ii. The HURDAT database was significantly modified around the 2011-2012 time period. The licensee stated that a review of the summary of changes (available at:
http://www.aoml.noaa.gov/hrd/hurdat/metadata master.html) indicated that significant changes were performed on the database, including changes to track and intensity information. Modifications to track and intensity information could potentially affect the storms that are selected and the specific data that is extracted from the database.
- c. The licensee stated that only events for which historical information is available for central pressure were used. Wind-pressure relationships were not used.
- d. The licensee stated that events not passing the line of demarcation were not included.
The NRC staff concluded that the information provided by the licensee was responsive to the information need request.
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED 2 Storm Surge - Storm Rate In response to the NRC staff information requests:
- a. The licensee stated that the rate originally used in the FHRR evaluation Evaluation of the effects of flooding from storm surge on water surface (based on a FEMA study) was subsequently considered not elevations at the PSEG site is requested in the 50.54 (f) letter. Section appropriate for the PSEG site.
2.4.3.4.2 of the FHRRs for Salem and Hope Creek (References [1] and
[2]) state the following:
- b. The licensee stated that only storm events that make landfall were "Although the FEMA storm surge study focused on surges for a considered in the event rate calculation. The licensee initially stated much lower range of return periods than those of interest in this that no further filters were utilized but later revised their response to analysis, the objective measure of storm rate within the angle indicate that a filter based on event categorization appears to have range of interest is used for the PSEG site, as the catalog of been used. The licensee stated that Fortran computer code used to historical storm data is the same. In the FEMA storm surge filter historical data identified "the storm events that crossed the area study the omnidirectional storm rate in the vicinity of the mouth of interest in order to define the storm characteristics, but not of Delaware Bay is approximately 0. 045 storms per year per necessarily only the storms that make landfall." The licensee identified degree [of latitude]." eight storms that make landfall within the area of interest: an unnamed It is noted that the above reference to the FEMA study is a draft report 1938 event, Agnes, Donna, Belle, Floyd, Gloria, Irene, and Sandy.
that provides the rate in units of storms/year/degree of latitude. The However, the licensee stated that Agnes and Floyd storm events have published version of the FEMA study provides the rate in units of central pressures less than the 980mb threshold at landfall and were storms/year/km, which (by unit conversion) yields an approximately thus deemed by the licensee to be too weak at landfall to produce a equivalent rate of 0.0004 storms/km/year. The omni-directional storm significant storm surge at the site and were excluded from the storm rate provided in the FHRR is subsequently multiplied by 0.1695% rate analysis.
(0.001695), which the licensee states is the percentage of the total omni-directional population considered to be "heading in the critical 1
- c. Based on information provided in response to Information Need 1 storm track needed to generate the 1o-s AEP storm surges." This yields (above), the NRC staff noted that the 1893, 1903, and 1934 storm 7.628 x 1o-s. events appeared to have been screened based on central pressure criteria. The licensee confirmed that these events were "were too weak In the licensee's September 2014 RAI response (Reference [3]), the within the area of interest and were excluded via the filtering process."
licensee states the following: The licensee further noted the following:
"[l]n this area (the coastal region defined in Figure RAl-5-2)
- The 1893 storm event had an observed central pressure of there have only been 6 landfal/ing storms (Unnamed 1938, 986 mb and an estimated wind speed of 75 mph at landfall Donna, Belle, Gloria, Irene, and Sandy) over the last 163 years. (storm AL041893 on August 24 at 1200 hours0.0139 days <br />0.333 hours <br />0.00198 weeks <br />4.566e-4 months <br /> GMT).
Thus, the frequency of storms in our range of angles is given by
- The 1903 storm event had an observed central pressure of 990 number of landfa/ling storms divided by the number of years mb and its estimated wind speed was 70 mph (storm (61163) divided by the number of degrees along the line defined AL041903 on September 16 at 1100 hours0.0127 days <br />0.306 hours <br />0.00182 weeks <br />4.1855e-4 months <br /> GMT).
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED as the crossing boundary here (5. 93) times the fraction that
- The 1934 storm event had an estimated central pressure of travel in this direction (0.0117), which results in a storm 989 mb and 65 mph wind speed at landfall (storm AL061934 frequency value at the PSEG site of 7. 26 x 1o-s storms per year on September 9 at 0200 hours0.00231 days <br />0.0556 hours <br />3.306878e-4 weeks <br />7.61e-5 months <br />).
per degree."
The licensee additionally stated that storm events occurring in calendar Thus, via the RAI response, the omni-directional rate appears to be years 1867, 1869, and 1879 had recorded central pressures less than reduced from 0.045 storms/year/degree to a value of 0.0062 980 mb but had estimated wind speeds at landfall of 80 mph, 100 mph, storms/year/degree.
and 90 mph, respectively. The licensee concluded that, because these In connection with the 50.54 (f) request, the licensee is requested to: wind speeds are all above the wind speeds for the 1893, 1903 and
- a. Describe the technical basis for modifying from the omni- 1934 storm events at landfall, the omission of these storms in the directional rate provided in the FHRR such that the rate is analysis was justified.
apparently reduced.
The NRC staff further identified another subset of storms (i.e.,
- b. Describe the criteria used to screen events so that only six events Unnamed events in 1869 and 1944 as well as the 1954 event Carol).
that were included in the omni-directional rate calculation. The licensee provided the following explanations for the exclusion of the following storm events:
- c. Describe the reason for exclusion of storm events that appear to have similar characteristics to the six events included in the event
- The September 7-9, 1869, storm (AL061869) data does not list used to calculate that the omni-directional rate (e.g., exclusion have any observation points that fall within the sample area.
of events such as Unnamed 1903, NotNamed 1893, and
- The September 9-16, 1944, storm (AL071944) data has a Unnamed 1934). single point that falls inside the sample area, but its wind speed is 75 mph (minimal hurricane). The event data has no
- d. Discuss whether the distributions for storm parameters are pressure observations inside the sample area.
defined consistent with the criteria used to define the omni-
- The event Carol data has two points which "clip" the sample directional event rate (e.g., whether input data used to develop area, but has no pressure readings inside the sample area.
distributions are consistent with the characteristics of events list The licensee noted that this storm data has a slightly-higher used to define the omni-directional rate). wind speed (85 mph) than the lowest storm in the sample
- e. Discuss whether the omni-directional rate is defined in a manner considered for computation of the recurrence rate (80 mph).
such that it is appropriate to subsequently multiply the calculated However, the licensee further stated the other storm events rate by a factor intended to include only events approaching in the included in the sample had pressure data recorded (unlike the "critical direction" (i.e., discuss whether it is consistent to (i) include event Carol). The licensee stated that there is a large error only landfilling events with limited range of headings in the event band in estimating central pressure from HURDAT wind rate calculation but (ii) include a wider range of events in speeds. The licensee concluded that, because the storm set is only used to specify the parameter distributions and not to estimate storm frequency, it was better to omit the event from
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED calculating the distribution of heading, including events that do not the storm catalog rather than to attempt to assign this weak make landfall). storm a pressure value and include it in the analyses.
- d. The licensee stated that "[t]his assumption is consistent with the state of the art in the field of surge prediction within JPM models." The licensee noted that the Sandy storm was viewed as an outlier with regard to storm heading. The licensee also noted (i) that examination of the correlation between storm heading and central pressures (excluding Hurricane Sandy) yielded a product-moment correlation coefficient of -0.4 and (ii) the t-Statistic is not significant for a two-tailed test at the 0.10 level. The licensee concluded that it would not be considered statistically-correlated in a typical test of significance.
- e. The licensee noted that a storm must make landfall to generate a large surge. However, in order to have a larger distribution for fitting the distributions of associated storm parameters, the licensee extended the sample area. With regard to the frequency with which these storms make landfall in the study area, the licensee concluded that the representation had to be quantified in terms of the frequency of landfalling storms. The licensee further stated that the frequency of landfalling storms with central pressures less than 980 mb can be combined with additional parameters (e.g. central pressure distribution, heading distribution, size and speed distributions) for storms with central pressures less than 980 mb, to form an appropriate estimate of the statistical annual rate of the occurrence of a combination of events (given that the storm makes landfall). The licensee stated that, had the analysis included non-landfalling storms, this would not have been the correct rate for landfalling storms.
The NRC staff concluded that the information provided by the licensee was responsive to the information need request.
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED 3 Storm Surge - Distribution of Rmax In response to the NRC staff information requests:
- a. The licensee stated that the distribution assumptions were changed in Evaluation of the effects of flooding from storm surge on water surface response to an error identified by peer reviewers.
elevations at the PSEG site is requested in the 50.54 (f) letter. The FHRRs for Hope Creek (Reference [1]) and Salem (Reference [2])
- b. During the audit, the NRC staff showed an illustration of a potential state that the following distribution is used for radius to maximum issue in which the computed probabilities for a fixed Rmax are not winds (Rmax) conditional on central pressure differential (llp):
invariant to the units selected for measurement. In response and after p(Rmaxl Cp) = Lognormal (Rmax),O" ln(Rmax)) = <t>(ln(Rmax),O" ln(Rmax)) performing some checks (as part of the current audit process) on the cumulative probability distribution, the licensee found that the latest version of the transformation did not conserve probability. As a result, where: the licensee changed to a kilometer-based Cumulative Distribution Rmax =exp(3.015-6.291 x1Q- 5(llp) 2 + 0.03374')/1.852 Function in the Fortran script prmaxnewf.for to address this problem.
The calculation for the new distribution was then carried out in units of O" ln(Rmax) = 0.44km = 0.24nm kilometers with the category boundaries given by the boundaries in In the above expressions, 4' corresponds to the site latitude in nautical miles converted to kilometers. The licensee stated that the degrees. output from this new code is used in the Joint Probability Method (JPM) code and is consistent with the reference (Reference [5]) cited in The RAI response (Reference [3]) indicates that: O" 1n(Rmax) =0.176nm. Revision 2 of the calculation (Reference [7]).
In a Fortran input file (designated "prmax3.for") associated with the RAI response, the modified standard deviation appears to have been The NRC staff concluded that the information provided by the licensee was calculated as: sufficient to address the information need request.
O" ln(Rmax) = j In (exp (0.44)/1.852) I In connection with the §50.54 request, the licensee is requested to:
- a. Clarify the reason for the change in distribution assumptions.
- b. Confirm that the computed probabilities are not affected by a change in units (e.g., computed probabilities for a fixed Rmaxare invariant to the units selected for measurement).
Storm Surge - Forward Scaling Velocity In response to the NRC staff information requests:
- a. The licensee stated that the expression in FHRR equation 2.4-6 was Evaluation of the effects of flooding from storm surge on water surface not correct but the calculations inside the JPM code are consistent 4 elevations at the PSEG site is requested in the 50.54 (f) letter. The FHRRs (References [1] and [2]) state that a velocity scaling
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED relationship is used in the generation of the response surface, which is with the intended scaling. The licensee stated that equation 2.4-6 based on an approximation in which forward velocity is scaled linearly should be corrected to read as follows:
with wind speed. The relationship provided in the FHRRs ~m = ~o * (1.09)m-mo (equation 2.4-6) is: where:
- 77 = surge in the mth forward speed 7J =770 + A. ovr
- 770 =surge in the reference forward speed (30 knots) where:
- m =the forward speed category counter (m=1 for 10 knots, m=2
- 77 = surge with forward velocity vr for 20 knots, m=3 for 30 knots, m=4 for 40 knots).
- 770 = surge with reference forward velocity of 30 knots The above expression yields multipliers for categories, 1, 2, 3, and 4
- A. =1.09 (a constant value) of 1.09-2 , 1.09-1, 1.09°, and 1.09+1 , respectively. The licensee stated that this is consistent with the multipliers 0.84, 0.92, 1.0, and 1.09
- ovr = difference between the storm forward velocity and a found in the JPM code.
reference storm forward velocity of 30knots
- b. The licensee performed a sensitivity study to identify whether the The input file Qpmzp2.for) associated with the September 2014 RAI response appears to define scaling factors (vfscla(i)) equal to 0.84, 0.92, 1.0, and 1.09. In application of the scaling factor after addition of the error term is addition, the input file appears to apply the scaling factor after the addition of inappropriately affecting the results. The licensee stated that the the error term (e.g., see line of input file containing expression difference in the estimated value 1 E-6 surge level was less than 0.01 isrg=vfscl*(srg(idx,irp,idp,iang)+tideadj+xdif)*10+1.0)). meter.
In connection with the 50.54 (f) request, the licensee is requested to:
- c. The licensee indicated that assumptions associated with the error
- a. Clarify whether the computed scaling factors in the input file term (i.e., standard deviation) in the JPM integration may implicitly Upmzp2.for) are consistent with the expression provided in cover error associated with this approximation. However, the FHRR equation 2.4-6. treatment is not explicit or quantified. The licensee and NRC staff
- b. Provide clarifying information if the above characterization is discussed this topic as potentially appropriate for a sensitivity study not accurate or describe the reason for the apparent (e.g., by increasing the standard deviation associated with the error application of the scaling factor after the addition to the error term). No further action was requested by the licensee as part of the term. audit.
- c. Discuss whether the use of the velocity scaling relationship The NRC staff concluded that the information provided by the licensee was introduces a source of error that should be included in the responsive to the information need request.
JPM integration.
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED 5 Storm Surge - Central Pressure Distribution In response to the NRC staff information requests:
- a. The licensee indicated that the method of moments (iterative solution)
Evaluation of the effects of flooding from storm surge on water surface was used to estimate distribution parameters. Three tests with the JPM elevations at the PSEG site is requested in the 50.54 letter. The code were performed by the licensee with various solutions to the FHRRs (References [1] and [2]) state that the distribution of central parameters of the distribution generated in three different manners (two pressure differential is a Gumbel distribution given by the expression:
self-developed codes and one commercial code) yielding the following:
F(z)=exp(exp(-z)) i. Iterative Method of Moments (based on licensee-developed code) where c = 0.159, a1=46.17, and a2 = 14.49 where z = I (/lp -ao)la1 I , ao = 36.68, and a1 = 14.67.
ii. EasyFit (commercial software package) with Maximum The FHRR states that the distribution was developed using input data Likelihood Method (with default settings) where c = 0.05095, that was screened from the HURDAT dataset using a line-crossing a1=49.474, and a2 = 11.931 approach. The post-screening dataset is provided in Table 2.4-10 of iii. Maximum Likelihood Method where c = 0.088, a1 = 49.64, the FHRRs. In the September 2014 RAI response (Reference [3]), the and a2 = 13.46.
licensee utilized a modified set of storms (Table RAl-5-2) and stated The licensee stated that the resulting stillwater values for the 1E-6 that the data exhibited significant departure from the Gumbel annual exceedance probability are 6.3m, 6.5m, and 6.5m, respectively.
distribution. As a result, a Generalized Extreme Value distribution for central pressure was developed: b. The licensee indicated that goodness of fit was checked via a graphic F(z) = exp(-(1-z/ff) assessment of the data (e.g., Q-Q plot). The licensee observed that the fitted distribution overestimates the highest value in the set, but where z = I (/lp -ao)la1 I, ao = 42.96, a1 = 16.77, and ( = 6.494. overall fits "the data quite well and cannot be rejected even at the 0.2 In connection with the 50.54 (f) request, the licensee is requested to: level of significance in the Kolmogorov-Smirnov test, the Anderson-Darling test and the Chi-Squared test."
- a. Describe the statistical approach used to select distributional form and estimate distribution parameters based on historical data. c. The licensee indicated the distribution form was changed because the
- b. Identify whether the distribution selected and associated previously selected distribution (Gumbel) was subsequently parameters are sensitive to the statistical approach utilized. determined not to adequately fit the data.
- c. Describe the reason for the difference between the distribution The NRC staff concluded that the information provided by the licensee was form and parameters identified in the FHRR and the RAI responsive to the information need request.
response.
6 Storm Surge - Distribution for Forward Velocity and Storm Heading In response to the NRC staff information requests:
- a. For parameters other than central pressure (see Information Need Evaluation of the effects of flooding from storm surge on water surface 5 above), the licensee stated that the small data set was not elevations at the PSEG site is requested in the 50.54 (f) letter. The
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED FHRR indicates that the distribution for forward velocity (vr) is based on considered adequate to justify deviation from previous works with previous FEMA studies and is a normal distribution (defined conditional larger data sets, which found heading to be reasonably on /:::,.p) with parametersµ= 6 + 0.4/:::,.p (knots) and a= 7 knots. The RAI represented by a Gaussian distribution. The licensee stated that response specifies that the distribution of forward velocity is developed introduction of skewness into this distribution would cause based on an analysis of data contained in Table RAl-5-2 and results in
- sensitivity to outliers such as the heading of Hurricane Sandy. As use of a normal distribution with µ = 23.6 knots and a= 9.6 knots. a result, the licensee opted to use the generalized form that is used in JPM studies with distributions based on observational data to The FHRR further indicates that the distribution of heading is based on date. The licensee emphasized the importance consistency a modification of a FEMA distribution and results in the use of a Normal between the analysis of the storm heading directions and their distribution with µ = 4 degrees east of north and a = 10 degrees. The application in the simulations performed. For storms that did not RAI response specifies that the distribution of forward velocity is make landfall, the licensee selected the heading at the time of developed based on an analysis of data contained in Table RAl-5-2 and minimum pressure. However, the licensee noted that, in the results in use of a normal distribution with µ = 70.9 degrees in a landfalling storms, it is important to form smooth tracks that are mathematical coordinate system (19.1 degrees east-of-north) and a=
similar in overall manner in which they are used in the simulations, 21.1 degrees.
which usually meant that the direction at landfall was used. For In connection with the 50.54 (f) request, the licensee is requested to: Hurricane Sandy, the licensee noted that the storm heading veered to the left about 120 miles off the coast (heading 148 degrees in a
- a. Describe the statistical approach and statistical tests used to mathematical coordinate system) and, about 35 miles off the coast, select distributional form and estimate distribution parameters it veered back to the north (heading 127.5 degrees in a based on historical data.
mathematical coordinate system).
- b. Identify whether the distribution selected and associated parameters are sensitive to the statistical approach utilized. In the RAI response (Reference [3]), the licensee used the landfall heading for this storm; however, the licensee stated that it is likely that the track location only reflects a "wobble" in the eye position within this large storm rather than an actual shift in the position of the entire large storm. For the purpose of maintaining an analogue to the simulations, the licensee opted to use a smoothed track with a heading of 137. 75. Use of this data point resulted in a re-estimation of the sample distribution parameters, yielding a mean heading of 71.59 degrees and a standard deviation of 22.43 degrees. The licensee observed that 1. 78 percent of the storms making landfall enter the simulation direction banks included in the simulations.
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED
- b. The licensee stated that they tried to "limit the storm set to move away from the conundrum created in the data analysis performed by FEMA in which storms with heading along tracks considered within our region would have been far less frequent." The licensee noted that an existing FEMA study uses a mean direction of 22 degrees and a standard deviation of 10 degrees.
The NRG staff concluded that the information provided by the licensee was responsive to the information need request.
7 Storm Surge - Pressure Differential Scaling In response to the NRG staff information requests, the licensee provided the following information:
Evaluation of the effects of flooding from storm surge on water surface
- a. The licensee provided a literature reference (Reference [6]).
elevations at the PSEG site is requested in the 50.54 (f) letter. The FHRR (Reference. [1 ], Reference [2]) states that a scaling function was 1 The licensee stated that NRG previously observed that the pressure used (in addition to interpolation and extrapolation) in establishing the b.
differential scaling relationship appeared to underestimate the surges response surface. The scaling relationship relates surge heights for for the 928-mb storms. In response, the licensee compensated for the events with differing central pressure differential but the same values of under predication by replacing the original scaling relationship:
other storm parameters (e.g., Rmax, vr, heading, landfall location, Holland B). The scaling relationship is specified as: ri2 = (!:::..p2 I !:::..p1)
- ri1 with the following (alternative) relationship:
ri2 =(!:::..p2 I !:::..p1)
- ri1 r]2 = (!:::..p2/ !:::..p1)
- ri1 * [1 - (k- ko)(0.046)]
where:
where:
- ri1 = surge associated with storm 1 (e.g., as estimated by a numerical model) l/1 = is the surge height for a storm with a pressure differential of
- ri2 = surge associated with storm 2 !:::..p1 ri2 = is the surge height for a storm with a pressure differential of
- !:::..p1 = pressure differential of storm 1
!:::..p2
- !:::..p2 = pressure differential of storm 2 ko =the counter for the reference pressure (918mb)
The NRG staff observed that comparison between values computed k =the counter for the alternate central pressure being estimated.
using the pressure differential relationship and those computed using numerical models suggest some variability and that the use of the The licensee noted that, to retain some conservatism in the estimates, relationship appears, in some cases, to be biased unconservative. As the scaling in the larger pressure differential direction to its initial value a result, the licensee was requested to discuss: is used. The licensee provided the following multipliers as a function of central pressure:
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED
- a. The applicability of the pressure-differential scaling for estuary Central 948 938 928 918 I 908 environments. pressure
- b. The potential for error (including bias) arising from the use of the 0.7966 I 0.8736 I 0.9414 I 1.00 I 1.10 Multiplier scaling function.
- c. Whether the use of the pressure differential scaling relationship introduces a source of error that should be included in the JPM c. The licensee and staff discussed this topic as potentially appropriate integration. for application of a sensitivity study (e.g., by increasing the standard deviation associated with the error term).
The NRC staff concluded that the information provided by the licensee was responsive to the information need request.
8 Storm Surge - Approximations and Heuristics in Computation Codes In response to the NRC staff information requests, the licensee noted that Evaluation of the effects of flooding from storm surge on water surface an observed rounding issue is expected to have a maximum impact of 0.05 elevations at the PSEG site is requested in the 50.54 (f) letter. The meters. The licensee further stated that the definition of the exceedance Fortran input file Upmzp2.for) associated with the September 2014 RAI value at each increment is estimated with an accuracy equal to the response (Reference [3]) appears to utilize some approximations or accuracy of the computer, since (i) for any value of surge greater than a heuristics (e.g., indexing structure used to compute bins probabilities for development of the hazard curve) that may have implications for the given integer value in tenths of a meter it will get added to the next category final computed results (e.g., due to rounding). without any rounding and (ii) for any value smaller than the integer, it will get placed in the category below the integer value in tenths of a meter, In connection with the 50.54 (f) request, the licensee is requested to where it belongs, with no rounding.
discuss the coding structure utilized in the Fortran input file jpmzp2.for associated with the September 2014 RAI response and the potential The licensee also stated that a heuristic approximation was used when implications associated with any approximations or heuristics utilized.
computing an uncertainty component of the storm surge analysis that In addition, the licensee is requested to be prepared (if assumed that the uncertainty could be written as a constant of 0.2 times needed/requested) to provide output for intermediate results generated the still water level. The licensee stated that subsequent studies have by the code.
shown that this is a poor approximation for the Salem-Hope Creek site because the ratio of the uncertainty to the pressure differential only reaches a value of 0.2 when the pressure differential exceeds 105 mb (i.e. the central pressure is less than 913 mb). The licensee stated that this was a very conservative approximation. Consequently, the value was replaced with an alternate approximation.
The NRC staff concluded that the information provided by the licensee was responsive to the information need request.
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED 9 Upon further review, the NRC staff determined that current staff guidance Local Intense Precipitation (LIP) - LIP 1-hr Front-Loaded Rainfall on the estimation of LIP duration is ambiguous and subject to multiple, Distributions equally valid interpretations. As a consequence, the NRC staff decided to Evaluation of the effects of flooding from LIP on water surface elevations withdraw this Information Need Request.
at the PSEG site is requested in the 50.54 (f) letter. The licensee quoted NUREG/CR-7046's conclusion that, "... local intense precipitation is, The NRC staff concluded that the information provided by the licensee was therefore, deemed equivalent to the 1-hr, 2.56-km 2 (1-mi 2 ) PMP at the responsive to the information need request.
location of the site .... " The licensee also stated that the PSEG site is bounded by the 1-mi 2 area, and that the National Weather Service Hydrologic Monitoring Report (HMR)-52 and NUREG/CR-7046 procedures were used to develop the rainfall event. The licensee stated that a front-loaded distribution was used because it results in maximum flood depths early in the event and minimizes the response time, which, the licensee stated, is conservative because once the water-tight doors are closed, the flooding from the LIP event is nonconsequential. The licensee also stated that, "procedural changes have been made to close the doors well in advance of a LIP event."
The NRC staff concurs that the front-loaded, 1-hr probable maximum precipitation (PMP) event used by the licensee minimizes the response time needed to prevent incursion of water at critical door locations, and thus provides a bounding case for warning time. However, the licensee's RAI response does not address the NRC staff's request for justification that the LIP analysis presented in the FHRR is bounding in terms of the flood depth and flood duration. The RAI stated that justification can include sensitivity analysis to localized PMP events up to 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> in duration. The licensee's response included a statement from NUREG/CR-7046, Section 3.2, that, " ... the amount of extreme precipitation decreases with increasing duration .... " This statement must have been intended to refer to the intensity of extreme precipitation, and not the depth of precipitation. The PMP depths provided in HMR 51 increase with duration (compare HMR 51, Figures 18 to 22). Longer duration LIP events can be constructed from the information in HMR 51 and HMR 52; NUREG/CR-7046, Appendix B, provides an example of a 6-hr LIP event. The NRC staff therefore requests that the licensee provide a revised response to the RAI.
INFO INFORMATION NEED DESCRIPTION ACTION (POST-AUDIT)
NEED 10 PMF Snowpack and Snowmelt Evaluation In response to the NRG staff information requests, the licensee stated that ANS 2.8 requires consideration of snowmelt only when it contributes to the Evaluation of the effects of flooding from LIP on water surface elevations controlling flood. At this particular site, the licensee stated that storm surge at the PSEG site is requested in the 50.54 (f) letter. The licensee stated is the controlling flood event and is thus a more significant flooding event that the potential for flooding of the site arises from a combined event than the PMF event. At the PSEG site, the licensee observed that snow with storm surge and tides being the main contributors, and stated that melt would occur only in the Spring and that storm surge-related flooding snowmelt is not a consideration in the combined event alternatives would occur later in the year- typically in the summer/fall timeframe so that described in ANSl/ANS-2.8. The licensee also stated that major there is little physical potential for a combined event involving these two snowmelt events would occur in the winter or early spring, while storm processes.
surge events would occur in the summer and fall. Finally, the licensee stated that PMP-based flooding of the Delaware River does not have a The NRG staff decided that no additional information would be requested major effect on water levels at the site, and that current procedures of the licensee and that the NRG staff would address the issue in the staff require that the water-tight doors providing flood protection be closed assessment.
when river elevation is well below the nominal site grade.
The RAI addressed the consideration of snowmelt in evaluating the probable maximum flood (PMF), not as a component of the combined events analysis. In ANS 2.8, Section 5.3, directs the consideration of (a) probable maximum precipitation on snow, and (b) probable maximum snowpack with rain. The FHRR does not discuss either of these conditions in evaluating the PMF on the Delaware River. The NRG staff therefore request that the licensee provide a revised response to the RAI.
Info Need No. 10 noted that ANS 2.8, Section 5.3, directs the consideration of (i) probable maximum precipitation on snow, and (ii) probable maximum snowpack with rain. The FHRR does not discuss either of these conditions in evaluating the PMF on the Delaware River. The NRG staff requested that the licensee provide a revised response to the RAI addressing this issue.
16 Table 2: Selected storms from the HURDAT database YEAR STORM STORM CENTRAL HEADING FORWARD NUMBER NAME PRESSURE (deg) SPEED (mb) (knots) 1867 AL021867 Unnamed 969 61.9 16.3 1869 AL061869 Unnamed 950 79.6 38.4 1879 AL021879 Unnamed 979 63.4 26.9 1936 AL131936 Unnamed 968 90.0 13.0 1938 AL061938 Unnamed 940 87.2 14.0 1958 AL041958 Daisy 970 58.6 20.0 1960 AL051960 Donna 965 47.6 30.0 1972 AL021972 Agnes 977 47.3 20.1 1976 AL071976 Belle 977 79.7 22.2 1985 AL091985 Gloria 951 71.0 30.0 1991 AL031991 Bob 953 58.0 26.7 1999 AL081999 Floyd 974 56.9 29.4 2011 AL092011 Irene 958 63.4 3.1 2012 AL182012 Sandy 940 137.75 1.3
17 ATTACHMENT 1 Salem/Hope Creek Audit Document List
- 1. PSEG Nuclear LLC, PSEG Nuclear LLC's Response to Request for Information Regarding Flooding Aspects of Recommendation 2.1 of the Near-Term Task Force Review of Insights from the Fukushima Dai-ichi Accident- Hope Creek Generating Station Flood Hazard Reevaluation, dated March 12, 2014, ADAMS Accession No. ML14071A511.
- 2. PSEG Nuclear LLC, PSEG Nuclear LLC's Response to Request for Information Regarding Flooding Aspects of Recommendation 2.1 of the Near-Term Task Force Review of Insights from the Fukushima Dai-ichi Accident- Salem Generating Station Flood Hazard Reevaluation, dated March 11, 2014, ADAMS Accession No. ML14071A401.
- 3. PSEG Nuclear LLC, PSEG Nuclear LLC's 90-day Response to Request for Additional Information Regarding Flooding Aspects of Recommendation 2.1 of the Near-Term Task Force Review of Insights from the Fukushima Dai-ichi Accident, dated September 23, 2014, ADAMS Accession No. ML14268A469.
- 4. National Oceanic and Atmospheric Administration (NOAA), "Hurricane Research Division Re-Analysis Project (HURDAT2)," [Online]. Available: http://www.aoml.noaa.gov/hrd/hurdat/Data Storm.html
- 5. D. W. Peter J. Vickery, "Statistical Models of Holland Pressure Profile Parameter and Radius to Maximum Winds of Hurricanes from Flight-Level Pressure and H*Wind Data," Journal of Applied Meteorology and Climatology, vol. 47, pp. 2497-2517, 2008.
- 6. N. Taylor, J. Irish, I. Udoh, M. Bilskie, & S. Hagen, "Development and Uncertainty Quantification Of Hurricane Surge Response Functions for Hazard Assessment in Coastal Bays". Natural Hazards, pp.1-21,2015.
- 7. PSEG Nuclear LLC & PSEG Power LLC, Calculation Number 2013-10436: Probable Maximum Storm Surge Using Joint Probability Method for ESPA Site, Revision 2, 2014.
- 8. PSEG Nuclear LLC & PSEG Power LLC, Calculation Number 2013-10436: Probable Maximum Storm Surge Using Joint Probability Method for ESPA Site, Revision 1, 2014.
- 9. PSEG Nuclear LLC & PSEG Power LLC, Calculation Number 2013-1.0436: Probable Maximum Storm Surge Using Joint Probability Method for ESPA Site, Revision 0, 2013.
18 ATTACHMENT 2 List of Salem Audit Participants Name Organization
- 2. Bob Henriksen PSEG
- 3. Tim Devik PSEG
- 4. Charlotte Geiger PSEG
- 5. William McTigue PSEG
- 6. Don Resio University of North Florida/PSEG Contractor
- 7. Mike Salisbury Atkins Global/PSEG Contractor
- 8. Dan Blount Sargent Lundy
- 9. Mehrdad Salehi Sargent Lundy
R. Braun If you have any questions, please contact me at (301) 415-6197 or by e-mail at Tekia.Govan@nrc.gov.
Sincerely, IRA/
Tekia V. Govan, Project Manager Office of Nuclear Reactor Regulation Japan Lessons-Learned Division Hazards Management Branch Docket Nos. 50-272 and 50-311
Enclosure:
Audit Report cc w/encl: Distribution via Listserv DISTRIBUTION:
PUBLIC JLD R/F RidsNRRJLD Resource JBowen, NRR ACampbell, NRO MShams, NRR ARivera, NRO MFranovich, NRR KErwin, NRO RidsNrrDorllpl 1-2Resource CCook, NRO MLee, NRO RRivera-Lugo, NRO KQuinlan, NRO RidsNrrPMSalem Resource ADAMS Accession No.: ML15364A073 *via email
!OFFICE I NRR/JLD/JHMB/PM I NRR/JLD/JHMB/LA I NRO/DSEA/RHM1!TR INRO/DSEA/RHM1fTRI INAME jTGovan jsLent jMLee* IMBensi* I IDATE I 0110512016 I 12/31/2015 11212212015 11212212015 I OFFICE NRO/DSEA/RHM1/BC NRR/JLD/JHMB/BC NRR/JLD/JHMB/PM MShams NAME CCook* TGovan (MMarshall for)
DATE 01/04/2016 01/04/2016 01/08/2016 OFFICAL RECORD COPY