ML20043B363
| ML20043B363 | |
| Person / Time | |
|---|---|
| Site: | Hope Creek  |
| Issue date: | 05/22/1990 |
| From: | Labruna S Public Service Enterprise Group |
| To: | NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM) |
| References | |
| NLR-N90104, NUDOCS 9005290188 | |
| Download: ML20043B363 (4) | |
Text
y a
Pubhc Service Electric and Gas Company Stanley LaBruna Pubhc Service Ekstric and Gas Company P.O. Box 236 Hancocks Bridge, NJ 08038 609-339-4800 May 22, 1990 NLR-N90104 United States Nuclear Regulatory Commission Document Control Desk WaJhington, DC 20555 Gentlement
+
REVISED RESPONSE TO REQUEST FOR ADDITIONAL INFORMATION REGARDING i
BAILEY SOLID STATE LOGIC MODULE RELIABILITY - 1988 AND 1989 DOCKET NO. 50-354 HOPE CREEK GENERATING STATION Public Service Electric and Gas Company (PSE&G) hereby provides a revised response to a request by Mr. Clyde Shiraki, NRC Licensing Project Manager for our Hope Creek Generating Station (HCGS).
Following several phone discussions with Mr. Shiraki, we have-modified the criteria used to monitor and trend Bailey Model 862 solid State Logic Module (SSLM) failure rates described in our letter, NLR-N90073 dated April 19, 1990.
Three concerns were expressed by Mr. Shiraki during a May 1, 1990 f
phone conversation.
The first related to our attempt to compare our statistics on the Bailey SSLMs to IEEE reliability standards for relays.
We quoted a value of 16.73 failures per thousand units per year from IEEE-500.
The NRC staff could not find this L
value in the standard.
PSE&G derived this value from page 155 of l
IEEE 500-1984 which gives a value of 1.91 failures per million hours of operation for " General Relays".
In discussions with NRC staff in 1987, reliability was discussed by NRC staff in terms of failures per thousand units per year.
PSE&G, in the interest of continuity, therefore, converted the 1.91 value to the 16.73 value by6 multiplying 8760 hours0.101 days <br />2.433 hours <br />0.0145 weeks <br />0.00333 months <br /> /vear and 1,000 units times the
~
1.91x10 value.
The resulting 16.73 failures per thousand units per year has been used by HCGS staff in subsequent comparisons to relay failures [i.e.,
in our monthly SORC reports during 1988 and 1989 and in our April 19, 1990 letter to NRC).
0 0'
~
- gas 2xta!!te8Bjig4 L iqs P
1 Document Control Desk 2
5-22-90 NLR-N90104 i
In 1987, during discussions between NRC staff and our engineering personnel (non-station) involved in Bailey SSLM reliability assessment, the comparison of logic modules to relays was viewed as inappropriate since a good engineering basis for that comparison could not be established, i.e., too many assumptions and variations existed.
Therefore, PSE&G proposed a value of 1.19 failures per million hours which is the value in IEEE 500 (page 721) standard for solid state computation devices.
PSE&G considered this a more realistic comparison even though our SSLM failure rates exceeded that standard.
The confusion appears to have been caused by a breakdown in communications wherein station engineering personnel continuing to use the " derived" relay standard instead of the statistically modeled IEEE solid state computation device standard for SSLM failure rate comparisons in the station's ongoing monitoring /
trending program.
That oversight has been corrected.
The second concern was regarding the inclusion, in our April 19, 1990 letter, of numerous types of failures other than in-service failures.
The data provided in that letter has been reevaluated based upon guidelines provided in our telephone conversation with Mr. Shiraki.
Attachment i now provides in-service failure rates and uses the 1.19 failures Der million hours of service standard for comparisons.
This is consistent with the data provided to NRC in our March 1988 submittal of Bailey 862 SSLM reliability /
failure rates for 1987.
In-service SSLM failure rates for 1988 and 1989 have shown an encouraging trend.
The number of failures ner million hours of ooeration has decreased to 1.41 and 1.30 in successive years.
When compared to our 1987 report rate of 1.8 failures per million hours of operation, a distinct improvement has been achieved, especially in light of the more extensive testing and more comprehensive test equipment / methodology in use after 1987.
The final concern related to the classification of the Bailey SSLMs.
This equipment, as well as the HVAC system which provides the environmental controls for the Bailey equipment are Class 1E.
This information is presented in the HCGS UFSAR, Sections 3.10, 3.11, and 9.4.
We believe that the data in this submittal
- 1) demonstrates that the Bailey Model 862 SSLMs have achieved a level of reliability that compares favorably with other types of equipment performing the same functions and
- 2) should resolve NRC concerns on this matter.
I Document Control Desk 3
5-22-90 NLR-N90104 Should you have any further questions, we will be pleased to discuss them with you.
Sincerely,
}
h#
Attachment C
Mr.
C.
Y. Shiraki USNRC Licensing Project Manager Mr. T.
P. Johnson USNRC Senior Resident Inspector Mr. T. T. Martin, Administrator USNRC Region I Mr. K. Tosch, Chief Bureau of Nuclear Engineering New Jersey Department of Environmental Protection
s i
ATTACHMENT 1
[
i BAILEY 862 LOGIC MODULE RELIABILITY FOR THE PERIOD 1988-89 MONTH 1988 FAILURES 1989 FAILURES January 1
1 l"
February 1
- 4 March
- 3
- 4 April
- 3 1
May.
3 1
June 1
0 July 4
2 August 1
0 September 2
- 2 s
October 5
- 3 November 2
- 8 December 2
0 Totals 28 Totals 26
- Maintenance / Refueling Outage periods.
1988 Avg. Module Population = 2268 1988 Annual Failure Rate Analysis:
1988ServiceHours=g760 Hours.
8760 x 2268 = 19.8x10 total module servico hours..
28 failures / 19.8x106. 1.41 failures per million hours failure rate, t
i 1989 Avg. Module Population = 2278 1989 Annual Failure Rate Analysis:
1989ServiceHours=g760 Hours.
total module service 8760 x 2278 = 19.9x10 hours.
26 failures / 19.9x106 = 1.30 failures per.
million hours failure rate.
-1988-89 failure rate compared to IEEE Standard 500, Section 8.24, computation Modules, page 721 of 1.19 failures per million hours failure rato.
Notest.
1.
Module population has increased due to design changes.
2.
Failures listed are actual in service failures, l
4 j
i.
1
.-.