Abstract Reservoir simulation for a full field heterogeneous model with millions of grid blocks demands significant computational time so improving the computational efficiency becomes crucial in designing a reservoir simulator. Graphics Processing Unit (GPU), a new high-profile parallel processor with hundreds of microprocessors, stands out in parallel simulation because of its efficient power utilization and high computational efficiency. Also, its cost is relatively low, making large-scale parallel reservoir simulation possible for most of desktop users.
In this paper several GPU-based parallel preconditoners, in conjunction with a new GPU-based GMRES algorithm, are proposed and coupled with an in-house black-oil simulator to speedup reservoir simulation. In particular, massively parallel ILU preconditioners (ILU(0), ILUT, block ILU(0), block ILUT), which are usually regarded as data dependence and highly sequential preconditioners, are developed on GPUs.
In the numerical experiments performed, the SPE 10 problem, a 3D heterogeneous benchmark model with over one million grid blocks, is selected to test the speedup of our GPU solver and preconditioners. On the state-of-the-art CPU and GPU platform, the new GPU implementation can achieve a speedup of over eight times in solving linear systems arising from this SPE 10 problem compared with the CPU based serial solver. Moreover, our GPU solver is successfully coupled with the in-house black-oil simulator to test the performance of the whole parallel simulation process, with a speedup of about six times. The excellent speedup and accurate results demonstrate that the GPU-based parallel linear solver and preconditioners have the great potential in parallel reservoir simulation.