Prediction accuracy improvement for neutronic characteristics of a fast reactor core by extended bias factor methods

Kugo, Teruhiko ; Mori, Takamasa; Yokoyama, Kenji   ; Numata, Kazuyuki*; Ishikawa, Makoto

The extended bias factor methods, the LC and the PE methods, are applied to the prediction accuracy improvement for criticality and sodium void reactivity of a traditional two-region homogeneous fast reactor core with utilizing various experimental results. The extended bias factor methods are more effective than the conventional bias factor method. The PE method is more effective than the LC method when a small number of experimental results are used because of the advantage of the former method, its higher degree of freedom in combining experimental results. The advantage hardly affects on the prediction accuracy improvement when a sufficiently large number of experimental results are used. The variances due to cross sections originally included in the design calculation values for the criticality and sodium void reactivity are almost eliminated by the extended bias factor methods with use of about 200 experimental results regarding various neutronic characteristics. The uncertainty of the criticality is considerably reduced, because the uncertainty due to cross sections largely occupies in the original total uncertainty. The uncertainty reduction in the sodium void reactivity is not so much, because the uncertainty due to cross sections is smaller than that due to calculation methods.