ECN-R SEPTEMBER \U Of. rl" 7 2Q. energy innovation ÜNR. A code for processing unresolved resonance data for MCNP A.

Transcription

1 SEPTEMBER 1994 ECN-R \U Of. rl" 7 2Q energy innovation ÜNR A code for processing unresolved resonance data for MCNP A. HOGENBIRK VOL

2 The Netherlands Energy Research Foundation ECN is the leading institute in the Netherlands for energy research. ECM carries out basic and applied research in the fields of nuclear energy, fossil fuels, renewable energy sources, policy studies, environmental aspects of energy supply and the development and application of new materials. ECN employs more than 900 staff. Contracts are obtained from the government and from national and foreign organizations and industries. ECN's research results are published in a number of report series, each series serving a different public, from contractors to the international scientific world. The R-series is for research reports that make the results of ECN research available to the international technical and scientific world. Het Energieonderzoek Centrum Nederland (ECN) is het centrale instituut voor onderzoek op energiegebied in Nederland. ECN verricht fundamenteel en toegepast onderzoek op het gebied van kernenergie, fossiele-energiedragers, duurzame energie, beleidsstudies, milieuaspecten van de energievoorziening en de ontwikkeling en toepassing van nieuwe materialen. Bij ECN zijn ruim 900 medewerkers werkzaam. De opdrachten worden verkregen van de overheid en van organisaties en industrieën uit binnen- en buitenland. De resultaten van het ECN-onderzoek worden neergelegd in diverse rapportenseries, bestemd voor verschillende doelgroepen, van opdrachtgevers tot de internationale wetenschappelijke wereld. De R-serie is de serie research-rapporten die de resultaten van ECN-onderzoek toegankelijk maken voor de internationale technisch-wetenschappelijke wereld. i O Netherlands Energy Research Foundation ECN P.O. Box 1 NL-1755ZG Petten the Netherlands Telephone : Fax : This report is available on remittance of Dfl. 35 to: ECN, General Services. Petten, the Netherlands Postbank account No Please quote the report number. Energieonderzoek Centrum Nederland Postbus l 1755 ZG Petten Telefoon : (02246) Fax :(02246)44 80 Dit rapport is te verkrijgen door het overmaken van f35,-- op girorekening ten name van: ECN, Algemene Diensten te Petten onder vermelding van het rapportnummer. Netherlands Energy Research Foundation ECN Energieonderzoek Centrum Nederland

3 SEPTEMBER 1994 ECN-R KS R- f 1 OE X ÜNR A code for processing unresolved resonance data for MCNP A.HOGENBIRK *DE X*

4 Abstract In neutron transport problems the correct treatment of self-shielding is important for those nuclei present in large concentrations. Monte Carlo calculations using continuous-energy cross section data, such as calculations with the code MCNP, offer the advantage that neutron transport is calculated in a very accurate way. Self-shielding in the resolved resonance region is taken into account exactly in MCNP. However, self-shielding in the unresolved resonance region can not be taken into account by MCNP, although the effect of it may be important in many applications. In this report a description is given of the computer code UNR. With this code problemdependent cross section libraries can be produced for MCNP. In these libraries self-shielded cross section data in the unresolved resonance range are given, which are produced by NJOY-module UNRESR. It is noted, that the treatment for resonance self-shielding presented in this report is approximate. However, the current version of MCNP does not allow the use of probability tables, which would be a general solution. \ Keywords Monte Carlo neutron transport MCNP unresolved resonance range ECN-R

5 CONTENTS 1. INTRODUCTION 5 2. METHOD 6 3. SAMPLE CASES Fe U 8 4. CONCLUSIONS 10 REFERENCES 11 APPENDIX A. INPUT DESCRIPTION FOR UNR 13 ECN-R

6 UNR ECN-R

7 1. INTRODUCTION Continuous energy Monte Carlo calculations of neutron transport offer the big advantage that self-shielding is implicitly taken into account. Besides, very few limitations exist in the field of the geometric modelling of a problem. Therefore, Monte Carlo codes are an indispensable tool in the accurate calculation of neutron transport in complicated geometries. The most widely used Monte Carlo code for neutron transport is MCNP [1], which can be used for shielding calculations as well as for criticality calculations. The cross sections needed in MCNP are represented by pointwise data. Although this is a correct representation for both low neutron energies (resolved resonance range) and high neutron energies (continuous cross section), problems occur in the unresolved resonance range, in which resonances often can not be observed separately. As the cross section still possesses a resonant behaviour in this unresolved resonance range, the effect of self-shielding should be taken into account. However, as the data are given in a smooth, pointwise representation on the MCNP library, no calculation of self-shielding is possible in MCNP. This implies that the cross section in the unresolved resonance range as used by MCNP is effectively too high. The overestimation of the cross section strongly depends on the density of the nuclide in question. The effect may be large in shielding applications (high material density of Fe) or in fast reactor applications (high flux in unresolved resonance range). In this report the computer code UNR is described, with whichproblem-dependent cross section libraries can be produced. The treatment is approximate, as only one value of the background cross section a 0 is used for the complete unresolved resonance range. A better solution for the problem mentioned above would be to sample cross sections in the unresolved resonance range in MCNP from a probability table of resonances. This method, which makes use of NJOY module PURR, is under development at Los Alamos [2]. However, it requires modification of both NJOY module ACER and MCNP, which may take some time. Therefore, the less general approach presented in this report is recommended in all applications in which self-shielding in the unresolved resonance range may be important. ECN-R

8 2. METHOD The method used in the code UNR in order to process cross section data for MCNP in the unresolved resonance range is described below. In the production of MCNP-libraries from ENDF data the cross section processing code NJOY [3] should be used. Usually, it is required to process the data in the modules RECONR, BROADR, UNRESR, HEATR and ACER. In many evaluations the unresolved resonance range is represented in MF2 by average resonance parameters and the distribution function for each of the parameters. In NJOY-module RECONR these parameters are used to compute the expectation value of the total cross section in an infinite dilute mixture. Only this infinite dilute cross section is available in MCNP, whereas an explicit treatment of self-shielding, as in the resolved resonance range, is not possible. This implies, that the effects of self-shielding in the unresolved resonance range, which are important in many applications, are neglected in MCNP. However, in module UNRESR it is possible to generate pointwise cross sections in the unresolved resonance range for specific values of the "background" parameter <r 0 and for a specific value of the temperature T. The parameter <r 0 is usually adopted in calculations in a multi-group scheme and is equal to oo if there is no shielding and equal to 0 for bulk shielding. An accurate determination of <x 0 can be made to determine the value of cr 0 in the unresolved resonance range of a specific isotope in a specific mixture (e.g. by using the code TRANSX [4]). Using this value of Ob as input for UNRESR it is possible to generate problem-specific cross section data in the unresolved resonance range. However, as these pointwise shielded cross section values are output on the MF2 section of the library, MCNP will not use these data. In the code UNR the following method is used to make the self-shielded cross section data available to MCNP: UNR calculates (on the energy grid of MF2) the difference between the infinite dilute cross sections and the self-shielded cross sections; UNR converts this difference to the energy grid of MF3; UNR subtracts this difference from the MF3 cross section data. The code system is visualized in fig The method is applied to MT1, MT2, MT18 (if present) and MT102 cross section data. Thus, a problem-specific MCNP-library results. The effects of taking into account self-shielding in the unresolved resonance range in MCNP-calculations will be most prominent for isotopes which are present in large concentrations: in shielding calculations a relatively large effect will be observed when the method is applied to cross sections for the structural materials, especially for 56 Fe. in criticality calculations the self-shielding in the unresolved resonance range of 238 U will be the main effect. Examples of the order of magnitude of the effect on the cross sections are given in section 3. The code has been made available and can be requested from the NEA Data Bank at Paris. ECN-R

9 Method RECONR BROADR NJOY UNRESR = o(q,=~)-a(a 0 ) on MF2 E-grid = c(cb= )-a(a 0 ) on MF3 E-grid I UNR ACER NJOY f ^--_ MCNP library MCNP Figure 2.1 Schematic representation of the code system used for the production of selfshielded MCNP-libraries. ECN-R

10 3. SAMPLE CASES In this section the use of the program UNR is demonstrated in the case of 56 Fe cross sections in a stainless steel mixture (AISI316L) 238 U cross sections in a standard LWR pin cell (Rowlands benchmark case) Fe In many shielding applications use is made of thick layers of stainless steel. As far as self-shielding is concerned, these layers can often be considered to be infinitely thick. In this section the effect of self-shielding on the cross section in the unresolved resonance range is shown for 56 Fe in a thick layer of stainless steel AISI316L. Self-shielded (a 0 = 1-85 b) and infinite dilute cross sections are given in fig Cross section data were taken from the EFF-2.4 evaluation. The unresolved resonance range extends in this case from E n = 862 kev up to E n = 3.0 MeV. The data were processed at T = 300 K. The value of cr 0 was calculated in the code TRANSX [4]; it amounts to 1.85b in the unresolved resonance range. The effect of this change of cross section in the unresolved resonance range on the neutron flux in this energy region may amount [5] to 10 to 20%, which is in many cases non-negligible U Self-shielded (CT 0 = 60, 23 and 14.9b) and infinite dilute cross sections for 238 U are given in fig Cross section data were taken from the JEF2.2 evaluation. In this evaluation the unresolved resonance range extends from E n = 10 kev up to E n = 300 kev. The data were processed at T = 300 K. Self-shielding in the unresolved resonance range in 238 U influences the value of k e ff and the value of spectral indices. In several applications effects in the order of 0.1% in k eff were observed for thermal reactor benchmarks. Larger effects are to be expected for fast systems. ECN-R

11 Sample cases 100 EFF2.2 56Fe Comparison sigma.o = inf with sigma J) = 1.85 b sigma_0 = inf Delta sigma & o i E_n (MeV) Figure 3.1 Infinite dilute 56 Fe total cross section at the beginning of the unresolved resonance range. The difference between self-shielded data (ao = 1.85 b) and infinite dilute data is given by the dashed curve. MCNP benchmark library 238 U; JEF2.2 data <W«VB0) o to,(o 0-23) o M (o ) 10 Figure U total cross section in the unresolved resonance range for several values ECN-R

12 4. CONCLUSIONS In this report the computer code UNR is described, with which problem-dependent libraries can be produced for the Monte Carlo code MCNP. In these libraries selfshielding in the unresolved resonance range is taken into account in an approximate way, by using only one value of the background cross section CTQ for the complete unresolved resonance range. It is shown, that significant effects may occur in shielding applications (thick layers of Fe) or in fast reactor applications (self-shielding in 238 U). The method used is approximate, and good results will only be obtained if the unresolved resonance parameters on the ENDF are correct neutrons will sample the complete unresolved resonance range in the MCNPcalculations. A better method would be to sample resonances from probability tables. However, this method will only be included in forthcoming versions of MCNP. 10 ECN-R

13 REFERENCES REFERENCES [1 ] J. F. Briesmeister (ed.), MCNP - A General Monte Carlo N-Particle Transport Code, Version 4A, Report LA M, Los Alamos National Laboratory, November 1993 [2] R. E. MacFarlane: How to NJOY ENDF-6, presentation given at the International Workshop on NJOY, Saclay, France, 6-10 April 1992 [3] D. W. Muir and R. E. MacFarlane: The NJOY Nuclear Data Processing System, Report LA-9303-M, Los Alamos National Laboratory, 1987 [4] R. E. MacFarlane: TRANSX 2: A Code for Interfacing MATXS Cross-Section Libraries to Nuclear Transport Codes, Report LA MS, Los Alamos National Laboratory, July 1992 [5] A. Hogenbirk: Improvements in the processing ofeff-2 data for MCNP using NJOY9L38, Report ECN-C , Netherlands Energy Research Foundation ECN, September 1993 ECN-R

14 UNR 12 ECN-R

15 APPENDIX A. INPUT DESCRIPTION FOR UNR The program UNR is interactive. The following self-explaining questions are asked: input and output unit number The same convention is used as in NJOY. Only positive (=ascii) unit numbers are allowed. mat Enter material identifier on ENDF. mtl8? (Y/N) Specify whether mtl8 data are present on the ENDF or not. If these data are available, the self-shielding treatment is carried out as well for these data. ECN-R