Parametric approach to LMFBR safety evaluation

The assessment of risk from LMFBRs has evolved to the present ''design basis accident'' approach. Of particular interest are consequences from low probability ''hypothetical'' accidents. Due to uncertainties involved in the accident progression, accident consequence spectra are needed to represent the probabilistic nature of the results of accident simulations. Many single simulations are needed in order to create probabilistic distributions of consequences. For these reasons, the accident simulation code PARSEC was developed. It runs more than two orders of magnitude faster than existing codes and is designed for use in parametric studies of uncertain input values. Parametric sensitivity studies show the relative impact of uncertainties on the results of the accident simulation. They indicate particular sensitivity to reactivity coefficients and core heat capacity. Many parameters, especially those associated with hydraulics calculations, show little sensitivity. An example of using simulation results for calculation of probabilistic accident consequence spectra shows wide variability of simulation results. The distributions appear to be largely symmetric, due to the lack of treatment of any highly nonlinear effects and the use of symmetric distributions for probabilistic input. The results show that the probability of realizing ''limiting'' or ''bounding'' case consequences is far out on the tail of the distribution, and are of vanishingly small probability. A comparison of PARSEC and SAS3A results for a 5 channel, unprotected loss-of-flow (LOF) accident shows that PARSEC predicts the core's behavior well. By applying a probabilistic treatment of uncertainties in six input values, the errors in the PARSEC models are shown to be small compared to the overall uncertainties in the results themselves