Determination of the Computational Bias in Criticality Safety Simulation of SMR version of Sodium Fast Reactor

Under the auspices of EURATOM, a series of European Sodium-cooled Reactor projects were started, developing a concept of a modern GEN IV reactor to further the nuclear energy industry in Europe. As a part of the third, ongoing project ESFR-SIMPLE, two sodium-cooled small fast reactor (SF-SMR) cores were proposed as a more economical option. An integral aspect of advancing the GEN IV nuclear reactor designs involves the evaluation of new and untested concepts by simulation to determine their safety and overall performance. Different code systems, such as SCALE, SERPENT, DYN3D and ERANOS, integrate various methods, each developed on a certain number of approximations and assumptions, inherently introducing bias into the process of simulation. Furthermore, the nuclear data utilised in the simulations also fundamentally introduces a certain bias, due to the evaluation and validation processes, and they also introduce a level of uncertainty, which stems from the imperfections of experimental methods. As the criticality safety is paramount in the new generation of nuclear technologies, the performance of individual simulation methods and nuclear data libraries must be assessed through benchmarking exercises. This paper seeks to assess the biases present in the calculation of the eigenvalue of the more conventional of the two SF-SMR cores. The ESFR-SIMPLE SF-SMR core model is presented, together with the calculated eigenvalues., TSUNAMI-IP and USLSTAT modules of SCALE were used to calculate sensitivity and uncertainty profiles, as well as similarity coefficients and the computational bias using the sensitivity profiles of experiments included in ICSBEP database.