Utilization of Experimental Integral Data for the Adjustment and Uncertainty Evaluation of Reactor Design Quantities.

摘要

Biases and uncertainties of calculated reactor design quantities caused by errors and uncertainties of basic parameters, such as neutron cross sections, fission spectra parameters, and prompt and delayed neutron yields, are large, and in most cases, exceed reactor design requirements. Errors and uncertainties due to models and methods approximations contribute as well. An extensive data base, with presently /approximately/300 experimental integral values from 28 critical assemblies, has been assembled at Argonne National Laboratory in order to provide improvements and to investigate both sources of uncertainties. Generalized-least-squares fitting is being used. The available large data base permitted the investigation of the influence of specific input data, the constraints of the covariance information, the selection of parameters, and the reliability of the predictions. It is shown that reliable improvements of calculated quantities like enrichment, breeding ratio, sodium void, control rod worth, power distribution, and material worth can be made. Substantial reductions of the uncertainties of these quantities, which are caused by the uncertainties of the basic parameters, are obtained in most cases. The FFTF uranium-metal-core conversion is the first application of the present effort. 21 refs., 2 figs., 10 tabs. (ERA citation 14:003012)