Research On Solution Of 2D Transport Equation Based On Discrete Ordinates Method

In the conditions of high temperature and high pressure, the unknown neutron fluxes of neutron transport equations were defined in energy, time, velocity phase-space and high dimensions geometry space. We need to deal with many problems such as multi-media, multi-group, and the other complex initial conditions, which insult in the large scale computing on solving those equations. So it is necessary to design a scalable parallel method on solving the those equations.A new load-balancing algorithm was presented to solve the unbalancing problem existing the current group-partition algorithm. Experiments on 12 CPUs cluster that has testify that the algorithm can achieve linear speedups.It's a hard problem to design and implement a parallel Sn algorithm based on geometry domain decomposition in unstructured grid. The communicating delay, sorting algorithm and inserting algorithm limited the performance of current algorithm, which decrease the scalable performance of CPUs. So we design a new parallel algorithm based on geometry domain decomposition to solve this problem. Experiments indicate that this algorithm get a better parallel efficiency.At last, a parallel program combining those two algorithms was designed. Experiments testify that the time cost of this program decrease 31% on single PC and decrease 16% on a 96 CPUs cluster than other similar program on solving the same scale problems.