ML20077B854
| ML20077B854 | |
| Person / Time | |
|---|---|
| Site: | Arkansas Nuclear  |
| Issue date: | 07/20/1983 |
| From: | Delgaizo T, Delgazio T, Overbeck G FRANKLIN INSTITUTE |
| To: | Chow E NRC |
| Shared Package | |
| ML20076G582 | List: |
| References | |
| CON-NRC-03-81-130, CON-NRC-3-81-130, RTR-NUREG-0737, RTR-NUREG-737, TASK-***, TASK-TM TAC-45273, TAC-45319, TAC-45376, TER-C5506-410, NUDOCS 8307250463 | |
| Download: ML20077B854 (58) | |
Text
_
.r ,
l -
TECHNICAL EVALUATION REPORT OPERATING REACTOR .
PORV REPORTS (F-37) 2
- TM1 ACTION PLAN REQUIREMENTS i
GENERIC RE.00RT -- BASCOCK & WILCOX DESIGNED UNITS 1
?
v m ,- __
..--**%E j
j NR C CCCXET NO. Various FRC FRC.;ECT'C5506 i
i FRC %SS:GNMENT7 l
I
, NRC CONTR ACT NO. NRC 03-81-1*.0 FRC TASK 410 Pr7:ared cy j Franklin Researen Cantsr Author: G. J. 0rerbeck lCth and Race Streets I. J. CelGaico Philacalonia. P t 3 03 .=?C Gccuo Laadar: G. :. C rsr':ac':
, Prepared rcr l Nuclear Regulatory Commission Lead NRC Engineer: I. Chev
. ] Washington, D.C. 20555 i j .
i July 20, 1993 I
l I
,1 This repcrt was crecared as an account of worx sconsored by an agency of the United States l Govemment. Neitner tne United States Government nor any agency thereof, or at:y of their e
employees, makes any warranty. expressed or implied, or assumes any legal liability or responsiblity for any third party's use. or the results of sucn use, of any infctmation. acpa-ratus. product or process disc!csed in this report. or represents that its use by such third party would not infringe privately owned rfgnts.
9 a
w p e w-
.%[j Copy Hos.Been.; SentlfoHDR.
. dl /
b
~
.J J. Franklin Research Center il
- 9 A Divisien of The Franidin Institute The Bernerran Frerman Parmway. PMa Pa. 19103 '21:1 u41000 1
Y r.
- h. -
TIOHNICAL EVALUATION REPORT OPERATING REACTOR FORV REPORTS (F-37)
TMI ACTION PLAN REQUIREMENTS GENERIC REPORT -- BABCOCK & WILCOX DESIGNED UNITS NRC CCCXET NO. Various FRC PRO, LECT C55CS FRC ASSIGNMENT 7
, NRC CCNTRACT NO. NRC43-81-1::0 FRCTASx 410 precared by Frank!!n Research Canter Author: G. J. Overbeck 20th and Race Streets T. J. DelGaizo Philadelphia. PA '9103 FRC Group Leacer: G. J. Overbeck Prepared for Nuclear Regulatory Commission Lead NRC Engineer: E. Chow Wasnington, D.C. 2C555
\ .
July 20, 1983
- This report was prepared as an account of work sconsored cy an agency of the United States
/3cvernment. Neitner the United States Government nor any agency thereof, or any of the:r emcloyees, makes any warranty, expressed or implied, er assumes any legal liability or l responsibility for any third carty's use. or the results of sucn use. of any information, acca-
' ratus, procuct or process disclosed in this report, or reoresents that its use by such third party would not intnnge privately owned rignts.
Prepared by: Reviewed by: Approved by:
I (. il$3d y/ sd 1 ff w A Prtr[cihal Authe'r ?:oject Manager 6epartment D/rectg'r Date 7 N* ' 3 Cate* T1o/ % Cate:
- 'O
^
. . . . FrankJin Research Center A Division of Tne Franklin Insttute
,, The Berymm Frenen Personey. PNa. 7s.191CJ (2:31448.;C00
TER-C5506-410 CONTEN"3 Secticn Title, ? ace 1 INTRODt%:TICN . . . . . . . . . . . . . 1 1.1 Purpose of Review . . . . . . . . . . . 1 1.2 Ceneric Background. . . . . . . . . . . I 1.3 ?lant-specific 3acxground . . . . . . . . . 4 2 ?2/IEW CRITERIA. . . . . . . . . . . . . 5 3 TEC'?.NICAL EVALUATION . . . . . . . . . . . 7 3.1 Review of the B&W Report for Completeness . . . . . 7 ,
3.1.1 Small-Break LOCA Probability calculaciens . . . . 7 3 .1.1.1 PORV Cpening Frequency. . . . . . 3 3.1.1.2 Small-Break LOCA Calculations . . . . 13 3 .1.2 Small-Break LC:A Prc= ability Based on operational Data . . . . . . . . . 15 3 .1.3 Percentage of PORV Cpenings Curing Cverpressure Transients . . . . . . . . . . . 16 3.1.4 Actions Taken to Reduce PORV Challenges. . . . 16 3.1.5 Small-Break LOCA from Stuck-Open Safety-Relief Valve (SRV) . . . . . . . . . . . 17 3.2 Evaluation of the B&W Report Submitted in Response to NUREG-0737, Items II.K.3.2 and II.K.3.7. . . . . 18 3.2.1 Small-Break LCCA Probability Calculations . . . 18 A. iii
..Lj F anicin Research Center
- cman= her n
, . . . _ . - ~, . . - . . ... .. , . - . . . . - . _ . - . . - . - . - , . .. . - . , .
TER-C550 6-410 FOREWORD This Tecanical Evaluat. ion Report was prepared by Franklin Research Cantar under a contract with the U.S. Nuclear Regulatory Comission (Office of Nuclear Reactor Regu1& tion, Division of Operating Reactors) for technical assistance in support of NRC operating reactor licensing actions. The technical evaluation was conducted in accordance with criteria established by tne NBC.
Mr. G. J. Cverbeck and Mr. T. J. DelGaizo contributed to tne technical preparation of this report tnrcugh a subcontract witn WESTIC Ser tices, Inc.
I l
I i
I Dur-
.??) Franklin Research Center V
A h ernwr .nn
.O , ,y.
_- ., .-r- ---
w-9 . . , - - y------- -.-
TER45506-410
- 1. INTRCDtI: TION 1.1 PURPOSE OF RE7IEW This technical evaluation report ("'ER) documents an independent review of a generic report of Babcocx & Wilcox (S&W) designed units prepared in response to NUREG-0 737 (1) , " Clarification of TMI Action Plan Requirements," Item II.K.3.2, " Report on Cverall Safety Eff ect of Power'Cperated Relief Valve Isolation System," and Item II.K.3.7, " Evaluation of Power Operated Relief l Valve Opening Procamility Curing Cverpressure Transient." This evaluation was pe.rformed with the following objectives:
o to ensure enat - tne B&W Report is ccmplete and properly documents tne information required by NUREG-0737, Items II.K.3.2 and II.K.3.7 o to ensure that the estimated procacilities of the 3&W Report satisfy the review criteria.
3 i
1.2 GENERIC SACKGRCUND In NUREG-0 565 [2] , " Generic Evaluation of Feedwater Transients and Small l
Break Loss-of-Coolant Accident Behavior in Babcocx & ,Wilcox Cesigned 177-FA Operating Plants," the Nuclear Regulatory Commission's (NBC) Eulletins and Orders Task Force recc:nmended the folicwing:
o "?covide a system which will assure that the bicek valve protects against a stuck-open PORV. This system will cause the blocx valve to close when BCS pressure has decreased to some value below tne pressure at which the PORV should have reseated. Tnis system should incorpcrate an override feature. Each licensee should perform a confirmatory test of the automatic blocx valve closure system.
o In order to minimize the opening of the POW, most overpressure transients should not result in the PORV opening. Licensees of S&W l Plants should document that the PORV will only open in less than five percent of all anticipated overpressure transients using the revised setpoints and anticipatory trips for the range of plant conditions which migne occur during a fuel cycle, o All failures of PORVs to reclose should be reported promptly to the NRC. All challenges should be reported in annual reports.
! A f./.Enkiin a :m a neResesren r, n . C. enter l
l
TER-Cf 506-410 o
Licensees should submit a report to the NE whien discusses tne safety valve failure rate experienced in E&W operating plants.
o All failures of safety valves to reclose should be reported promptly to the NBC. All challenges should be reported in annual reports."
These recommendations were later included in NUREG-0660 (3], "NBC Action Plan Developed as a Result of the TMI-2 Accident." The first reconnendation was incorporated into NUREG-0660 as Item II.K.3.1, " Installation and Testing of Automatic Power-Operated Relief Valve Isolation System," and the remaining recommendations were included in Items II.K.3.2 and II.K.3.7. In Reference 1, the staff delayed implementation of Item II.K.3.1 until the pending PORV reliacility analysis of Item II.K.3.2 confirmed the necessity of an autcmatic isolation system. Specifically, NUREG-0737, Item II.K.3.2 stated:
" (1) The licensee snould submit a report for staff review documenting the various actions taken to decrease the probability of a small-break loss-of-coolant accident (LCCA) caused by a stuck-open pcwer-operated relief valve (PORV) and show how those actions constitute sufficient improvements in reactor safety.
(2) Safety-valve failure rates based on past history of the operating plant designed by the specific nuclear steam supply system (NSSS)
- vendor should be included in the report submitted in response to (1) above."
In addition, Reference 1 further clarified that:
i
" Modifications to reduce the likelihood'of a stuck-open PORV will be considered sufficient improvements in reactor safety if they reduce the probability of a small-break LOCA caused by a stuck-open PORV such that t
it is not a significant contributor to the probability of a small-break l LOCA due to all causes. (According to WASE-1400, the median probability j
of a small-break LOCA S2 with a break diameter between 0.5 in. and 2.0
! in. is 10-3 per reactor-year with a variation ranging from 10-2 to 10-4 per reactor-year.)
The above-specified report should also include an analysis of safety-valve failures based on the operating experience of the pressurized-water-reactor (PWR) vendor designs. The licensee has the l option ot preparing and submitting either a p'ar.t-specific or a generic l
report. If a generic report is submitted, each licensee should document the applicability of the generic report to his own plant.
Based on the above guidance and clarification, each licensee should perform an analysis of the probability of a small-break LCCA caused by a 4 O Frankun 4c= a. w a Resear.en w Center
_ - - - . . - . . . , _ . , _ . - . _ . - _ . . , ~ . ~ , - .- ~ . . - - -,
l e e l
i TER-C5506-410 stuck-open PORV or safety valve. This analysis should consider scdifica-r tions which have been made since the TMI-2 accident to improve the probability. This analysis shall~ evaluate-the-effect of an automatic POW isolation system specified in Task Action Plan Item II.I.3.1. In evaluating the automatic POW isolation system, the potential of causing a subsequent stuck-open safety valve and the overall effect on safety (e.g. , effect on other accidents) should be examined.
Actual operational data may be used in this analysis where appropriate.
The bases for any assumptions used should be clearly stated and justified.
The results of the probability analysis should then be used to determine wnether the modifications already implemented have reduced the
, probability of a small-break LOCA due to a stuck-open PORV or safety valve a sufficient amount to satisfy the criterion stated above, or whether the automatic PORV isolation system specified in Task Action Item II.K.3.1 is necessary.
In addition to the analysis described above, the licensee should ccmpile operational data regarding pressurizer safety valves for PWR vender designs. These data should then be used to determine safety-valve failure rates.
The analysis should be documented in a report. If this requirement is implemented on a generic basis, eacn licensee should review the appropriate generic report and document its applicability to his own plant (s) . The report and tne d,ccumentation of applicability (where appropriate) should be submitted for NIC staff review by the specified date."
- Wita regard to Item II.K.3.7, NUREG-0737 stated
" Based on its review of best-estimate calculations performed by Babcock and Wilccx (B&W) , the NIC staff believes that the frequency of PORV challenges has been reduced using the revised POR7 and hign-pressure reactor trip setpoints and assuming that the anticipatory reactor trips function as designed. At this time, however, the staff is unable to make a quantitative judgment of the expected frequency. Tnerefore, licensees with B&W-designed plants should perform additional analyses of antici-pated transients which indicate the sensitivity of PORV challenges to (1) the variation in core physics parameters whien may occur in the plant i
cycler (2). single failures in mitigating systems; and (3) transients which do not actuate the anticipatory reactor trips. Analytical assumptions should include those specified in the plant final safety analysis reports (FSARs) . The results of these more-detailed and extensive analyses should be used to determine the expected frequency of PORV openings for overpressure transients. This frequency should be less than 5% of the total numoer of overpressure transients, thereby confirming the findings of the staff's review.
A ..U Franedin 4 o- n a ne nResearch n in. C. enter
, . _ , . , . _ - - - . , , . _ _ . . , _ . . - , _m-. -. -. . _ _ _ - _ . _ _ ~ _ , . _ - - _ _ - - . ,
TER-C550 6-410 i
The results of this study should be documented and submitted for staf f review by the scheduled date."
1.3 PLANT-SPECIFIC BACKGROUND In response to NUREG-0 737, Items II.K.3.1, II.K.3.2, and II.K.3.7, licensees of 3&W-designed plants endorsed and submitted to the NaC a report entitled " Report en Power Operated Relief valve Opening Probability" (4] . A preliminary review of the report resulted in the NBC's sending a request for additional information (RAI) to one of the licensees on Decemoer 16, 1981 (5] . The Licensee responded to the RAI in a letter to the NBC dated October 4
22, 1982 (6]. This TER evaluates the information provided in References 4 and 6, along with other information pertinent to the topic of a small-break LOCA from a stuck-open PORV or safety valve.
l t
I t
l I
O "Ill Franklin Research Center'
- A cm w w rr en =
I.
% +4~= y- -i-g , w - -- y -- r* .-,- y e --+- .er--- --w--- e e
TER-C5506-410
. 2. REVIEW CRITERIA The B&W response to NUREG-0737, Items II.K.3.2 and II.K.3.7, was evaluated against the acceptance criteria provided by the NRC in a letter dated July 21, 1981 (7), which outlined Tentative Work Assignment F.
Specifically, the response to NUREG-0737, Item II.K.3.2 was to contain the following information:
"1. The report shall list the actions taken by -Jte licenses to decrease the prob. ability of a small-break LO::A caused by a stuck-open PORV.
- 2. The report shall include an analysis of safety-valve failure este based on the past history of the operating plants designed by the licensee's NSSS vender. This may be a plant-specific report or a generic report showing the applicability to the specific plant.
- 3. The report shall have an analysis of the procability of a small-break LOCA caused by a stuck-open PORV or a stuck-open safety valve. This
. analysis shall evaluate the effect of an automatic POW isolation -
system. In evaluating this system, the licensee shall evaluate the potential of causing a subseqcent stuck-open safety valve and the
, overall effect on safety.
- 4. Actual operational data may,be used. The basis for any assumption ,
7 should be clearly stated and justified.
l 5. The automatic POW isolation system is not required if the licensee's actions constitute sufficient improvements to reactor safety in t reducing the probability of a small-break LOCA due to a stuck-open PORV or a stuck-open safety valve such that it is less t.%1tn 10-3/ reactor year, the median probability of a small-break LCCA S2 with a break size between 0.5 in. and 2.0 in. due to all causes."
Cn July 26, 1982 (8] and September 17, 1982 (9], the NBC clarified that the pecbability of a small-break LCCA due to a stuck-open PORV ce safety valve
~
did not necessarily have to be less than 10 per reactor-year. Instead, a comparison of pre-TMI and post-D1I data should demonstrate that plant modifications have reduced the probability of a small-break LOCA due to a stuck-open POW or safety valve and that this reduction should be sufficient to approach the WASII-1400 (10] median probability of a small-break LCCA S 2
i with a break diameter between 0.5 and 2.0 in.
{
l 4 J.J Framlin Resenten Canter 4c nh. r
a.
TER-C550 6-410 The response to NUREG-0737, Item II.K.3.7 was to contain the following
. _ , _ information: *
"1. Licensees with B&W-designed plants shall do an analysis of anticipated overpressure transients which give the sensitivity of POW cnallenges to the following:
- a. variation of core physics parameters during plant cycle;
- b. single failures in mitigating systems;
- c. transients whica do not actuate the anticipatory reactor trips.
- 2. The exoected frequency of POW openings for overpressure transients shall be less than 5% of the total numoer of expected overpressure transients.
- 3. Analytical assumptions shall include tnose specified in the plant FSAR."
4 '
..h Franidin Research Center a w w w amen m
. - . -r__ . . _ . , _ , . . , . - , . _ . . - - . , . . - - . . . _ _ - - . _ _ . - _ _ . , _ _ . . - - . . .
TER-05506-410
- 3. TECHNICAL EVALUATICN 3.1 REVIEW OF THE B&W REPORP FOR CObfLET."lESS In Reference 4, B&W approached the question of PORV reliability by first determining the probability of the PORY opening and ccabining it with the probability of the valve failing to close. The PORV opening procability was determined by two methods, an analytical approach and an operating data approach. To determine the frequency by which a PORV will fail to cicae on demand, a combination of operating date and analysis was employed. B&W deter:sined that the probability of a small-break LOCA from a stuck-open PORV was 3.04 x 10 per reactor-year, by the analytical methcd, and 4.7 x 10 ~4 per reactor-year, based on operating data.
B&W concluded the report of Reference 4 by saying:
"3oth the analytical prediction and the estimate based on historical data result in values for a stuck-open PORV from all causes which meet the requirements given in II.K.3.2. Note that no credit had been assigned for the operator closing the block valve given an open PORV. Analytical predictions (given proper auxiliary feedwatea. respons e) result in a value less than .01% of PORV openings,for overpressure transients (taking into account the most limiting non-anticipatory trips) and historical data sncws the frequency to be less than 1.6% which satisfies the criterion (less than 5%) specified in II.K.3.7.
Since the requirements of II.K.3.7 and II.K.3.2 are met with the current PORV configurations and set point, it is not necessary to address the requirement for an automatic block valve closure system per II.K.3.1."
With regard to the pressurizer safety valves, it was stated in Ref erence 4 that no estimate was made of failure rates because there were so few data points. In Ref erence 6, however, the frequency of a small-break LOCA due to a stuck-open safety valve was estimated to be 2.6 x 10 per year.
l 3.1.1 Small-Break LOCA Probability Calculations The first method used in Reference 4 to determine a recurrence frequency of a small-break LOCA from a stuck-open PORV involved determining the PORV s
4_ 2.2 Franklin Research Center a ca==== (w== a nm. ,
l N l
'll ,,
x '- . > .alL. ---- -
s TER-C550 6-410 opening probability, based ups. t analysis, and then multiplying by the probability that the POW fails to close. . . .
3 .1.1.1 PORV Coening Frequency In order to analytically determine a POW opening frequency, B&W first determined a PORV opening frequency during a loss-of-feedwater (LOW) or turbine trip event to be 3.9 x 10 per reactor-year, with a PORV setpoint of 2450 psig. This frequency was based on the assumption that the high pressure trip setpoint was 2300 psig with a standard deviation of 1.4 pai and that the actual setpoint at wnica reactor trip occurred was a random variable which is normally distributed. The small standard deviatien was based on the fact that the PORY and reactor protection system (RPS) actuation points are not completely independent in that they share a common source, i.e., sensor and inst;rument string. Thus, parts of the string errors were perfectly correlated and cancelled one another in the analysis. Other parts of the relevant string error were not correlated, and it was upon these tnat the 1.4 I
psi standard deviations were based.
In Reference 6, further justification was provided for using such a small standard deviation:
"We difference between the setpoints for the high pressure trip and the PORV actuation is of interest and one contribution to this difference is due to electronic module accuracies. Accuracy of individual modules were obtained from the manufacturer (BMCo) and are .1% of range. Se range of interest is approximately 1000 psi resulting in a value of .001 x 1000 or 3
1 psi. The standard deviation is derived as E i=1 (( Accuracy) x (Range)] 2 .
Both the pressure trip and the PORV share common modules that need not be included in this assessment as errors will cancel out (e.g., if module error is high then both the trip and the PORV are high but the difference is not affected) . mere are four non-common modules in these two strings, a bistable in the RPS channel and a buffer amp (from either aC3A-PTl or HC35-PTI), an inverter (RC 3 PIC) and a H/L monitor (RC 3-P58) in the POW string. D e SD is therefore /1 ps i + 1 ps i + 1 ps i + 1 ps i .
~
or 2 psi. The reference incorrectly used 1.4 as the standard deviation.
Note however that the standard deviation of the overall calculation h(2)2 + (2)2 + (27.52)2 is dcminated by the third term which is
! associated with the collove~r data. In fact the module accuracies can be as large as 10 psi without impacting the standard deviation."
l A :
Franxii j 4o- n e.Resear.en e . = Center l
l
TER-C5506-410 With regard to the trip setpoint of the PORV, a similar analysis was undertaken, which also assumed this trip to be a normally distributed random variaole. ':he assumption of normality for the actuation of either the high pressure trip or the POW was just an assumptions no data are available to justify or deny the validity. The aCs pressure rise above the RPS high i pressure trip setpoint (referred to as " pressure rollover") during a LOW or turbine trip was determined by a combination of plant data and engineering analysis. Pressure rollover data from the operating plants were compiled from available data. However, these data points represented situations in unich the POW could open, thus decreasing the amount of pressure overshoot.
Therefore, it was necessary to correct for the PCRV opening, since the situation of concern was when the POW remained closed. This was acccmplished by conchmarking the CADD code to a transient in which the POEV was isolated.
Af ter satisfactcry duplication of this transient, the code was rerun, acdeling proper functioning of the PCRV. The resulting pressure correction to the rollover data was 17.4 psi. The rollover data were tested and were statisticaaly acceptable as normally distributed, with a mean of 9.2 and a standard deviation of 27.52 psi.
Using the above data and assumptions, a Monte Carlo simulation of the relation
{ POW - RPS - EXCESS - BI AS = SAMPLE was conducted. The terms in the above relation are defined as folicws:
l PCW = POW setpoint, a normally distributed random variable
- RPS = Hign pressure trip setpoint, also a normally distributed random variable l EXCESS = Pressure rollover, a randomly distributed normal variable .
BIAS = A constant (17.4 pei) defined by analysis which compensates the rollover data for the fact that the POW will remain closed.
Six thousand sample values of the above alogritha expression were calculated using the SAMPLE code. A negative value of the above expression O O Franxiin .nw.
Research Center 4a r, n
_ ~ _ , . _ _ - - _ ___ _ _ _ - , . _ ._ _ - - . _ - , _ . . . - _ . --
6 TER-C550 6-410 implied that the PORV opened. In the computer trials, no negative values in 6000 instances were observed.
It was then assumed that the random variables described above were independent in the probabilistic sense, so an analytic approacn was applied. !
The sum or difference of several independent normal distributions is also a normal distribution with mean equal to the algeoraic sum of the means and with standard deviation equal to the square root of the sum of variances. In this case, the mean was 2450 - 2300 - 9.23 - 17.4 = 123.37 and the standard deviation was 27.59. The frequency by which the PORV will open during over-pressure transients, under those conditions, was calculated to be 3.9 x 10 per reactor-year. In Reference 6, it was noted that this frequency was incorrectly stated as 3.9 x 10 events per reactor-year rather than 3.9 x 10 events per overpressure transient.. Estimating 10 transients per year, i
it was determined in Reference 6 that the frequency of PORV openings during
~
- overpressure transients was 3.9 x 10 events per reactor-year.
In order to determine the frequency by which the POW will open due to any cause, the above calculated frequency of opening during overpressure
~
transients had to be added to other possible initiating events. B&W grouped PORY opening events into five categories with annual probabilities as follows:
9 Per Reactor-vear l 1. PORV opening on overpressure transient 3.9 x 10-6 l
- 2. PORV opening on transient with delayed 1.4 x 10-3 auxiliary feed f 3. PORV opening on operator action under 1.58 x 10-2 A2CG guidelines l
- 4. PORV opening due to instrumentation 5 x 10~3 control faults
- 5. PORV opening from additional considera- 1.8 x 10~3 tion from II.K.3.7 l
- Total 2.40 x 10~2 l
- In Reference 4, brief discussions were provided regarding the derivation l
- of each of the above probabilities. In Refer
- 7ce 6, the additional information l
l O 'N F anklin Research Center
{ 4c-===.# w wn %
l
. .._. . .~_.___ ___ _ _ _ , _ . _ . . . _ . , . . . . _ _ _ _ . . . _ _
TER-C550 6-410 was submitted in specific areas in response to NaC questicns. A brief sumnary of these derivations follows:
Transient wita Delayed Auxiliarv Feed
~
The value of 1.4 x 10 per reactor-year was determined by comoining EPRI data for loss of main feedwater and loss of offsite power (ICCP) with B&W auxiliary feedwater unavailability. It was assumed that if auxiliary feedwater was unavailable, the PORV would open (i.e. , probacility of 1.0) .
In Reference 6, a revised value of 7.6 x 10 per reactor-year was l deter:sined by updating the analysis with data obtained since Reference 4 was completed.
Oceration Action Under ATOG Guidelines The demand on the PORV, given a steam generator tube rupture, depends upon the availasility of offsite power. If pcwer is available, only one PORV opening is required. If offsite power is not available, as many.as 23 PORV openings are required. However, since there is no causal conne.: tion between steam generator tube rupture and LCCP, the WASH-1400 value of LCCP (1 x 10" ) was used.
Since there have been no tube ruptures in B&W experience, a chi-square
~
l i
50% confidence value with zero failures of 1.54 x 10 per reactor-year was used. The initatior event frequency was calculated as follows:
1.54 x 10-2/rx-yr x 1 demand. = 1.54 x 10-2/ex-yr 1.5 4 x 10-2/rx-yr x 10-3 x 23 demands = 3.54 x 10-4/rx-yr 1.58 x 10 '/ex-yr Instrumentation and Control Faults i
Reference 4 estimated POR7 openings due to instrumentation and control
~
failures at 5 x 10 per reactor-year. Reference 6 justified this value by taking the following failure rates that will cause tne PORV to open from IEEE Std 500-1977:
I A
..Enxiin 4 cm .e Researen.
w r- Center
l t
TER-C5506-410 '
Pressure transmitter fails high 0.25 x 10-6/hr Bistable functions without signal 0.206 x 10-6/hr I/E converter fails high 0.31 x 10-6/hr Summer module fails high -
0.31 x 10-6/hr Short across contacts 0.01 x 10-6/hr The sum of these failure rates (approximately 0.8 x 10 per hr) was I
multiplied by the number of hours per year, of plant operation (assuming an average annual plant availability of 80%) to arrive at the initiator frequency
~3 of 5 x 10 per reactor-year.
However, Reference 6 stated that the value of 5 x 10 per reactor-year ~
had been incorrectly summed with the other categories of Reference 4. In this case, since the PORV opens due to instrur int failure, it must be assumed that tne failure will keep the PORV open'. Therefore, these failures must be treated differently than the other initiator frequencies which assume a calculated closing probability.
Cvercooling Transients That Initiate High Pressure In4ection (HPI)
In Reference 4, it was estimated that the POW opening frequency from '
overcooling transients that initiate HPI (with operator failure to throttle or terminate flow before the PORV setpoint) is 1.8 x 10 ~
per reactor-year.
In Reference 6, however, the frequency was revised to be 8.4 x 10 per reactor-year, stating i
"The prediction value of 1.8 x 10~3/rx-yr for overceoling events that
! initiate HPI and result in subsequent PCRV actuation was determined from operating experience and operator failure probabilities. The calculation consisted of 8 overcooling events in 392 reactor trips x expected number i
of reactor trips per year x probability of operator failing to throttle
{
HPI given the overcooling event. There have been 8 HPI initiations due to overcooling events (exclusive of PORY initiated events): Oconee-1, 02/14/78, Davis Besse-1,10/23/77, Rancho Seco, 01/05/79, TMI-2, 03/29/78, TMI-2, 04/23/78, TME-2,11/07/78, THI-2,12/02/78. In addttion there have been two events that could have started HPI (according to pressure trace of transient) but did not. Conservatively including these two events (Oconee-105/05/73 and Davis Besse-1 11/29/77) results in 10 events in 392 RX-trips. Seven of these were due to auxiliary feedwater cooling. As pointed out previously, ANO-1 is upgrading the E2W system
' which will preclude auxiliary feedwater overcooling. The expected i
" equency of overcooling events then is 3/392 per reactor trip Reactor i
( gh '
..% aFranklin ca .e N Researc.h.C.
n . enter 1
TER-C550 6-410 trip frequency in 1981 for ANO-1 was 6. The capacity factor was .658 for 1981. Since other calculations in this report assumed a future capacity fac ce of .80 the trip frequency used here is 7.3 (i.e., 8/.658x6). The operator failure probability is 1.5 x 10-2 demand (see attached event tree) . The overall probability is therefore 3/392 x 7.3 x 1.5 x 10-2 8.4 x 10-4/rx-yr."
In summary, the determination of a PORV opening frequency, from all ,
causes, is as follows:
Reference 4 Reference 6 Event Per Reactor-year Per Reactor-vear Overpressure transients 3.9 x 10~6 3.3 x 10-5.
Transients with delayed auxiliary 1.4 x 10-3 7.6 x 10~4 feedwater Cperator action under ATOG guidelines 1.58 x 10-2 1.58 x 10-2 Instrumentation and control faults 5.0 x 10-3 1. 7 x 10 -3 *
- Overccoling transients that initiate EPI 1.8 x 10 -3 8.4 x 10-4 Total 2.4 x 10-4 1.9 x 10-2 3.1.1.2 Small-Break LOCA Calculations In ceder to determine the probability of a small-break LOCA from a stucx-open POKV, S&W multiplied the total POR7 initiator event frequency of Ref erence 4 (2.4 x 10-2 per reactor-year) by the probability of the POR7 failing open
~
on demand. S&W estimated the failure rate to be 2.1 x 10 failures per demand, based upon operational dats of 5 (non-installation) failures in 250 recorded demands, combined with an analysis of other mechanical failures wnich
- Assumes 10 overpressure transients per year.
- Includes only the component attributed to failure (high) of the pressure transmihter. The remaining ins trument failures tapproximately 5.6 x 10-3 per reaptor-year) are treated separately because these initiator events must also be, presumed to result in failure of the POR7 to close.
Aik ...; Fmnidin Research Center a c-w ni r,ww.n m
f TER-C5506-410 would prevent valve closure. The probability of a small-break LOCA from a stuck-open PORV was established ass (2.4 x 10 -2 events per reactot-year) x failures per demand) = 5.04 x 10 -4 per reac'or-year.
(2.1 x 10 t This estimate took no credit for operater action to terminate the LOCA by closing the block valve.
In Reference 6, however, it was stated that the probability of an open PORV flow path was the product of a stuck-open PORV and a failure of the block valve to close. The formula used in Reference 6 was:
((Sum of PORV Initiator Events) x (Failure of PORV to Close)
+ (POR7 Openings due to Instrument Fault)]
x (Failure of Block Valve to Close] = Open PORV Flow Patn The following values were applied to the equation:
Sum of PORV initiator events: 1.3 x 10-2 per reactor-year - from Section 3.1.1.1 above Failure of PORV to close: 1.7 x 10-2 per demand - This figure was revised frcm the 2.1 x 10-2 per demand of Reference 4 by including data from C-E plant experience and EPRI tes ting .
POHV open'ing due to 5.6 x 10-3 per reactor-year (0.3 x ins trument faults: 10 -6 x 0.8 x 8760) from Section 3.1.1.1 above, excluding the component
' attributed to the pressurizer pressure transmitter failing high.
Failure of the block valve to close: 2.4 2 x 10-2 per demand - 2.22 x 10 -2 per demand for valve-celated failure + 1.95 x 10-3 per demand for loss of motive power.
With regard to d2e block-valve failure rate, Reference 6 stated:
"Two dominant contributors were identified which would not allow the block valve to close. These were valve related faults including local power and the absence of 480 VAC motive power. The dominant instance of l motive power unavailability will occur as a result of LOOP (DsCPxdiesel l fails; 1.9 5 x 10-3 ) . The conclusion is conservative since if this condition existed (i.e. , LOOP) some of initiator events could not occur.
The block valve failure rate was determined using a Bayesian updating l 4 dl! Franxiin Researen Center a cm n a ne rwa.n mau.
TER-C550 6-410 procedure. A value of 8.1 x 10-4 for failure to close per demand was calculated frca Ref.11. This failure rate was used to construct a lognormal distribution (mean = 8.1 x 10-4, range factor = 10) , wnich was then used as the prior in the Bayesian analysis. A review of Reference 12 produced 34 failures in 1433 demands, whicn was then implemented to update the prior distribution. This resulted in a posterior mean of 2.22 x 10-2 with 5th and 95th percentile values of 1.63 x 10-2 and 2.89 x 10-2 respectively."
Inserting these values into the formula, it was concluded that the probability of having an open PORV flow path (i.e. , the procacility of a I
small-break LOCA from a stuck-open PORV) was 1.43 x 10 per reactor-year:
[ (1.9 x 10-2) (1.7 x 10-2) + 5.6 x 10-3] 2.42 x 10-2 1.43 x 10-4/rn-yr
, Reference 6 was concluded by stating:
"The results of tais study indicate that the probacility of having an open POW ficw path is 1.43 x 10-4/Rx-yr. This value does not significantly impact ene small break LOCA probability for all causes. A sensi ivity study was also conducted in order to determine the effect of multiple PORV cnallenges with certain initiator frequency groups. As mentioned in the responso to Question 7, multiple PORV cpenings could
. cecur with causes 2 and 5. To illustrate the potential impact of these increased PORV demands, causes 2 and 5 were assumed to initiate 10 PORV openings. The results of this investigation demonstrate tnat the small break LO:'A pecbability would only be perturbed 2.1% in botn cases."
l 3.1.2 Small-Break LCCA Pr'obability Based on Ocerational Data Reference 4 stated that there had been ISO reactor trips of S&W units which would have lifted the PORV with the old POW / reactor trip setpoints, but that only 3 of these events would have reached the new POW setpcints. In addition, modifications had been made so that PORV action would new be precluded given those three initiating events, as well. Consequently, 3&W ccnservatively estimated that there could be one PORV actuation in the
~
l 45-year life of a Esw plant for a resultant probability of 2.22 x 10 per t
l- reactor-year. Combining this initiation frequency with the POW failure rate
~
in Reference 4 of 2.1 x 10 failures per demand, B&W concluded the
~
l probacility of a small-break LOCA from a stuck-open PORV to be 4.7 x 10 per reactor-year (2.22 x 10' x 2.1 x 10 ). ~
A -15
.0.1 Franklin Resesren Center
%*w
[
e TER-C550 6-410 In Reference 6, further justification of the estimate of one PORV actuation in 45 years was provided:
"From the pressure responses associated with various actual transients, three transients could have actuated the PORV with the revised setpoints (Oconee-3, 04/30/75, Rancho Seco, 03/20/78, Crystal River-3, 02/26/80) .
However, changes have been made to the plant that would have precluded the initiating events that caused these three transients. Even with the revised setpoints and.other changes it was assu.ned that if one event (not i specified) could occur in the 45 vears of operation then the probability of occurrence would be 2.22 x 10~2/Rx-yr. Although this assumption was made, a closer estimate of 0 events in 45 reactor years is believed to be a better indicator of future event frequency."
3 .1.3 Percentage of PORV ocenings curing over=ressure Transienes As indicated in Section 3.1.2, B&W stated in Reference 4 that 3 overpressure transients out of 190 would have lifted the PORV with the new setpoint. Further, plant modifications had been installed to preclude even these three events. Nevertheless, even with 3 of 190, less than 1.6% of the overpressure transients would cause a PORV opening. B&W stated that this satisfies the 5% criteria of Item II.K.3.7.
3.1.4 Actions Taken to Reduce POW Challenges In Reference 6, the following statements were made relative to reducing the probability of a small-break LQ'A from a stuck-open PORV:
"In addition to tIhe elevated PORV setpoint AP&L has taken many steps to reduce the ' probability of a stuckwpen PORV. These actions have been directed in three major areas: reducing the PORY challenge potential, equipment upgrades that rectify past problem areas, and an increased l emphasis on operator awareness.
The potential for challenging the PORV has been greatly reduced by
! incorporating two anticipatory trips and improvements in auxiliary feedwater control. ANO-1 has installed anticipatory reactor trips on loss of feedwder and on turbine trip. They are also upgrading the EFW system which will preclude auxiliary feedwater overcooling.
A review of B&W operating history has identified three transients which cculd have challenged the PORV at its elevated setpoint: (Oconee 3, 04/30/75, Rancho Seco, 03/20/78, Crystal River-3, 02/26/80) . An investigation into the failure mechanisms which caused these pressure A M Fesnkfin Research Center 4s n as n. n
. . . _ . . ~._ . - _ . . _ _ . _ , _ . . _ . _ _ . . _ . . _ _ . . _ - - _ . _ . _ _ , . . _ , , . . , , _ . . _ . _ - . _ _ - . .
. e TER-C5506-410 s
excursions has led to a variety of equipment upgrades. As a result, actions have been taken to avoid short circuits that would permit POR7 opening, to enable pecper response on loss of single power supplies to NNI control circuitry, and to upgrade power supply reliability. Changes have been made in the PCRV control system along with power upgrades.
In the event of a small break LOCA, measures have been taken to increase operator awareness to permit valid diagnosis and actions. The presence of an alarmed acoustic monitor at the outlet of the Posv will facilitate the action of d2e operator closing the clock valve. In addition APSL has implemented the ATCG program. The training the operator receives in this program is very extensive; areas wnich pertain to this discussion are:
For overcooling events the operator is instructed to throttle EPI to prevent pressarizer filling in the presence of both subccoled reacter coolant and the return of pressurizer level, Recognition of pressuri:er steam space breaks, Quench tanx pressure / level changes are an indicator of PCR7 discha rge."
l 3.1.5 Emall-Bream LOCA from Stucx-Ocen Safetv-Relief valve (SRV)
In' Reference 4, B&W stated there have been three cases in which SR7s have lifted on B&W plants. None of these resulted in failure of a valve to l
reseat. Because of the lack of data, no estimate of SR7 failure rate was made.
In Reference 6, however, an SR7 failure rate was established of 3.12 x 10-2 per demand for an SRV water discharge. This rats was calculated by using main steam safety valve (MSSV) data (1 partial failure in 2950 demands) ,
estimating a water-relief failure rate to be 100 times larger, and adjusting l for recent EPRI test data and an actual lift of an SR7 at the Crystal River plant in 1980.
In Reference 6, it was estimated that the predominant events leading to SR7 challenges were overcooling and overheating events. Overheating events were not considered due to the reliability of the existing auxiliary feedwater system design. Overcooling events were estimated by reviewing B&W transients l leading to reactor trip. of the 392 trips reviewed, a list was created of
- those which resulted in low pressure ESFAS initiations. Of these events, many were no longer possible because of plant modifications. Three remained applicable. Ihe probability that the operator would fail to throttle or O_ _ '
...h Franklin Researen Center ca aw w a
._ . _ _ _ ~ . . . _ . _ _ _ _ . _ . . . . - _ _ _ _ _ . . _
.. \
l TER-C550 6-410 terminate HPI flow prior to reaching the SRV lif t point aas estimated to be 1.49 x 10' per demand.
The resulting probability of a small-break LOCA from a stuck-open SR7 was then calculated to bes (3 events /392 reactor-trips) (7.3 reactor-trips / year) X (1.49 x 10-2) (3.12 x 10-2) = 2.6 x 10 -5 per reactor-year.
3 .2 EVALUATION OF THE B&W REPORT SUBMITTED IN RESPONSE TO NUREG-0737, ITEMS II.K.3.7 AND II.K.3.7 The following sections evaluate the information provided in References 4 and 6, and summarized in Section 3.1 above, as related to Items II.K.3.2 and II.K.3.7 of NUREG-0 737.
3.2.1 Small-Break LCCA Probabilltv Calculations The evaluation of the analytical approach to determining the probability of a small-break LOCA from a stuck < pen PORV will be conducted in two parts:
(1) an evaluation of the PORV opening frequency as developed by B&W, and (2) an evaluation of the probability that the PORV will remain open following any given demand opening. ,
3.2.1.1 PORV Ceenina Frequency
~
B&W determined that the PORV will be opened with a frequency of 1.9 x 10 openings per reactor-year. The frequency was determined by summing i
five initiator events. Three of these initiator events (transients with delayed auxiliary feed, instrumentation and control faults, and overewling transients) are minor contributors to the resultant frequency. Detailed discussion of these events has not been provided below because detailed information has been provided by B&W as to the derivation of the initiator frequencies and also because it is reasonable to assume that.these events would not have a major impact on the total opening frequency. The remaining two events (overpressure transients and operator action under ATCG guidelines) are further discussed below because (1) overpressure transients have
-ta-e$ Franklin Researen. Center 4 o==en w n. r
TER-C550 6-410 historically been tne major contributor to POR7 cpenings (before the setpoint changes) and (2) operator action is currently the major factor in the determination of a ccmoined initiator frequenef.
Overpressure Transients Prior to the TMI accident, overpressure transients were a major factor in the lif ting of the PORV. In Reference 4, S&W stated that there were 190 w ' 4 events on B&W plant.s that would have lifted the PORV with the pre ':MI setpoins.. 'n Reference 6, however, it was concluded that the pecbability of lif ting the to. a single overpressure transient was 3.9 x 10 per event. Estimating .. 'rpressure events per reactor-year, Reference 6 concluded that the recuru "requency of POR7 openings was 3.9 x 10
~
per reactor-year.
This dramatic decrease in the number of PORV challenges due to overpressure transients is the result of changes in both the high pressure f
reactor trip and POW actuation setpoints following the TMI accident. The pre- and post-TMI setpoints are as follows:
Pr e-T4I Post-TMI Cperating Pressure 2155 2155 PORV Opens 2255 2450 HP Reactor Trip 2355 2300 SW Opens 2500 2500 As can be seen from the above table, not only has the PORV setpoint been l
increased by nearly 200 psig, but also the hign pressure reactor trip, which previously did not occur until after the PORV had been signaled to open, occurs 150 psig before the PORV setpoint is reached. Both changes significantly affect the probability that a given overpressure transient will be terminated prior to reaching the revised PORV setpoint. It is raasonable to expect, therefore, that a change in setpoints of these two trip functions will greatly reduce' the recurrence frequency of a PORV opening during an j overpressure transient.
I Part of B&Wfs calculation that the probability of a PORV opening during an overpressure transient is 3.9 x 10 per event, however, depends upon a 4 -t9-hd Franklin Researen Center
% at m e -
TER-05506-410 standard deviation of 2 psig for the high pressure reactor trip setpoint and the PORV actuation _setpoint. B&W iustifies the use of such a small standard deviation by identifying four non-common modules in the two strings (each mcdule having a 1-psig accuracy) and by stating that the errors in the common modules exactly cancel each other in the analysis. A possible problem with this analysis is that the high pressure reactor trip signal is generated by a 2-out-of-4 logic circuit, whereas the PORV is actuated by selecting one of two possible input signals. Consequently, even if the pressure transmitter selected to actuate the PORV is one of the two transmitters that will eventually actuate the high pressure trip logic, a second and completely independent (of the PORV) transmitter must operate in order to cause the high pressure trip. Therefore, it appears that more than just the four medules identified by the B&W must be activated. Furthermore, if the transmitter selected to the POW is calibrated well below the remaining three pressure transmitters, peak system pressure will be muca closer to the POR7 setpoint than the combined normal distribution with a mean of 123.37 and standard deviation of 27.59 used by B&W in the calculations.
Taking the above logic one , step further, the authors of this m
> developed the following normal distributions by taking data from Reference 4 and also from the EPRI PWR Safety Valve and Relief Valve Test Program (11]:
l Standard Mean Deviation (os ia) (psig) P4ference(s)
Peak System Pressure 2327 27.5 4 PO W Opening 2450 21.7 4/11 SRV Opening (bo th) 2500 29.5 4/11.
Using these distributions, the authors of this TER calculated the probability of a PORV opening given a reactor trip on overpressure and also the probability of an SRV opening given a reactor trip on overpressure (with no PORV opening). The calculation essentially ir.volved a determination of the interference of one normal distribution with another (e.g. , the interference of the reactor trip distribution with the PORY distribution) . The calculations indicated the following opening probabilities:
l 4 .
J'2 4Franklin ch n om. r=i-n Research
. C. enter
, , _ , ., _.m,_-._..~.,.____,..._-__m. _
_ _ . _ _ , ..o _ , - _ _ . , , . . . . . - _ _ . . . _ _ _ , , . . . _ , . . - _ . _ . _ , . . . _ , . . , . . _ , , . , _ -
a-. ._4 _
g .a TER-C550 6-410 PORV 2.2 x 10-4 per overpressure event SRV 9.0 x 10-6 per overpressure event
- Assuming 10 overpressure events per year, PORV openings would occur with
~
a frequency of 2.2 x 10 per reacter-year and SW openings with a frequency
~
of 9.0 x 10 ' per reactor-year (assuming the PORV to be blocked or otherwise inoperative) . These values :nust be considered to be extremely conservative,
- particularly since they assume complete independence between the reactor trip and POW opening or because the effect of the PORV is not considered in the SRV analysis. Nevertheless, they show that the contribution of overpressure events to the total PORV openings remains small, even in view of the conservative assumptions involved.
C=erarer Action Under ATOG Guidelines Since there were no steam generator tube ruptures in B&W experience, the determination of an initiator for operator action under ATOG guidelines in Reference 4 depends upon the determination of a chi-square 50% confidence value with :ero failures. Calculating back from the B&W value of 1.54 x 10 -2 events per reactor-year, it was determined that B&W assumed a. 45-year
! plant lif e. This is consistent with average plant life as estimated in other
, . parts of the report and is also consistent with the typical design lifetime of modern nuclear plants. In view of the fact that there have been no tube ruptures in the history of S&W plants, this approach appears to be a reasonable method of establishing an initiator frequency.
I B&W has assumed that, in a steam generator tube-rupture event, there will be one demand on the PORV if offsite power is available (e.g., reactor coolant pumps are available to transfer heat to the steam generators) and that there will be 23 PORV demands without of fsite power (e.g. , natural circulation conditions) . With regard to this portion of the analysis, several observations should be mader i
- 1. Although it is possible to complete a post-tube-rupture cooldown and depressurization with only one PORV cycle (of fsite pcwer available) ,
it is core likely, and experience at PWRs other than B&W units indicates, that multiple PORV cycles may be necessary. It does not A fSnklin Resesreh Center 4 c== a. r -
{
w, -,w-e-- -
e-, . ,..,y- -
. TER-C550 6-410 appear unreasonable to assume that as many as five cycles may be '
required.
- 2. In a situation where as many as 23 POW cycles may be required in a fairly short period of time, a failure rate somewhat larger than the I
predicted failure rate for the POW should be used. Reference 6 indicates that a sensitivity study by B&W shows that the failure rate is increased by 2.1% af ter 10 cycles of the POW. It is not logical
' to assume, however, that -Jnis value can be linearly extrapolated to a value such as 23 cycles. At some point, the increasing number of cycles in a short time period will greatly affect the probable failure rate. It appears that an assumption of a failure rate increased by 50% would not be unreasonable.
- 3. Finally, the probability of the operator's potential failura to close the block valve in a situation in wnich the operator has opened the POW is not as great as the condition when the POW has automatically lifted. Since the operator is controlling the POW, he is mucn more likely to be aware of the effect of the POW on plant parameters, and therefore he should immediately be aware of a malfunction of the POW. The expectation that an operator who is controlling a POW which sticks open will immediately shut tne block valve was recently demonstrated during a steam generator tube rupture at the Ginna Power Plant. Consequently, when considering the probability of operatcr action to shut the block valve in response to a stuck-open POW under steam generator tube rupture conditions, only the probability of mechanical or electrical valve failure need be considered.
In view of the above discussion, the number of POW openings per reactor-year due to steam generator tube ruptures is calculatad as the frequency of a tube rupture event times the pecbability of retaining or losing offsite power:
1.54 x 10-2 x 0.999 x 5 = 7.69 x 10-2 1.54 x 10-2 x 0.0 01 x 23 = 3.54 x 10-4 7.73 x 10-2 In view of the above discussions, a comparison of the B&W-calculated PORV opening recurrence frequency and that recalculated in this TER is provided below:
B&W TER Event (Rx-yr)~1 (Rx-yr)-1 l
Overpressure transients 3.9 x 10-5 2.2 x 10-3 Transients with delayed 7.6 x 10 d 7.6 x 10-4 auxiliary feedwater l
i 4 Rese
.DJ4 FranWin ow m.w. arch C. enter
.. - . . , , _ . . _ . . _ . _ . . , .._._ -- - . ~.___. _ ,_._-._~ _ _ . . . - . , _ . . - _ _ . - . - _ - - . .
TER-C550 6-410 B&W TER Event (Rx-yr)-1 (Rx-yr)-1 operator action under ATCG 1.58 x 10-2 7.73 x 10-2 guidelines Instrumentation and control 1.7 x 10-3 1.7 x 10-3 faults overcooling transients that 8.4 x 10-4 8.4 x 10-4 initiate HPI Total 1.9 x 10-4 8.28 x 10-4 3.2.1.2 Small-Break tcCA Calculations Ha'?ing determined a POW opening frequency, B&W calculated the probability of a small-break La:A from a stuck-open POW by multiplying the opening frequency by the probability of the PORV sticking open and the probability of the blocx valve failing to close.
~
B&W calculated the failure rate of the PORV to be 2.1 x 10 per de:nand using actual valve operating history. In Reference 6, this value was revised
~
to 1.7 x 10 per demand by including additional operating history from Comcustion Engineering-designed plants and EPRI test data. mese failure rates can be justified because they incorporate actual valve operating experience and because they are in keeping with similar ratas established in other studies and reports relative to similar relief valves. Use of the
~
failure rate of 1.7 x 10 is considered both justified and appropriate.
In the case of potential failure of the block valve, hcwever, B&W does not consider the possibility that the operator could fail to recognize the
~
PORV as having failed open or, even if the PORV failure is correctly diagnosed, that the operator could ' fail to take appropriate action. The potential failure of the block valve to close on demand must be combined with the potential of the operator to fail to initiate the demand, for all cases except those for which the operator is intentionally operating the PCRV.
~
In Reference 6, a failure rate of 2.4 x 10 per block valve demand was determined by B&W based upon two dominant contributors to failure: valve-related failures and absence of 480 7AC motive power, the contribution of 4 '
...a Franidin Research Center a c.a .# n r, a
. TER-C5 50 6-410 valve-related failures was determined using a Bayesian updating procadure.
The prior in the Bayesian analysis (8.1 x 10~4 mean failure rate, taken from Reference 12) was updated with data of 34 failures in 1433 demands (13] .
Although B&W did not thorougnly justify the use of the 8.1 x 10 mean failure rate' prior, the update data of 34 failures in 1433 demands totally dominate the Bayesian prediction, and a classical statistical estimatie predicts a failure range similar to the B&W analysis. Consequently, the B&W anaysis is considered to provide a reasonable assessment of block valve reliability with respect to valve-related failures. '
With regard to the absence of motive pcwer, B&W makes the statement that dcminant instance of motive power unavailability occurs as result of a loss of offsite power (IDCP x diesel failure) . B&W states that this is a conservative conclusion because under LCCP conditions, scme of the initiator events could not occur. This statement is interpreted to mean that under LCCP conditions, some of the valve-related failures (associated with electrical faults) could not occur. There are undoubtedly certain block valve failure modes whica
. require the unavailability of electrical power, and the combination of motive power failure and valve-related faults in the B&W analysis is considered to be conservative. In summary, the block valve failure rate of 2.4 x 10 per ~
demand determined by B&W appears to be a reasonable assessment of failure probability. However, when combined with the probability of the operator failing to close the valve (1.5 x 10 derived from NUREG/CR-1278 (14]), a ~
i ccabined probability of the block valve remaining open per event of 3.9 x
~
10 is obtained.
As shown in Figure 1, B&W determined the recurrence frequency of a maall-break LOCA from a stuck-open PORV by multiplying the probability of the POW remaining open with the probability of the block valve remaining open.
The probability of the PORV remaining open is the sua of openings due' to instrumentation faults and openings due to other demands multiplied by the probability of the PORV to fail to close. Figure 1 shows that B&W determined the probability of a small-break LOCA from a stuck-open PORV to be 1.4 x 10 per reactor-year.
p CJ Franklin 4 :> w n Re. r===search
= Center
- . . _ . . . . , - . . _ . _ _ . _ _ _ . . . _ . _ . _ . _ , . , _ . . _ _ _ . _ . _ _ _ , . . . , ~ _ . _ . . . _ , _ . . _ _. _ . . _ _ . . _ -
e .
nn a uo$8fO 5E uV A A L P
V n
oE l S AO RtC g E
. PO E
V ;
OT L
A V L K A a C C E l
O l i E H -e f VO l L
B I
lA8 l AN
=
2 N a CVA E f 2 E -
P 0 1
M O a E 4 ES G _
2 V t 4 U 4T N
AC t H '
VO O N. P S 'es KT OSl cd .
Cs nl
_ Ol A a 4 L
a El e B fA vil E O a M Ib El I I A V s a
-g n
l o a lWnA g i
4 OO ,
t 2 MPAV a g
g a t
i l t
u l
ll EE t i l c
Il A
n(i uit l u 'o a
_ P sMI s
C s0S
- W tt f4 M i U
O iAA I E
- e i PHF l
a i I D 4
V I
F 1
1 1 a f t L n E
P O KC J e O PAEt Ff 3 r
8 u
'0 g e 1 i n n '
F i Olt 'o A40 OG I i a . I 3e
-e t il I a l
= A N tC A rA s
s M 4 O S
EE D it P
_ V O
_ H U
P s
-0
'01 3
_ m =
8 S G
_ 3 I 4
- D N lE VD
- V E
P a AE r
t i L E e t
i O SET i
_ o D aDx t
s
_ l P a
l A
Aiu nt i t
i ti A Et M
_ P E W O D a
-e t
8 a 8
6 E
- it e
t uS t NS bl a
- _ I St e auOsL Et s
_ t i t h
J HA .
t AA P a: o 5T F 4
HAs t s N l s VI O
8Nu. I i
ty Es g3:a:r Or3 o
>n a." I
TER-C5506-410 Figure 2 shows the same calculation revised in accordance with the comments and discussion developed in this section of the report. It includes changes made to the initiator frequency for overpressure trancients (2.2 x
~3 ~
10 rather than 3.9 x 10 ), the initiator frequency for opera:or action under ATOG guidelines (7.7 x 10" rather than 1.58 x 10
~
): and the probability of the operator failing to shut the block valve .chen requirad (1.5 x 10-2 per demand rather than 0) . As result of this recalculatice, ale probability of a small-break LCCA from a stuck-open PORY is 2.53 x 10 -4 per reactor- year. While this value is semewhat larger than the value determined by B&W, it is quite comparable and is also well below the median WASE-1400 frequency for a small-break LCCA of 1 x 10" .
3.2.1.3 Multiple POW Cycles With regard to the question of more than one PORV opening per initiator event. the folicwing statement was made in Reference 6: ,
"As mentioned in the response to question 7, multiple PORV openings could occur with causes 2 and 5. To illustrate the potential impact of these increased PORV demands, causes 2 and 5 were assumed to initiate 10 PORV openings. One results of this investigation demonstrate that the small break LOCA probability would only be perturbed 2.1% in both cases."
Although not expressly stated, B&W has apparently discounted the i
possibility of multiple PORV cycles under overpressure transient conditions because of the new high pressure trip /PORV setpoints. Similarly, multiple ,
I cycles of the PORV are not a consideration under conditions of failure of the pressure transmitter. In the case of operator action (ATOG), the matter of multiple valve cycles has already been considered (e.g., 23 PORV openings if offsita power is not available) . This leaves cause 2 (transients with delayed auxiliary feed) and cause 5 (overcooling transients which initiate HPI) as initiator events with the possibility of more than one PORV opening.
B&W's statement that the investigation of these two causes with 10 multiple cycles eadt yielded a 2.1% increase in both cases considered each cause separately, regically, to determine the overall peccability of a E Frannlin Research Ce :ter a c- at m e
TER-C550 6-410
- W .
~ Wa2 t t.*a
's.
~
s
~
n a
-a>:
s
~
. t .
> a
- ~ , 9
~
's. a va I
v
= 5:
~
=
( ,i .
3 g c ., .-
6, i ew s e f ? - * . o 2 . 5 ~
~ -
3 n.
- =
9 m
3 C .
- s s
m
? o . 1
>-w . ::.
- s. -
saf - -
a ~
- C .i. c -
N
-(-
- w iw 8 -'
= *
= .
z3 55 e5<
s.
~
53<
~
. .:J Franklin Research Center acm=.nor ner n ..
e e .
TER-C5506-410 3
?
f*
7
't 5
s =c -
N 5 b8
.E __
i
.r_1
.r . 5sf -
5 I 's 2 w
? #8
~
f3 O si a:
\d
? w 2 -
> *5 *
~ -e 5
= 5 a
- 3: -
=
.c.e
-,, 5 s Q
3 2 2
g E=k
& a= =
y o % 3 g .oe ow ? >
,z -
A S. 3 y
L53 s1
~
- .E 3 2 = q:g=
- . 3 e
n, : R -
3 <
a n
$ =
'I. 2.e
!N ~ '
3 C_ s
, g
=
_ 2 N ,
l e . o l *3 w g hb.
i g ._ b ?. -
3 w
1 2
\d lai3 i
b I i
{
'S.
W g $2 ca .
i g .,-
15
=
3-2 mz "
l 3 5&
5-l t
l I
l D
l ' JUhranklin Researen Center J
A Consen af he fw moeuwse i
TER-C550 6-410 small-break'LOCA from a stuck-open PORV (considering possible multiple r
cycles) , multiple cycles during these events shoulcLbe_ analyzed concurrently.
The revised calculation of Figure 2 (discussed in Section 3.2.1.2) has been updated to include the effect of multiple POPV cycles. The results of this updated analysis are given in Figure 3. Figure 3 shows a small-break LOCA probability of 2.61 x 10 -4 per reactor-year, a 3.3% increase over the
, value calculated in Figure 2. These calculations show that multiple PORV cycles is not a primary concern for B&W-designed units.
The results of these POW small-break LOCA calculations are summari:ed below:
Probability of Small-break LOCA from Stuck-coen PORV (per reactor-year)
Figure 1 (B&W Calculation) 1.4 x 10-4 Figure 2 (Revised calculation 2.53 x 10-4 i
described in Section 3.2.1.2)
Figure 3 (Revised calculation 2.61 x 10-4 of Figure 2 with additional l consideration of multiple i cycles) -
l , 3.2.2 Small-beeak LCCA Probability 3ased on Coerational Data In Reference 4, B&W determined the recurrence frequency of a small-break LOCA from a stuck-open POKV to be 4.7 x 10 , based upon operational data.
S&W used a tseory that of the 190 reactor trips which had lifted the POR7 in the history of B&W plants, only 3 of these events would lif t the PORv with the new setpoints. Further, B&W reasoned that subsequent modifications had been made to B&W plants since these three events such that none of them would cause a POW lift today. Consequently, S&W estimated that if one PORV lift occurred in the 45 years of life of a S&W unit, the probability of POR7 openings would be 2.22 x 10 -2 per reactor-year. Combining this frequency
~
with the previously determined PORv failure rate (2.1 x 10 per demand) ,
the small-break LCCA frequency was calculated. It should be noted that this 4J0dO Franidin Researen Center 4 o-=en as n. er n -
- . . - - _ - - . , , , , - . - - - _ , _ . _, . , - . _ , .,_ , - .~..., - - - - - - _ , . . - - - ,
- h, 2 64 a le * .
e :-
OYL N D7 d POANitOW 0'Alta 3
cs b
Y" 1
3 It a te-s 3 3. ga" 1.# T I OPENFtOw PAllt gag g 3 2 3 = 80" b
t as a so a 3 g a gg-a i I OPEN POHy Orth alt OCK VAtVE l _
i OPEstAIOet tesV AClaOra usage M l.WalWL &sLCHAldeCAL f Ad S to FutM ft ATOG C4 eX,L VAtWE UNAVAR AastE f Ad Utlk p p . ge-*
e i a so-a l e s Bea
, 2 2= Wa ei to
- o A
u.
k'igure 3.
(l' age 1) Revised Calculation, considering thaltiple l'OlW Cycles t>e o
1 O
, e TER-C550 6-410 "W
f-
-i ?
8
. -5 : e
's. Ed -
o
- a 3- -*
o
- s.
O 3 -
- e Wgt ". %
s _ t= - ~
$ , 555 h S 8 1 e
's. W -.
- 15
- ~ ga .c.
so a g-oC. 3
- 3'
. m g55 . ?, 5* *
- l -
-si 52 7 I $
i 3 -
- a. -
e a = 3
= < at w, 7 O 6 u g ( =T.
"i 6
r :,12 = c o
o
" 3 8- -
g o. . M
= -
2-8 's 3
- s 5 . 2:00 s - o
. n
. '2 *
- U w ~e m=
A "3
's. .,
m
{
3 5.1
?.
< l w== -
2$ r ! 212~
T 7
f%
k Q- l j 2 . ~
a
- e
-g .
e m l Teo -
2 Y
s - l O i 3~5d 2 3; s ~
= 9
\% \ =_-
~=< ~
g g E m o
. w
's.
w A i
/4 = i 1
r g @== ;
5
- c"5 -
E*2
- 53 5= 1 A
uda Franklin on a er Th. 7 Research. Center
4 *
. - =
TER-C5506-410 I
calculation does not take credit for or.erator action to terminate the small-break LOCA by shutting the associated block valve.
' While this analysis is theoretically based en operational data, it should he noted that the actual data (148 reactor trips with PORV actuation prior to the TMI-2 accident and 3 reactor trips with PORV actuation af ter the setpoint changu.) are not used in the calculation. Rather, B&W estimates that, of the 190 reactor trips which would have lifted the old setpoints, only 3 would have lif ted the PORV with the new setpoints and even these would not cause FORV actuation today because of additional plant changes. Consequently, B&W is using an analysis to determine that, under present plant conditions, no actual PORV lif ts would nave occurred; hence, the estimate of one PORV lif t in a 45-year plant life is used..
l In view of the dependence of B&W's analysis on theoretical information, the authors of this TER determined that an independent confirmatory analysis, based on operational data to the maximum practical extent, was desirable. In order to do this, it was necessary to consider only the cperational data
- recorded after the TMI setpoint changes. Although there are only limited operational data since the TMI setpoint changes were initiated, there was one significant event (at Crystal River Unit 3 in February 1980), during which both the PORV and an SRV were opened.
It is not the intent of this TER to provide a detailed review of the i
Crystal River event. However, the following significant facts are relevant to this report:
- 1. The loss of a non-nuclear instrumentation (NNI) bus caused the PORs to open and remain open.
- 2. The erroneous control input signals (due to loss of NNI) caused a high pressure transient which resulted in a reactor trip (2300 psig) .
- 3. Decreasing BCS pressure caused initiation of HPI (1500 psig) .
- 4. *he operator isolated the open PORV by shutting the block valve.
5.' The HPI system filled the BCS to a solid water condition, lifting l code safety valve BCV-8 (SRV).
L l
i
_renidin Rese_ _ ._arch _ Center
. s -
TER-C550 6-410
- 6. After establishing emergency feedwater to one steam generator,.HPI ficw was throttled to establish an PCS pressure of 2300 psig, researing the SRV.
It must be noted that certain plant modifications have been made to preclude NNI power supply failures since the above incident at Crystal River Unit 3. These :nodifications were.made pursuant to IE Bulletin 79-27, published in response to the Crystal River event. Nevertheless, failures of NNI power supplies subsequent to these modifications, resulting in at least partial loss of non-nuclear instruments (e.g., partial loss of NNI at the cavir-Besse plant on June 24, 1981), indicate that PORV/SRV challenges as a result of NNI failures are still credible events. Ccnsequently, ccasideration of the Crystal River event in an analysis of post-TMI POR7/SRV operating data remains valid, although the results of the analysi's will be fairly conservative in view of the above-mentioned plant adifications.
Since the Crystal River was not a classical overpressure transient but a
was caused by failure of instrument power supplies, the event tree used to analyze these data has been specifically tailored to an overcooling situation (i.e. , where HPI operation challenges the PORV/SRV) . In this scenario, operator control of the bicek valve'and of HPI discharge pressure are key censideratioEs in challenging the PORV/SRV. (Note: Even thcugh the PCRV cpens on loss of NNI power supply, a probability is assigned to the PCRV reseating which presumes that NNI pcwer is regained within a short pericd of '
ti:ne. )
i As shown in Figure 4, the small-break LOCA pecbabilities resulting frca this analysis are:
SRV (eer ex-yr) PORV (per ex-yr) 3.17 x 10-3 7.12 x 10-5 1.55 x 10-3 6.38 x 10-5 i
! 4.72 x 10N 1.35 x 10~4 l The results indicate that the probability of a small-break LOCA frem a
~
stuck-open PORV or SRV remains within the WASH-1400 range of 10 to 10 O
d]!}4 Franklin m e n. r, Resear.en. Center
a e e
TER-C5506-410 w *
?
a 2 2 ? ?
- 2 2 s 2 s
~ - ~ n.
g $
- $_ m 2
- 2 3 2 g
- a . 3 % a. g = 2 a a =. -
I n
[n2 .
7 v
a a w
. =.
a
.. . . U w
a -
a o
- s Q j o
. s e: r v o g2- ,3 2 -
2 2 a
'. E 2 -
3 -
c Q
a u
,a-
- 7 w a =3 2 - e a
U ag . 2 -
o w
5
- c 7 W 1 ,2 2 o ga .A- m w
5 .-
< 1"se:
a .
9
- :35 5
5=w %
3E3 :
2
[35 I t
l i
f i
r 1
O_
.l.!,c' Franklin Research Center 4 w or N r = me.e
u r +
p on .
,3..
- .-', t - i
-( ,
f f
,y ;
'. 5
\1 l > + I s~h ~
4x, a r P TER-C5506-410 w
t 3.-
- #. . t . - :- r i ,t i
.;A; * ,
' 1 r <
c ,
3 NCTES TC FIGURE 4
./ . . , , y s.
, ( . i ' >
1., PCRVaccens ensNNI Failure: '
..q- q s do, of' events: .l_qcrystal River Unit 3, NUREG-0667)
No,qct reactor-years of. B&Wunit cperation frca April 1980 to June 1983:
- fj2 (assumers a 60'6 avuilability of 7d: nits throughout the peried)
No. of(cpenires, set--reactor year: 1/3.2 = 0.109 trents/rx-yr C. hicek valve Initially Open: Yes (Conservstive assumption)
Note: The -effect of assuming that the block valves are open less than 100%
of the time is_ to decryase the calculated pecbability of a small-break LOCA fr:m a stuck-open POK/ 'while the calculated pec ability of a small-break
, , UIA .frem a stuck -cper. SRV remains essentir.lly c: , nstant as folicws:
, ,i nitiEl 'si.cci valve Ceenino PCRV (eer ex-vr) SRv (cer ex-yr)
=
','f h i 100 % 'l.35 x 10-4 4.72 x 10-3 O ' /- 50% 6'.75 x 10-3 4.80 x 10-3 1
t-r 0% '
&0 4.39 x 10-3
-% t. , 3. 31cck Valve Shut: 0.961 (Figure 2, page 2 of this report) t
,ll ,
~
- 4. SRV ,Cpens: 0.985 gror all plants except Cavis-Besse)
,\
With the PCRV isolated, the SM/ setpoin is exceeded by HPI system pressure unless the operator throttles HPI sufficiently prior to reaching the setpcine. The aame pechability is; assigned for operator failure to j throttla. as-for ~ cperatcr failure to shut the block valve (1 5 x 10-2 per demand; see paga 2 of , Figure 1-on page 28 of this report) .
1: 10 (f ce !cavis-Besse) is ,
I ' -
EPI system pressure at shutoff head does not exceed the SRV setpoint at the Oavis-Sesse plant. A prcbability' of 1 x 10-3 is assigned to account for possible premature opening at elevated RPS system pressures.
- he. effect of,thirl reduces the pecbability of an SRV LCCA at the Davis-20sse plant to 4.79 x 10-' per reacter-year.
4.
s . t
~
The operator failure probability folicwing SRV opening is assigned the same
- value as the procability of failure to throttle EPI prior to SRV cpening.
1 E. SdV Fails to Cicse on comand: 3.1 x 10-2 (Reference 6).
-^ 7. PCRV fails to Cicse on Demand: 1.7 x 10-2 (Section 3.1.1.2 of this repor t) ~.
'. s-A. ,. ,
12 Frarklin Cen oman uResesren.wam,ter
- n. mn.en i
5 $
Y
[
,b. - -~ - -. , ,
TER-C5506-410
~~
for small-break LOCA. Furthermore, when considering that plant modifications have been made which will substantially lower the probability of the PORV opening and remaining open on loss of NNI power supplies and that less l
instrumentation will be lost on less of NNI power supplies, the small-break LOCA probabilities will be icwiered to the bottom of or below the WASH-1400 range. Since the modifications to the NNI power supplies affect the event initiating frequency, these.:nodifications will affect the small-break LOCA pecbabilities of both the PORY and SRV.
In summary, both the B&W analysis based upon operating data and the conservative independent analysis of this TER using only post-TMI setpoint change data indicate that the probability of a small-break LCCA from a stuck-open PORV or SRV remains within the WASH-1400 range for small-break LCCA.
3.2.3 Percentage of PORV Ocenings curing Over;ressure Transients In Reference 4, B&W stated that with new setpoints, the POR7 would have opened in 3 of 190 overpressure transients. B&W further stated that plant modifications had been installed which would preclude even these PORV openings. Nevertheless, even if the three openings were not precluded, the percentage of PORV openings per overpressure transient would still be 1.6%,
which met the 5% criteria of Item II.K.3.7.
It was noted in Section 3.2.2 of this report that since the TMI accident, there have been 59 reactor trips of which cne PORV actuation occurred. This information indicates that 1.7% of overpressure transients cause PORV openings. In addition, in Section 3.2.1 of this report,. a value of 2.2 x
~
10 PORY openings per reactor-year due to overpressure transients was determined out of a total of 8.28 x 10" PORV openings per reactor-year from all sources (other than certain instrument failures) . These values lead to a determination that approximately 2.7% of all openings will be associated with overpressure conditions.
l l The compilation of this information indicates that the criteria of Item II.K.3.7 have been met.
M Frank!!n Researen Center 4 ce===i as n. 7.w n was
. TER-C5506-410 3.2.4 Actions Taken to Reduce PORV Challenges _
As has been developed previcusly in this report, prior to the TMI accident, approximately 148 reactor trips resulted in PORV actuations. Since the TMI accident, there have been 59 reactor trips, 42 of which would have lifted the PORV with the old setpoints but only one of which actually caused the PORV with new setpoints to lift. Subsequent to that singular event, additional modifications were made to preclude a repeat of that lifting. The primary reason for the substantial reduction of PdRV actuations at B&W plants is the resetting of the high pressure reactor trip setpoint and the PCRV actuation setpoint. Prior to the TMI accident, during an overpressure situation, the PCW lifted before a high pressure reactor trip would occur.
The POW was used to reduce the numcer of high pressure trips. Since the setpoint revision, the roles of these two functicns have been reversed. The reactor trip now occurs 150 psig prior to reaching the POR7 setpoint. The
, PORY new performs only its primary function of limiting safety-relief valve operation under overpressure c.nditicas.
Other measures taken at B&W plants to reduce recurrence of PCW openin'gs, l such as increased operator awareness, installation of certain additicnal
! anticipatory trips, improvements in auxiliary feedwater centrola, and other l
l circuitry changes, have undoubtedly contributed to the cbserved reduction in PORV openings.- There is no need to attempt to quantify these centributicns to the overall reduction, however, since the everall reduction has been so substantial. At the same time, it is censidered that these additional measures have had only a minor effect in comparisen to that of the setpoint i
i changes. .The setpoint changes are censidered the primary centributor to ensuring that the probability of a small-break LOCA from a stuck-cpen POW
, remains belcw the WASE-1400 median frequency of a small-break LCCA.*
l
,
- Notes Reference 6 stated that ANO-1 had installed two anticipatory reactor
! trips as measures to reduce PORV challenges. It is not known if these
! trips have been installed at all asW units; however, this particular information is not considered relevant because of the substantial reduction in the PCRV cnallenge rate resulting from the setpoint caanges.
l
-3~.-
C'.')4::a FranMin
.a w n r,Research
. C. enter
TER-C5506-410 3.2.5 -
Small' Break LCCA frem Stuck-ocen safetv-Relief Valve (SRV)
In Reference 6, the pecbability of a small-break LOCA from a stuck-open
~
SRV was determined to be 2.6 x 10 ' per reactor-year per SRV. The calculation used a failure rate taken from main steam safety valve data, correlated to water relief valves. The calculation also relied upon data from overcooling events that resulted in high pressure engineered-safety-feature system activations which could challenge the SRV, if cperator actions to limit the pressure increase were not taken. The other initiator event which could lead to initiation of high pressure safety injection, overheating, was not
, censidered in this calculation because the improved reliability of the auxiliary feedwater system essentially precludes overheating events (see Section 3.1.5 of this report) .
In addition to this calculation by B&W, a calculation based upcn only operational data since the change in setpoints folicwing the T.MI event was provided in Section 3.2.2 of this repcrt. That analysis indicated an extremely conservative probability of a small-break LOCA from a stuck-open SRV
~
was 4.72 x 10 per reactor-year. The primary focus of that analysis was en overcooling events in which the SRV is lif ted by pressure head from the HPI system because this is tne only actual operational event to cause an SRV cpening since the TMI accident. In order to investigate the probability of an SRV opening during a possible overpressure transient, the authors of this TER conducted the theoretical' analysis described below.
The pecbability of a small-break LCCA from a stuck-open SRV initiated by an RCS system overpressure transient was determined by considering the interaction of three normal distributions; the RPS trip setpoint (with pressure overshoct), the PORV setpoint, and the SRV setpoint. The mean values and variances for each setpoint were previously described in Sectica 3.2.1.1.
The values are:
Mean fesig) Variance (esig)
RPS trip (with overshoct) 2327 27.5 PCR7 setpoint 2450 21.7 SRV setpoint 2500 29.5 Ad)4cmF anidin
.en.r Research
. C. enter
,,ww a_ -- -- -
l i
l TER-C5506-410 i
By integrating throughout the entire normal distribution of the RPS trip, the following precabilities were developed regarding PORV and SR7 openings:
PORV opening per overpressure RPS trip 2.2 x 10-4 SRV opening per overpressure RPS trip ,
(no PORV opening) 9.0 x 10-*
3cth PCRV and SRV opening per overpressure RPS trip 8.3 x 10-7 SRV opening given that the PCRV has opened 4.0 x 10-3 Applying these probabilities to the event trees of Figures 5 and 6, the folicwing small-break LOCA probabilities were determined:
Block Valves 100% Eleck valves 100%
Ocen (Figure 5) Closed (Figure 6)
Stuck-cpen PORV per ex-yr 1.5 x 10-5 Not applicable Stuck-open SRV per ex-yr 3.1 x 10-0 2.8 x 10-6 The results of this analysis must be tempered by the follcwing censiderations:
- l. The assumption of normality is at its worst in the tails of tne distributions, the areas of primary fccu;s of this analysis, 9
- 2. The dependence of the PCRV and the SRV is not accounted for, i.e., if the PCRV opens, the ICS peak pressure will probacly be less than if it does not open.
- 3. The variances (standard deviacions) of the PCRV and SRV setpoint distributions are based upon a limited amount of test data and are not based upon actual in-plant operating conditions (111 In addition, the resulting probability of a small-break LOCA frem a stuck-open PORV is based upon only the overpressure transient, which is not the controlling initiating event for this' accident. Consequently, there is no direct' correlation between the above results and those discussed in Sections 3.2.1 and 3.2.2 of this report. Nevertheless, in considering the pecbability of a small-break LOCA frca a stuck-open SRV due to an overpressure conditicn i.
23 Franklin Research Center a cs ner ne rrm en m
TER-C5506-410 t
in the SCS system, this analysis shows that the probability is quite small and that it is amply bounded by the predictions of WASH-1400.
The possibility of two says opening during a given initiator event (overpressure or overcooling) has not been considered (for plants with two Savs) because the opening of either SRV will terminate the pressure increase.
The possibility of both SRVs opening simultaneously is considered to be extremely remote.
O b
l l
t I
a Adj Franklin a w w w Resear.ch r, . % Center
TER-C5506-410 h
~
8 . . . .
's. i.~ w b.
8, .
.~ . . ~
< < < < < Q s 5 8 8 - 8 8 8 :
8 5 4 :
.s > ; > _ ;
8 8
. ; 3 a
w o
= a
_ $ .z.
e
=
o
= l o
= . e.
o
. g. ,
O C3 t'3 e C D
- e U io ?
>? 2 2 8aO
.d3 .
., I a e 3 a@
ew e >
b#4
.c >
s J J
>w *
=
I Q hd 5 0 a :
o:
a G M w
0 "C 25 o *
= .
o*
~
3 2 '.
so O
'e
- - bf ,
a ; - I o
. an
- w C
G O
- * *E
,e= . .
.. e o
==
02 =*
g *e4 V
w d e ? O 2: S w
@ %. =
25 :
- I
. 3.3
.u
. d ~
l 7= "
l *
>s*b
$3 .3
- e 5 u w
3
$" 1 w: 1 EI. i o
O._
e C.") Fran) din Research Center s m a % r anan raamae
TER-C5506-410 w
3
+
5 *
.r
$ 8 < < < <
.s 8 -
8 <
. a ; .'
8 8 w 8 w I 9 9 i -
8_ :: -
- - 2 5 3 9 5 2 2
- 6 - .E %
. 9 3
IO Oh
?.
- D
?.
] .-
5
- w' a $.
r hN ?
go ?
2 4 s' G ~
- a. - 0-m
~
O
= , .
es( ,
2 5" o
? be 30
- QydN a .
-t. 's
- s. v.
N s ..:
e , aa '.1 53 ,
- Jy~ ; 2 e ~ M
- ] 's 3 0
- A
' g QO w== "M b
>Q ? A E*
3 kk- A v "&
u
~ k
. ,$ 5 y ..
v-k .8.
- g$r*3 ~
aC I ,
oa w
5 3 c.n W mEd 5
wz
-w a.
a
~
1 5 -
1
...; Franklin a r.m a Researen.C.e.nter
= N r, .
- d. O
. e TL'i-C350 6-410 NOITS TO FIGUPES S AND 6 Overpressure Events: 10 per year - Conservative B&W estimate (Reference 6)
Block Valves Initially Open: Yes (Figure 5)
No (Figure 6)
. PORY Cpens: Yes: 2.2 x 10-4 per demand Determined by integrating (frca negative to positive infinity), the expression of the normal distribution of the RPS trip pressure (PRPS) times the probability that, for any RPS trip pressure, the PORV setpoint (Ppog) is less than PRP S . The probability that '
Ppog is less than PRPS for any given reactor trip was expressad as:
e 9/
e d:
- 1 ~ ~"
?ORV o -
wnere = = 'PORY - y?ORV PORY SRV Cpens (POW Cpen): Yes: 4 x l'0~3 per demand Determined by integrating (from negative to pcsitive infinity) , the expression for the normal distribution of P RPS times the pr bability that for any PRPS, both PPCRV and P 3 gy (the SRV setpoint) are less than P3pg, divided by the probability of PPCRV being less than P3 pg, i.e. ,:
EN **d <
PORY RPSlRPSk SRV RPSlRPSf3 P#fPOR7 R?S RPS SRV Opens (No PORV Cpening): Yes: 9 x 104 per demand Determined similarly to the calculation of PORV opening (above) , using the normal distribution for the SRV rather than the PCRV.
SRV Closed: No: 3.1 x 10-2 per demand (Reference 6) j POW Closed: No: 1.7 x 10-2 per demand (Reference 6)
Block Valve Closed: No: 3.9 x 10-2 per demand (Figure 2, page 2 of this repcrt)
~43-4 dSW Franidin Research Center 4c== .eni e
. _ _ . . _ . . . . _ . . . _ , _ - . _ _ _ . . - . _ . . . . ~ ... . . _ _ _ _ - , , , _ . _ . . _ . _ . _ _ . , _ . . _ _ - , . _
TER-C5506-410 I
- 4. APPLICABILITY '
4.1 APPLICABILITY CF THE B&W REPORT In considering the applicability of the B&W report, two main facters need '
to be considered: (1) the type of PCRV installed in tha varicus B&W units and (2) the value of the revised setpoints at the different units. In the first instance, all B&W units have De- 1ser PCRVs insallad except for the Davis-Besse plant, which has a Crosby PORV installed. With regarri to the revised setpoints, all B&W units have reset the PORV to 2450 psig with exception of the Davis-Besse plant, which is set at 2400 psig. Each of these facters is considered separately below.
4.1.1 Tvte of PORV In Reference 4, B&W made the following statement regarding the type of PORVs installed at the different units:
" Note that all plants except Davis-Besse (Crcsby POW) have Dresser valves; however, the entire B&W operating plant experience was used to arrive at a generic PCRV sticking open probability as folicws: There have been ten stuck open PORV events, five of which could be classified as mechanical failure of the POR7 (the other five were basically installation errers) . Using all of these five failures in determination of future frequency is censidered conservative since two of the failures
- CC-3, 6/13/75 and C3-3, 11/75) were rectified by design changes, another (TMI-2, 3/28/79) cause is unknown. CC-2, 11/6/73 could be considered as a burn-in failure and the DB-1,10/13/77 event is a Croshy valve. Using five failures in 250 demands results in a value of 2 x 10-2 to fail to reclose on demand. This value is considered conservative not only due to the inclusion of all five failures but also the number of demands is probably much higher than 250. There have been 148 documented PORV '
openings on reactor trips; however, there is not a listing of PORV demands when the reactor did not trip (e.g., ICS runback) nce is consideration given to transients that could have actuated the PCRV numerous times during an event. The value of 250 demands is conservatively used here. An analysis was also performed to include values for other than mechanical failure that keep the PORV open. The results of this analysis is summed with the mechanical contributcr (2 x 10-2/d) 10-2/d)."to arrive at the value for failure to reclose on demand (2.1 x The aoove statement by B&W does not appear to address the rationale for develcping a single failure rate for both the Dresser and Crcsby valves. The M
dj Frankiin Research Center s o- a w n. re
. , _ ,.~ __. _
TER-C5506-410 implicaticn is that because a.censervative failure rate has been determined, it is representative of both of the valves concerned. This implicaticn would be correct if the failure mechanisms and the failure characteristics of each of the valves are similar enough that the ccmcined data adequately represent 4
each valve individually. Ecwever, no information has been presented relative to such a conclusion.
The 148 dccumented PCKV cpenings on reactor trips discussed by S&W are identified in NUREG-0667 (15] . A review of Reference 15 shows that 29 PCRV openings on reactor trip ccrurred at the Davis-Besse plant prior to the TMI accident, one of which resulted in a PCRV failure (Octcber 13, 1977). In crder to cecpare these data to the B&W-determined PCK7 failure rate of 2.1 x
~
10 per demand, however, the nu=ter of PCRV cpenings =ust be increased cy the facter 1.68 (250/148), unica accounts fer PCRV openings that did net result in reacect trips. Consequently, 49 pre-TMI PCRV openings at the cavis-Besse plant (29 x 1.68) =ust be considered. Cne failure in 49 openings yields a failure rate of 2.04 x 10~ per, demand. This value coincides closely with the S&W value determined using the comcined Dresser and Crosby data. In view of the fact that 3&~p considers the 250 openings to be a
, ecnservative value, it is concluded that ccmaining the cresser and Crescy data and determining a single 20RV failure rate is technically justified and that tne S&W report is applicacle to both the cresser valves and the Crcsby valve at the Davis-Besse plant.
4.1.2 PCK7 Set:cints In Reference 4, B&W determined that the probability of the PCR7 lif ting
~
during a loss of feedwater is approximatey 3.9 x 10 per overpressure event for PCRVs with a setpoint of 2450 psig. For the Davis-Besse plant [16] , this
~
probability was determined to be approximately 3.9 x 10 per overpressure event, since the cavis-Besse plant has a PCRV set at 2400 psig. Assuming 10 cverpressure events per year -(as assumed in Reference 6), the pecbability of 1
O -
.2d sFranMn em w w r, Research . . C. enter
- - - _ _ _ - ~ _ _
l
. TER-0550 6-410
. the POBV lif ting during a loss of feedwater is approximately 3.9 x 10 per ~
reactor-year. Substituting this value into the determination of the POR7 cpening frequency shown on page 13 of this report, the FORV cpening frequency for the Cavis-Besse plant beccmes:
Event Per Reactor-Year Overpressure transients 3.9 x 10-2 Transients with delayed auxiliary feedwater 7.6 x 10-4 Operator action under ATCG guidelines 1.58 x 10-2 Instrumentaticn and centrol faults 1.7 x 10-3 Cvercooling transients that initiate HPI a .4 x 10-4
- Total 5.3 x 10-2 Using the Cavis-Besse PORY opening rate of 5.3 x 10 _o per reacter-year, the calculation of Figure 1 of this report was repeated. The resulting probability of a small-break LOCA frem a stuck-open POK7 was as follows:
All B&W units except Davis-Besse Davis-Besse Figure 1 -
Figure 7 1.4 x 10-4/rx-yr 1.7 x 10-4/rx-yr Next, the calculation shown in Figure 2 of this report was repeated for a unit with a PORV setpoint of 2400 psig (e.g. , Cavis-Besse) . In this case, a reacter trip setpoint of 2300 psig with an overshcot of 27 psig and a standard deviation of 27.52 psig was compared with a POR7 with a 2400 psig setpoint and a 21.7 psig standard deviation. The analysis snowed that the PORV would cpen
~
with a frequency of 1.9 x 10 per event, given an overpressure transient causing a reactor trip. This value was then inserted into Figure 2 and the resulting small-break LOCA probability was determined as follows:
i "Although not discussed by B&W, the Cavis-Besse safety injecticn system is a low-head system and cannot challenge the PORV. Never theless, this transient is never a significant contributor to the total opening frequency.
-4 6-
<<f31s '
Jdd Franklin Research Center A Onsmen d The Franuun nimen.se
, . _ , .,. . , _ . - - - - , ,- . . - . . , _ . _ . _ ~ . . . _ . -, - , . , , - , - ,
F 6
4 W I
TER-C550 6-410 '
aw i 2>
. <=
b4 3
oAw e
.O z it e w-W e 4O Q a=
a
, W
- 8 =>s e 5
5<a ~
=
C ? ws > *<
- s 3 2
Q a w v W d
~
58 3, n -
g's .
- 5. -<
=
= .2 g .= .
Ob
? l 3 Q o
f aq * -
a .k g=
a, 'a o cy 6
~ 2a, . ,.I e
m
< I 55.3
=r3 ,a a ,
42- i 3 N % V o =c a e;h - -
2
- a e
3 1
% as pug Od 7 4 E a 34 e
~
i g e
EM3 (, ~
s 2 x
w r <= =c -
W een G
? 3
)
3: a C
- 5
.e
- C;
- w, e
n 4 1 ,,
n e.- =~
30< 13 m e
22 /
., s< /
r t C G
= ww g O '
3 x i ae C l
, Y b 21
. i e
o
- 3 5 ma 3
e
=-3 w<w ?
a 3 o 5-t i-
. [e[ S
~ E 3_ .5 E
~
m I
$ d 5 o -
?
2 2
o.e 5n og ?
2Es d-wm ZA
-3 d=2 << a=g m W"h
=
i g g $
e f
w
-4 7-d - %.. - .
. . . '. eranxh.n mesearen Center 4 h of The Fween esense
TER-C550 6-410 All B&W units except Davis-Eesse Davis-Besse Figure 2 Figure 3 2.5 x 10-4/rx-yr 3.7 x 10-4/rx-yr As a result of these calculations, it can be concluded that, altnougn the probability of a small-break LOCA frcm a stuck-open POW at the Cavis-Besse plant is greater than the same procability at the other B&W units, the difference is not significant and is precably less than a factor of 2 larger.
Furthermore, since all values determined are less than the median WAL'I-1400
~
i value for small-oreak LCEA frca all causes of 1 x 10 per reacter-yeau. the B&W report can be considered applicable to all 3&W units, whether the PC.%
setpoint is 2450 psig or 2400 psig.
Finally, with regard to the safety-relief valves, the pecbacility of a small-break LOCA from a stuck-open SRV at the cavis-Besse plant will be icwer 4
i than the prcbability of the same event at the other 3&W units. This is because at the Cavis-Besse plant, the POW opens soccer than at the other units, making it less likely that the SRV setpoint will be exceeded.
Ccnsequently, any determinaticas er cenclusions relative to s all-break LCCA frca a stuck-cpen SRV are equally applicable to the Davis-Besse plant.
l I
l I
i
-4 8-
..b F anMn Research Center aen===w nebenen
[ , .__ ,
. - - , , ~ . . -- -- - - - - - - - - ~ ~ - - - ' ~ ~ ~
~ ~ ~ ' ~
. ~
"'E't-C550 6-410
, 2 .
' g a 4
x
=1
<.5 -2
,2
= a
~
s
~ ,
I !-
-t a:
~
- a y
I
.e. :-
> .i-
.i at _ 8
=
M "
- I 5 2 23c l e
1 I
2 a
a a
(
- i j act -
=
} .
2
- go E 'g
- c l .
=
- '( a r <,z 3
a,,
~ ~ .-
' o
- 3 C 3,
- (% .
Z O
$ no
?
,*o o
2 ,
. -.C a
- 5^3 : >
- Lia~
~
- 3 4
- m 4a i
z d
o i 3 3 :
1
- 5=W .
i 733-2 m
Q<
bg ~
v
< w a
- n o
.s.
I t
i l
i i
r.
1 i
4 J.l.) .* rank!!n Researen Center
- o am wr aw L.
TER-C550 6-410 a.
=>
4- 4 .
3= -
. . o.,
e 50 5o t
= <
& 'N w h
=>-- a s 12 .
z .
c><
a
~
~ - o 3 . s u m
o
. .23 ~a -
y
~
tG zo 0 7 s
$h h:
3 .< i =
. i .
!. ! !, ,e .
=
=c.!
^t> .
- e 2
sf4 . .
s
~
a 2
2
=, . .m
,=, .
x -
u s.n- b*
3 I = -
< =- 3
." S r' U s
O D
I l $5' /
D p i o= ?
l o a - -J s ,
=
- 3=- .
..a.
b sidx -
1 o
, I .::
'o a.
~
' , i:2 )?2 *4 a - 35:
s a o
, ; a .
2- ~ m s=
g-p -.
sa.-
/- c.
a
>D z ( ~
o
?.- u
. t. :o I
= -
=
. g
~
. m .o
= .
- - m 3
=
o l :~
5 :< :ais
- = 4 N
': 55i i
3 8
Q i 2
3 s
's.
= $a a: ..
y3a 2
=-3
=<<
e-a s.
R-5 ci -
w A_
.1 Franidin a cm or n.Resear.ch r .in a Center 1
TER-C5506-410
- 5. CONCLUSICNS The conclusions that result from evaluation of the S&W report against the criteria of Section 2 are as follows:
o The B&W report documents .the post-TMI modifications instituted at B&W-designed NSSS plants wnich have made a significant reduction in the expected frequency of a small-break LOCA frem a stuck-open PCRV cr safety valve.
o The methods and results of tne B&W report have been reviewed, and the expected frequency of a small-break LCCA frca a stucx-open POK7 or safety valve is less than the median WASE-1400 frequency for a small-break LOCA 1 x 10-3 per reactor-year. The following is a tabulation of the resules of this evaluation:
Recurrence Frequency (per reacter-year)
Independent Repor:
Event 3&W 7erificatien Secticn
, Small-breax LCCA from stuck-cpen PORV (analytical calculation) ,
1.4 x 10-4
- 2.5 x 10-4 " 3.2.1
-Small-break LOCA frcm ,
stucx-cpen Poav (operating data) 4. 7 x 10-4 *
- 1. 4 x 10 -4
- 3.2.2 9
Small-break LOCA frcm stuck-open SRV 2.6 x 10-5 4.7 x 10-3 3.2.2 3.1 x 10-6*** 3.2.5
- Includes credit for operator action to shut block valve.
- No credit fcr operator action to shut block valve.
- Censiders overpressure events cnly.
o Multiple PORV cycles en a single initiator event has been considered and is not significant for B&W units. The effect of 10 cycles per each applicable initiator event raises the pecbability of a small-creak LOCA by about 3%.
o The perec.ntage of PORV cpenings during overpressure transients is less than 2% of the total number of cverpressure transients.
o Information used to verify the assumptions and calculations submitted by B&W has been taken frcm generic sources applicable to B&W-designed NSSS units. This repcet is considered applicable to all S&W plants, including Oavic-Besse, wnich has a different PCRV than other units and a lower PCRV setpoint.
- __s L.J F anklin Researen Center 4 ra or m r
TER-C5506-410
- 6. REFERENCES
- 1. " Clarification of TMI Action Plan Requirements" NRC Office of Nuclear Reactor Regulation, November 1980 NUREG-0737 2.
" Generic Evaluation of Feedwater Transients and Small Break Loss-of-Ccolant Accidents in Babcock & Wilccx Cesigned 177-FA Cperating Flants' NRC Office of Nuclear Reactor Regulation, January 1980 NUREG-0565
- 3. "NRC Acticn Plan Developed as a Result of the TMI-2 Accident" NRC Office of the Executive Director for Cperations, May 1980 NUREG-0660
- 4. C. C. Trimble (APL)
Letter to D. G. Eisenhut (NRC)
Sub]ect: Arkansas Nuclear Cne Unit 1, NUREG-0737, Items II.K.3.2 and II.K.3.7 (File 1510.6)
January 8, 1981
- 5. J. F. Stol: (NRC)
Letter to W. Cavanaugh, III (APL)
Susject:
Request for Additional Infor:ation in Regard to Cperating Reactor PORV and ECCS Reports Decaccer 16, 1981 '
- 6. J. R. Marshall (APL)
Letter to D. G. Eisenhut (NRC)
Subject:
Arkansas Nuclear Cne Unit 1, NUREG-0737, Items II.K.3.2 and
- II.K.3.7 - PCRV Reliability Cctcber 22, 1982
- 7. J. N. Ocnchew, Jr. (NRC)
Letter to S. P. Carfagno (FRC)
Subject:
Centract No. NRC-03-81-130, Tentative Worx Assignment F July 21,1981
- 8. NRC Meeting Between NBC and FRC Sub]ect: Status of Reviews of TMI Action Items II.K.3.2, II.K.3.7, and II . K.2 .17 l July 26, 1982
- 9. NRC Telephene Conversation Between E. Chow (NRC) and G. J. Cverbeck (WESTIC Services for FRC) l Sub]ect: B&W Plant Generic Report l Septemcer 17, 1982 i
l l
O d.2) 4c Frankiin iesearch Center
% . m r,-.a -
t'
, e-TER-C5506-410 10 . " Reactor Safety Study, An Assessment of Accident Risxs in U.S. Ccamercial Nuclear Power Plants" United States Nuclear Regulatory Ccamis31cn, October 1975 WASE-1400 (NUREG-75/014)
Research Project V102 Interim Report, April 1982 12 . . "Cata Summaries of Licensee Event Reports en Valves at Cc=mercial Nuclear Power Plants" NBC Office of Nuclear Regulatcry Researca, June 1980 NUREG/C2-1363 13 . Reliability Availability Data Collection and Analysis System NPGD-TM-597, Babecck & Wilecx, Lynchburg, VA Marca 1982 14 . "Handbeck of Human Reliability Analysis with Empnasis en Nuclear Pcwer Plant Applications" NBC Office of Nuclear Regulatory Researca, Cctcher 1980 NUREG/CR-1278 1
l i
I I
l t
I i
! O l
l EEJ Franidin Research Canter 4 ca a h. % -
. - - - - . - .