Sensitivity and uncertainty analyses of the Accident Sequence Precursor (ASP) report were undertaken. The sensitivity analysis was performed using a parametric study code developed by the Oak Ridge National Laboratory. The uncertainty analysis was performed using the ASAP code developed in this study and the MOCARS code developed by EG&G Idaho. The sensitivity analysis included estimation of the sensitivity of the calculated core damage frequency to the following parameters:

  1. loss of function failure and recovery factor data recommended by the Institute of Nuclear Power Operations
  2. recovery factors
  3. degraded function multiplier
  4. loss of function probabilities
  5. vendor, reactor type, etc.
The sensitivity calculations were performed for both boiling water reactor and pressurized water reactor precursors, individually. Numerous case runs were made and the results summarized. The uncertainty analysis included estimation of the 95 % confidence limits on the point estimate calculations of the ASP study. Two basic approaches were used to calculate the uncertainty bounds. The procedures used and the results obtained were compared. The point estimates in the ASP study for which uncertainty limits were calculated were
  1. individual significant precursors (52 event trees, each representing one precursor)
  2. the total core damage frequency, both including and excluding the three major precursor events.
The data used for the uncertainty calculations were obtained from the licensee event report data in Table C-l of the ASP study. Binomial or Poisson distributions were used to represent the initiating event frequencies and loss of function probabilities that appear in the data base. To minimize the computation time, for cases where the point estimates were small (i.e., p < 0.05) and the demand number large (i.e., N > 200), the equivalent Poisson distribution was used.