Simulation of multiple propellers using a GPU-optimized overset grid-based Immersed Boundary method.

Overset grid-based Immersed Boundary methods effectively concentrate finer mesh near the body while keeping a coarse background mesh. This reduces the total grid required for achieving

a well-resolved flow resolution near a body. The presence of multiple bodies is catered to using multiple overset grids. Researchers have used multiple CPUs in parallel to reduce computational time. However, this comes with an additional overhead cost due to the communication required between the CPUs. So, it also affects the performance when

the number of bodies increases. Also, to get substantial speed-up, many CPU cores are needed, and the data management and transfer between them must be explicitly monitored. We hereby propose a GPU-based overset grid Immersed Boundary method, which obtains speed-ups similar to large multicore systems, thus effectively able to simulate multiple

bodies having very fine near-body grid resolutions whilst using minimal computational resources like a single GPU card. We demonstrate this capability of our solver with an example of multiple propellers.