On the need for changing the ENDF/B convention for the representation of cross-sections in the unresolved resonance region of fertile and fissile nuclei

Abstract

A more fundamental scientific basis for the prediction of the Doppler effect in fast reactor systems demands that the usual ENDF/B and KEDAK conventions for the representation of cross-sections in the unresolved resonance region should be changed. This conclusion is based upon the experience acquired in evaluations of statistical mean resonance parameter sets, made at RRC Kalpakkam, for use in neutronic calculations of fast reactors in the recent past for 235U, 239Pu, 232Th, 233U and 238U. This paper gives a brief account of several sensitivity studies and neutronic calculations made with these mean resonance data sets. The existence of non-uniqueness of mean resonance data sets has been clearly established and an uncertainty due to the choice of the mean resonance data set is found to be associated with the theoretical calculations of neutronic parameters of reactor systems. This uncertainty, identified at Kalpakkam, has not received attention in the past and is found to be significant in the case of the calculation of the Doppler coefficient in fast reactors. It should have been taken into account in the interpretation of the Doppler effect experiments performed in fast critical facilities for both fissile and fertile samples. The present method of statistical representation of cross-sections leads to a higher statistical uncertainty in the prediction of the Doppler reactivity effect under coolant voided conditions in fast power reactors than in the normal case. Also the effect of inclusion of intermediate structure in the fission cross-section for 239Pu on the temperature derivative of self-shielding factors is significant and cannot be modelled satisfactorily at present. If the high resolution cross-section measurements, as suggested by de Saussure and Perez, are made available for the main fissile and fertile isotopes, thinning and direct Doppler broadening of the cross-section data by the preprocessing methods developed by Cullen and his coworkers appear to be a promising approach. Until such time the evaluated nuclear data files should attempt to contain self-shielding factors directly evaluated with the support of self-indication and transmission measurements instead of the present convention of having mean resonance data sets in the unresolved resonance region.