ML18057B363
| ML18057B363 | |
| Person / Time | |
|---|---|
| Site: | Palisades  |
| Issue date: | 10/24/1991 |
| From: | Brian Holian Office of Nuclear Reactor Regulation |
| To: | Office of Nuclear Reactor Regulation |
| References | |
| TAC-M75059, NUDOCS 9111140129 | |
| Download: ML18057B363 (14) | |
Text
- ..I
~....
UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, D. C. 20555 October 24, 1991 Docket No.
50-255 LICENSEE:
CONSUMERS POWER COMPANY FACILITY:
PALISADES POWER
SUBJECT:
SUMMARY
OF MEETING WITH CONSUMERS POWER COMPANY (CPCo)
PALISADES PLANT - INCORE ANALYSIS (TAC NO. 75059)
I"\\
On September 23, 1991 NRC staff from Reactor Projects and Reactor Systems met with Palisades Reactor Engineering personnel to discuss the new incore analysis methodology Palisades has proposed for cycle 10 (beginning in March 1992). A list of meeting attendees is included as Enclosure 1.
This new incore analysis methodology had previously been submitted in support of cycle 9 (the current cycle); however, questions regarding the new code and its associated uncertainty analysis resulted in Palisades committing to supplement their TS application, and "run" the current cycle with both codes.
Palisades is requesting to convert from its currently authorized one-eighth core symmetry program (INCA, supplied by Combustion En~ineering) to a Palisades Incore Detector Algorithm quarter-core symmetry model (PIDAL).
Palisades representatives responded to additional questions from the staff as documented in Enclosure 2.
These questions had been previously issued to the utility in preparation for the meeting, and covered a variety of issues, concentrating on the controls that would be used in conjunction with the new code.
Of particular note, the NRC staff recognizes that Palisades does not currently.plan to re-use rhodium detectors, thereby eliminating this uncertainty component.
The utility is also planning on updating its incore uncertainty analysis each cycle based on reviews of previous cycle data (until sufficient data warrants discontinuation).
The final topic of discussion centered on the revised uncertaintie~ that CPCo is requesting be used with this new code.
CPCo proposes reducing the uncertainties for Fr(delta h) and Linear Heat Rate (from 5% to 4.55%, and 10%
to 6.23%, respectively). The NRC staff requested clarification on the potential impact to plant operations if the uncertainties valves were not relaxed.
CPCo responded that if the older INCA uncertainties are applied to the new PIDAL methodology, the Fr(delta h) calculation may limit reactor power to 98.5% power for a few weeks at the beginning and end of cycle 10.
CPCo has stated that there is a large margin available with Linear Heat Rate, and rl/luWt operation this cycle would not be restricted with the more conservative uncertainty value.
9111140129 911024' PDR
'ADOCK 05000255 p
.j
'\\
J
_i --
The NRC staff is concerned that Palisades is 11 chipping away 11 at historical uncertainty values tha.t h*ave provided for satisfactory margins in measuring reactor variables, withqut providing sufficient data to ensure that reactor Technical Specific~tion Limiting Conditions are accurately measured and maintained.
The NRC *staff is currently _reviewing the uncertainty analyses and is evaluating CPCo's requ~st.in.accorqa.nce with standard uncertainty values previously allowed l:iy the.staff.
Enclosures:
As stated cc w/~n~losures:
See next page DISTRIBUTION See attached page OFC
- PM:PDlll-1
-* Odgtnai 'sig.hed. By-:*'*
.~
...... \\
' -~
- .Brian Holian~ )roje,ct Manager
- ,, *, Project Directorate III-1 Div1sion of Reactor Projects III/IV/V Offi.ce o_f-Nudear Reactor Regulation
~ _,
'/ '
NAME--~sH~ii~~~;~------~EK;~ci;i~k-~~~LM~;;h--------=--------------:--------------
:---------~---:---~----------=-------------- --------------:--------------
DATE
- ;o/-z\\/91
. : 1~/ iy91
- tO/i.'¥91 OFFICIAL RECORD COPY Document Name:
PALISADES MEETING
SUMMARY
Mr. Gerald B. Slade Consumers Power Company cc:
M. I. Miller, Esquire Sidley & Austin 54th Floor One First National Plaza Chicago, Illinois 60603 Mr. Thomas A. McNish, Secretary Consumers Power Company.
212 West Michigan Avenue Jackson, Michigan 49201 Judd L. Bacon, Esquire 212 West Michigan Avenue Jackson, Michigan 49201 Regional Administrator, Region III U.S. Nuclear Regulatory Commission 799 Roosevelt Road Glen Ellyn, Illinois 60137 Jerry Sarno Township Supervisor Covert Township 36197 M-140 Highway Covert, Michigan 49043 Office of the Governor Room 1 - Capitol Building Lansing, Michigan 48913 Mr. Patrick M. Donnelly Director, Safety and Licensing Palisades Plant 27780 Blue Star Memorial Hwy.
Covert, Michigan 49043 Resident Inspector c/o U.S. Nuc*lear Regulatory Commission Palisades Plant 27782 Blue Star Memorial Hwy.
Covert, Michigan 49043 Palisades Plant Nuclear Facilities and Environmental Monitoring Section Office Division of Radiological Health Department of Public Health 3423 N. Logan Street P. 0. Box 30195 Lansing, *Michigan 30195 Gerald Charnoff, P.C.
Shaw, Pittman, Potts &
Trowbridge 2300 N. Street, N.W.
Washington, D.C. 20037 Mr. David L. Brannen Vice President Palisades Generating Company c/o Bechtel Power Corporation 15740 Shady Grove Road Gaithersburg, Maryland 20877 Roy W. Jones Manager, Strategic Program Development Westinghouse Electric Corporation 4350 Northern Pike Monroeville, Pennsylvania 15146
NAME Brian Holian G. A. Schwenk E. D. Kenrick Thomas E. Hollowell Gregory Bau st i an Dick Smedley MEETING ATTENDEES September 23, 1991 CONSUMERS POWER AND NRC AFFILIATION NRR/PDII I-1 NRR/PMSB NRR/SRXB CPCo/Palisades CPCo/Palisades CPCo/Palisades PHONE 492-1344 492-0814 492-0891 616-764-8913 Ex 0593 616-764-8913 Ex 0460 616-764-8913 Ex 0973
9 Enclosur:e 2
CPCo Respons o NRC PIDAL Related Comments/Questions Dated 9/11/1991 PIDAL Methodology/Software Specification Comment 1 - The number of references to octant locations in the software specifications imply that substantial coding changes may be required to implement quarter-core modeling. Please comment on QC-specific coding revisions, if any, that may be necessary.
Response - As part of the reload design process for the current fuel cycle 9, our fuel vendor Siemens (formerly ANF) revised their methodology so as to provide CPCo with PDQ based quarter-core W-primes (detector signal-to-power conversion factors) instead of the previous PDQ based octant-core W-primes. During the PIDAL conversion process from EOC 8 to BOC 9, all PIDAL octant-core W-prime related coding was converted to quarter-core. As part of this effort, all of the additional octant-core dependent coding was updated to quarter-core coding. For convenience, some of the octant-core edits were retained since cycle 9 was loaded as an eighth-core symmetric core. In these instances, analogous quarter-core edits are available as a user option.
The impact of conversion from octant to quarter-core W-primes on the existing PIDAL Uncertainty Analysis was quantified as documented by Palisades Engineering Analysis EA *QAB*9o*os. "Cycle 8 PIDAL Uncertainty Analysis. 1/8th Core and 1/4th Core W-primes". The purpose of this analysis was to ascertain the accuracy of the new 1/4th core CASMO/PDQ W-prime based PIDAL methodology. The goal was to show that the power distribution uncertainties associated with the new model were equal to, or lower than those determined previously for the 1/8th core XPOSE/PDQ W-prime based PIDAL model.
It was found that for all three radial power distribution uncertainty components the uncertainties associated with the new 1/4th core W-primes are lower than or essentially equivalent to the corresponding values for the 1/8th core W-prime PIDAL model. The numerical results are summarized in Table # 1. Therefore, it was concluded that the new model is bounded by the uncertainty analysis previously submitted to the NRC for approval.
Table #1 - PIDAL Cycle 8 Uncertainty Analysis Summary Uncertainty Cycle 5,6,7 Cycle 8 Value Cycle 8 Value Cycle 8 Value Component Value 1/8th Core 1/4th Core 1/4th Core Entire Data XPOSE XPOSE CASMO Base W-primes W-primes W-primes SF(s) 0.0277 0.0258 0.0246 0.0254 SF<sa) 0.0194 0.0139 0.0124 0.0140 SF(r) 0.0022 0.0020 0.0020 0.0020 1
Comment 2 - With respect to coding changes, describe the software configuration-control quality assurance procedures and the validation and verification testing procedures employed to ensure correct design and implementation.
Response - The working level procedure governing the software quality assurance and use of the 'PIDAL code is Palisades Nuclear Plant Software Quality Assurance Plan Procedure SOAP-029. Revision 2 "lncore Analysis Systems". This software quality assurance plan describes the controls required for the maintenance, documentation and use of the systems of computer programs used to calculate the Palisades Plant reactor power distributions. This software quality assurance plan covers both the mainstream INCA and PIDAL systems. Auxiliary programs used to update, correct, back up, retrieve and plot data associated with the INCA and PIDAL systems are considered part of these systems. This procedure satisfies CPCo and Palisades software quality assurance procedures and follows the guidelines put forth in IEEE Standard 730. "Control of Computer Software".
The PIDAL software quality assurance plan, Palisades Nuclear Plant Software Quality Assurance Plan Procedure SQAP-029. Revision 2 "lncore Analysis Systems",
requires that a Test Plan and Test Analysis report must be written for any system change and that these be included as part of the Engineering Analysis associated with the change. The Test Plan and Test Analysis serve as the verification and validation of PIDAL. This procedure requires that the Test Plan be devised to test the sections of the program that were changed and to ensure that unchanged portions of the system still function correctly.
Comment 3 - In the software spec and other submittal documents the terms theoretical/predicted/ calculated appear to be generally equivalent as do measured/inferred/ derived. Please clarify any uniquely separate definitions.
Response - The terms are equivalent as noted, however some clarification between measured and inferred may be warranted. The term "measured" when referring to the power distribution, can refer to the assemblies which contain detectors as opposed to assemblies which do not in which case the power distribution is "inferred" based on the theoretical/predicted/ calculated powers as well as the "measured" powers in neighboring assemblies. The term "measured" can also refer to the overall final PIDAL power distribution which encompasses both the "measured" and "inferred" assembly powers.
Comment 4 - How is calculational convergence verified and what is the procedure*
followed for non-convergent cases.
Response - Convergence is checked automatically by the coding for each of the routines which utilize numerical methods (i.e. matrix solution routines, simultaneous equation solution routines and curve fit routines) as part of their algorithm. Appropriate error messages are produced automatically by the program which in most cases result in an abnormal early termination of the run sequence. The user would be quite aware that a 2
f
- problem existed an9ould follow the problem reporting. corrective action guidelines established by Palisades Nuclear Plant Software Quality Assurance Plan Procedure SOAP-029. Revision 2. "Incore Analysis Svstems".
3
PIDAL Uncertainty Analysis Comment 5 - Are standard statistical techniques used to determine the independence of variables and the normality of distributions that are assumed in the analysis.
Response - Following the statistical model provided by Siemens (ANF) in XN-NF 0l(P). "Exxon Nuclear Analysis of Power Distribution Measurement Uncertainty for St.
Lucie Unit l", histograms for each of the different uncertainty component variables were constructed. For each component or variable the resultant distributions were normal based on the plotted histograms.
Comment 6 - In determining the allowable number of inoperable detectors, in addition to a limit on individual detector failures, what consideration is given to failures by string, by level and by quadrant.
Response - In addition to the requirement that 75% of the detectors be operable, the previous requirement of at least two detectors per level per quadrant has been retained.
In evaluating the uncertainties with failed detectors the conservative approach was taken in that 11 whole strings of 5 detectors were failed randomly. This leaves "holes" in the core where no measurement of power at any axial position is available. The alternate scheme would be to fail 25% of the individual detectors randomly. Also, five different sets of eleven failed strings were included in the uncertainty data base. There is no reason to expect detectors to fail in other than a random manner.
Another feature which improves the inferred power for failed detector positions is the availability of a significant number of symmetric detectors between quadrants. The core is loaded in a 1/4 core rotationally symmetric pattern and in the solution process for the full core, a 1/ 4 core XTG solution is used to produce an interim inferred/measured 1/4 core solution. Failed detectors in one quadrant are replaced, if possible with a measurement from a symmetric detector position in this 1/4 core solution.
By comparison, the INPAX-11 monitoring system requires that 75% of the detector strings have. at least three out of four detectors operable. PIDAL is more restrictive.
Comment 7 - For the proposed quarter-core symmetric loading pattern designs are there "critical" detector strings which may increase the uncertainty of monitoring fresh bundles with gadolinia.
Response - With anticipated future reload cores the number of gadolinia assemblies includes nearly all the fresh assemblies. For example in cycle 10, 60 of 68 assemblies will contain gadolinia. During their first cycle in the core and the beginning of the second cycle, many of the gadolinia assemblies will be close to the highest power assemblies.
4
Therefore there are' single detectors which will always! "critical". The cycles used in the PIDAL uncertainty analysis, i.e. 5,6,7 and 8 included a significant number of gadolinia assemblies thus the core monitoring uncertainties will also be representative for gadolinia assemblies.
Comment 8 - Does the historical individual detector failure distribution at Palisades appear to be random or string dependent.
Response - Usually random: Palisades is different from most plants in that all 43 detector strings are replaced with new detectors each cycle. Failure rates have decreased steadily as the design and manufacturing processes have matured.
In fuel cycles 6 and 7, an aggressive program of plotting hourly incore detector signals from the then 225 incore detectors (now 215 due to loss of two strings to reactor vessel level monitoring system instrumentation) was established. It was found that there were a few strings, most notably two which exhibited abnormal behavior from two or three of their respective detectors. There were also four or five other detectors which exhibited abnormal behavior on an individual basis. The causes for these abnormalities were found to be cabling problems which have been corrected. During cycle 8 we operated with only three individual detectors inoperable. For cycle 9 no detector cabling problems have been identified.
Comment 9 - When declaring a detector failed (or inoperable), especially using the DEV (XTG vs PIDAL) criteria, what measures are taken to ensure that a modeling problem or a real operational asymmetry will not be overlooked.
Response - When detectors are failed by the code, they are carefully scrutinized by looking at trends of the hourly signals of the detector in question, in conjunction with the other detectors in that string, near neighbors and symmetric partners if any exist.
Generally failures of this type tum out to be cabling problems (most notably during cycles 6 and 7). For cycle 9, these types of comparisons were used to verify the validity of detector readings from three incore strings which exhibited problems during physical insertion of the incore strings during refueling.
Comment 10 - Would the inclusion of Cycle.8 benchmarking results into the statistical data base change the uncertainty results... (remainder of comment illegible)
Response - A complete uncertainty analysis for the cycle 8 PIDAL model was performed as documented in Palisades Engineering Analysis EA *GAB*90*08. "Cycle 8 PIDAL Uncertainty Analysis. 1/Sth Core and 1/4th Core W-primes". A summary table from this analysis showing how the cycle 8 uncertainty components compared with the cycle 5, 6 and 7 components follows (see Table #2). This analysis shows that the cycle 8 values are bounded by the PIDAL Uncertainty Analysis submitted to the NRC. Inclusion of the cycle 8 values in the statistical data base would lower the PIDAL measurement uncertainties currently being proposed by CPCo.
5
Table #2 - Individual Cycle PIDAL Uncertainty Summary For Cycles 5,6,7 and 8 all values using 1/8th core W-prime based models Uncertainty Cyde 5 Value Cycle 6 Value Cycle 7 Value Cycle 8 Value Component SFfs) 0.0293 0.0272 0.0259 0.0258 SFfsa) 0.0233 0.0125 0.0195 0.0139 SF(r) 0.0023 0.0023 0.0021 0.0020 Comment 11 - Will additional criteria be used to qualify the reuse of partially depleted rhodium detectors, since this uncertainty was determined based on a single cycle of operation.
Response - At this time it is not anticipated that detectors will be reused again at Palisades. This is due to the design of our upper guide structure and the problems we have in reinserting used detectors. The reinsertion of reused strings is an extremely dose intensive task. The cycle 7 reused detector uncertainties were calculated and incorporated in the PIDAL Uncertainty Analysis for completeness in the event that a method could be found for successfully reinserting incores. H reinsertion were to be tried again, it would be on a limited basis with one-fourth of the strings or less reused.
This would be consistent with the cycle 7 reuse attempt and therefore it is believed that these uncertainties would be useable as-is until additional measured data could be obtained. Note that appropriate penalties for the limited size of the cycle 7 data base were incorporated into the final* reused detector uncertainties.
6
.. 'It>~*
PIDAL Cycle Dependent Updates Comment U - What cycle dependent coding changes will be required with respect to constants imbedded in the coding, constants updated by BLOCK DATA, etc.
Response - During the EOC 8 to BOC 9 PIDAL conversion process, all cycle dependent coding of constants was removed. All constants are now input to PIDAL via two data files whose generation and update are controlled by Palisades Nuclear Plant Fuels Technical Manual Procedure ORP-P-03. Revision 0. "Palisades Incore Detector Algorithm (PIDAL)".
Comment 13 - What recompilation/reverification procedures to ensure coding QA from cycle to cycle.
Response - Test plans and test analysis reports are written for each change made to the PIDAL system whether it be to software coding or the data base. This PIDAL Program Documentation is in accordance with Palisades Nuclear Plant Software Quality Assurance Plan Procedure SOAP-029. Revision 2 "lncore Analysis Systems".
Comment 14 - What audit/verification procedures are applied to the vendor-supplied library (W', peaking factors, control rod correction factors) to ensure the validity of the data for the new operating cycle.
Response - The vendor notebooks which document the generation of the W-primes, peaking factors and control rod correction factors are audited and reviewed by CPCo personnel as part of the CPCo review of the vendor core reload package.
Comment 15 - What is the frequency of the vendor library updates (W', etc) required during an operating cycle; how is the changeover handled with respect to overlap and check calculations.
Response - For cycle 9 the vendor library information was transmitted once at BOC to cover the entire fuel cycle. It is anticipated that this will also be the case for future fuel cycles. In past fuel cycles, partia:! libraries have been installed, usually a short library to cover only the first few months of a new cycles operation. Overlap has never been a problem with the updates since the update libraries always contain data back to BOC.
When a new library is installed, test cases are run and documented in accordance with PIDAL quality assurance procedures in order to verify the correctness of the library.
Comment 16 - How is the exposure-history file shuffle/update to a new reload cycle handled for fuel, control rod and detector exposure arrays, etc.
7
°(.*
. e e
Response - The exposure history file is shuffled from EOC to BOC using a support program which is controlled per Palisades Nuclear Plant Software Quality Assurance Plan Procedure SQAP-029. Revision 2 "lncore Analysis Systems". Use of this program is described in Palisades Nuclear Plant Fuels Technical Manual Procedure ORP-P-03.
Revision 0. "Palisades Incore Detector Algorithm (PIDAL)".
8
_.) *-
Miscellaneous Comment 17 - How many "fuel-type" indicators (IGAD, ISP, ?) are used to identify neutronic/thermal hydraulic bundle/zone differences.
Response - The number of these fuels types varies from reload to reload but is typically between 8 and 11. The PIDAL coding is capable of handling up to 20 different fuel types per fuel cycle.
- Comment 18 - What is the core follow /update frequency plarined for production PIDAL 0usage.
Response - INCA and PIPAL are typically run three times per week each for core follow and exposure update purposes when the reactor is at power. INCA and PIDAL must be run on a weekly basis, as a minimum, per Palisades Nuclear Plant Technical Specification Surveillance Procedure DWf-12A "Monitoring Reactor Parameters".
Comment 19 - Has Cycle 9 operation been modelled with PIDAL in a "production" mode to generate exposure history for cycle 10..
Response - Yes. A production PIDAL model for fuel cycle 9 has been in place since initial cycle 9 startup. This model has been run concurrently with INCA for the entire fuel cycle. This was also the case for cycle 8.
Comment 20 - What procedures are used for the input setup, execution and interpretation of results.
Response - Use of the PIDAL system is covered by Palisades Nuclear Plant Fuels Technical Manual Procedure ORP-P-03. Revision 0. "Palisades Incore Detector Algorithm (PIDAL)". Several sections of this procedure apply directly to the Palisades Reactor Engineer who has responsibility for the use of the PIDAL system.
Comment 21 - What procedures are used for the loading and implementation of the Tech Spec limit outputs on the PIP system.
Response - Loading of the Technical Specification incore detector alarm limits into the PIP is covered by Palisades Nuclear Plant Technical Specification Surveillance Procedure DWf-12B. "Monitoring Reactor Parameters (Incore Alarm Updating)".
9
~-:::---~. ;-*
I DISTRIBUTION FOR MEETING
SUMMARY
DATED: October 24, 1991 Facility: Palisades DuG:k&t~
NRC PDR Local PDR T. Murley F. Miraglia B. Boger J. Zwolinski L. Marsh B. Holian E. Kendrick P. Shuttleworth OGC" E. Jordan ACRS ( 10).
G. Grant*
12-G-18 12-G-18 15-B-18 MNBB-3302 P-315 17-G-21 f
f
,(_,'I
- 1
. '