Theoretical Study on New Bias Factor Methods to Effectively Use Critical Experiments for Improvement of Prediction Accuracy of Neutronic Characteristics

Abstract

Extended bias factor methods are proposed with two new concepts, the LC method and the PE method, in order to effectively use critical experiments and to enhance the applicability of the bias factor method for the improvement of the prediction accuracy of neutronic characteristics of a target core. Both methods utilize a number of critical experimental results and produce a semifictitious experimental value with them. The LC and PE methods define the semifictitious experimental values by a linear combination of experimental values and the product of exponentiated experimental values, respectively, and the corresponding semifictitious calculation values by those of calculation values. A bias factor is defined by the ratio of the semifictitious experimental value to the semifictitious calculation value in both methods. We formulate how to determine weights for the LC method and exponents for the PE method in order to minimize the variance of the design prediction value obtained by multiplying the design calculation value by the bias factor. From a theoretical comparison of these new methods with the conventional method which utilizes a single experimental result and the generalized bias factor method which was previously proposed to utilize a number of experimental results, it is concluded that the PE method is the most useful method for improving the prediction accuracy. The main advantages of the PE method are summarized as follows. The prediction accuracy is necessarily improved compared with the design calculation value even when experimental results include large experimental errors. This is a special feature that the other methods do not have. The prediction accuracy is most effectively improved by utilizing all the experimental results. From these facts, it can be said that the PE method effectively utilizes all the experimental results and has a possibility to make a full-scale-mockup experiment unnecessary with the use of existing and future benchmark experiments.