Assessment of Lattice-Boltzmann velocity set and floating-point precision for ship airwake analysis compared to RANS–LES and experiment

All modern naval fleets operate ship-based aircraft and rotorcraft, and perform ship deck landings regularly. Despite its frequency and commonality, rotorcraft ship deck landing is still one of the most challenging operations a modern naval fleet faces. To reduce possible operational losses, naval pilots can be trained in realistic flight simulators (if available), which require an accurate yet fast model of the aerodynamic interactions between the ship and rotor airwakes. Previous studies using the Lattice-Boltzmann method (LBM) have shown good results for ship airwake simulations at low computational cost through GPU acceleration. Further improvements to the computational efficiency can be obtained by using a smaller LBM velocity set, as well as reduced floating-point arithmetic precision. In this study, the effects of different LBM velocity sets and floating-point precision on the NATO Generic Destroyer ship airwake were investigated. Results obtained from the LBM were compared to hybrid RANS–LES results and experimental wind tunnel data, and good correlation was found. Results showed that reducing the LBM velocity set and the floating-point precision from double to single provided results that correlated equivalently well with the experimental data, while computing three times faster and reducing the GPU memory requirement by 65%, offering computational time savings of two orders of magnitude compared to the RANS–LES. These findings show the potential for the LBM to be used for large-scale naval aviation problems that require quick turn-around simulations.