GPU-accelerated FVM-RANS Hybrid Solver for Simulating Two-phase Flow beneath a Hovering Rotor

摘要

The particle-laden flowfield beneath a hovering micro-rotor is simulated using a hybrid FVM-RANSrnsolver coupled with a Lagrangian sediment tracking algorithm. The present work attempts to examinernthe capabilities of a high-fidelity CFD analysis together with a sediment tracking algorithm to analyzernthe evolution of brownout clouds. Both the single-phase and dual-phase flowfields are simulated andrncompared to available experimental data. The variation of the phenomenological attributes of the cloudrnevolution as a function of particle size was analyzed and found to agree with experimental evidence. Tornenable single-desktop simulations, all components of this methodology are designed to run on doubleprecision,rnprogrammable Graphics Processing Units (GPUs). This hybrid implementation uses a multigranular,rnparallel approach to demonstrate significant performance gains over equivalent single-corernCPU simulations.