Simulation of a research reactor reactivity transient with deterministic and GPU-assisted Monte Carlo reactor kinetics codes

Reactor kinetic codes are crucial in safety assessment. Validating spatial and high temporal resolution kinetic solvers without thermal feedback is problematic as measurements seldom involve detailed spatial and fine temporal resolution. Benchmarking of deterministic codes thus often resorts to code-to-code comparison against Monte Carlo codes, which can only recently treat direct time dependence. In this paper, we have attempted to compare results from the GUARDYAN directly time-dependent Monte Carlo code and the SEnTRi transient driver developed for the PARTISN deterministic transport code to low power transient measured at the BME Training Reactor. Code-to-measurement comparisons were successful, despite a major uncertainty in the actual timing of the reactivity insertion and withdrawal originating from the instrumentation of the pneumatic rabbit system. Code-to-code comparisons concluded that time dependence was correctly implemented in both GUARDYAN and SEnTRi; furthermore, a hypothetical scenario was set up involving an instantaneous insertion of a negative reactivity into the BME TR core in order to compare spatially and temporally dependent fluxes. The simulations demonstrated the appearance of higher-order modes, and results showed a relatively good match, although fidelity of the comparison could be further improved by reducing the statistical uncertainty of the results provided by GUARDYAN.