Abstract
Abstract The simulation of neutron transport equation (NTE) has become one of the main trends of nuclear reactor engineering and design owing to its high accuracy, wherein one of its limitations is the low computational speed. To improve this condition, this paper provides a GPU accelerated neutron transport lattice Boltzmann method (NTLBM) for solving neutron transport problem. The simplified finite Boltzmann scheme and the GPU CUDA technique are combined to achieve the spatial parallel calculation and the space-angle full parallel calculation. Due to the ease of implementation and high parallelism, the GPU accelerated NTLBM (GPU-NTLBM) can be very easily and efficiently to realize. Besides, by using the multi-GPU acceleration, the large-scale NTLBM engineering calculation can be efficiently speeded up, and the properties of NTLBM makes the internal boundary treatments convenient. In considering the properties of the NTE, the space-angle full parallel acceleration is applied to further speed up the GPU-NTLBM for the first time, in which all the spatial and the angular dimensions are calculated in parallel. Simulation results show that the NTLBM can be applied to solve the multi-group neutron transport problems accurately and the GPU acceleration can be applied to effectively speed up the NTLBM calculation. The space-angle full parallel acceleration can further improve the GPU-NTLBM acceleration. This paper could provide some parallel perspectives for solving the large-scale neutron transport engineering calculation.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call