ML20206A053
| ML20206A053 | |
| Person / Time | |
|---|---|
| Issue date: | 07/10/1988 |
| From: | Catton I Advisory Committee on Reactor Safeguards |
| To: | Boehnert P Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-CT-1935, NUDOCS 8811140401 | |
| Download: ML20206A053 (3) | |
Text
... u..c
'o UNITED STATES dT-/ 35 g
f, y
NUCLEAR REGULATORY COMMISSION 3-f E
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS 4
WASHINGT ON, D. C. 20655 July 10, 1988 RECEi"ED wmcs! coi!2iTEE Ou p,;;.Cich SMEges USNRC.
MEMORANDUM TO:
Paul Boehnert g j g ggg p
n1 h Ivan Catton
, M 2i3i 5iG 4
FROM:
/
SUBJECT:
TECHNICALPROGRAMGROUP(TPG) MEETING-CODESCALING APPLICABILITY AND UNCERTAINTY (CSAU) METHODOLOGY, SAN JOSE, CALIFORNIA, JUNE 9-10, 1988 In rny sumary of what I learned at the latest TPG meeting, I will try to key to the Westinghouse review exercise that is now underway. As you will see, the NRR Staff review of the W code was weak in a number of ways that could lead to underestimating the PCT.
During the initial portion of the meeting, some interesting infonnation came to light. As a result of the ACRS review of Westinghouse, the question of LOFT external thennocouples is being revisited.
It is thought by some that the LOFT peak clad temperature may be up to 150*F low, due to enhanced heat transfer. This was not considered by Westing-house when they compared tneir results with LOFT measurements. There has been some interesting interplay between the US and FRG regarding the UPTF missing mass. The TPG need for the data and the possibility that the CSAU process would lead to the FRG image being damaged by resorting of a poor experiment lead to critical review and prompt action t1at eliminated the source of error in the mass balance.
An action item at the previous TPG meeting directed LANL to see if there were any time-step and convergence studies of TRAC.
It was reported back that none could be found. A cardinal rule in the code business is that such studies be made, documented, and reported before a given code be used. Most of the professional societies require such, studies before work can be published.
NRC and LANL management were clearly ~ delinquent in this regard. Time-step studies by the FRG show that one runs out of dollars before one reaches convergence. This has special significance at this time with the Westinghouse Best Estimate code under review.
As you are aware, scaling is, and will remain, one of the big issues in rationalizing the use of computers to explain plant response to acci-dents - whether they are Design Basis LOCAs or transients of some kind.
A given code must be able to predict within certain limits over the I
range of relevant scales before it can be considered a Best Estimate code.
To do this, one has to compare predictions over some range of l
ee11140401 000710 g%%
l
$19NN PNV Q
s A
~
CSAV 6/9-10/88 Meeting July 10, 1988 4 : ';.y scales with experiments. Although limited, such data exists to a di certain extent. As a part of the TPG effort. :. number of counterpart ' $
tests from SCTF, CCTF, FLECHT, and UPTF have tven identified. These counterpart tests should be calculated with any code that is to be considered Best Estimate to put scaling issues to rest. Use of counter-part experimental data was a no recomended by the CSAV Peer Review Group and could be done, at least for TRAC, within the framework of the 2D/3D program.
The role of nitrogen has been clarified as a result of searching for an explanation of the calculated reflood PCT double peak.
It was found that the surge of nitrogen following the emptying af the ECC accumula-tors shoves water into the core (the core water mass increases from 25%
to 70% of its maximum).
The result is rapid cooling leading to water entrainment into the steam generators and steam binding which pushes the water out of the core.
The result is a second reflood peak. Under some circumstances, the second reflood peak is higher than the first.
Westinghouse does not calculate a second reflood peak because their treatment of the nitrogen is too simplistic for Best Estimate modeling.
A logic error was found in TRAC that emphasizes the need for line by line checking of a code.
It also points to the need for comparisons of the elements of a code with experimental data. Here it was found that TRAC-calculated heat transfer coefficients were too high as a result of the comparison with experimental data. At the outset, the comon thinking was that TRAC was conservative - whereas in reality it was non-conservative by approximately 150'F. This kind of review was not given to the Westinghouse Best Estimate code and must be before it can be acceptable as a Best Estimate code.
It can not be overemphasized that the problem in TRAC was found by looking at heat fluxes, heat transfer coefficients, heat transfer mode, temperatures, and flows and noting inconsistencies then searching for their cause in a comon sense way.
Study of the reflood PCT is well on the way to completion. There are still some questions to be further resolved dealing with entrainment (steam binding) effects, and whether or not it is properly accounted for.
In-core interfacial drag discussions took up a great deal of time as it will detemine core exit flow of water and, as such,* has a direct influence on steam binding. To make results look reasonable when comparing code predictions with SCTF data LANL increased the inter-facial drag by a factor of ten. That such measures are needed does not say a great deal for the TRAC developers.
If the interfacial drag coefficient must be changed by an order of magnitude in the core, what must be done in other parts of the primary system? It seems to me that there is a lot of room for research in the two-fluid modeling arena.
- a
~
CSAU 6/9-10/88 Meeting July 10, 1988 V'?a In an attempt to see where we (the TPG)were at, a table was produced that y,'
contains a preliminary estimate of the PCTs and their associated uncer-tainties.
I have reproduced the table below for your review.
PCT in 'F Ave Kw/ft = 5.8, PF = 1.6, 17X17 fuel in a W 4-Loop plant First Second Blowdown Reflood Reflood Peak Peak Peak Nominal PCT 1084 903 660 Maximum PCT at 95%
1437 1370 1301 probability Biases due to various factors Hot channel effects 68 68 68 Logic error 80 140 180 Steam binding scale effect 0
-20 90 ECCS bypass scale effect 0
-50
-50 148 133 288 Maximum PCT at 95% plus 1585 1508 1589 biases Delta added to nominal PCT 501 605 929 Several conclusions can be drawn from the above table.
First, the second peak and its magnitude clearly shows the importance of modeling the nitrogen that rushes into the core following ECC accumulator empty.
ing. Second, the different values of the biases shows the importance of propagating the uncertainty as the transient behavior during the LOCA progresses.
Finally, the magnitude of the delta that is added to the nominal or Best Estimate value of the PCT in each of the phases of the event are dramatically different, clearly showing that one can not lump them together as was done by Westinghouse.
-