ML20204C180

Requests Transmittal of Encl Summary Evaluation of Pending NRU Severe Fuel Damage Tests to ACRS Subcommittee on NRC Safety Research,Per C Siess Request

ML20204C180
Person / Time
Issue date:	06/02/1986
From:	Silberberg M NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:	Duraiswamy S Advisory Committee on Reactor Safeguards
Shared Package
ML20204C177	List: ML20204C180 ML20204C184
References
ACRS-2427, NUDOCS 8607310040
Download: ML20204C180 (4)

Text

-

j o rcrug'o UNITED STATES gg ,g NUCLEAR REGULATORY COMMISSION 3.c , g g }{)#

g, f E WASHINGTON, D. C. 20555

. o, - _ -

1. p

\ ,,

i g . 21986 MEMORANDUM FOR: S. Duraiswamy, ACRS  ;

FROM: M. Silberberg, DAE

SUBJECT:

May 7, 1986 MEETING OF THE ACRS SUBCOMMITTEE ON THE SAFETY RESEARCH PROGRAM ,

Please forward the enclosed summary evaluation of the pending NRU severe fuel damage tests to the members of the ACRS subcomittee on the USNRC safety research program.

This evaluation was requested by Dr. C. Siess, the chaiman, to address the following question posed by Dr. D. Okrent "What regulatory issue would remain unanswered if the NRU facility would be unavailable for testing?"

h at the subcommittee meeting of May 7, 1986.

WN f W f Melvin Silberberg, Chief Fuel Services Research Branch, DAE

Enclosure:

Statement of key regulatory issues supported only by the pending NRU Reactor tests cc: C. Siess, ACRS D. Ross Z. Rosztoczy D. Okrent, ACRS W. Kerr, ACRS hj@73gjjo860622 2427 PDR ATTA C H M &H T' d D-1

I REPLY p THE May 7, 1986 ACRS COMMENT _0N THE NRU. TEST PROGRAM Although we have addressed this question several times over the past three years, we may not have made the important issues covered by the NRU program perfectly clear to the ACRS. The NRU program for testing LWR fuel bunules under severe accident conditions was initiated to define clearly the total amount of hydrogen that would be released in a core uncovery accident boiling down at prototypic rates over a prototypic fuel length, i.e., twelve feet. The NRU facility is the only fuel testing reactor which allows testing of full length rods under severe accident conditions.

The regulatory issue being addressed by the program is the issue of hydrogen production during the in-vessel phase of a severe accident, with particular attention given to the production after the initial relocation of the molten unoxidized Zircaloy and liquified TueT Current models used for NRC source term analyses allow the user to input a very wide range of unoxidized Zircaloy surface area to be in contact with steam after clad melting. This leads to a very large uncertainty in (a) the post-relocation production of hydrogen, and (b) the amount of unoxidized Zircaloy exiting the vessel. (A range of 30% to 78% for the oxidized Zircaloy in the Surry TMLB' Accident, QUEST study, 1 SAND 84-0410, v.2.) The large uncertainty in hydrogen production can and does

result in a large uncertainty in containment failure time estimates and, for J .

some plants and sequences, is the major contributor to this uncertainty. Such

~

( a large uncertainty in containment failure timing can and does lead to orders of magnitude differences in the prediction of the source term magnitude, particularly for BWR Mark I plants. Input of a large available Zircaloy surface area has 1

i i even generated sufficient hydrogen to cause early containment failure in a SURRY-type PWR without a hydrogen burn. (QUEST study, SAND 84-0410,v.2.)

i The corresponding large uncertainty in the amount of unoxidized Zircaloy I . before and after relocation has a large effect on the calculation of the release of certain types of fission products. It has been verified by experiments at ORNL, INEL(PBF), and BCL that if less than 25% of the Zircaloy is oxidized during the in-vessel phase of the accident, little or no tellurium release will occur. While the mechanism producing this effect is not certain, it is hypothesized that surface oxidation of the zirconium frees and concentrates Zircaloy tin near the oxide boundary, and that this tin becomes available to trap the tellurium as tin telluride. This effect leads to extensive Te release during the ex-vessel phase of the accident and, depending on the containment failure time, corresponding high Te releases to the atmosphere. If extensive oxidation occurs in the vessel, then most of the Te is released early and deposits in the upper plenum, and can not reach the atmosphere unless revaporization occurs much later in the sequence.

Current models of the ex-vessel fission product release during core / concrete .-

interactions depend strongly on the amount of unoxidized Zircaloy. The Zircaloy is oxidized by the water released from the concrete resulting in the formation of hydrogen bubbles which migrate upward through the melt. This hydrogen reduces normally non-volatile fission products such as La,30, to the much more volatile form La0. It is this effect which caused higher thHn expected releases of lathanides in some of the source term calculations performed in the preparation of NUREG-0956.

I cD4

Therefore, if the NRU test reactor were shut down tomorrow (as the ACRS question postulated), this agency would be left with a very large uncertainty in our ability to model the predicted source terms for certain plants and sequences. We should also add that current IDCOR modeling of in-vessel oxidation of Zircaloy and hydrogen production assumes that at the onset of rod collapse sufficient blockage of steam occurs to preclude any further hydrogen production. This result of the IDCOR model produces significantly different source terms with respect to tellerium and the lanthanides and a factor of 2-3 lower hydrogen for certain BWR sequences. Such a large discrepency between current best-estimate NRC and industry models pose a significant issue resolution problem for the NRR staff charged with implementing the Commission's Severe Accident Policy Statement.

We would again like to address another ACRS comment which suggests that the

.NRU tests would not be very useful, since there are so many possible sequences that might lead to core melt that the results of a few tests would be meaningless.

The answer is that although many sequences may indeed lead to core melt, they almost all have the same initial situation at core uncovery; namely, a simple boiloff of the core water at some pressure and decay power level. Therefore, the major parameters that affect the boildown rate and melt progression phenomena are pressure (easily modeled), power level (also easily modeled), and the molten core morphology, composition, viscosity and interaction with steam (not so easily modeled and the reason for the PBF, ACRR, and NRU programs). Although the latter effects are difficult to model (due to lack of phenomenological information),

they are essentially independant of the many sequences which can cause core uncovery in the first place. We feel, therefore, that a few well-designed tests in the full-length prototypic NRU test reactor will give sufficient data to considerably reduce our uncertainty on this vital issue.

It is difficult, as the ACRS suggests, to obtain information on detailed local behavior from large integral in-pile tests. We note, however, that most of

- our current understanding of in-vessel severe-accident behavior has come from the series of four Severe Fuel Damage tests in PBF and the analysis of these results. The primary purpose of the NRU tests is to provide the necessary integral data on hydrogen generation, particularly after the onset of molten Zircaloy relocation, to resolve the current significant uncertainties and the substantial differences (up to a factor of 2-3) between NRC and IDCOR on this issue, and for validation of the hydrogen models in the mechanistic MELPROG and SCDAP codes. Because of the complex multiple interactions involved, integral tests under close to prototype conditions are required. Only in NRU can the required boildown tests with full-length 12-foot fuel bundles be performed.

These experiments provide very accurate measurements of hydrogen generation during the boildown transient, and good local measurementsResults of cladding of early temperature up to thermocouple failure (at about 24C0K).

tests up to the peak of the initial oxidation transient have demonstrated our ability to provide accurate hydrogen measurements prior to extensive Zircaloy relocation. Future tests will be held at power after the initial oxidation transient for twenty minutes to determine the additional hydrogen produced during extensive relocation of molten Zircaloy and liquified fuel within the 4

2 g3

~

bundle. Comparison of these later results with those from the quickly cooled early tests should narrow the uncertainty range and help to resolve the NRC-IDCOR differences on this issue. The Severe Fuel Damage (SFD) program of experiments and analysis has been successful in achieving its objectives and in improving our understanding of the early stages of in-vessel core-melt progression. The gaps in our knowledge of the later stages of core-melt progression represent one of the "last frontiers" in developing an understanding of the governing aspects of severe-accident behavior. These gaps produce the largest of the risk uncertainties, and closing them also has the highest pay-off potential for achieving closure of key issues.

i i

l 3

3-tf