ML19241C061
| ML19241C061 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 04/27/1979 |
| From: | Audette R Office of Nuclear Reactor Regulation |
| To: | Rosztoczy Z Office of Nuclear Reactor Regulation |
| References | |
| NUDOCS 7907260503 | |
| Download: ML19241C061 (5) | |
Text
.
a 'f %
.?p uNtTED STATES J
.a.
y '9 WUCLEAR REGULATORY COMMisslCN
$.gn4 3
j WASHINGTcN, O. c. C555
/ :,
su s.
.np Apq 2 7 379 MEMORAt! CUM FCR:
Z.R. Ros: toc:y, Chief, Analysis Branch, CSS FRCM:
R.F. Audette, Systems Analysis Secticn, Analysis 3 ranch, CSS hP.E.florian,SectionLeader,SystemsAnalysisSecticn, TH R'J :
Analysis Branch, DSS SU3 JECT:
A?lALYT! CAL SUFFCRT TO.'lRC VAtlAGEME:lT FCR CFERATING PL2:li ::lCICENTS Of the many lessens to be learnec frca the recen: Three Mile : stand Uni 2 (TMI-2) accident, one of the ore im::crtant cnes was tnat only limited infor ation was available for NRC management ccncerning the clant status. The infor aticn tnat was initially available was largely inadequate, c:nfusing, and often contradictory from all accounts, as would be ex;:ected in any first-of-a-kind accident.
This accident has, as incicated by Ccmmissioner Ahearne, served to spot-lignt many procedural weaknesses whicn are : rrectible to varicus degrees, and the folicwing are scme suggestions fer carrecting cne of the wea<nesses that I believe falls within our branch cnarter res;:cnsibilities.
Infor.aticn as to the sequence of events that occurred at NI-2 was quite sicw ( ~ 3 days) in reacning the A3 staff, and what information was furnished was incomplete in terms of being able to clearly define wnat had occurred, and what the status of ne plant was a: the time when management decisions were needed. As a result, our succcrt and contributions to the decision process was hampered, wnen critical decisions had to be made compared to what I believe wculd be cssible with scme upgrading of our Brancn ca:: abilities. To rectify this inadequacy, I believe that the follcwing suggestiens for u:: grading of our capacilities and functions should be implementec over One short and long term; A.
Short Term Uccradinc 1.
Incident Infor nation Transmittal to NRC Headcuarters To obtain arcmat and as comolete infor ation as is available en recent events or events in crogress at any acerating :lant, it is suggestec :na: a plan: incicent data ac;uisition (?:DA) system be imolementec at NRC headcuarters. This system would 790726
- 3 G4S205 g 4
Z.R. Ros:::c:y consist of digital ccm: uter taces and tape drives designec to receive and st:re data transmitted frcm any selected plant site over teleonone lines. As each plant site new contains some form of digitized legging of plant measurements wnen events occur in acdition to aut mated logging of events, such infor-mation could be readily transmitted electronically to NRC headquarters for staff review and analysis.
2.
Cata Conversion for Review and analysis Frcm the ?!CA ta:es, charts of events anc 010:2 Of selected parameter transients for tne incident could be generated on existing NRC headcuarters equipment aithin minutes of trans-mittal frca the site. This would recuire premet, Or i m ediate, access to ccmcuter services to convert the tape data to gracnics equipment c:= ands.
Information converted to such a format would serve as a casis for prcmot and timely inci-dence assessments by the staff and management.
3.
Incident Analvsis Fr:m the incidence secuence of events and parameter transients data, incut data mcdificaticns to an existing plant mcdel acercximating the plant in puestien eculd be inclemented, and attemots to approximate tne incidence transient c uld be made for the pur:ose of assessing un-measured or un-measurable plant parameters, and to pre-predict suggested courses of action for an on-going incident.
From these pre-precictions, a rec menced course of action for an on-going incident could be mace.
This phase of the AB c:ntribution to the TMI-2 accident evaluaticn was not initiated en a timely basis due to the lack of accurate and c:mplete infor aticn an the accident chronology as was indi-cated earlier.
!malementaticn of the first step in the crocosed uegrading process wculd obviate this prooiem, and would place tne AB in a position to res;cnd within a few hours, or in One day to accidents such as occurred at T'iI-2.
This phase of our support to management in such incidences would necessarily be the most time censuming, and would be limited by
- m:: uter availability, problem run times, and the numcer of cre-predictions of possible courses of acticn.
543206
n.
Z.2. Ros:: c:y 4.
Information Feecback to NRC Site Recresentatives Information feecback to NRC site recresentatives follcwing incicents, or during on-going incidents to assist their assessments and selection of actions to follcw could ce transmitted verbally, and enar:s and gracnics genera:ed by Headquarters facilities C0uld be transmittec by telec Dy to equip them with all informa".icn available to the staff u:cn wnicn staff recommendations are made.
3.
L0nc Term U0 grading The shcr: :erm u grading proposed abcVe is considered feasible for implementaticn in a to 6 months as only tne information acquisition portion of this pnase of upgrading will invcive some ccmcuter programming and utility information access arrangements.
For a truly adequate information and analytical base for assessing incidents in coerating piants, and for recommending courses of action for en-going incicents, a more fully ccmouter based and automated system resembling scace-age telecommunica:icns and control systems to a minor extent is recuired. The principles involved are cuite simple as has been cutlined in the short term u grading anase.
The principal difference for the long term ucgrading will be in the degree of automated data transfer between the site and NRC headcuarters, and the use of plant specific models for the analytical portion of the brancn support.
The central feature of the crcccsed system will be the use of a design data base containing all pertinent design data recuired to racidly assemoie a LCCA or systems transient analysis mcdel.
Current plans to develco such a data base in c:nnection with the WRAP system in devele: ment at SRL are being formulated for i=clementatien in Ff'30, and will be described separately, elsewhere.
The basic plan for use of such a data base is to provide the design data source, cata handling, and programming to permit a fully automated and balanced plant specific mcdel for analysis of any operating plant or plant in licensing review.
Storage of'sucn design data, much of wnich is commen to several plant designs, in a readily retrievable da a base such as provided by :ne JCSHUA system 3: $RL will :ermit pr0=ct access for manipulation and develc: men Of an analytical mccel. The self-initiali:ati:n fea ure of the WRAF system for pWRs anc SWRs will provide ne Ore-transient steady-state :alanced c:ndi:icns recuired for transien: analyses from any specified initial Operating condition.
5492C7
Z.R. Ros::Oc:y With a fully automated cdel assemoly procedure, such as planned for WRAP using ne JOSHUA system data base, the addition of a sequence of inout events from an incident chronology cbtained frcm a plant ccmcuter events log would be straignt foraard, and could be transmitted directly as an input perturbation secuence to WPAP once the plant in puestion nas been identified, initial conditions speci-fied, and the staady-state balance ccmcuted. VeMfication that the automatically assemcled model was a:proprice and adecuate c:uld be evaluated :y running a ccmcariscn of c:=u:ec ut:u fecm the model for :ne prescribed incut events si:n the measured :arameter cutouts from tne plant anica nave been transmitted feco the -ite and :Ic::ec as in stec 2 of :ne shcr: term ucgrace. The c =aris:n snaula be run in parallel with the model calculations anc be disclayed en terminal output graphics to provide a running assessment of the automated. cdel adequacy. The ccmparisen will indicate any need for user intervention in changing model carameters to obtain an acceotable ccmcarisen, or :0 fine tuning as it is otheraise kncwn.
From sucn a mccel and c:= arisen calculations, :ne basis for assessing possible actions to be :aken in on-going incicents such as cccurred at TMI-2 could be developed on a timely basis to assist management decisions.
Informaticn transmittal to the NRC recresentatives at the site could be cirect from the ccmcuter ccmcarison over to terminal graphics equipment at the site cc= arable to that at the NRC ccmcuter so that simultanecus assessments could e made by staff and site representatives.
The long term concept for craviding an utgrading of our support caca-bilities to NRC management in the event of accicents such as occurred at TMI-2, or for less severe incidents that may be of interest to NRC staff and management is based on existing ecuitment cacacilities much of which is already in place or planned for the near future.
One significant feature of the short and long term precosals described above is the s sff depencence On CCE La: oratory com uters which is considered to be the primary weak link in the entire process. This weakness stems from difficulties in c = uter access en a high priority basis when needed for NRC werk on COE mchines, the need for two staffs, NRC and CCE Labs, to be involved when output results are needed en an excediticus basis, and the need for out:ut data transmission to *wo te ninals, NRC headcuarters anc the plant site frem a CCE location wnen management decisions are neeced promp !y.
For a viaale racid rescanse ca:acility sucn as is neecec unen oiant accidents or inci-dents occur, the avaiiacility of an in-hcuse ccm uter main-frame cacaole of :erforming all our analytical studies uncer staff ;uicance G482C8 M
I.R. Ros: toc:y and control is almost mandatory, anc is strongly recommended for procer imolementation of this entire support capacility.
Many more details of the support senemes, short and Icng term, are neeced to comoletely describe each plan, out tne general outlines are presented in the foregoing discussicn, and will :e cescribed more fully if this proposed plan is of further interest.
f&/*;
_5 i
g,e,--
. /, @juvdIW R.F. Audette Systems Analysis Section Analysis Branch Division of Systems Safety R..v ttscn cc:
a
'/. Stallo H. Thornburg R. Tedesco
- 5. Hanauer C. Tecutman P. Check P. Norian L. Phillips
- 5. Fabic L. Shotkin M. Lauben J. Guttmann R. Audette 5432c3
_