ML20198K559

Safety Evaluation of TR BAW-10167,Suppl 3, Justification for Increasing Reactor Trip Sys On-Line Test Interval. Rept Acceptable

ML20198K559
Person / Time
Issue date:	01/07/1998
From:	NRC (Affiliation Not Assigned)
To:
Shared Package
ML20198K546	List: ML20198K542 ML20198K559
References
NUDOCS 9801150040
Download: ML20198K559 (9)

Text

. _ _ _ __ _ ____ . ___ _ _ _ _

1 ra atg

p. .\

UNITED STATES NUCLEAR REGULATORY COMMISSION i i! W ASHINGToN, o.C. 302H001

%,*...+f" SAFETY EVALUATIO'l BY THE OFFICE OF NUCLEAR REACTOR REGULATION S&W OWNERS GROUP TOPICAL REPORT BAW 10167 SUPPLEMENT 3 JUSTIFICATION FOR INCREASBlG THE REACTOR TRIP SYSTEM ON LINE TESTINTERVAL

1.0 INTRODUCTION

By letter dated June 7,1996,(Reference 1), the B&W Owners Group (B&WOG) submitted Topical Report BAW 10167 Supplement 3, " Justification for increasing the Reactor Trip System On Line Test intervals." This report was prepared by B&W Nuclear Technologies and provides the technical basis to justify increasing the on line surveillance test interval (STI) from the current one month to a six-month interval, for reactor trip system (RTS) trip devices consisting of reactor trip breakers (RTBs), reactor tr!p modulea (RTMs), and I electronic trip relays By letters dated September 18,1996 and January 3,1997 (Reference 2), the B&WOG provided additional information to substantiate the topical report request. Subsequently, the B&WOG in their letter to the NRC dated November 5, 1997 (Reference 4), submitted an amendment to the topical report to change the requested STI for RTS trip devices from the proposed six months to a three month interval.

ENCLOSURE 9901150040 990107 PDR TOPRP EMY3W j

C POR ,1

2-The methodology and models used in this topical report are the same as those previously used in Supplement 1 of BAW 10167," Justification for increasing the Roactor Trip System On Line Tost intervals," which was submitted to justify the RTS instrument string STI extension from one month to a six month interval. At that time, the B&WOG chose not to include the RTS trip devices in their request for the STI extension because the RTB front-frame, active shunt trip, and lubricant upgrades (installed in response to Generic Letter (GL) 83 28," Required Actions Based on Generic implications of Salem ATWS Events") were new, and it was prudent to test the trip devices more frequently until additional operating experience was obtained. The staff approved Supplement 1 of BAW 10167 and suggested some specific improvements in the methodology. Supplement 3 of BAW 10167 used an improved methodology and concentrated on the changes to the modeling and data that are necessary for examining the sensitivity of the RTS reliability to the RTS trip devices STI including updating of operating experience data. The taliability models used in this analysi::

are representative of both the Oconoe (Oconee Units 1,2 & 3, Crystal River Unit 3 and Arkansas Nuclear One Unit 1) and Davis Besse RTS design groups and do not include Three Mile Island which was not represented by the B&WOG on this issue. The unavailability of each of the t-'o RTS trip device design groups is rnodeled in the report using reliability block diagrams for both the current one month STI and the originally proposed six-month STI. The analysis ovaluated the impact of the proposed STI extension on core rnolt troquency and RTS unavailability to demonstrate that the proposed STI change did not significantly increase plant risk when compared with the current technical specification requirements.

~. ,

3-The following evaluation addresses both the acceptability of the probabilistic risk analysis presented in BAW 10167,Suppiament 3 and the acceptability of the orginally proposed and amended STI extension.

t 2.0 EVALUATIQN The staft evaluation included the following two aspects of the probabilistic risk analysis (PRA) performed by B&W Nuclear Technology to justify the proposed extension of the RTS trip devices STI:

(1) Models and data used for the reliability analysis (2) Quantification of the analysis models The methodology and models used in the BAV/10167, Supplement 3 analysis are the same as those used in the staff approved Gupplement 1 of BAW 10167 including time-dependent, common modo failure and uncertainty analyses. Emphasis was placed on the use of operating experience for the data source in the derivation of both random and common mode failure rates. Nuclear Plant Peliability Data System (NPRDS), Licensee

8. vent Reports (LER), a Sandia National Laboratory (SNL) research report, and the technical judgement of the maintenance technician or engineer were the ::ources for the RTS component failure database, improvements were made in the Reliability Block Diagram (RBD) models of the RTS trip devices to incorporate increased details of the RTO f ailure data. Specifically, the RTB portions of the RW wers divided by f ailure mechanism into two components corresponding to the failures caused by cyclic (i.e., demand) stress and

4 time-in-service-related (i.e., standby) stress. The RTB failure data reflected reliability improvements cnd reduction in the potential for common mode failures due to the implementation of the guidelines of GL 83 28.

in order to assess the sent,itivity of RTS reliability to the trip device STl, the instrument string portion of the model was held constant (i.e., with six month test intervals), and the testing frequency in the trip device portion of the model (RTMs, RTBs, and Slectronic trip talsys) was varied from one-month to six months. An error f actor of 10 (the largest error factor listed in WASH 1400,"An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants, for instrumentation," and suggested in the staff safety evaluation report on Supplement 1 of BAW 10167)was used for the RTS trip devices landom failure rate (lambda factor). When a common mode failure rate could not be determined from the component failuie history, a beta factor (fraction of lambda factor in which two or more components are involved due to common mode failutel was used as suggested by NUREG/CR 580;," Procedure for Analysis of Common cause Failures in Probabilistic Safety Analysis." A beta factor of 0.05 was assumed in this analysis because the recent failure history of the RTS trip devices showed no evidence of multiple failures since the generic letter 83 28 upgrades were irnplemented.

A time-dependent and time averaged RTS unavailability calculation was performed by B&W Nuclear Technologies using a reliability block diagram and computer codes for both the Oconee and Davis Besse RTS trip devices designs (Oconee design class plants use silicon centrol rectifiers (SCRs) to trip the re0ulating rods, groups 5 through 7 while the Davis Besse configuration uses the SCRs to trip both safety and regulating rods, groups 1

.s.

through 7). "All seven rod groups must trip" was used as the mission success criterion in the reliability rnodel which made the quantification results more conservative (the most conservative success criterion for reactor trip used by the staff was defined in SECY 83-293 as in,sertion of half of the control rodr,into thn core in a checkerboard pattern to shutdown the reactor). The quantification calculation also included plant spurious trip evaluation results which were directly attributed to sun oillance testing of the RTS trip devices. This represented a net improvement in the r.smber of scrams / plant / year and a reduction in core melt frequency (CMF) due to relaxation of the RTS trip devices STI.

Table 4 2 cf BAW 10167, Supplement 3 (henceforth called Supplement 3), prosents the f ailure rate data of the GE model AK RTBs used in B&W and Combustion Engineer: rig designed plants. These data are primarily based on a NPRDS search (between 1988 and 1993 to update the data af ter implementation of GL 83 28 improvements) and on Sandia National Laboratory Report SAND 93 7027," Aging Management Guideline for Commercial Nuclear Power Plants - Electrical Switchgears". Figure 41 of Supplement 3 compares the failure rates of the RTBs before 1984 and after 1989. The number of RTB failures after 1989 is about one sixth the fuilures before 1984. Considering that the results of the NPRDS search depends significantly on the search command used, the NRC staff conducted an independent search for this data base and came up with comparable results.

Using the Table 4-2 data of Supplement 3, the B&WOG calculated the RTS unavailabilities (failure / demand) for one month and six month test intervals using the computer program SAMPLE (Reference 3), The results of the computer run are presented in Table 61 and Figures 6-1 through 6-11 of Supplement 3. Figure 61 indicates that af ter 1989, the RTS

f C

t

.g.

f availability went up seventeen times. This improvement in RTS availability was used to i calculate the incremental risks of core damage frequency (CDF) from the extension of the  !

RTS STI from one month to six months and the results are presented in Table 6 3. Table i 6 3 indicates a not incremental risk in CDF if 3 2x10E 9 for Davis Besse and 2.6x10E 8 for Oconee type plants. From a risk slanificaw", point of view these values are acceptable.

However, the staff could not directly verify the results without performing a computer analysis similar to the one performed for BAW 10167A, Supplement 1 by the Idaho  ;

' National Laboratory. Therefois, the staff requested the B&WOG to provide an ,.

extrapolation of the data from BAW 10167A, Supplement 1 using a direct correlation l methodology to show the reasonableness of the Supplement 3 results without reliance on computer analysis. The B&WOG's response in Reference 2 provided additionalinformation including the following RTS failure probability per demand to establish the reasonableness of the Supplement 3 results:

DATA DAVIS BESSE GROUP OCONEE GROUP One month test interval from 9E 9 1.1 E 6 Supplement 1 >

3ix month test interval 9.01E 9 2.6E 7 t extrapolated from Supplement 1 Six-month test interval from 9.05E 9 3.3E-7 Supplement 3. .

Worst case sensitivity analysis 9.12E 9 1.1 E 6 of Supplement 3 data -

__The staff review of the information presented in References 1 and 2 indicates that '

Supplement 3 adequately demonstrated a negligible change in CMF and overall plant risk, and the conclusions drawn therein are reasonable. However, the staff determined that the au we , , - - - g . . , , , - - , , , ,- ,,,,w,, + . - - - - - -- .-,- -- ,.+ --

.- - ~ , - - . - . - - - - . - ~

t

. 7 availability of approximately five years of data collected between 1988 and 1993 indicating a six fold improvement in the numbet of HTB. failures since implementation of GL-28 is not sufficient to offset the uncertainties associated with the actual potential.

breaker teilure modes. As such, the staff did not find adequate operating history with test intervals foncer than one month to support a change of STI from the current one month to six months.- The B&WOG agreed with the staff concerns and submitted an amendment to Supplement 3 in Reference 4. The amendment requests a more conservative three moi.th STIinstead of the originally proposed six month interval.

The basis for techincal specification STI for any safety related component or system is to ,

ensure that the probability of an undetected failure existing within the component or system is small and to reduce the potential for spurious trips which may unnecessarily challenge the plant operators and safety systems. The B&WOG determined that the monthly test of the RTS trip devices was an unnecessary burden on utility resources be.cause the reliability of the RTBs has considerably improved since implementation of the GL 83 28 recommendations and because there are sufficient data to confirm that the breaker upgrades were effective. Additionally, a three month STI has been approved by

' the staff for CE plants which use GE Model AK RTBs, Combustion Engineering Owners Group (CEOG) Topical Report CEN 327 A, "RPS/ESFAS Extended Test interval Evaluation" dated January.1989 (Reference 5) which was approved by the staff included a three

, month STI for_ CE plant RTBs. Although the CEOG has not maintained a verifiable failure experience record of a three month STl for their RTBs, the B&WOG believes that a

- ' con bined effort of the B&WGG and CEOG willincrease the data pool for generatin0

-- cxperience with tho extended test interval. Nevertheless,:n J.eterence 4 the B&WOG e n , -

- ._ . __ _ . - _ ~ . _ .- . _ ._ . _

. .g .- '

1 identified that'the laboratory tests performed by GE (manufacturers of the RTBs used'at CE and B&W power plants) on the RTBs have shown that the upgraded lubricant will retain'its'

' stability well beyond 90 days. Further,' the B&WOG has committed to monitor e

performance of the RTS trip devices to ensure tha' degradation does not occur as a result of.the proposed STI extension. If performance criteria are exceeded, the B&WOG has committed to rerform an evaluation and the feedback mechanism would alert the utilities.

l to taka corrective action, including more frequent testing,

3.0 CONCLUSION

Based on the above, the staff concludes that the analyses in BAW 10167, Supplement 3, .

as amended, adequately demonstrated a negligible change in CMF and overall plant risk, and thus the preposed extension of the RTS trip devices STI from the current one month to three months, is acceptable. Since ti;e proposed change does not adversely impact plant sefety, it is approved as an appropriate plant specific technical specification change for

- those B&W plant . licensees covered by the BAW 10167, Supplement 3 analyses.

E" li.

f 1

1 A.-

di 4 + er 9 e -e *ur -

Y

i,. ., , - ,

e.

-_ 9-

' 4.0 -- REFERENCES -

1. Letter, J. H. Taylor to NRC Document Control Desk, dateo June 7,1996.

...i 2.. Letters, David J.- Firth to J. L. Birmingham, dated, September 18,' 1996 and January 3,,1997. >

3. SAMPLE, General Purnose Computer Program for Uncertainty Analysis by Monte ,

Carlo Simulation, original by WASH 1400 Reactor Safety Study Group, and modifications by E. Oelkers and T. L. Wilson, NPGD TM 501, Rev.1, Rabcock &

Wilson, Lynchburg, VA, April 1980.

~4.: Letter, M. W. Epling to J. L. Birmingham, dated November 5,1997.

5. CEOG'fopical Report CEN-327-A, "RPS/ESFAS Extended Test Interval Evaluation",

datM January 1989.

c

-- - _ _ _ _- - - l