| dc.description.abstract | A model to quantify the temporal failure probability for a nuclear power station’s fleet of multiple, redundant, emergency diesel generators (EDGs) is developed and demonstrated in this thesis. The initiating event for this problem is Loss of Offsite Power (LOOP). This model calculates the probability that the load on the system overcomes (LOOP duration) the capacity of the system (time until the EDGs fail), as a means to quantify system safety margin; this concept comes from The United States Department of Energy (DOE), the Idaho National Laboratory (INL) and the Electric Power Research Institute (EPRI) collaboration on the “Risk-Informed Safety Margin Characterization” (RISMC) approach. The ultimate application of this model is to quantify improved safety margin for an originally two-EDG system that has been upgraded with an additional, reinforced, FLEX diesel generator (DG). Some unique features of the Non-Recovery Integral (NRI) (main model of this thesis) are that it can account for dynamic timing of the EDG failures, model both hot and cold standby EDG arrangements, and accept time-dependent hazard function inputs for hot standby cases (when the hazard functions meet certain conditions). Nuclear industry and Standardized Plant Analysis Risk (SPAR) model data are used as inputs to the NRI to create six specific system model cases. The results from these cases are compared to see how different EDG arrangements affect the overall system reliability. The three main conclusions drawn from the various result comparisons are the following: (1) adding a FLEX DG to an originally two-EDG system makes the system three times less likely to fail for LOOP durations of 24 hours (further improvement in system reliability is seen for longer LOOP durations); (2) the specific model of load placed on the system has a major impact on the system failure probability quantification; and (3) the most effective way to increase safety margin (for the most likely LOOP duration scenarios) is to reduce the likelihood of common-cause failure events. | en |
