Issue 148: Smoke Control and Manual Fire-Fighting Effectiveness

DESCRIPTION

Historical Background

This issue was raised in SECY-89-170[1] and addressed the effectiveness of manual fire-fighting in the

presence of smoke. This concern arose as a result of an NRC-sponsored Fire Risk Scoping Study[2] which focused on existing fire protection practices for control rooms, remote shutdown areas, control transfer areas,

and local control areas. In addition, Item 2.3c, "Smoke Control," identified in NUREG-1251[3] expressed concern over smoke propagation from one unit to an adjacent unit.

In general, lubricating oils and cable insulation are the primary fire sources found in nuclear power plants. Both of these sources represent the most prolific smoke-generating fuel. Experimental evidence indicates that burning

such fuels in a typical nuclear power plant enclosure would obscure the entire enclosure in about 10 minutes.[4] In actual experience, fire-fighters have had difficulty in seeing the fire source because of smoke (Browns Ferry, 1975) and equipment is known to have failed in smoke-filled environments.

Safety Significance

Smoke can impact plant risk in several ways:

(1)          Smoke can reduce manual fire-fighting effectiveness, cause misdirected suppression efforts, and subsequently damage equipment not directly involved in the fire.

(2)          Electronic equipment can be damaged or degraded by smoke resulting in functional loss or spurious response. Very little experimental data on equipment response in smoke environments were available at the time this issue was evaluated in August 1992 and the methodology for including smoke in PRAs had not been adequately developed. Additional research efforts were believed to be required to fully address the risk impact of smoke on safety-related systems.

(3)          Smoke can hamper an operator's ability to safely shutdown a plant by causing evacuation of control centers and subsequent reliance on backup shutdown capability.

(4)          Smoke can initiate automatic fire protection systems in areas away from the fire, potentially damaging safety systems and components. (This item was addressed separately in Issue 57, "Effects of Fire Protection System Actuation on Safety-Related Equipment.")

NUREG/CR-5088[5] focused primarily on Item 1. Using information developed as part of the Risk Methods Integration and Evaluation Program (RMIEP) on the response of fire-fighters to specific areas of the LaSalle plant, sensitivity studies were performed on four PRAs. These studies showed the variation in CDF as a result of fire-fighting response time and misdirected suppression efforts. A discussion of the methods used and results of the study are provided below.

Impact of Manual Fire-Fighting Response Time: Smoke can increase fire risk by prolonging fire-fighting response time. With the LaSalle nuclear plant as a model, walkdowns by fire protection engineers as part

of the Fire Risk Scoping Study[6] established bounds on the time to detect, apply suppression agents, and successfully suppress fire for all critical plant areas. This information was then applied to the previously reviewed fire-initiated core damage scenarios in the four selected PRAs (Oconee, Seabrook, Limerick, Indian Point-2).

Thirteen plant areas were grouped by area, equipment contained in the area, available suppression equipment, and type of detection. These areas were partitioned into the following five groups:

(1)          Oconee (Cable Shaft), Indian Point-2 (Electrical Tunnel, Cable Spreading Room), Seabrook (Cable Spreading Room)

(2)  Seabrook (Control Room)

(3)  Seabrook (Turbine Building)

(4)  Limerick (13 kv Switchgear Room), Oconee (Electrical Equipment Room), Indian Point-2 (Switchgear Room)

(5)          Seabrook (PCC Pump Area), Limerick (Safeguards Access Area, CRD Hydraulic Equipment Area, General Equipment Area)

However, only the analyses of Groups 1, 4, and 5 specifically considered the effect of manual suppression efforts on the mitigation of critical damage. The control room area, Group 2, did not allow successful suppression, and Group 3, the Seabrook Turbine Building, did not lead to core damage. Group 1 corresponded to the LaSalle cable spreading room, while Groups 4 and 5 corresponded to the LaSalle essential switchgear room and

large areas of the reactor building, respectively. The results are shown in Table 3.148-1. The minimum and maximum times are representative of the most and least effective fire brigades, respectively, and the average time represents a typical fire brigade. Although the time to detect the fire, report to the suit-up area, and suit-up are all important contributors to the response time, the major time elements (up to 75%) include: (1) response to scene; (2) set-up at scene; (3) scene search; and (4) time to suppression or substantial control. Given a smoke- filled environment, times associated with each of these four elements can be prolonged substantially.

Impact of Misdirected Suppression Efforts: NUREG/CR-5088[7] assessed the effect on CDF of a fire brigade damaging equipment not directly involved in the fire. The assessment included:

(1)  Identification of components susceptible to spray, flooding, or temperature within the fire area.

(2)  PRA re-quantification, assuming susceptible components fail by suppression efforts.

(3)          Identification of important areas and probability of spraying essential equipment not involved in the fire but located in those areas.

(4)          Combined with fragility information, determination of the conditional probability of suppression-induced failure.

Table 3.148-1

Effects of Fire Brigade Response and Extinguishment Time on CDF Due to Fire

 

Plant	Area	CDF/Year Response and Extinguishment Time
		Minimum	Average	Maximum	Original PRA Value
Seabrook	Cable Spreading Room	8.0 x 10-7 (mean)	4.9 x 10-6	8.9 x 10-6	4.1 x 10-6
	PCC Pump Area	1.4 x 10-5 (mean)	6.2 x 10-5	1.0 x 10-4	7.2 x 10-5
Oconee-3	Cable Shaft	5.3 x 10-6 (point estimate)	1.1 x 10-5	1.4 x 10-5	1.0 x 10-5
	Electrical Equipment Room	5.4 x 10-9	1.5 x 10-8	2.0 x 10-8	1.6 x 10-8 (point estimate)
Indian Point-2	Cable Spreading Room	1.1 x 10-7	1.2 x 10-6	2.5 x 10-6	1.9 x 10-6 (mean)
	Electrical Tunnel	9.3 x 10-6	4.7 x 10-5	7.6 x 10-5	5.0 x 10-5 (mean)
	Switchgear Room	2.2 x 10-6	3.0 x 10-5	6.3 x 10-5	5.6 x 10-5 (mean)
Limerick	13 kv Switchgear Room	6.0 x 10-7	4.7 x 10-6	2.8 x 10-5	6.2 x 10-6 (point estimate)
	CRD Hydraulic Equipment Area	5.0 x 10-7	8.3 x 10-6	2.1 x 10-5	6.4 x 10-6 (point estimate)
	General Equipment Area	3.9 x 10-7	3.6 x 10-6	1.5 x 10-5	1.9 x 10-6 (point estimate)

The Limerick PRA contained areas in which safe shutdown would be lost if fire and/or fire suppression activities failed all components in the fire area. To determine the significance of failing equipment by misdirected fire suppression efforts, the following methodology was used:

(1)  Compare the screening value of CDF from the original PRA for a fire area to its final adjusted value.

(2)  Determine the method(s) of fire suppression available in the area.

(3)  Determine access routes to the area.

(4)  Assess the probability of accurate, location-specific detection of a fire within the fire area.

(5)  Assess the potential for smoke buildup and visible obscuration of the fire.

(6)  Determine what method would be used to discover fire location.

The upper bound (screen value) [8] for the potential impact of misdirected fire suppression efforts for each area of concern was compared to the PRA estimate below. The reduction factor (Screening Value divided by the PRA Value) shows the reduction in CDF, as a result of successful fire mitigation.

 

Fire Area	Screening Value	PRA Value	Reduction Factor
13kv Switchgear	2.5 x 10-3	6.2 x 10-6	403
Safeguard Access Area	3.8 x 10-3	6.0 x 10-6	633
CRD Hydraulic Equipment Area	2.5 x 10-3	6.4 x 10-6	390
General Equipment Area	2.8 x 10-3	1.9 x 10-6	1473

Because of the large area (approximately 10,000 square feet) and large open equipment hatchways (200 square feet) for mitigating smoke buildup, certain fire areas were screened from further analysis on the basis that fire- fighters would identify the source of the fire through its generation of a smoke plume.

Although Limerick's design features reduced the risk of misdirected fire suppression efforts, two important safety concerns were raised:

(1)          Fire and suppression damage (or smoke if equipment is susceptible) confined to a single fire area can lead directly to core damage.

(2)          A large reduction factor (up to 1473) is needed to reduce the fire-induced core-melt frequency to a reasonably low level.

In summary, the above sensitivity studies indicate the safety significance of smoke. Through variations in the fire- fighting environment, CDFs ranged from 1.4 x 10^-6/year to 3.8 x 10^-5/year. In addition, the impact of misdirected suppression efforts because of smoke (or the effects of smoke directly if the equipment is susceptible) could

be substantial, i.e., a CDF on the order of 10^-3/year, if no credit for fire suppression efforts is given. This issue affected all operating and future plants.

 

Possible Solution

A possible solution was to use the above methodology to search for plant-specific vulnerabilities to smoke and smoke propagation from area to area or unit to unit. This information would then be used to: (1) select effective smoke removal means to preclude potential equipment damage and enhance fire-fighting capability; (2) select appropriate detection and suppression systems in various fire areas (with due consideration of Issue 57); and (3) provide guidance in developing fire response plans.

PRIORITY DETERMINATION

Assumptions

It was assumed[9] that the issue affected 134 operating and future plants with an average remaining life of 28.3 years.

Frequency/Consequence Estimate

The safety significance of this issue was evaluated[10] by PNL as well by SNL.[11] Comments provided[12] by NRR were also considered in this evaluation. The discussion presented by NRR pointed to a number of

important elements of fire protection at nuclear power plants. Although there were weaknesses in the program and areas of potential improvement, the description and potential merits of fire protection programs at nuclear power plants are based on an ideal implementation of all elements and additional prudent steps taken by

licensees beyond those already mandated by regulatory requirements.[13] However, the objective of this issue was to assess the effectiveness of certain elements of fire protection, namely, smoke control and manual fire-

fighting effectiveness. Hence, the insights and data developed as part of NUREG/CR-5088,[14] as well as operational experience, were taken into account in evaluating this issue.

There were many differences in the models used in the PNL[15] and SNL[16] analyses. Therefore, the absolute values of CDF were not used, but the changes in CDF resulting from the sensitivity studies were. The analyses and data used by PNL and SNL are summarized below.

The Oconee-3 PRA was the basis for the PNL analysis with three large-fire-initiated accident sequences

dominating the risk associated with this issue.[17] However, as pointed out by SNL, certain assumptions made and data used in the PNL analysis should be adjusted to more realistically reflect operational experience, as well as the results of the SNL analysis contained in NUREG/CR-5088.[18] Based on the PNL approach and taking into account the insights of SNL and cognizant NRC staff, the following adjustments to the PNL assumptions were made to obtain a more realistic assessment of the potential risk associated with smoke control and manual fire- fighting.

The PNL analysis[19] assumed that the base case mean fire suppression time (T_s) was 14 min utes. Furthermore, it was assumed by PNL that the postulated resolution would reduce T_s to 11 minutes. PNL also used a value of

6.7 minutes for T_g, the time required for fire growth and equipment damage based on the Oconee-3 PRA. Based on these assumptions, the following values for CDF due to smoke control and manual fire-fighting and risk were calculated by PNL.

 

	CDF/RY	Public Risk (man-rem/RY)
Base Case	1.0 x 10-5	2.3 x 10-1
Adjusted Case	8.6 x 10-6	2.0 x 10-1
Change	1.4 x 10-6	3.0 x 10-2

 

The data developed in NUREG/CR-5088,[20] however, provided a different set of values for T_s: the base case mean suppression time (T_s) was 42 minutes and the solution reduced this value to about 11 minutes. These

values were more realistic based on the ranges of T_s developed by SNL.[21] Specifically, the following ranges of suppression times were available: (a) 5 to 60 minutes, based on fire protection expert analysis of specific plants; and (b) 2 minutes to 5 hours, based on LER data. For T_g, a value of 15 minutes was deemed more realistic for the time required for fire growth capable of substantial damage. Based on these assumptions, the following respective values for CDF and risk were calculated:

 

	CDF/RY	Public Risk (man-rem/RY)
Base Case	4.1 x 10-5	9.5 x 10-1
Adjusted Case	1.6 x 10-5	3.5 x 10-1
Change	2.5 x 10-5	6.0 x 10-1

The potential public risk reduction associated with the issue was (134)(28.3) x (0.6) man-rem or 2,275 man-rem.

Cost Estimate

Industry Cost: Resources for implementation were estimated to be required for two major activities. The first was to search for the potential vulnerabilities identified in this analysis. This search was estimated to require approximately 0.5 man-year or $50,000/plant for reviewing plant drawings and existing fire hazards analyses and a walk-through inspection of potentially susceptible areas. The second main activity was to install improved smoke removal, fire detection, and suppression systems where necessary. A nominal $10,000/plant equipment procurement cost plus an additional 4 man-weeks to install the improved equipment were estimated. This 4 man-weeks labor estimate was increased to account for inefficiencies in nuclear power plant labor productivity resulting from access and handling difficulties, work in radiation zones, congestion and interference (factor

of 1.7) and equipment removal (factor of 2.7). At $2,270/man-week, these labor costs were estimated to be approximately $40,000/plant. Thus, the total implementation cost was estimated to be $100,000/plant and $13M for all affected plants.

There were no major new requirements for periodic inspection or maintenance activities that were not already in place. A nominal 1 man-day/RY or $454/RY was added to account for increased operation and maintenance of the improved smoke control, fire detection, and fire suppression systems that were proposed to be installed to replace existing vulnerabilities. The total operation and maintenance cost was estimated to be $1.7M for all affected plants. Thus, the total industry cost was $(13 + 1.7)M or $14.7M.

NRC Cost: NRC development costs were estimated to be incurred for development of fire protection program guidance in the area of smoke control and for preparation and issuance of a generic letter that would transmit the new guidance to all licensees. A nominal 2 man-years or $200,000 were estimated for development of the fire protection guidance. The NRC labor needed to prepare and issue a generic letter was estimated to be

approximately 4 man-weeks or $10,000.[22] Thus, the total cost associated with development of a solution was estimated to be $210,000.

It was estimated that it would require 5 man-weeks/plant to review and approve implementation of the enhanced smoke control program and improved equipment and 7 man-weeks/plant to prepare a safety evaluation.[23] At

$2,270/man-week, implementation costs were estimated to be $3.65M for all affected plants.

Review of licensee operation and maintenance was estimated to require 1 man-day/RY or $454/RY. For the 134 affected plants, this cost was $1.7M. Thus, the total NRC cost was $(0.21 + 3.65 + 1.7)M or $5.56M.

Total Cost: The total industry and NRC cost associated with the possible solution was estimated to be $(14.7 + 5.56)M or approximately $20.26M.

Value/Impact Assessment

Based on a potential public risk reduction of 2,275 man-rem and an estimated cost of $20.26M for a possible solution, the value/impact score was given by:

 

CONCLUSION

Based on the above results, the issue fell in the high priority range, on the basis of CDF, and in the medium priority range, on the basis of risk. However, the safety significance was likely to vary greatly from plant to plant

and it appeared unlikely that any cost-effective generic resolution could be identified. Thus, it was believed that plant-specific reviews would most likely be required. Such reviews were already required as part of the IPEEE Program. However, the staff had little or no guidance for the review and acceptance of IPEEE submittals in this

area. Therefore, the issue was classified[24] as a Licensing Issue in August 1992 to allow the staff to develop guidance to improve its effectiveness in the review of licensee IPEEE submittals. The issue was later closed out

when the staff completed the review guidance and incorporated it into the overall IPEEE review guidance.[25]