ML19322A408

Submits That Questions of Philosophy Re TMI Investigation Must Be Settled Before Specific Recommendations for Practical Implementation Are Given

ML19322A408
Person / Time
Site:	Crane
Issue date:	06/26/1979
From:	Theofanous T PURDUE UNIV., WEST LAFAYETTE, IN
To:	Plesset M CALIFORNIA INSTITUTE OF TECHNOLOGY, PASADENA, CA
References
ACRS-CT-1138, NUDOCS 7910240425
Download: ML19322A408 (3)

Text

_ __ _ _ ____ __ _ _._________________

<i-9 PURDUE U N I V E R S ITY O g p A se 7 ais t M T OF NWCLEAIt E N G IN E E seeN G wf S T L A s A V E T T E, GN DI A *e a 4 7,0 7 C 7-H 3 8 June 26, 1979

[g kJ3[7<f LLCunD ADVISORY cor.::ITTEI 03 EEACToR SAFEGUL*2S U.1 Nu

(.h. 319K P!

F"Y n

n:

Professor M. Plesset Division of Engineering &

Applied Science California Institute of Technology Pasadena, California 91125

Dear Professor Plesset:

Inis letter is to express my deep concern for the manner in which we are responding to the TMI-2 accident. My thoughts are along the lines presented in my June S letter to Professor Okrent of which you are already aware. The June 19, 20 meeting of your ECCS subcommittee reinfo ced my views and I would like to continue with the present letter.

In my June 8 letter I argue on matters of philosophy anc overall approach.

Both you and Professor Okrent indicated, independently, that you would like to see more specific recommendations for practical 4mplementation.

I believe that coming up with such unilateral recomendations, before the matter of philosophy has been settled, may distract from the rc.ajor goal by bringing up what I consider as the detailed aspects of the r:atter.

I will venture into the area of specifics in this letter with the understar; ding that there are many ways to obtain the desired goal which can be equally acceptable as long as they satisfy the requirements of clear philosophy towards establish-ing and synthesizing mechanistic accident sequences, speed, and completeness.

I emphasize the philosophy because I am convinced no one within the NRC leadership appears eager to spearhead such an effort. As I have listed in ny June 3 letter, I have been trying to instigate such an effort for a few years now working through the ACRS as well as through review group meetings of the WRSR. The fact that nothing came of it serves to support my claim. Along the same lines, I would also like to mention that the initial augmented budget i

as drawn by WRSR was presented to the ACRS THI-2 subcomitt.ee with a very mini-mal effort in this area. One month later as presented to the ECCS Subcomittee the budget was changed, in response to my criticism, from 0.4 to 1.4 million.

This is an inadequate change especially since it became clear that the problem j

has not even been approached in the planning stage.

Dr. Murley implied a game l

of " aimless" search when he made reference, while responding to my question of I

timetable for this activity, to a slow " turning stones around" process. More importantly Dr. Fabic, whose branch is suppossed to be involved in this, pre-l l

sented "endfeu " lists of codes " developed," "under development," and "to be developed," he even discussed the rudiments of a code assessment methodology but did not volunteer any thought in the most important subject I am talking about, that is, code application.

I want to make it clear that these comments 7910340 W i

e

..... a:'.. i. '.......,.. v...

Professor Plesset June 26, 1979 Page 2 are not meant to criticize Stan Fabic; instead to point out that his branch has b%n charged with analysis (it 4 called System Analysis) and indeed it has sone aft admirable job in directing the development of the crucially in-portant tiest estimate codes. On the other hand I am talking about synthesis o# accident scenarios (sequences) including (but not limited to) man-systems itteractions during the accident progression. Although I have argued many times in the past that some measure of synthesis can be very helpful in the analysis efforts as well, it would require a considerable shif t of gear for the full blown effort that I consider necessary and urgent now.

The point of all the above is 6nat TMI indicated the existence of cracks in our safety assessments. Many of us were aware of the petcittial existence of such cracks but misjudged the urgency of the problem in relation to achieving adequate R & D results (analytical tools) that could help pin-point the deficiencies. Now we have the proof that this effort is overdue.

Also we must realize that the job can be done well with less than perfect predictive capability. The iterative process of synthesis and analysis coupled with MJstema. tic presentation of results for continuous scrutiny in the tech-nical community can only provide a degree of completeness ever-approaching the high level required in this business.

Two months have gone by since the TM-1 accident and as far as I can see we spend our tine in tafhLng rather than doing.

In my opinion the nature of the problem is such that the analyst must be out in the field getting to know his system, and thus further diminish the opportunity for errors in predicting its response. This then brings me to my recormiendations.

(a) The Task.

Establish realistic accident sequences (scenarios),

by postulating initiating events and man-machine-accident interactions; ca rr y out to various degrees of degraded core cooling conditions, and systematize the area. Although probabilistic aspects have to be taken into account the emphasis should be in physical aspects and better understanding of the mechanistic accident sequences. No two reactors are exactly alike and a method to systenatize and study the differences needs to be established.

Identify weaknesses and deleterious circumstances, if any, with a goal to improving safety through improving:

the systems, the diagnostics, the operator training anc responses.

(b)TheLeadership. Establish a Synthesis, or Code Application Branch in Dr. L. S. Tong's organization, to carry out the direction of the above task.

Since a good direction will be benefitted by significant in-house work, this branch must be staffed with atypically large personnel. Establish a review group of experts including people knowledgable in systems as well as in codes and modeling, and preferably in both. This review group should be willing to devote substantial time to the effort, meeting no less frequently than once a month.

Its members should be willing to also carry out independent assign-ments at home.

Perhaps a 30'; commitment would be appropriate.

Strong inter-action with licensing and inspection and enforcement would be essential A counterpart effort should be established by the NRC staff.

~.

Professor Plesset June 26,1979 Page 3 in addition to the Application Branch Staff (c) The Aledas Opetanti, and the Review Group each of the major contractors should of the <tefuiia to support future calculations should be arrived at by consens groups to this area.

with all participating, initiative of groups, and individuals, should be Efforts should be made to encouraged for exploratory type investigations. systematize ac it should atic incorporation of the lessons to be learned to improvin to draw implications and make recommendations for 4hott tetm actions.

The above efforts should be established immediately, The WRSR should (d) The Timing.

with funds made available from current lower priority programs.

make recommendations for such reallocations together with the impact projec The charter of this effort must include significant milest Further in-depth studies are to continue for the subsequent within a year.

several years.

Having established a realistic organizational struc-(c) The Funding.

ture wi;h the above goals in mind a budget projecti9n should be made, making generous allowances for computer calculation costs and/or computation As a minimum this should form the focal poiat of future budgets.

It is expected that results from this task will be useful to direct futureThe total LOCA/

ties.

efforts for R & D by providing a perspective on priorities.

Transients budget should he examined thoroughly with a view for efficiency an econoqy.

If I hope the above explains my ideas and goals sufficiently well.

you have any questions please call me.

Sincerely.

T. G. Theofanous Professor TGT:wb 4

1

. ~

• ~ ~

. ;.......:. :.: c.W..

.... a. - ~ - >

P ll R Li LJ E UNIVERSITY o,

,~s~,

ua nucus ~ s ~... t s. ~ c WL b f b 4 0 A % $ f f f, l Df e. A 4 I44, June 8, 1979 g

b Professor David Otrent fuergy 1. Kinetic Engineerinn UCLA Los Angeles, California 90024

Dear Professor Okrent:

Tnis letter is in response to your request for coments following the 5/31-6/1 meeting of your Subconraittee on the TMI-2 accident implications.

There are many detailed points that I would iike to nake at some later time.

However, for the purpose of emphasis and to avoid dilution I would like to enncentrate, in this lettt:r, in the major point of philo.ophy of approach in responding to TMl-2.

According to the office of Nuclear Regulatory Research the T'11-2 accident has brought up the need to better study the area between design basis accidents and core melt accidents. P.ased on this an extensive but rather dif fuse list of task areas ranging from "accelev ating developront of transient and sciall LOCA codes" to " containment integrity under fuel melt conditions,"

was prepared with a total price tag of s 30 uillion dollars. Althougn ! still believe, as I did in a 1971 letter to the ACRS, that better understanriing

(" probability of occurrence anr1 consequences") of this intermediate area

~

(" partially degra,1e.1 conditions") ir, warra ted I think it will be wrcng to cd.e it the initial focal point of our res; nnse to T'11 Instead, I believe, we need to l'Tr Idi a n. ore generic tashion) for safety deficiencies priniarily 6

responsible for Till.

In my opinion the answer to this question is lack of sufficient understanding of accident sequences (of not only small LOCAs but of the whole spectrum of sizes) including the whole breadth of physical phenomena associated with LOCAs and system / human interactions. This is tike constructing event trees except with the emphasis in mechanistic details of accident pro-gression (as determined from physically grounded analysis. tools) togc.ther with the usual probalistic oriented aspects of compnnent/ human behavior.

A major and diligent effort would be required to produc.e useful results in this' area. On the other hand such results would be instrumental in a number of areas:

(a) indicating areas whcre further fundamental research and/or empirical information would have the greatest impact on safety; (b)

I providing a background against which operator training may be made substantially I

complete, including better elucidation of the type and kind of instrumentation crucial for correct operator responses; (c) provide the necessary basis for a realistic approach to the advanced code verification (assessment) efforts that is about to commente, and finally; (d) such studies will provide us with a better basis (than that available today) for a realistic approach to striking the a;ipropriate balance between p*,cuentl..;, mifipa t.icn, (uf rivention, and m

~ Q

' p f\\ plin'".e various lines of esfonatien v/t ecuseyncoces for reartur accidents.

Alnn w ?'

n f A14 ] $

79/9 /s #/ 57 6

s.a.:.b. v

. ?l ':

.?

'..... l..

Prof. David Okient Oh June 8, 1979 Page 2 defense verification of projections becomes increa'ingly more difficult.

Unscheduled. complications may always arise, hence major uncertainties in the projected behavior develop. Clearly, therefore, the research emphasis should be in the above order. However the demarcation lines and the absolute degree of emphasis need better definition.

I would also like to take issue with the attitude, expressed by many these days, that the main licensing thrust has been on " conservative" analysis while TMI points to the need of " realistic" evaluations.

I think, in a more appropriate view of this situation, we should recognize that unless one knows the true phenomenology and sequence of events one is hard pressed, in mary situations, to make " conservative" choices in the analysis. This has been well known in the past. The ECCS hearings and Appendix K implementation hee many tio.es pointed to this fact. This is precisely the reason that such large efforts have, and are continuing to be devoted to the development of best estimate Camputer Codes for LOCA (and transients).

Further it should be obviot s that as the size of the break decreases and the time sequence of the accident increases, there is more opportunity for phase separation and large degrees of nonequilibriun (i.e.

injecting cold water etc.) both being comolicating and ill-characterized factors affecting the thermal-hydraulic response of the system.

Also there is more opportunity for human and systcm interactions (i.e.,

actuating /deactuating systems and random systems failures) further complicating the sequence.

I do not think we have failed to recognize the importance of all these things in thc : at.

We failed instead in carrying out the relevant analysis, thinking and scrutinizing the results, to better understand the system response and identify weak links in systems and troublesome areas in human interactions.

Sach endeavors are difficult and not precisely definable in detail at the outset.

The response of the system can be very complicated indeed.

There has been a " natural" hesitation, therefore, to undertake major efforts in this direction in favor of a plug-and-caug approach with code com-putations carried out primarily for the purpese of obtaining a peak clad temperature. The excuse has been, at least given in response to my asking for such applications, that the analytical tools have not teen adequately developed l

as yet. This may have been true five years ago, but it has becone less true during the past 1-2 years.

If we wait until the tools are completely perfected it cill take forever.

I believe that it is now urgent that any further analysis

)

tools and code development be guided by approoriate " synthesis" of accident sequence studies. Like I mentioned in a 1977 letter to the ACRS we need to put major emphasis in scrutinizing " code results and accident seo,uences to pro-vide the basis for an iterative synthesis-analysis proce s converging fu the l

ac to,n' phenemeneceyy of entenest to safe ty."

One can think of situations where operator intervention would be essential during the course of an accident.

This will be particularly true for small breaks. Hence as the break size decreases the accident sequence becomes more couplicated. llence it is less clear what represents a conserva-tive analysis choice and most importantly it becomes more difficult formulating a reasonably compact set of recucunended operator actions. For this latter task, in any cae'. it is absciutalv necessary that the oneratnr have the appropriate

. :. u. ;... cia.... u.,

Prof. David Okrent June 8, 1979 Page 3 diagnostic tools and to relate,thrcugh analysis, the indicatien of these tools to the physically occurring processes. Hence the need for predicting actual The (vs. consecuative) system response increases as break size decreases.

Incidentally I difficulties of modeling and computations also increase.

find the current B&W effort to provide a " plausible" explanation for the TMI events a step consistent with their capabilities but rather inadequate.

Finally it will prove, I think, rather difficult to find appropriate facilities for assessing (or verifying) the adequacy of such computations. This is because There are reasons to doubt, for example, the scaling problems become more severe.

that we can expect to learn much abrut small breaks from Semi-scale. Since such experimental programs need long lead times, I suggest that this issue also receive concentrated attention in conjunction with the accident sequence studies mentioned above.

Due to the time available between our subconnittee and the full ACRS meeting I am afraid this write-up is not as well organized or as clear as I would have liked. Please call me if you have any questions, and I will do my best to attend the meeting of 6/14-6/16.

Sincerely, T ZLAs T. G. Theofanous Professor TGT:wb i

i e-l