The accident sequence precursor program: Methods improvements and current results

Changes in the U.S. NRC Accident Sequence Precursor program methods since the initial program evaluations of 1969-81 operational events are described, along with insights from the review of 1984-85 events. For 1984-85, (1) the number of significant precursors was consistent with the number observed in 1980-81, (2) dominant sequences associated with significant events were reasonably consistent with PRA estimates for BWRs, but lacked the contribution due to small-break LOCAs previously observed and predicted in PWRs, and (3) the frequency of initiating events and non-recoverable system failures exhibited some reduction compared to 1980-81. Operational events which provide information concerning additional PRA modelling needs are also described. (orig.)