Evaluation of Boundary Conditions for Realistic Ship Airwakes Using a GPU-Accelerated Lattice-Boltzmann Solver Compared to High-Fidelity RANS–LES and Experiments

Recent studies have shown that the mid-fidelity computationally-efficient lattice-Boltzmann method (LBM) may be a good alternative to traditional Navier–Stokes-based CFD methods for ship airwake analysis, providing results of comparable accuracy but at far less computational cost. Multiple previous ship airwake studies focused on the Simple Frigate Shape-2 (SFS-2) notional ship geometry that oversimplifies modern combat ship designs. To better represent modern combat ships, the more curved NATO Generic Destroyer (NATO-GD) geometry was analyzed. Several previous ship airwake LBM simulations used the simple bounce-back boundary condition to model the ship surface, which represents the ship as a voxelized solid, hence not approximating curved shapes well. Therefore, this study investigated the effects of the Mei-Luo-Shyy (MLS) and Grad curved-wall LBM boundary conditions for the ship surface and compared results with hybrid RANS–LES results as well as experimental data. Results showed that both the time-efficient LBM and MARIN hybrid RANS–LES results were able to accurately simulate the ship airwake. The Grad and MLS boundary conditions performed better than the bounce-back boundary condition, but due to numerical instability of the MLS boundary condition, the Grad boundary condition was deemed favorable for ship airwake simulations. Reproduced by a synthetic eddy method approach in the LBM simulation, the effects of realistic turbulent atmospheric boundary layer (ABL) inflow conditions were investigated and compared to particle image velocimetry (PIV) measurements. Good correlation was found for the predicted flow upstream of the ship, and for the flow field over the landing deck, relevant to rotorcraft launch and recovery operations.