Comparison of results from the GO methodology and fault tree analysis

The GO methodology and fault tree analysis techniques for reliability analysis were used to model the electrical portion of a reactor trip system. The models were prepared independently after agreement on the problem boundary and data base. Results for the probability of removal of control rod holding power on trip demand were compared for single and multiple parameter trips, with and without maintenance. Minimal cut sets of second order were determined for the fault tree and sensitivities were calculated for components in the GO model. The study was intended to demonstrate the overall capabilities, features and limitations of the GO methodology and to introduce the methodology to NRC as an additional analytical resource. The numerical results were intended to be used for the methodology comparison rather than as an accurate reliability evaluation of the Reactor Protection System. Both methods produced the same results when minor differences in modelling were reconciled. The GO results included higher order failures which were significant only for single parameter trips. Intermediate event probabilities and sensitivities for all component kinds were computed from the GO model