An OpenMP-enabled parallel simulator for particle transport in fluid flows

By using C/C++ programming and OpenMP parallelization, we implement a newly developed numerical strategy for simulating particle transport in sparsely particle-laden fluid flows. Due to its highly dynamic property of the chosen numerical framework, the implementation needs to properly handle the moving, merging and splitting of a large number of particle lumps. We show that a careful division of the entire computational work into a set of distinctive tasks not only produces a clearly structured code, but also allows taskwise parallelization through appropriate use of OpenMP compiler directives. The performance of the OpenMP-enabled parallel simulator is tested on representative architectures of multicore-based shared memory, by running a large case of particle transport in a pipe flow. Attention is also given to a number of performance-critical features of the simulator.