Comparison of GPU-accelerated Navier–Stokes and Lattice-Boltzmann Method approaches for mid-fidelity aerodynamics simulations

Mid-fidelity quick-turnaround CFD methods are attractive for aerodynamics characterizations. Reduced-order techniques such as immersed boundary models of bluff bodies and actuator-line models for propeller or rotor blades are of high interest, and these models can be run highly efficiently using structured-block Cartesian GPU-accelerated solvers. In particular, the Lattice-Boltzmann Method (LBM), an alternative fluid simulation method on Cartesian grids, has shown strong performance on GPUs. Yet, comparisons of the LBM with traditional Cartesian finite-volume Navier–Stokes methods on GPUs are limited. In the present work, the accuracy and computational performance of LBM and finite-volume Navier–Stokes methods was assessed for mid-fidelity aerodynamic applications, including an isolated actuator-line rotor and a ship airwake characterization. Both solvers were run using identical computational grids and GPU hardware, allowing for a controlled comparison between the two methods. The results showed that LBM was generally much less dissipative than a 2 nd -order accurate finite-volume method and provided speedups of around 4–6X in time to solution.