Best Estimate Plus Uncertainty Analysis of LBLOCA for a Pickering B CANDU Reactor

The Pickering B Nuclear Generating Station consists of four CANDU reactors. These reactors are horizontal pressure tube, heavy water cooled and moderated reactors fuelled with natural uranium. Under a postulated large break loss of coolant accident (LOCA), positive reactivity results from coolant void formation. The transient is terminated by the operation of the safety systems within approximately 2 seconds of the start of the transient. The initial increase in reactor power, terminated by the action of the safety system, is termed the power pulse phase of the accident. In many instances the severity of an LBLOCA can be characterized by the adiabatic energy deposited to the fuel during this phase of the accident. Historically, Limit of Operating Envelope (LOE) calculations have been used to characterize the severity of the accident. LOE analyses are conservative analyses in which the key operational and safety related parameters are set to conservative or limiting values. Limit based analyses of this type result in calculated transient responses that will differ significantly from the actual expected response of the station. As well, while the results of limit calculations are conservative, safety margins and the degree of conservatism is generally not known. As a result of these factors, the use of Best Estimate Plus Uncertainty (BEPU) analyses in safety analyses for nuclear power plants has been increasing. In Canada, the nuclear industry has been pursuing best estimate analysis through the BEAU (Best Estimate Analysis and Uncertainty) methodology in order to obtain better characterization of the safety margins. This approach is generally consistent with those used internationally. Recently, a BEAU analysis of the Pickering B NGS was completed for the power pulse phase of a postulated Large Break LOCA. The analysis comprised identification of relevant phenomena through a Phenomena Identification and Ranking (PIRT) process, assessment of the code input uncertainties, sensitivity studies to quantify the significance of the input parameters, generation of a functional response surface and its validation, and determination of the safety margin. The results of the analysis clearly demonstrate that the Limit of Operating Envelope (LOE) results are significantly conservative relative to realistic analysis even when uncertainties are considered. In addition, the extensive sensitivity analysis performed to supplement the primary result provides insight into the primary contributors to the results.