ML20205A924

Forwards Description of Qualitative Analysis to Provide Basis for Acceptability of two-minute Background/Reference Check Every Hour During Time One Monitor Is Oos,Per NRC RAI Re TS Change Request NPF-38-211

ML20205A924
Person / Time
Site:	Waterford
Issue date:	03/25/1999
From:	Ewing E ENTERGY OPERATIONS, INC.
To:	NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM)
References
W3F1-99-0051, W3F1-99-51, NUDOCS 9903310100
Download: ML20205A924 (7)

Text

.

e Of

. Ent perations. Inc.

Killona. LA 70066 Tel 504 739 6242  ;

I Early C. Ewing, ll1 l

'* ucIsar hfety & Regulatory Affairs W3F1-99-0051 i A4.05 PR 1

March 25,1999 i

' U.S. Nuclear Regulatory Commission Attn: Document Control Desk Washington, D.C. 20555

Subject:

Waterford 3 SES Docket No. 50-382 License No. NPF-38 Request for AdditionalInformation Technical Specification Change Request NPF-38-211 Broad Range Gas Detection System Gentlemen:

In response to a request from Ms. Carol Lauron of your Staff, Entergy is hereby /

providing the attached description of the qualitative analysis to provide the basis for the acceptability of the two-minute background / reference check every hour during the time one monitor is out .of service. By Letter W3F1-98-0167 dated October 1,1998, Entergy submitted Technical Specification Change Request (TSCR) NPF-38-211. The I N proposed change modifies TS 3.3.3.7.3 and Surveillance Requirement 4.3.3.7.3 for the broad range gas detection system. A change to the Technical Specification Basis 3/4.3.3.7.3 was included to support this change. This change to the Technical Specifications is necessary for the installation of a new, more reliable broad range gas detection system.

l

~

9903310100 990325 ADOCK 05000382 ,

PDR j P PDR , t

l J

' Request for AdditionalInformation .

Technical Specification Change Request NPF-38-211 Broad Range Gas Detection System i W3F1-99-0051 I

- Page 2 March 25,1999 l

1 Should you have any questions or comments concerning this request, please contact

. me at (504) 739-6242 or Curt Taylor at (504) 739-6725.

Very truly yours,

/ E.C. Ewing Director j Nuclear Safety & Regulatory Affairs 1 ECW/ CWT /rtk

Attachment:

cc: E.W. Merschoff, NRC Region IV C.P. Patel, NRC-NRR C.L. Lauron, NRC-NRR ,

J. Smith I N.S. Reynolds  ;

NRC Resident inspectors Office Administrator Radiation Protection Division (State of Louisiana)

American Nuclear insurers

! 1 l

l l

1 l

1 L-

r- .

i - .

l O l .

W3F1-99-0051 ATTACHMENT 1 AdditionalInformation l

3 i

1 i

I I

l l

l l

. .R

Information Reauested With one channel of the BRGM out of service (OOS), when the second monitor performs its background check, both channels will be unable to perform the function of isolating the control room for the short period of time of the background ' check. The TSCR states that this is acceptable because:

" Qualitative analysis based on a quantitative risk assessment has shown that the impact on operator incapacitation and subsequent core damage risk of the background checks while one monitor is out of service for its 7 day allowed outage is negligible." A description of this qualitative analysis sufficient to provide the basis for the acceptability of the 2 minute background check every hour during the time one monitor is OOS was requested.

l

Response

l The qualitative analysia was based on the results of a toxic chemical external events probabilistic risk assessment (PRA) developed by Waterford 3 in 1997.

This PRA started with the significant toxic chemical hazards from tP design I basis calculations and performed an external events PRA on th' sgnificant hazard chemicals (chlorine, ammonia, and hydrogen chloride). The PRA used a functional event tree approach which included the functions of monitor actuation, control room isolation, operator donning of emergency breathing air masks, decay heat removal, and offsite recovery. Fault trees were used to represent failure of the monitor, isolation, and decay heat removal functions, and a human reliability model was used to represent the failure of the breathing mask and offsite recovery functions, which are primarily operator actions. (The unavailability of the emergency breathing air system was added to the operator failure probability, however, to capture that failure mode.)

The purpose of the BRGr1 system is to alert the operators to the presence of a hazardous chemical at the control room air intake by alarrning in the control room and to actuate the control room isolation function. In normal operation, outside I air is brought into the control room at the rate of about 0.6 air exchanges per l hour. Although after isolation no air is brought in from outside, the PRA conservatively assumed the pressurized makeup inflow rate of about 0.06 air  ;

exchanges per hour for the isolated condition. This meant that even with  !

isolation, the toxic chemical concentration could increase inside the control room. j The effect of isolation was to slow the rate of increase and give the operators I more time to don their emergency breathing air masks. l There were three major operator incapacitation scenarios considered in the l event tree:

1. Both monitors fail and the operators fail to don their emergency breathing air masks before they are incapacitated. Because the monitors fail, the

control room is not isolated and, under worst case spill and meteorological conditions, the control room concentration could increase rapidly. In

, , addition, since neither monitor actuates, the operators are not alerted to the hazard until they smell the chemical in the control room. This gives them little time to respond and produces the highest operator failure probability.

l

2. At least one monitor operates, but the isolation function fails. This scenario is similar to the first, but because the opemtors receive an alarm in the control room, they have an earlier waming of the hazard and the probability of their failure to don emergency breathing air masks is somewhat lower.

3. At least one monitor operates and the control room isolation succeeds, but the operators are not able to don their emergency breathing air masks in time. Because of the successful control room isolation, the

( concentration in the control room would build up more slowly, giving the operators more time to respond. This scenario has a lower probability of operator incapacitation than the first two scenarios. )

l For each of these scenarios, the probability of operator incapacitation was

calculated. Incapacitation was represented as a failure to make the decision to .

don emergency breathing air masks in time to actually don the masks before the -

control room concentration reaches the incapacitation level. The probability of the decision failure was calculated using a time reliability correlation, a standard l human reliability method used in PRAs. The time required to don emergency breathing air masks (2 minutes or less from emergency drills) was subtracted J from the time to incapacitation to give a time available to make the decision.

Then the decision failure probability was calculated as a log-normal function of the time available for the decision.

The time to incapacitation was calculated from the time-dependent control room concentration using the incapacitation modelin NURF 'CR-1741. The control I

room concentration was estimated by calculating a tir.. dependent vapor release rate from the spill, then using a gaussian plume dispersion model to compute the toxic vapor concentration at the inlet to the control room, and finally determining the control room concentration using the air exchange rate. (For the third scenario, in which isolation was successful, the monitor actuation time was calculated from the outside air toxic vapor concentration, and after the calculated l monitor actuation, the air exchange rate was reduced to the isolated rate.) The vapor release rate, plume dispersion, and control room buildup calculations used methods similar to NUREG-0570.

l l Because the atmospheric stability class and wind speed are two key parameters

( of the release rate and dispersion equations, the stability class and wind speed ranges were divided into bins (i.e., elements of a matrix of u ~ bility classes and t

i

w l

wind speed intervals), and the control room concentration / operator incapacitation calculation was done for each bin. The total operator incapacitation probability

, was calculated by weighting the incapacitation probability for each bin by the joint probability of the bin (the probability of the stability class-wind speed interval combination), and summing the resulting weighted probabilities over all the bins.

The probability of core damage given operator incapacitation was estimated from i the Level 1 (core damage) PRA model by setting the reactor trip initiator probability to 1.0 (i.e., a reactor trip would definitely occur) and all operator I

actions to failed. The possibility of recovery of the plant by the next shift was  !

included (with a conservatively high failure probability), since it would most likely I take many days for a reactor trip and subsequent core damage to occur following I operator incapacitation.

The result of the toxic chemical PRA was that the operator incapacitation 4

frequency was 1.7x10 per year for the scenario in which both BRGMs failed. '

The core damage frequency was 1.2x10-12 per year for this scenario. These values included the actual monitor unavailability's (fractions of the time OOS) .

from plant experience: 5.0x10-2 for channel A,8.6x10-2 for channel B, and j 1.7x10" for both monitors at the same time. I The impact on operator incapacitation and subsequent core damage risk of the background checks while one monitor is out of service for its 7-day allowed outage time was estimated as follows. Assuming the limiting case of a 2 minute background check every hour, and conservatively assuming one BRGM channel out of service PERMANENTLY, the unavailability of the BRGM system (both channels) would be 2 minutes OOS / 60 minutes = 3.3x10-2 ,

Adding this unavailability to the probability of both monitors failing (7.0x10' ) from the event tree increases the monitor failure probability by a factor of

( 7.0x10 + 3.3x10-2 ) / 7.0x10' = 5.7. )

l Increasing the operator incapacitation frequency for the scenario of both monitots failed by a factor of 5.7 gives an incapacitation frequency of 4

1.7x10 x 5.7 = 9.7x104per year, or an increase of 4 4 4 9.7x10 - 1.7x10 = 8.0x10 per year. j This increase is negligible.

l l

l

[

l'

.- 4 l

l Since the core damage frequency is 3 orders of magnitude lower than the operator incapacitation frequency, the effect on CDP is about 3 orders of

, magnitude lower than this, or about 10'" per year. Regulatory Guide 1.174, "An Approach for Using Probabilistic Risk Assessment in Risk-Informed Decisions On Plant-Specific Changes to the Licensing Basis," identifies a CDF increase of 4

less than 10 as very small and acceptable for license amendments. Since the effect of the background check, even assuming one monitor is never available, is l 5 orders of magnitude below the Regulatory Guide 1.174 criterion, the background check is acceptable.

l l

i