Evaluation of desirable accuracy for nuclear data in fast reactors by optimization method

The desirable accuracy of nuclear data for minimizing the design margin in fast reactor nuclear designs is evaluated through an optimization method, in which a parameter termed the ''degree of difficulty'' has been introduced in order to represent the relative degree of difficulty expected to be encountered in improving the accuracy of the data. This parameter serves to determine the desirable accuracy, by which the design margin could be reduced the least onerously, to obviate uselessly severe requirements being imposed on the accuracy demanded of data with high sensitivity coefficients. Application of this method to the equilibrium core of a prototype fast reactor leads to the conclusion that nuclear data that should to best advantage be improved in accuracy are: sigma sub(f)sup(41)(400 -- 0.5keV), sigma sub(f) sup(49)(1.4MeV -- 10keV) and sigma sub(c)sup(28)(100 -- 0.5keV). Furthermore, it is indicated that the ratio between the desirable accuracy and the present uncertainty level nuclear data is roughly inversely proportional to the square root of the product of the sensitivity coefficients and the present uncertainty level of nuclear data. (auth.)