Selecting benchmarks for reactor simulations: An application to a lead fast reactor

For several decades reactor design has been supported by computer codes for the investigation of reactor behavior under both steady state and transient conditions. The use of computer codes to simulate reactor behavior enables the investigation of various safety scenarios saving time and cost. There has been an increase in the development of in-house (local) codes by various research groups in recent times for preliminary design of specific or targeted nuclear reactor applications. These codes must be validated and calibrated against experimental benchmark data with their evolution and improvements. Given the large number of benchmarks available, selecting these benchmarks for reactor calculations and validation of simulation codes for specific or target applications can be rather tedious and difficult. In the past, the traditional approach based on expert judgement using information provided in various handbooks, has been used for the selection of these benchmarks. This approach has been criticized because it introduces a user bias into the selection process. This paper presents a method for selecting these benchmarks for reactor calculations for specific reactor applications based on the Total Monte Carlo (TMC) method. First, nuclear model parameters are randomly sampled within a given probability distribution and a large set of random nuclear data files are produced using the TALYS code system. These files are processed and used to analyze a target reactor system and a set of criticality benchmarks. Similarity between the target reactor system and one or several benchmarks is quantified using a similarity index. The method has been applied to the European Lead Cooled Reactor (ELECTRA) and a set of plutonium and lead sensitive criticality benchmarks using the effective multiplication factor (keff). From the study, strong similarity were observed in the keff between ELECTRA and some plutonium and lead sensitive criticality benchmarks. Also, for validation purposes, simulation results for a list of selected criticality benchmarks simulated with the MCNPX and SERPENT codes using different nuclear data libraries have been compared with experimentally measured benchmark keff values.