ML20217G934
| ML20217G934 | |
| Person / Time | |
|---|---|
| Issue date: | 02/24/1997 |
| From: | Houghton J NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | |
| References | |
| NUDOCS 9804290241 | |
| Download: ML20217G934 (24) | |
Text
1 l
i l
/
NUCLEAR ENGINEERING INTERNATIONAL i
FIRE & SAFETY CONFERENCE London, England,24-26 February 1997
)
I i
FIRE EVENTS - FEEDBACK OF U. S. OPERATING EXPERIENCE James R. Houghton j
{
Office for Analysis and Evaluation of Operational Data ]
U. S. Nuclear Regulatory Commiedon Washington, DC 20555 1
Q
~- l D '
./
Submitted to Nuclear Engineering International for publication in Fire & Safety Conference proceedings in February,1997.
,n 9804290241 970224 PDR ORG NEXD PDR
\ .
FIRE EVENTS - FEEDBACK OF U. S. OPERATING EXPERIENCE ABSTRACT This paper characterizes the frequency and nature of fire event data from U.S. operating plants and determines the potential impact this updated data could have on fire risk assessments for U.S. nuclear power plants. This paper provides a review and revision to the present U.S.
fire events database (Sandia National Laboratories for 1965 through mid-1985). With fire event data from Licensee Event Reports (LERs),
component failure histories from the Nuclear Plant Reliability Data System (NPRDS) and the Electric Power Research Institute (EPRI) databases, both a reconciliation of the 1965-1985 database and an extension of the database for 1986-1994 were made.
With the combined and updated data for 1965-1994, the following was performed: Listing of fire events data in a Fire Events database; apportionment of fire events by number, major cause, and plant location; evaluation of the duration and frequency of fire events that occurred during power and shutdown operations; grouping of fire events by severity during power operations; comparison of the updated power operations fire frequency with Probabilistic Risk Assessment (PRA) data and recent industry and NRC sponsored studies for effect on fire induced core damage frequency estimates; and comparison of the duration and frequency of shutdown fire events with power operations fire events.
- 1. INTRODUCTION The U.S. Nuclear Regulatory Commission (NRC) report, " Procedural and Submittal Guidance for the Individual Plant Examination of External Events (IPEEE) for Severe Accident Vulnerabilities" (NUREG-1407, Reference 1), referenced the NRC sponsored Fire Risk Scoping Study l (NUREG/CR-5088, Reference 2), as a confirmation that fire continues to ;
represent a dominant risk contributor. One of the six items listed in the " Internal Fires" section of Reference 1, item 2. addressed fire frequency data as follows: "Most initiating event frequencies were increased based on a much more complete data base available on fire occurrences in nuclear power plants. Under currently applied risk assessment methodologies, this increase in initiating event frequency alone results in a direct increase in overall fire-induced core damage '
frequency with all other factors remaining constant." l The fire events data used for the Fire Risk Scoping Study was limited to previous industry gathered data, such as the Sandia Laboratoria- d'L base (1965 through mid-1985, NUREG/CR-4586, Reference 3). Therefore, an update of this data was done to evaluate whether PRA insights about fire frequency and consequences were consistent with a comprehensive review of available U.S. operating experience.
I Significant attributes used in the estimation of fire induced core damage frequency, that could not be included in the scope of this paper due to the limitations of data were as follows: the determination of fire suppression probability from the reported fire durations, the updated information on the ignition and damage thresholds of cable insulation, and plant modifications as a result of the U.S. code of I federal regulations, 10CFR 50, Appendix R, " Fire Protection Program for Nuclear Power Facilities Operating Prior to January 1, 1979."
Evaluation of fire precursors was also excluded from the scope of this review.
2 l
- 2. APPROACH AND METHODOLOGY Fire Events Database Update The Sandia Fire Event data base contains 354 fire events during a period from 1965 to mid-1985. The listed fire events occurred during the constructicn phase, preoperational testing phase, and operational
- phase for the majority of U.S. nuclear power plants. Most of the information provided was sufficient to identify the plant location
! (i.e., building, room, area), estimate the duration of the fire, determine the cause of the fire, and the effect on plant power operation. However, the data was sparse in its identification of safety-related train (s) needed for safe shutdown of the plant in the event of a fire. The fire event data base update included a review of existing Sandia data for the 1965-mid 1985 period, using the NPRDS database (1965-1985), LER database (1980-1985), and other fire event updates made by Sandia and identified in later reports. Construction phase fires were excluded in the update. The 1986-1994 period was added to the data base,'using fire event data from LERs and from component failure histories (NPRDS).
The proprietary EPRI Fire Events database (1968-1988) included the SANDIA database fire events through mid-1985, except for construction phase fire events. Generally, the EPRI fire events that were included in this paper had the following attributes:
Resulted in or were associated with a SCRAM or Loss-of-Offsite Power (LOOP).
Explosions, except for Recombiners identified below.
Fire events with duration of 5 minutes or longer in plant areas that have safety-related systems or systems necessary for continued power operations.
EPRI database fire events (predominantly previously unreported fire events from plant-specific questionnaires) were excluded based on the following: {
Reviewed and evaluated as not a fire or explosion (i.e., smoke or odor).
Not significant, due to one of the following: (1) Duration less than 5 minutes (except explosion, SCRAM, or loss of safety-related train or syrt= , ;2) located in a plant area that is not ad;meent to or could not propagate to a plant area that contains a safety-related system or system that could cause a plant shutdown (such as a stockroom, hallway trash can fire, warehouse, trailer, temporary building, etc.); (3) located outside the plant area without affecting the switchyard area (i.e., a non-encroaching forest fire or brush fire); or (4) recombiner explosions (with no corollary fire 5 minutes or longer) that did not affect any plant power operations or safety-related systems that are needed for plant shutdown.
Apportionment of Overall Fire Events by Cause and Location An apportionment of overall fire events was performed to identify the major cause categories of fires and how each cause category was apportioned by the plant locations (i.e., buildings, rooms, etc. ) . In addition, fire events (all causes) were also apportioned by plant location.
1 l 3
I Overall Fire Frequencies Overall fire frequencies (i.e., number of overall fire events divided by number of calendar operating-years within the time period) were plotted to provide: (1) overall fire frequencies over the 1965-1994 period for possible trends, including a comparison of the 1965-1985 average overall fire frequency and the 1986-1994 average overall fire frequency for significance of difference and (2) overall fire frequencies by plant location over the 1965-1994 period for possible trends.
Power Operations Fire Events and Duration of Power Operations Fire Events by Plant Location The number of power operations fire events and apportionment of these fire events by duration intervals were determined for the two periods by plant location. Plots were made of these power operations fire event duration intervals, including means, to provide a basis for:
(1) evaluating the effectiveness of suppression at power operations, (2) comparison between the two periods for significance of differences when used in risk insights, and (3) comparison with fire event duration at shutdown.
Plant Average Unit Availability Factors and Power Operations Fire Frequencies It was necessary to determine the power operations fire frequencies by plant location to allow comparison with fire PRAs and other studies, where fire frequency is an attribute in the calculation of fire induced core damage f requency (CDF) estimates. For this determination, the plants Unit Availability Factor for each year of the two periods (1965-1985 and 1986-1994) was established, including averaging the overall 1965-1994 period. These averages were used to convert calendar years to reactor-years of power operation. This paper determined and plotted the power operations fire frequencies to provide:
a means to identify trends in power operations fire frequencies over the 1965-1994 total period by plant location and a comparison between 1965-1985 and 1986-1994 average fire frequencies by plant location.
Eayes 90% intervaAs were developed for the power operations fire frequencies for both periods and the overall period, by plant location for use in comparison with plant fire PRAs and industry studies, using a Jeffreys noninformative prior and the area specific data to develop the updated posterior interval.
Risk Insights Fire risk assessments generally use the following steps to produce an estimate of core damage frequency due to fires: (1) estimate fire frequencies for particular locations in the plants; (2) estimate the probability of non-suppression of the fire in a given location; (3) determine the effects of the fire, including a list of all damaged equipment assumed to be unavailable; (4) quantify the conditional probability of core damage; and (5) combine the frequency of fires with the probability of non-suppression and the conditional probability of core damage to produce the core damage frequency estimate due to fire events.
The operating experience data on fire events in this paper is suited to providing estimates of fire frequencies and some insight into non-suppression probabilities. However, it does not contain models and 4
I .
data for the other steps. Therefore, this paper used two means of evaluating risk significance as noted in the next two subsections. One was to group operating experience into severity groaps and the other was to evaluate existing PRA results (prior to IPEEE submittals) by a sensitivity analysis based solely on differences in fire frequency estimates.
Power Operations Fire Events Severity Grouping The apportionment of power operations fire events among five severity group categories was performed to obtain insights from the risk significance of these fire initiators. The five categories used in this paper are:
Category A - Fire events that caused loss of more than one train of a safety-related system or loss of multiple single-train safety-related systems.
Category B - Fire events that resulted in a SCRAM and LOOP gr resulted in a SCRAM and a loss of one train of a safety-related system and had a duration of 5 minutes or longer, EI resulted in a SCRAM and a loss of one train and had an explosion, regardless of the fire's duration.
Category C - Fire events that resulted in a SCRAM, regardless of the fire's duration, but without loss of a safety-related train.
Category D - Fire events that resulted in a loss of one train of I safe shutdown equipment, regardless of the fire's duration, but without SCRAM.
Category E - All other reported fire events.
Comparison of Power Operations Fire Frequencies with Selected Plant PRA Data and Other Data by Plant Location In order to determine whether updated fire frequencies support the Fire Risk Scoping Study (FRSS) contention that fire events continue to be a dominant contributor to overall plant risk, a sensitivity analysis was performed. Since the data on fires did not provide sufficient !
information to assess the probauAlities of detection and suppression, this sensitivity analysis only addressed potential changes in the fire event frequencies.
The fire locations in PRAs and other analyses (such as the .
Kewaunee IPEEE and the FRSS) are often different from one analysis to l another. These differences may be due to plant-specific features, ;
analysis assumptions, or screening criteria. However, five plant locations (Control Room, Cable Spreading Room, Switchgear Room, Reactor j Building (BWR), Auxiliary Building (PWR), and Turbine Building) appear !
to be common to most analyses.
The first step in the sensitivity analysis was to compare the operating experience to the point estimates used in the PRAs and other I analyses. A generic fire frequency for each of the areas noted above was derived using a Bayes 90% interval, based on a Jeffreys noninformative prior and the pooled data from all plants. Point estimates from PRAs and other analyses that fall completely outside this range are indicative of a statistically significant difference between l the generic frequency and the PRA estimate. Conclusions regarding the '
generic frequencies being higher or lower than the point estimates in 5
j
l the PRAs and other analyses are dependent on a finding of significant difference as noted above.
The next step in the sensitivity analysis was to compare how the resulting CDF of a particular analysis would change if the generic mean for the post-1986 period was substituted for the plant-specific PRA (or other specific analyses) fire f requency point estimate.
The last step in the sensitivity analysis was to determine the significance of the resulting CDFs. Many of the plant-specific analyses used screening criteria such == that found in the Fire Induced Vulnerability Evaluation (FIVE) methodology when calculating fire induced CDF for a specific plant location. This paper used the FIVE methodology threshold of 10+ as the basis for concluding whether the resulting change in the existing CDF for that plant location would be significant.
Shutdown Fire Events Fire events that occurred during plant shutdown were excluded from plant PRA estimates and from IPEEE internal fire CDF estimates However, fire events during plant shutdown may directly affect shutdown risk. This paper reviewed: mean fire durations during shutdown, by plant location, and compared these means with fire duration means during power operations for each of the two periods (1965-1985 and 1986-1994);
and average fire frequencies during shutdown, by plant location, and i compared these averages with average fire frequencies during power I operations for each of the two periods (1965-1985 and 1986-1994). l l
l
- 3. RESULTS !
Overall Trends and Apportionment l
No trend is evident in the total number for fire events after 1975 l for each of the plant modes and overall fire events (see Figure 1),
although the number of plants and their operating times increased during the period.
The overall fire events were apportioned to gain insights in fire causes and plant Jocations, regardless of plant power levels. This data is intended for overall fire event concerns, but not directly applicable to risk without adjustment to power operations only. The following provides the apportionment determined by this study:
Overall fire events were apportioned among four major causes:
l Electrical Failure, Overheated Material, Explosion, and Welding.
For the overall period (1965-1994), Electrical Failures and l
Overheated material comprised 69% of the causes (see Figure 2).
l
- Electrical Failure (38%) was the predominant cause of failure during all plant operations (i.e., overall) for the combined period 1965-1994, with an increase in apportionment (to 50%)
during the update period, 1986-1994; while the apportionment of the other causes (Overheated Material, Explosion, and Welding) decreased slightly during the update period (see Figure 2).
Overall fire events, apportioned by location only were similar in apportionment for each of the two periods (1965-1985 and 1986-1994). For the overall period, the fire event locations apportionment was predominantly in Auxiliary Building (PWR-15%),
6
l Turbine Building (18%), Diesel Generator Building (15%), and Reactor Building (BWR-12%) (see Figure 3) .
1 Frequencies and Duration As in the review of overall fire events, the overall fire frequency provides the inclusion of operating (calendar) years in the I overall fire review. The following provides a comparison between the j two periods, 1965-1985 and 1986-1994, for both total plant and specific internal plant locations:
The average total plant overall fire frequencies for the update ,
period (1986-1994) were approximately one-third less than for the l 1965-1985 period (see Figure 4). j The overall fire frequencies in the majority of the internal plant locations reviewed appeared to have a lower average frequency for the updated period (see Figures 5 and 6)
For the 19M- D85 period, the majority of fire events (78%) in plant locations with safety-related systems were less than ten minutes duration, while 68% of these fire events were less than five minutes duration. Mean durations were longer than 10 minutes in some areas due to the occurrence of a few long duration fires. The following provides i
a summary of the durations at the plant locations with safety-related !
systems: 1 TABLE A Plant No. Fire Events Mean location <5 Min. <10 Min. 10 Min. or > Total Duration I i
Control Room 3 3 0 3 2.5 Containment Blds 3 5 1 6 5.6 Reactor Bldg (BWR) 12 13 5 18 15.3 l Auxiliary Bldg (PWR) 26 31 9 40 8.6 l Cable Spreading Room 0 0 3 3 52.7*
Switchgear Room 7 7 2 9 17.6 Battery Room 3 3 0 3 2.5 Diesel Gen. Bldg 26 30 7 37 6.4 Serv. Water Pumphse .] 2 0 .__,1 2.5 Total No. Events: 82 94 27 121 Percent of Total: 68 78 22 100
- Includes Browns Ferry fire, but limited to 100 minutes.
Two nonsafety-related locations, the Turbine Building and Switch Yard, had a higher r._.oge duration of approximately 20 minutes. :.t most plants these locations have less risk significance. However, for some plants,the Turbine Building includes safety-related equipment or fire safe shutdown equipment.
For the 1986-1994 period, the majority of fire events (63%) in plant locations with safety-related systems were less than ten minutes duration, while 46% of these fire events were less than five minutes duration. Mean durations were longer than 10 minutes in some areas due to the occurrence of a few long duration fires. The following provides a summary of the durations at the plant locations with safety-related systems:
7
TABLE B Plant so. Fire Events mean i Location <5 Min. <10 Min. 10 Nin. or > TRIAL Duration Control Room 0 0 0 0 0 Contairment Bide 1 1 1 2 18.2 Reactor Bide 3 3 6 9 12.8 Auxiliary Blds 4 6 4 10 9.7 Cable spreading Room 0 0 2 2 13.8 Switchgear Room 3 3 1 4 14.4 Battery Room 0 0 0 0 0 Diesel Gen. Blds 5 10 1 11 6.6 3 Serv. Water Pumphse 4 4.8
~
4 1 5 Total No. Events: 20 27 16 43 Percent of Total: 46 63 37 100 Two nonsafety-related locations, the Turbine Building and Switch Yard, had a higher mean duration of approximately 24 minutes.
The comparison of power operations fire events by plant location between the 1965-1985 and 1986-1994 periods showed somewhat similar durations, except the latter period was lower for the Cable Spreading Room and higher for Other Buildings and Containment. The lower duration was due to no long-term fire (i.e., no Browns Ferry fire type duration),
while the higher duration in the Other Buildings was due predominantly to charcoal fires in the waste and off-gas treatment building. The high duration in the containment was due to one event caused by welding.
The power operations fire frequency for the raajority of plant locations showed a decrease during the update period (1986-1994) when compared with the 1965-1985 period. The following depicts the ratio of 1965-1985 period to 1986-1994 period for these plant locations (see Figures 5 and 6, except for Other Buildings and Switch Yard):
TABLE C Plant tocatigg) Batie l Control Room (no fire events, 1986-1994)
Battery Room (no fire events, 1986 1994)
Reactor Building (BWR) 3:1 l Other Buildings 3:1 j Auxiliary Building (PWR) 2:1 I,lesel Generator Building 2:1 j Cable Spreading Room 2:1 ;
Switchgear Room 1:1 j contairment 1:1 -
Switch Yard 1:1 Turbine Building 1:1 Service Water Pumphouse 1:3 Severity Grouping Severity Grouping of Power Operations Fire Events - Figure 7 and Tables I and II depict the plant operational data severity grouping for j power operations fire events. The following provides some risk !
perspectives relating to these fire events:
Of the 341 power operations fire events during 1965-1994, one, the Browns Ferry fire in 1975, was not suppressed in time to prevent propagation to other safety-related trains or systems (Category A).
Loss of function occurred to multiple systems.
8
l 1 l
For the 199 power operations fire events in the 1965-1985 period:
Ten fire events caused SCRAM and a loss of safety-related train and were suppressed without propagation (Category B)
Forty-one fire events caused a SCRAM, but with no loss of a safety-related train and were suppressed without propagation (Category C). l Twenty-two fire events resulted in a loss of one train of safe shutdown equipment, regardless of the fire's duration, but without a SCRAM (Category D).
For the 142 power operations fire events in the 1986-1994 period.
1 l
One fire event caused a SCRAM and loss of a safety related train, without further propagation (Category B). A second fire event (Oyster Creek), initiated by an offsite fire, I resulted in a SCRAM and Loss-of-Offsite Power (LOOP), but l caused no loss of function to safety-related systems l (Category B due to SCRAM with LOOP).
Forty fire events caused a SCRAM, but with no loss of a l safety-related train and were suppressed without propagation (Category C)
Twenty-one fire events resulted in a loss of fire safe sigtdown equipment, regardless of the fire's duration, but without a SCRAM (Category D).
The balance of power operations fire events were evaluated as less severe (Category E)
Sensitivity Analysis The sensitivity analysis for the PRAs and other analyses, based on i 1986-1994 operating experience fire frequencies for each of the five l plant locations, resulted in the following (see Figures 8, 9 and 10).
Control Room - No fire events occurred during the period 1986-1994. Using a Jeffreys uvainformative the updaLed power operations mean fire f requency (8. 5x10')prior, was lower than used in i the selected PRAs and the industry studies. The majority of Control Room fire frequencies were above the Bayes 9C% interval upper bound (3.3x10 4) However, when the update frequency is taken into account, the majority of the Control Room CDFs were l still in the 10* range (except Diablo Canyon which fell to the 10 7 !
range).
Cable Spreading Room - The updated power operations fire frequency mean (4. 3x101 ) varied from slightly higher to slightly lower than used in the selected PRAs and the industry studies.
However, with the exception of Millstone 3 (2. 6x10 )4, the majority of Cable Spreading Room frequencies were within the Bayes 90%
interval upper bound (9. 5x103 ) and, therefore, differences between the PRAs and the update are not statistically significant. The majority of Cable Spreading Room fire-induced CDF estimates were still in the 10+ range.
9
t 1
. l l
l l -
Switchgear Room - With the exception of the Kewaunee IPEEE l (1. 8x10 ) , the updated power operations fire frequency mean fire 2
2 f requency (1. 3x10 ) was higher than used in the few PRAs and industry studies that had developed Switchgear Room fire frequencies and fire-induced CDF estimates. For those PRAs where the Switchgear Room fire-induced CDP was quantified, the new estimates were generally higher, but still remained in the 10 5 range (above the FIVE threshold of 10' range) .
Reactor Building (BWR) - The updated power operations mean fire f requency (5.4x10 2) was approximately the same as the two PRAs (Cooper and LaSalle) used for this study. For LaSalle, the fire-induced CDF was still in the 104 range, while the fire-induced CDF for Cooper was not quantified.
Auxiliary Building (PWR) - With the exception of the Kewaunee IPEEE (7. 3x103 ) and the FRSS (6.4x102 ) point estimates, the updated power operations mean fire frequency (4. 9x10 3 ) was approximately the same as all other point estimate fire frequencies used for specific PRAs and other studies. For these PWR PRAs there were no fire-induced CDFs quantified.
Turbine Building - The updated power operations fire frequency 3
mean (6,9x10 ) was higher than all the point estimate fire frequencies used for specific PRAs and ether studies. With the exception of the Kewaunee point estimate (6. 6x10 2) , all PRAs and other studies' point estimates were less than the Bayes 90%
2 interval lower bound (5.2x10 ) . The new fire-induced CDF estimate was higher for LaSalle (increasing from the 10' range to the 10*
range) and for the FRSS (but still in the 10* range).
Although the majority of the updated power operations fire frequency means were somewhat lower than the requantified 1965-1985 fire frequencies, there was little significant change in the specific plant location fire-induced CDFs (the majority were still in the 104 range).
Therefore, the following summarizes the comparison and addresses the major purpose of this review:
- The updated fire events database provides new information for assessing generic fire frequencies. The 1986-1954 fire event frequencies at power operations were lower for the Control Room and Cable Spreading Room, approximately the same for the Auxiliary Building (PWR) and Reactor Building (BWR), and higher for the Switchae,r Poom and Turbine Building than those assumed by PRAs used in this report.
e The updated fire events database and estimated frequencies should provide an improved source of fire initiation information for use in estimating fire frequencies for risk assessment or for reviewing such assessments, o The use of these updated fire frequencies has the potential for proportionally lowering generic fire-induced CDF estimates for the Control Room and Cable Spreading Room than those previously estimated in PRAs; while indicating a potential for proportionally raising generic fire-induced CDF estimates for the Switchgear Room and Turbine Building. However, when a 104 range threshold is used, there appears to be no significant effect on any of these plant locations fire-induced CDFs.
10 i
- Updated fire frequencies were estimated for major plant locations in this report. To the extent that fire PRAs (and other risk-based fire analyses) contained similar areas where fire-induced CDP was calculated, sensitivity analyses were conducted in this report to determine the impact of changing the fire frequency based on operating experience. However, the lack of a one-to-one correspondence between areas analyzed in this report and those in the fire PRAs precludes making a plant specific inference on the overall fire-induced CDF.
Power Versus Shutdown Comparison Except for the Containment, the mean duration of shutdown fire events for plant locations was approximately the same to lower than for mean durations occurring during power operations for both the 1965-1985 and 1986-1994 periods. The Containment fire events were predominantly caused by welding.
For the 1965-1985 period, when compared with power operations average fire frequencies, the shutdown average fire frequencies were approximately the same or lower for most risk significant plant locations used in PRAs, while higher fire frequencies were calculated for the Containment and Reactor Building (BWR) .
For the 1986-1994 period, when compared with power operations average frequencies, the shutdown average fire frequencies were varied, with the majority approximately the same or lower for most risk
- significant plant locations used in PRAs; while higher average fire l frequencies were calculated for the Containment, Reactor Building (BWR),
! Auxiliary Building (PWR), Switchgear Room, and Diesel Generator Building. A more detailed review of shutdown fire events and shutdown fire frequencies, resulting in lower system-specific fire frequencies is as follows:
I For the containment fire frequency, fires were predominantly caused by welding operations and did not involve decay heat removal systems.
For the Reactor Building (BWR), Auxiliary Building (PWR), and Switchgear Room, the fire events that were applicable to functional operability of Residual Heat Removal (RHR) and Decay Heat Removal (DHR) system trains were limited in number and resulted in lower fire frequencies than for power operations fire frequencies.
For the Diesel Generator Building, fire events that were applicable to the functional operability of the Emergency Diesel Generator (EDG) trains were also limited in number and resulted in approximately the same fire frequency as at power operations.
The system-specific fire frequencies at the above plaat locatiens are as follows:
TMEE D No Systems Plant Point Estimmte Location System Fire Events
- Shutdonat-Yrs Fire Freesency Reactor Bldg RHR 4.5* 90.4 5.0x10' Auxillary 8tdg RHR & DHR 2.5* 139.8 1.8x10' Switchgear Room RHR 1.5* 230.2 6.5x10' Diesel Gen. Bldg EDG 6.5* 230.2 2.8x10'
- Indicates a noninformative prior was used, j 11 j l
Therefore, the operating experience indicates that the frequency and duration of shutdown fire events appears to be less significant than for fire events occurring at power operation.
- 4. REFERENCES
- 1. NUREG-1407, " Procedural and Submittal Guidance for the Individual Plan t Examination of External Events (IPEEE) for Severe Accident Vulnerabilities," dated June, 1991.
- 2. NUREG/CR - 5088 (SAND 88- 0177) , " Fire Risk Scoping Study:
Investigation of Nuclear Power Plant Fire Risk, including Previously Unaddressed Issues, " Ja ted January 1989.
- 3. NUREG/CR - 4586 (SAND 86-0300), " User's Guide for a Personal Computer-Based Nuclear Power Plant Fire Da ta Base, " da ted August, 1986.
I l
l 12
NUMBER OF POWER OPERATIONS NUMBER Of SHUTDOWN FIRE EVENTS TOTAL PtANT FIRE EVENTS - TOTAL PLANT fee. el Power Operesens Fke 8vente he of $5 Phe Evenes
- 1. -
1 1.
1 g .
I
,, m mll ,m mm _. mRR 111111111111111111111111111111 111111111111111111111111111111
- f????????!!::::::::::::::
......is.... .. 1 ....... 1 3. :::: f??!!!!!!? ::::::::::::::
.. ...i 3...... 1,.... .. 1 3.
Calender veers Casender Years W Fkaivente .6-.. C Feaevente M C Fkeevents .. .. C Fkeevene M-M No. Pwe Oper.Fke Evenes,1MS IM.: 1 No. Shutdown Fhe Events,1 bt.B.: .
No.
- w,,Pwre.Oper. Fke f .iM
., even vents.1.M
.i 1
1 M.M.2. No. S.hutdo,wn 6,e e. Fke Events,1
.71, im e. . ev.ni..i.6.1.M.Mi i
l NUMBER OF OVERALL RRE EVENTS l TOTAL PLANT
! No, of Total Fire Events 35 j 30 -
25 -
20 -
)
15 -
10 -
5 -
0 111111111111111111111111111111 999999999999999999999999999999 666667777777777888888888899999 567890123456789012345678901234 Calendar Years Revised:19651985 L__J Update:1986-1994 Rre Events.65-85: 319(inct 23 PreOpTst)
Fire Events, 86-94: 233 (includes 10/yr entrapolated fire events,89-94).
FIGURE 1 1
1 1
I l
OVERALL FRE EVENT APPOHilONMENT OVERALL FRE EVENT APPORTIONMENT SY CAU$E
- 1965-1985 BY CAUSE - 1986 1994 O$m og OVERMT MATL 32%
EMMO$h in 2,,.. + gg
':Q;s. 'e ILEC F AIL 32%
N' 9:.!:
'N.
OW 11%
fLIC F AL 60%
NRW WELDM416%
~ . . . . _ , , .' Encludes .strapolated dete. 1989-1994 OVERALL FIRE EVENT APPORTIONMENT BY CAUSE - 1965-1994 EXPLOSION 9% OVERHT MATL 31 %
k?>
ELEC Fall 38%
\ OTHER 10%
WELDING 12%
i No. Fire Events: 492 FIGURE 2
! Excludes extrapolated date, 1989-1990.
1
OVERALL FIRE EVLNT APPORTIONMENT OVERALL FIRE EVENT APPORTIONMENT BY LOCATION - 1965-1985 BY LOCATION - 1986 1994 l
Aus5Mg 18% A"*N II%
n,3gg g gg
/ -
ReectEndg 12%
Contam 10% /
Twhsdg 14% ; / Content 7%
., ['
k NjV Offwte 5%
$wetyRm $% SwchgrRm 6%
Mec 7% Ewitchyd 5% Misc 6%
Sewd 14 No FWe f vente 319 Toter he. Fee Evente- 173 Meena Cond Rm, CW Sped Am. SW Pawhee, Mier., m CeW Spad Rm. Bett tett mm. Offene, & Temp Owes Room ord Servka Watee Pumphouse.
OVERALL FIRE EVENT APPORTIONMENT BY LOCATION - 1965-1994 i
i AuxBidg i 15 % l j ReactBldg 12 %
TurbBidg 18% , Contain.
N.
s wch Rm l O hB gs 6
Misc \
6% \ -
Swch Yd DG Bldg 8%
15%
No. Fha Events: 492 Misc: Cnti Rm, Cbl Sprd Rm. Butt Rm. FIGURE 3 Serv Wtr Pmphse, & Temp Bldgs.
OVERALL FIRE FREQUENCIES TOTAL PLANT Overall Fire Frequency x E-1 7
65-85 '
AVE.
6 -
(3.8E-1) ; ,
COMBINED 5 -
- : , 65-94
, AVE. 86-94
- 4
, (3.3E-1) AVE.
4 -
1 ! - (2.8E-1)
- 2 ; ::
k C 3 -
6 ; ; : 4 <
1
> , y < 1 i ',
G j $ ; I 6 . . , . .
8 8 8
'. 0 8 -
~
2 ; a G -
! E. i 1 i j $
8 -
B . . . : ; ; ; ; -
s' s; 1 1 ! s s ,
, ' l. ; 1 1 s s s j j i 6 l '! ! ! 4 s ; ! ! i !
] ; ! 6 6 4 ; : $ -; 3 ! ! ! I j ! ! !
, i, < <
0 111111111111111111111111111111 999999999999999999999999999999 l
666667777777777888888888899999 567890123456789012345678901234 l Operating Years Overall Freq:65-85 I I Overall Freq:86-94 No. fire events: 65-85, 319; 86-94, 233 (inci 60 extrapolated events).
Cal.oper.-yrs:GS-85, 850.4; 86-94, 816.3 FIGURE 4
j POWER OPERATIONS FIRE FREQUENCIES POWER OPERATIONS FIRE FREQUENCIES CONTAINMENT - 1965-1994 and 19861994 REACTOR BLDG (BWR) - 1965-1994 Power Operations Fire Frequency a E 2 Power Operatione Fire Frequency a E-2 6 50 5 -
86 94 40 -
65 86 AVE, AVE.
4 65 86 A vf. 30 -
3 ,it OE 2) 9 AVE.
(2 9E 2)--
2 -
, i 1' 10 -
I\
7 n m,. .I O ' ' '
0*' '* * ' '
111111111111111111111111111111 111111111111111111111111111111 999999999999999999999999999999 999999999999999999999999999999 666667777777777888888888899999 666667777777777898888888899999 5678901234567090123456789b1234 567890123456789012345678901234 Operating Years Operstmg Years 1965-1985 1986-1994 @ 1965-1985 1986 1994 l
POWER OPERATIONS FIRE FREQUENCIES POWER OPERATIONS FIRE FREQUENCIES AUXILIARY BLDG (PWR) - 1965 1994 TURSINE BLDG - 1965-1985 AND 1986-1994 l
Power Operedone F6te Frewency a E-2 Power Operatione Fire Frepency a E 2 80 30 1
25 -
~
65-85 AVE. 20 -
65,85 (1.1E 1) AVE.
l
~
40 -
86-94 15 -
86-94 AVE, AVE, (t4.8E 2) , (6.8E-2) 10 -
, , , 3 5
0 ' ' '
0 111111111111111111111111111111 111111111111111111111111111111 999999999999999999999999999999 999999999999999999999999999999 668857777777777888888888899999 666667777777777888888888899999 567890123455789012345678901234 567890123456789012345678901234 Opereung Yeare Operating Years VA 19651985 C 1986-1994 -
1965-1985 C 1986-1994 FIGURE 5 i
i
POWER OPERATIONS FIRE FREQUENCIES POWER OPERATIONS FIRE FREQUENCIES CONTROL ROOM - 19651985 AND 19861994 CABLE SPREAD RM- 19651985 AND 19861994 Power Operations Fire Feequency a E 2 Power Operations Fire Frequency a E 2 5 3 _
4 - EO ~
2 -
~
65 85 15 1 E-3) ,
- 65 85 2 -
80 8" (5.1E 3) 3
, AVE.
(3 4E 3) 1
, 0.5
[
0*' ' ' ' ' '
0"' ' ' ' "'"' * * '
111111111111111111111111111111 111111111111111111111111111111 999999999999999999999999999999 999999999999999999999999999999 666667777777777888888888899999 666667777777777888888888899999 567890123456789012345678901234 567890123456789012345678901234 Operating Years Operating Years
' ' 1965 1985 C 19861994 '
1965 1985 C 1986-1994 No Are events dunng 1986-1994 penod.
POWER OPERATIONS FIRE FREQUENCIES POWER OPERATIONS FIRE FREQUENCIES OlESEL GEN. BLDG- 1965-1985 AND 1986-1994 SWITCHGEAR ROOM- 1965-1985 ANO 1986-1994 Power Operadone F6re Frequency a E-2 Power Operemone F6re Frequency a E-2 25 25 20 -
20 -
65-85 g AVE. u II 15
' (8.3E-2) -
86 94 65-85 10 -
ql AVE. 10 -
AVE.
86 94 (2.7E 3) - 11.5E 2) AVE.
l (1.2E-2) y 5 -
jf 5 -
' l
)
0 0 '
111111111111111111111111111111 111111111111111111111111111111 s 999999999999999999999999999999 999999999999999999999999999999 '
666667777777777888888888899999 666667777777777888889888899999 567890123456789012345678901234 567890123456789012345678901234 Operat6ng Years Operating Years
@ 1965-1985 1986 1994 @ 1965 1985 1986-1994 FIGURE 6
l RISK INS 4GHYS - SEVERITY GROUPING RISK INSIGHTS-SEVERITY GROUPING POWER OPERATIONS FIRE EVENTS- 1966 1985 POWER OPERATIONS FIRE EVENTS- 1986 1994 0 13 %
C 20%
c ass o its yaNNcs s
.g ' i e5% s Ig' p'"'t
.,3lO.k
- ,',*>,,,,'>,' 8'%
b ?$$'[;) ;'Ify? ' > 7' >' ' O, ';',
< < n >v . ~>,'
8"'
s ses No. of Fire Events at power: 199 No. Fire Events at Power: 142 Category A: 1: Category B: 10 Category A: 0: Category B: 2:
Category C: 41; Cate0ory D: 22 Category C: 40.; Category D: 21 RISK INSIGHTS - SEVERITY GROUP POWER OPERATION FIRE EVENTS - 1965-1994 l l
D 12% C 24%
2 f
, B3%
Afl %
v; ,. ; < ,. + ,, . < >
- .:;+ o
' . gfs '* - ,.
,y s ' s , :
'?',,
v.
E 60%
l i
Total No. of Fire Events at Power: 341 Category A: 1: Category B: 12; Category C: 81: Category D: 43 FIGURE 7 l
l
TABLE I RISE INSIGHTS - SEVERITY GROUPING OF P0bER OPERATIONS FIRE EVENTS - 1965-1985 ITEM PLANT SAFETY STS PohER SEVERITY LOCATION & QQ, JNA DUR TRAIN EFF EFFECT CATEGORY Switchgear Am: 4 1 ++ 1Hr- 1 train SCRAM B 45 min. (Evaluated)
Auxiliary BldB: 5 206 San 1Hr- 1 train SCRAM B Onofre 1 45 min. (Evaluated) 19 409 Lacrosse 20 min. 1 train SCRAM B (Evaluated)
Reactor BldB: 20 265 Quad 2 Hrs 1 train SCRAM B Cities 2 (Evaluated) 45 249 Dresden 3 <5 min. 1 train SCRAM B (Expl) (Evaluated)
D. G. BidB: 49 333 FitzPatrick 45 min. 1 EDG train SCRAM B Cable Sprd Re: 51 259 Browns Ferry 1 7 Hrs- Multiple SCRAM A 30 min. Systems 99 278 Peach 45 min. 1 train SCRAM B Bottom 3 (Evaluated)
Other BtdBs: 66 325 Brunswick 2 <5 min. 1 train SCRAM B (Expt) (Evaluated) 74 249 Dresden 3 <5 min. 1 train SCRAM B (Expt) (Evaluated)
Switch Yard: 265 3 ++ 3 Hrs- Plt AC Pwr SCRAM / B 45 min. (att) LOOP l MOTES:
- 1. Power Operations fire events data (i.e., at power operations) is from Appendix A, Table I (1%5-1985).
- 2. Severity Grong Category A: Fire events at power operations that caused loss of more thant one train of a safety-related systan or loss of multiple sirgle-train safety-related systems.
- 3. Severity cros , Category B: Fire events at power operations that resulted in a SCRAM E d LOOP g resulted in a SCRAM and a toss of one train and had a & ration of 5 minutes or tanger, g resulted in a SCRAM and a loss of one train and had an explosion, regardless of the fire's duration.
- 4. Severity Group Category C: Fire events at power operations that reutted in a SCRAM, regardless of disation, but no loss of a safety-related train occurred (41 fire events - not shown in this table.
See Appendix A - Tables I and II).
- 5. Severity Grois Category D: Fire events at power operations that resulted in a loss of one train of fire safe shutdown ecpipment, regardless of the fire's duration, but without a SCRAM or Reactor Trip (22 fire everits - not shown in this table. See Appendix A - Tables I and II)..
- 6. Were *1 traits (Evaluated)" is listed, the specific safety-related train was not identified in the initial SAleIA database or other indJstry database.
I
- r ,
l l
TABLE II l
RISK INSIGHTS - SEWRITY GROUPING OF POKR OPERATIONS FIRE EVENTS - 1986-1994 ITEM PLANT SAFETY STS POER SE WRITY LOCATION A DE NME E TRAIN EFF EFFECT CATEGORY Reactor 5tde: 17 341 Fermi 2 >10 min. HPCI Sys SCRAM B offsite: 139 219 Oyster freek 17 Hrs Plt AC Pwr SCRAM / 8 (all) LOOP NOTES:
- 1. Power Operations fire events dets from Appendia A, Table II (1966-1994).
- 2. Severity Grote Category 8: Fire events at power operations that resulted in a SCRAM a_n_d LOOP PI resulted in a SCRM and a toss of one train and had a duration of 5 minutes or tonger, er resulted in a SCRM and a loss of one train and had an emplosion, regardless of the fire's duration.
- 3. Severity Grote Category C: Fire events at power operations that resulted in a SCRM, regardless of duration, but no loss of a safety-related train occurred (40 fire events - not shom in this table.
See A g endix A - Tables I and II).
- 4. Severity Crote Category D: Fire events at power operations that resulted in a loss of one train of fire safe shutdeun equipment, regardless of the fire's daration, but without a SCRM or Reactor Trip (21 fire events - not shoun in this table. See A @ endia A - Tables I AM II).
1 I
i i
e
f CONTROL ROOM POWER OPERATIONS FIRE FREQUENCY COMPARISON DATA SOURCE 1965 1985 1986 1994 -I -
1965 1994 Stlucie 1&2 l Turkey Pt. 3&4 -
l l
Quad Cities 1&2 l Cooper -
l Pt Beach 1&2 -
l l Indian Pt. 2&3 l Mellstone 3 - No Data Diablo Canyon i Newaunee No Data LaSalle l Seabrook(PRA) l l
l Seabrook(EPRI) l Pch bottom (PRA) l PCH Bottom (EPRI) -
l FRSS -
, ,l , , ,
0 2 4 6 8 10 12 14 FREQUENCY a E 3 CABLE SPRD ROOM POWER OPERATIONS FIRE FREQUENCY COMPARISON DATA SOURCE 1965 1985 -
1986-1994 1965 1994 -
Stlucie 1 &2 l Turkey Pt. 3&4 -
l Ouad Cities 1 &2 -
l Cooper l )
Pt. Beach 1&2 l Indian Pt. 2&3 -
l Millstone 3 -
l Diablo Canyon - No Data Kewaunee - l LaSalle l SeabrooktPRA) -
l Seabrook(EPRI) F Pch Bottom {PRA) l PCH Bottom (EPRI) l FRSS -l , , , , ,
0 5 10 15 20 25 30 l
FREQUENCY a E 3 FIGURE 8 i
l
a SWITCHGEAR ROOM POWER OPERATIONS FIRE FREQUENCY COMPARISON DATA SOURCE 1966 1985 1986 1994 -
1965 1994 -
Strucie 1&2 g, p ,,
Turkey Pt. 3&4 No Data Quad Caties 1&2 No Data Coope' No Data Pt. Beach 1&2 l Indian Pt. 2&3 No Data Millstone 3 No Data Diablo Canyon No Data 1(owaunee l LaSalle l Seabrook(PRA) No Data SeabrooklEPRI) No Data Pch Bottom (PRA) l PCH Bottom (EPRil l FRSS , No Data ,
0 6 10 15 20 25 30 FREQUENCY a E 3 REACTOR BLDG (BWR) POWER OPERATIONS FIRE FREQUENCY COMPARISON DATA SOURCE 1965-1985 i i 1986 1994 i i
- $;.394 - = i Quad Cities 1&2 No Data Cooper f
LaSalle -
1 No Data Pch Bottom (PRA) l l
PCH Bottom (EPRil No Data l 0 20 40 60 80 100 120 FREQUENCY a E-3
AUXILIARY BLDG (PWR) POWER OPERATIONS FIRE FREQUENCY COMPARISON DATA SOURCE 1965 1985 l 1986-1994 lC 1965 1994 l St Lucse 1&2 -
l Turkey Pt. 3&4 l Pt. Beach 1&2 -
l Indian Pt. 2&3 l Millstone 3 No Data Diablo Canyon No Data Kewaunee l Seabrook(PRA) l SaabrooklEPRI) No Data FRSS
,l , , ,
l 0 20 40 60 80 100 120 140 160 FREQUENCY a E 3 l
TURBINE BLDG POWER OPERATIONS l
FIRE FREQUENCY COMPARISON DATA SOURCE 1965 1985 1986 '9a4 1966-1994 St. Lucia 1&2 l Turkey Pt. 3&4 -
l Quad Cities 1&2 No Data Cooper -
l Pt. Beach 1&2 -
l Indian Pt. 2&3 No Data Millstone 3 -
g, p ,,
Diablo Canyon - l Kewaunee l LaSalle l Seabrook(PRA) l Seabrook(EPRI) No Data l
Pch Bottom (PRA) - No Data PCH Bottom (EPRI) -
No Data FRSS -
0 20 40 60 80 100 FREQUENCY a E 3 FIGURE 10