Meshless Large Eddy Simulation of Rotorcrafts Using GPUs

This paper presents a meshless large eddy simulation approach for rotorcraft wake prediction, using a vortex particle method accelerated on GPUs. The solver couples a rotor model with a vortex particle wake model, employing the Fast Multipole Method for computational efficiency and implementing viscous diffusion through Particle Strength Exchange and Core Spreading Methods. GPU acceleration achieves speed-ups of up to 10x compared to CPU execution. The solver’s predictions are validated against experimental data, showing excellent agreement. Effects of time step size, numerical integration schemes, viscous models, and particle overlap factors on simulation accuracy and computational cost are systematically analyzed. This GPU-based vortex particle framework provides a fast, accurate, and scalable tool for rotorcraft wake simulations.