Bias extension for CANDU reactor applications using ZED-2 critical experiments

Validation of reactor physics codes requires the determination of bias and uncertainty in calculating parameters of importance (such as reactivity coefficients) in the analysis of nuclear power reactors. A challenge of this process is extending bias and uncertainty from validation against measurements in a test facility (such as AECL's ZED-2 critical facility at the Chalk River Laboratories) to the power reactor. AECL, with help from ORNL, is developing the use of sensitivity/uncertainty (S/U) analysis for bias extension for reactivity coefficient calculations (including coolant void reactivity, which is not a coefficient per se) to CANDU-type power reactors. This builds on methodology for validation of keff in criticality safety analysis [1]. An underlying requirement of this method is that the physics codes model the reactor accurately with respect to the underlying nuclear theory, geometry and material composition. With a Monte-Carlo code and a high-fidelity model in which modelling uncertainties are minimized, the largest contributor to uncertainty in the calculation is from the nuclear cross section data. The main S/U analysis function is to assess the ZED-2 experiment representativity of the power reactor and to generate cross-section adjustments that increase consistency of results for the purpose of determining bias: Assessment of uncertainties and sensitivities of keff and reactivity coefficients, due to uncertainties in nuclear cross-section data libraries used in MCNP5 [2] simulations. Estimation of representativity of the power reactor by the ZED-2 critical core experiments. This is assessed and quantified by the sensitivity of the keff calculation to variations of cross-section data within their uncertainties, as well as the sensitivities of reactivities. Extension of MCNP5 computational biases in keff and various reactivity coefficients based on simulations of ZED-2 experiments to the conditions of the power reactor.