ML20038C332

Forwards Detailed Comments on Draft Interim Assessment of March Code.Addl Discussion Re Future Uses of Code Is of Paramount Importance.Related Info Encl

ML20038C332
Person / Time
Issue date:	07/29/1981
From:	Cunningham M NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:	Rivard J SANDIA NATIONAL LABORATORIES
Shared Package
ML20038C330	List: ML20038C329 ML20038C332 ML20038C336
References
FOIA-81-320 NUDOCS 8112100441
Download: ML20038C332 (14)

Text

v.

^

[

g C 01981

!!r. J. B. Rivard Division 4421 Sandia !!ational Laboratories Albuquerque,ICi 87185

Dear Joe:

I have received your draft of the interim !%RCH report and have read it with great interest.

In doing this I have attempted to consider the text buth from the point of view of a person who has ser.e responsibility in tolding the critique into future work and one who has soaa rudinentary knowledge of the code and its history. My detailed coments on specific points made (or not made) in the text are attached; a few more general ones are provided below.

I tust note that you and others have obviously spent a great amount of time in collecting all the data in the text.

You are to be comended for completing this thankless job in a tisaely canner.

Hy greatest concern about the report is that of somewhat of an icbalance between the past and the future. When we requested that you perfom this critique of that has been done, we also charged you with providing recom cadations regarding VMat needs to be done. While you have addressed this aspect to scoe degree, I believe that significant additional discussion is of paramount importance. Through this program and others you have developed a good understanding of the l%RCH code. What I'm requesting is a set of explicit recc=endations, based on this understanding, on what the next generation of a !GRCH-type code should look like, what weaknesses in !MRCH we should try to improve upon first, etc. Ne here at i:RC sorely need such explicit recom.cndations spelled out at the front cf this report.

!!y second general con-ent is very ruch related to the first. You have compiled cany corrents on I'MCH and put thm together in Chapter 4.

Uith your knou-ledge of the code, you surely must have developed sore.fudgcents on uhich ones are important and which are not. (It's not clear fron the report uhether or not pu included each and every coment received.)

I believe such a prioritization is essential to =aking sense out of the litany of cor: ents.

To include relatively trivial corrents interspersed anong those with distinct significance only serves to obscure uhat's being said.

\\

gjj2 g di sto91e o" " >......

NEISSB1-320 PDR,I

!............l...

l l

l c o n >;..

macscavsia mer vevc2a OFFICIAL RECORD COPY f

i

l'r. J. E. Rivard 2

Fy last cort.:ent also relates to your general conclusiors. You raake a statc=cnt regarding hau !*P.CH results should be used and not used. Your conclusions in this area are obviously very inportant and uill be videly read, rece2cred, and cited. For such conclusions, the required degree of clarity cust be very, very high.

From 3our conclusion, it is difficult for te to decide if you don't believe FARCH should be used at all directly in regulatory ratters, trat it should only be used with great care, etc.

I am by no ceans telling you what to conclude; I am saying, however, that 1;h2tever 3cu conclude, be very deliberate in writing it down.

I hope all of these coments will be of use.

If you want to discuss thea further, don't hesitate to call ee (FTS 443-5960).

Sincerely.

Orig:n:!e -

Mark T t, -t

_...:n Fark A. Cunningham Peactor Risk Branch Division of Risk Analysis Office of f,'uclear Regulatory Research

Enclosure:

Detailed coments and draft Interim Technical Assessment of the l'. ARCH code cc:

R. Curtis, RES F. Portsone, RES R. Bernero, RES DISTRIBUTION:

Central File RES CHRON/ CIRC RRB RDG/SUBJ MCunningham RDG l

o'.,,ced RES/Df,A RRB R

D V"R B W6 ifbi

....."r'.

. ' O.".

Ef h

... /. )..[........ \\i11" ' "' ' "

"F

• ""* * ' > it

.....f./81

~..

.~.

i 7

c re) 7/24/81 r..c r:n n., ie..o. a:o c a' OFFICIAL RECORD COPY

D: tailed co.ments on draft Interim Technical Assessment of the FARCH Code

p. 5.

In the discussion of the organization for maintaining and improving MARCH, it's not explicitly clear that anyone has a responsibility to actually do the improvements to the code.

p. 2.1-3 The first sentence of the last paragraph implies PARCH is not yet out, in contradiction to the rest of the paragraph.

p. 2.3-5 Section 2.3.2 is.so detailed that it breaks up the' flow of the chapter.

I suggest the detail of the standards might be better in an appendix.

I agree completely, however, that the approach of structured programming is essential._ However the next generation of the FARCH-like code comes about, it will have to be developed in such a way (if I have anything to say about it).

Section 3.1 This section tends to mix sequences important to risk and those important to the core melt probability. Some additional words clarifying this point might be appropriate.

Ch. 3,_p. 4 In the discussion of accident sequences involving loss of heat removal, I suggest the last sentence might be modified as such:

... " containment will overpressurize and may eventually result in containment failure and is predicted for some desians to lead to core melt."

Ch. 3, p. 6 The discussion of fuel melting and its progression downward is written in a manner implying somohat more certainty in our knowldege of these processes than perhaps we truly have.

Section 3.2.4 In the second paragraph:

"but vessel breach and discharge of the core material downward (?) will probably result"...

Section 3.2.5 A reference for your information in the last paragraph (i.e.,

liquid pathway) seems appropriate.

Section 4.1 In the top paragraph, if YBRK is input-specified, why is it that

p. 4 only fully-open or fully-closed val.es can be modeled? Perhaps this is.just faulty sentence structure, but it would seem like one could input a YBRK equivalent to the size of a half-open valve or whatever size / degree-of-openness one desired.

Section 4.1 The containment-pressure - steam generator pressure sensitivity

p. 8 is based on quite 'a wide band of secondary pressure (750 psia to 1200 psia).

Such an example says to me not that MARCH necessarily has a problem, but that reasonable secondary pressures should be selected.

If PARCH really does have a problem, I suggest using an example that proves it more,strongly.

i Otherwise, this doesn't seem to deserve the visibility it gets.

2 Section 4.1 Af ter reading the discussion on how pump flows are calculated,

p. 10 I was left with the question of "so what"? Is this an out-right ermr, a bad approximation, or a good one, or what?

The reader isn't in a position to judge the significance of this.

Section 4.1 With respect to the top paragraph, for what size breaks might

p. 11 this be a problem? This sounds like it is an attempt to use MARCH for the non-mechanistic, highly artificial, and risk-unimportant large LOCA. Perhaps this is also of concern in other situations.

If not, I'd suggest it be deleted because a criticism resulting from a desire to model some of the more unimportant facets of DBAs are as irrelevent and immaterial to MARCH as anything can be.

Section 4.1 Since you had just previously discussed the effect of steam

p. 13 generator dry-out calculations in MARCH versus TRAC and RELAP, it would be helptful to clarify if this was coepensated for in these calculations relating to Y3RK.

If not, the importance of the latter becomes much more difficult to discern.

Section 4.1 If I recall correctly, MARCH uses input from more detailed codes p. 13 to define the power history during an ATWS event.

I know for a fact that these detailed calculations are very uncertain, for nany physical reasons. To imply (as the commentor apparently does) that doing these calculations of reactor kinetics within MARCH will reduce the crudeness of the code, and thereby making the results more accurate, is ridiculous.

It is, however, consistent with the philosophy you take note of elsewhere that a more detailed code is assumed by many people to be by definition more accurate.

Also, to fault MARCH for not modeling rod ejection accidents is ludicrous, for reasons discussed above.

I suggest you delete such comments.

p. 4.2-4 One could also conclude at the end of section 4.2.1 that although none of this is convincing, but absent much more detailed analysis in the code, MARCH doesn't seem to do very bad.

p. 4.2-11 With respect to the first full paragraph, can you indicate in l

any way how much of an effect this might be? I find it hard i

to believe that radiolytic and thermal decomposition of water l

are going to contribute much hydrogen, relative to the zirconium l

oxidation.

p. 4.2-15 As I've noted for other sections, section 4.2.4 ends inconclusively.

What judgements can be made based upon all the cited data? From my viewpoint, the relative importance of this concern (and there-fore the relative priority I assign to fixing it) is a critical l

aspect.

p. 4.2-21 As just above, Section 4.2.5.4 ends without making some judgement on the importance of this concern relative to others.

l-i

3

p. 4.2-22/23

'In the f_irst full paragraph, the con'clusion is. reached that "a more detailed model of radiation interchange is needed."

I submit that unless it is first shown that the MARCH modeling introduces.significant errors, one cannot conclude (as the commentor apparently does) that such a change is necessary.

In any code developed to handle the entire course of a melt-down accident which is also in any.way practical, one cannot add / change models to include unimportant phenecmena for the sake of " truth and beauty".

p. 4.3-7 This kind of discussion is just what I've been locking for.'

From this section's conclusion, I have a start on where to go next. More like this is essential for me to do my job and get the next generation of code development under way.

.Section 4.4.4 This Pratt and Bari discussion in the third full. paragraph-

.p.

12 leaves one hanging. Can a bit more detail be provided?

~

p. 4.6-7

., With respect to the coment-(TR-13) at the top of the page, if the sprays are on,'their direct heat transfer effect must overwhelm that through the walls. This again apparer,tly suggests somebody's desire for " truth and beauty" being at odds with the practical need for a integral, fast-running code.

p. 4.6-7 With respect to Section 4.6.3, I'm not sure I can agree that "the impact of their absence cannot be adequately assessed." Some sort of quick calculations seem possible -

to give at least a first approximation. Also, as I noted above, I find it hard to believe that' the sources noted in c, d, and e are of any importance, except perhaps in very long time periods.

l It might help to clarify the last sentence; it's difficult

p. 4.6-9 to tell whether your indicating a problem of MARCH or noting experimental evidence.

i With respect to the last sentence,_I know of many MARCH results

p. 4.6 which indicate the inability to burn hydrogen because of steam inerting. Perhaps I'm missing the point, but this sentence seems-incorrect.

In the first paragraph, can you add what type of plant has the

p. 4.6-15 pool under the cavity? I presume they mean a BWR Mark II design.

It is difficult to figure out what the objective of Chapter 5 Ch. 5 is. It might make sense to have an introduction to the -

chapter explaining what is intended, etc. Also, the BCL write-ups seem to be left somewhat hanging.

It's not obvious whether you are concurring or not.

If you are, perhaps you can briefly explain and comment.

If you

_ _. ~ _ _ _...

I

v

-4 don't concur, tell us what you think.. As the chapter is now, it's simply not very coherent..

Ch. 5, p. 26 In the next to last line, RSSMAP is mentioned for the first time.- A reference seems appropriate.

\\

t 4

I i

o r

l i

l i

I

.+

I i '.-

+

. + e

. + -

n.

a.

.n n-_--,,..-

m I

@Bafelle Co'cr: cu laboratories 303 k.mt kr ue e

cmc i, cm.o um

• :epnort :tM 424-9 24

.i e

Ich 24-3434 b

AugustO 1$8) a Mr'.7 Joseph B. Rivard Division 4421 Sandia National Laboratories Albuquerque, New Mexico. 87115 Decr. Joe:

We have reviewed the draf t copy of the Interim Technical Assessment of the MARCH Code transmitted by your letter of July 6,1981. We are

~ impressed by the level of. detail reflected in the assessment and have had called to our attention a number of potential problem areas that we had not recognized. At the same time we are disturbed by what we perceive a's a. lack of needed perspective in a number of areas; the latter relate both to the MARCH code as well as to core meltdown phenom-

~ ~ '

enology. A number of the more significant point:; of concern are discussed below.

As we have pointed out previously, the capabilities and limitations of.

MARCH, ~or any other code, cannot be assessed without considering its

,e~. - intended aphlications.

While parts of the ~ draft assessment seem to indic' ate an: appreciation of the scope and intent of MARCH development, this appreciation appears to be lost in the body of the assessment.

THe latter appears to be a "wish list" of what might be included in a

. meltdown accident analysis code with little perspective on the relative importance of the various phenomena or the availability of a data base

~~to describe themri'n detail. MARCH is not intended to be a first prin-

'l ciples code for giving a uniqua description of meltdown accident scenarios, if indeed such a code.were possible.

Considerable criticism is attached to the fact that MARCH requires a number of user selected options as input and that the choice of different options may lead to different results. The latter is, of course, to be expected, except for the most trivial of exercises.

(Inthisconnection, it is interesting to note the MARCH-RELAp-TRAC comparison that is given in the. draft assessment; considerable differences are seen between the results of RELAP and TRAC which cover in great detail a substantially narrower. range of phenomenology than is considered by MARCH.) The need i

for a number of user selected options stems from several reasons, inclu--

~

ding:

the l'ack of a detailed understanding of certain_ phenomena, the possible stochastic nature of certain events, and the need to examine

~

4 4

D C

Mr. Joseph B. Rivard 2

August 7, 1981 different paths for any given accident sequence. These input options provide a great deal of flexibility and permit the examination of broad ranges of phenomenology and accident assumptions. Thus, we view them as a strength rather than a limitation of the MARCH code. While it is undoubtedly true that some of the user selected options could be elimi-nated by more detailed modeling of the phenomena (where that is possible),

it is neither possible nor desirable to eliminate all such options. The existence of these user selected input options does require the exercise i

of. informed judgement for the meaningful use and application of the MARCH code, implying at least some understanding of the physical phenomena in-volved.

If ever there is a place for blind exercising of computer codes, the area of degraded core cooling accidents is not it.

In a number of instances, the draft assessment addresses, sometimes at great length, the default input options in MARCH. As is clearly stated in the MARCH User's Guide, these default values are included as a conven-

)

f ence in the initial implementation of the code and should not be construed as recommended values. The choice of input parameters should be considered l

by the user on a case-by-case basis; it is clearly up to the user to justify his choices of input parameters for any particular application.

The recommendations for further meltdown code development should be much more focused and specific. As the report stands, one can only conclude that more code development is needed.

It is not at all obvious which par-ticular areas of development would fulfill which specific needs. ' Tables 6.2-1, 6.2-2, and 6.2-3 are apparently an attempt to provide some indication of what can as well as what should be done. They are not altogether success-ful, however, particularly since many of the assignments in these tables are.

highly debatable.

A number of additional specific coments on the draft assessment are attached.

In addition to these comments, we are enclosing a marked-up copy of the draf t assessment that includes a number of other suggested changes and corrections of apparent misinterpretations of MARCH or its documentation.

In summary, it is clear that you and your colleagues have gone to considerable lengths in your assessment of the MARCH code.

The assessment, however, seems to us to lack balance, with a number of items being highlighted that are shown to be relatively unimportant. The document could be substantially improved if it were more focused on areas that are really significant, making clear the basis for the inferred significance.

In particular, the recommen-dations should be much more specific; as they stand, one can only conclude that everything needs further development.

While it has been somewhat disconcerting for some of the contributors to MARCH to be submitted to this level of critique, we do recognize the need for the review in view of the current usage of MARCH. We have tried to C

August 7, 1981 3

Mr. Joseph B. Rivard i it and j

make our coments on the draft assessment in a constructive sp r We are, of course, hope that we have not been too defensive about it.

e lengths you may willing to discuss these points with you at whateverThank you for

[

desire.

1 Sincerely, m,-

r, f/

i i

4,J L.*, ce.,A

" 2

// s.

Peter Cybulskis Richard S. Denning PC/R$D/llj Enclosures R. Bernero, NRC cc:

M. A. Cunningham, NRC^

C. Kelber, NRC R. T. Curtis, NRC

1 DETAILED COMENTS ON INTERIM TECHNICAL ASSESSMENT OF THE MARCH CODE (1)

The middle paragraph on page 2.2-2 does not reflect an understanding of the role of the analyses of physical pre. esses in reactor accident risk assessment. While the detailed consideration of many rate pro-cesses is not important, the relative timing of key events, particularly that of containment failure relative to core melt, has always been a prime consideratioit.

(2)

It should be recognized (e.g., page 2.3-1) that some comparisons between the containment analysis portions of MARCH and KESS are available and indicate good agreement.

(3)

The initiation stage of the accident is given substantial discussion in Section 4.1.

For core meltdown sequences, the initiation stage is not particularly important, but does establish the time frame of the overali accident sequence.

The initiation stage is very significant in consideration of recovery short of complete core melting; the latter, however, was not a significant concern in the development of MARCH.

-i (4)

There appears to be some misunderstanding on the intent and use of steam generator secondary side pressure, PSG. The discussion in the draft assessment (page 4.1-8) seems to imply that this is an arbitrary input; it is not. This variable is intended to represent the secondary side relief value setting and is thus quite well defined; a constant value is quite appropriate in this context.

(5)

The logic in the discussion of fission product treatment on page 4.2-4 and 4.2-5 is flawed.

The model cannot at the same time overpredict and underpredict the heating of the primary system inventory or, con-versely, underestimate and overestimate the heating of the containment atmosphere.

t f

2

(

(6)

The discussion of gamma ray escape in Section 4.2.4 concludes that j

l the decay heat transferred from the core is 1-2% of the total. This is a specific example of a number of " deficiencies" that were con-

}

sidered in some detail. We would classify this magnitude of approx-i mation as a trivial concern, particularly in view of many other phenomenological uncertainties.

(7)

The discussion on page 4.2-16 and 4.2-18 regarding the potential for a factor of fifteen error in core heating rate needs qualification.

If more heat is available for steam generation, less of it is avail-able for core heating during any time period.

If such an error should occur, it would be limited.in time and localized in space.

Clearly the overall time for core melting could not differ greatly from that predicted by PARCH.

(8)

The discussion of in-vessel recovery in Section 4.3 is interesting but may be out of place in this assessment since PARCH is not intended to stress these types of situations.

(9)

The discussion of a fragmented melt, Section 4.4.1.2, is not well founded in that it ignores the initial steps of particle bed formation, concentrating instead on the events that might follow after the bed is fomed.

I, the core is molten as it drops'into the water in the i

bottom head, the core debris must first be solidified in order to form a particle bed. The energy transfer and steam generation assoc-iated with debris solidification would precede particle bed fomation and must be taken into account in the analysis. Just the latent heat of fusion may in some cases be sufficient to evaporate all of the available water.

Only after the debris are solidified and settled would the particle bed considerations discussed in this section apply.

i Such considerations as interstitial water volume, counter-current liquid / vapor flow, etc., would not really apply during the initial molten fuel contact with water. While PARCH does not of course provide I

i

1

}

I 3

a detailed treatment of the dymanics of this initial contact, neither does a particle bed model; the latter would just repre-a

'~

sent the substitution of one set of assumptions for another.

I t

Which is more ' appropriate is clearly open to discussion.

i:

'li 4,

I (10)

The ideal gas calculation of the containment pressure rise due to primary system breach discussed in Section 4.5.1 is done only to give an indication of potential containment pressurization; the actual evaluation is included in the MACE calculations and is not subject -

to the ideal gas approximation.

L (11) The discussion of the heat transfer coefficient used in HOTDROP (Section 4.5.2, page 5-6) does not correctly represent the MARCH Further, it treatment which does include a radiative component.

is stated "...the steam spike resulting from rapid boiling can be J

l The latter could be true if the heat transfer coeffi-much larger."

cient were the only limitation; that is not the case.

In practice, we have found that the magnitude of the largest steam spikes, given.

I near optimum quantities of water, is limited by the heat content of 1

the debris.

With regard to the melt-concrete interaction, Section 4.5.3, the history (12) and limitations of INTER are well recognized.

It should be recognized, however, that at the time that MARCH was being developed INTER was the only model available. WECHSL was developed somewhat later, but is CORCON has only still not generally available in the United States.

recently become available in draft form.

It would undoubtedly make sense to incorporate the latter in a future version of MARCH.

l i

The discussion of the atmosphere to wall heat transfer, Section '4.6.2, (13) fails to recognize that conduction tends to govern heat losses to Thus structures for most of the duration of an accident sequence.

the coefficient used would be imaterial except in special situations.

i The experiment noted on page 4.6-6 is such a special situation, having been conducted in a metal vessel at an initially very low temperature l

p 4

i:

where the surface heat transfer coefficient could be important.

Even in this experiment, however, the difference.between the pre-I dicted and measured peak pressure cannot by any means be ascribed b

entirely to the heat transfer coefficient used.

1]

Section 4.6.3, Hydrogen Generation Transport and Burning, fails (14) to grasp the intent and 'use of the hydrogen burning models in PARCH.

It is not the aim of' PARCH to provide.a rigid or unique hydrogen burning model that will be invoked whenever certain conditions are satisfied; rather, it is the intent of the FARCH modeling to provide a framework in which the potential for and the effects of hydray:n burning can be examined under a variety of conditions and assumptions.

Whether hydrogen burning will be actually included in any particular The approximate flam-analysis is intentionally lef t up to the user.

mability region for hydrogen-air-steam mixtures included in the code.

can be used to guide the analyst in selecting when burning should be l

It must also be remembered that the existence of a -

considered.

Additionally, flammable composition is not synonymous with burning.

it may be noted that:

The work of Jahn and Langer was conducted for controlled LOCA (a) conditions and may not be relevant to core meltdown accidents.

Laminar burning considerations are probably irrelevant to the I

(b) types of accidents and conditions considered by PARCH.

While FARCH does not include an explicit suppression of burns (c) at high steam concentration, the approximate effect is predicted indirectly as a result of conditions on the < ther constituents of the atmosphere.

Burn times longer than one MACE timestep will lead to difficulties (d) with the containment model; our experience would indicate that burns of a few seconds duration do not differ appreciably from adiabatic burns.

The FARCH hydrogen burn model does take into account the evapor-(

(e) ation of water droplets in the atmosphere; the effect on post-burn conditions is obviously a function of the water content of the atmosphere.

e

6 (18) Appendix 4.6.3 addresses the variation of heats of combustion of hydrogen and carbon monoxide at temperatures up to 6000 K.

It concludes that PARCH underestimates the energy release by 1.8% at temperatures of 3300 K.

We would be interested in knowing how the containment atmosphere could get to 3300 K prior to burning.

G e

9