A STUDY ON EFFICIENCY IMPROVEMENT OF HYBRID MONTE CARLO-DETERMINISTIC METHOD FOR BULK CALCULATION IN ISOL FACILITY OF RAON

Abstract

Variance reduction techniques in Monte Carlo (MC) method are generally used to estimate reliable calculation results since the number of particles to be transported is very limited. In this regard, the computational procedure for estimating the desired quantities by generating the parameter of variance reduction technique using the deterministic method and reflecting it to the Monte Carlo calculation is called the Hybrid Monte Carlo-Deterministic method. In most neutronics shielding calculation, its validity and good efficiency of calculation are verified by authors. However, the statistical uncertainty of estimate is relatively high at the outward region of interest in bulk calculation such as the dose map evaluation. In this study, the spatial correlation of detectors inside entire region of interest is suggested, and total adjoint source based on FW-CADIS methodology is reformed. This modification is applied to the ADVANTG code which is generates variance reduction parameters. By using variance reduction parameters generated, MCNP code is used for radiation transport calculation based on Monte Carlo method. In this study, simple test problem is assumed for validation of proposed method and results of bulk calculation are compared to ISOL facility of ROAN as a practical problem. Neutron fluence evaluated in simple test problem by using two Hybrid Monte Carlo-Deterministic methods existed within 95 % confidence interval of reference value which is result evaluated from analogue Monte Carlo (AMC) calculation with histories of 1.0×109. From bulk calculation with same histories of 1.0×105, average FOM factor of proposed method is higher by 2.24 times from 383.2 to 857.3 than FW-CADIS methodology. Neutron dose rate are evaluated in the ISOL facility of RAON by using two Hybrid Monte Carlo-Deterministic methods with 2.0×108. Representatively, neutron dose rate distributions are compared in laserbaem corridor placed in edge region of interest. Average FOM factor is higher by 1.1 times from 0.01687 to 0.01856 than FW-CADIS methodology. Through this study, it is noted that particles are transported preferably outward region of interest by considering spatial correlation of detectors; it causes more uniform statistical precision of estimates in entire region of interest. Therefore, it is expected to be utilized for efficient bulk calculation by using the Hybrid Monte Carlo-Deterministic method.