ML19309E073

Submits Addl Info Re 791018 Amend Request.Describes Range of Allowable Nonmoderator to Moderator Ratio & Computer Codes Used for front-end Calculations for Analysis of Max Credible Accident

ML19309E073
Person / Time
Site:	Lynchburg Research Center
Issue date:	04/11/1980
From:	Olsen A BABCOCK & WILCOX CO.
To:	Reid R Office of Nuclear Reactor Regulation
References
NUDOCS 8004180355
Download: ML19309E073 (4)

Text

'

Babcock &Wiicox aesearcn ono oeveionment oivision P.O. Box 1260, Lynchburg, Va. 24505 Telephone:(804) 384-5111 April 11, 1980 Mr. Robert W. Reid, Chief Operating Reactors Branch #4 Division of Operating Reactors U. S. Nuclear Regulatory Commission Washington, D. C. 20555 License: CX-10, docket 50-13

Reference:

Letter to Mr. Robert W. Reid from A. F. Olsen, dated October 18, 1979

Dear Mr. Reid:

It has been brought to our attention by Mr. Harold Bernard and Ms. Jocelyn Mitchell of your staff, that our amendment request forwarded to you with the referenced letter, needs clarification of the license change being requested. We desire to widen the range of the allowable non-moderator to moderator ratio, and suggest the following wording for the specification:

"The non-moderator to moderator volume ratio of any core loaded with aluminum-clad 2-1/2% - enriched UO 2 fuel r ds shall be between 0.7 and 9.7".

In addition, Mr. Bernard and Ms. Mitchell have requested other technical information which follows:

Codes Used Front end calculations for the analysis of the maximum credible accident-were made using the computer codes XSDRNPM and KEN 04. XSDRNPM1 is a program for solving the one-dimensional Boltzmann transport equation by the discrete ordinate method (the SN method). XSDRNPM is one of the modules of the AMPX system or code package developed at the Oak Ridge National Laboratory. The 123 energy group XSDRN cross section set was used with resonance processing being performed by the AMPX computer code NITAWL. NITAWL uses the Nordheim

1. M. N. Greene, et al. , AMPX-A Modular Code System for Generating, Coupled Multigroup Neutron-Gamma Libre. ries from ENDF/B, ORNL/TM-3706'. ,

March 1976.

O

Babcocks3Milcox April 11, 1980 ,

Page 2 2

method to calculate resonance self shiciding for U-238. KEN 04 is a multi-group, three-dimensional Monte Carlo solution of the Boltzmann transport equation. This computer code like the AMPX code package was developed by the Oak Ridge National Laboratory.

XSDRNPM and KEN 04 were selected because they are state-of-the-art for calculating neutron multiplication factors (k) and reactivity effects. They are both widely used throughout industry, at DOE contractor installations, and by the NRC to evaluate the wide range of nuclear criticality safety situations encountered in the manufacturing, testing, storing, and trans-porting of nuclear fucis.

In addition to extensive benchmarking by industry and the national laboratories, B&W has carried out its own benchmarking of these codes to insure that the intended code versions were operational at B6U cnd to provide internal assurance of the adequacy of the codes. Over 20 benchmark cases have been performed for both the SN and Monte Carlo codes. These bench-mark cases bracket experimental conditions and configurations planned in the described experimental program. Temperature dependence of the resonance parameters calculated by NITAWL have been compared to experimental work of both He11 strand and Pettus.

The XSDRNPM and KEN 04 are thus appropriate codes for use in support of the accident analysis and their validity can be supported by appropriate benchmark work which brackets the proposed experimental effort. The effect of anticipated biases established from the benchmark effect is discussed below. .

Uncertainty in do/dt The Nuclear Criticality Safety Group of our Lynchburg Research Center has benchmarked the KENO package with over 30 critical assemblies ranging from very wet configurations to dry metal spheres. In examining all of these benchmark cases, the largest discrepancy between experiment and calculation is 1.9% AK/K. If our calculation for pex in Table A-1 for the loose-packed square case is assumed in error by 1.9%, then the dp/dt value in Table A-2 is raised to 0.055%/sec. Utilizing this value in the point kinetics model-of the maximum credibic accident raises the predicted clad temperature from the previously determined value of 611 F to 732 F. This is still considerably below the acceptabic value of 10500F given in amendment 5.

2. F. G. Welfare, Editor, B&W Version of ORNL Code KEN 04 (L. M. Petric and N. F. Croso, "An Improved Monte Carlo Criticality Program", ORNL-4938, November 1975) NPGD-TM-503, April, 1979-

. e

BabcocksAMilcox April 11, 1980 Page 3 Uncertainty in Fuel Temperature Coefficient A formulation of the variation of resonance integral with temperature used by both Hellstrand3 and Pettus4 in fitting their experimental data is: ,

R = Ro (1 + 6 ( 8 - do~ ))

where T, is a reference absolute fuel temperature R, is the resonance integral at tht reference fuel temperature R is the resonance integral at absolute fuel temperature T.

S is an experimentally determined constant How well 6 is known, is the key to how well we can predict the temperature  :

coefficient since 0

a' = K /f~ (eq. 13-43 Lamarsh5) where ap is the fuel temperature coefficient K is a proportionality constant This latter equation shows that for any given temperature, if the uncertainty is expressed in %, then the uncertainty in o p is the same as the uncertainty in B.

For an S/M of 0.3239 -(which corresponds to the surface to mass ratio of the pellets in the fuel pins), the Hellstrand experimental work gives a value for 8 of 0.0074 + 0.0006, and the more recent Pettus data gives 0.0076 +

0.0002. The code NITAWL was used to evaluate 8 and gave a value of 0.0075.

NITAWL provides the resonance parameter used in XSDRNPM to determine the fuel temperature coefficient in our recent amendment request. Thus we conclude that NITAWL matches the experimental value to within 1.5%.

3. E. Hellstrand, P. Blomberg, and S. Horner, "The Temperature Coefficient of the Resonance Integral for Uranium Metal and Oxide", Nuclear Science, and Engineering 8, 497-506 (1960).

4. W. G. Pettus, M. N. Baldwin, and C. Samuel, " Resonance Absorption in U-238 Metal and Oxide Rods", BAW-1244, (1962).

5. John R. Lamarsh, " Nuclear Reactor Theory," Addison-Wesley Publishing Company, Inc. (1966) .

e -

w AO 9 -

a 4

Babcock &Wilcox April 11,'1980 Page 4 Being very conservative, we have examined the effect on the maximum credible accident if the absolut. value of the temperature coefficient is reduced by 10%. Utilizing this value in the point kinetics model of the maximum credibic accident raises the predicted clad temperature from the previously determined valut of 663 F to 7150F. This is still considerably below the acceptable value of 1050 F given in amendment 5.

Thank you for your prompt consideration and if you require additional information, please contact me.

Yours very truly, Babcock & Wilcox Company Lynchburg Research Center

)

. . gwu A. F. Olsen License Administrator rbf Attaciment Subscribed and sworn to before me this /YM day of , 1980.

00ll Lim) lW Notary Fublic My cotetission expires h ary d [ y 1

O