Abstract
Safety analyses for nuclear reactors involve inherently various uncertainties. The uncertainties come from the input parameters, physical modeling, and numerical methods employed. In the present study, a stochastic analysis of the decay heat removal of a liquid metal reactor is performed. The output from the stochastic analysis is to obtain the tolerance limit and the confidence level of the computational output, and to identify the sensitive input parameters to the output. The tolerance limit and the confidence level are evaluated using order statistics technique for the computational results. The importance of the individual input parameters are estimated using variance of the conditional expectation. Furthermore, a methodology has been developed for evaluating dependencies among input parameters. It is found that the proposed methodology is effective for understanding the stochastic safety analysis and the results.