Fast-Running Lattice Boltzmann Method for Prediction of Real-Time Helicopter and Shipboard Environment Aerodynamic Interactions

Navy rotorcraft shipboard launch and recovery operations encounter complex aerodynamic interactions between the rotor wake and ship airwake which can significantly affect handling qualities. Accurate modeling of these effects is thus essential for predictive ship launch and recovery, and urban environment flight simulations, but real-time piloted flight simulations typically use simplified aerodynamic models that neglect these important interactional aerodynamic effects. High-fidelity Navier-Stokes CFD methods may be used to accurately model the effects, but the computations are significantly slower than real time. This compromise in fidelity is a barrier to using modeling and simulation for launch and recovery envelope development and for pilot shipboard certification. Recent advances in the Lattice-Boltzmann Method (LBM) offer the possibility of calculating the evolving flowfield in the loop within a piloted flight simulation because the equation set is reduced and the algorithms are highly scalable on modern GPU computer hardware. This project will develop and deliver an innovative fast-running LBM solver and computational architecture for prediction of real-time helicopter and shipboard environment aerodynamic interactions. The proposed approach leverages the multidisciplinary Loci software framework and builds upon an existing vectorized Boltzmann equation solver that makes use of Single Instruction, Multiple Data (SIMD) style fine-grained parallelism to accelerate the Boltzmann propagation and collision operations. Loci provides an abstraction whereby computational kernels may be mapped either to CPU or GPU depending on computing requirements, and this infrastructure will be adopted and expanded upon during this project to enable a fast-running GPU-accelerated Loci/LBM solver. Phase I will focus on developing the Loci/LBM physical modeling capability and demonstrating accuracy for ship airwake predictions. Additionally, a piloted flight simulation of a hovering rotor in close ship proximity will be carried out using the GAZEBO open-source flight simulator. Working with faculty in the Department of Computer Science and Engineering at Mississippi State University, our STTR collaborator, the Loci/LBM implementation will be expanded upon to support thread-level SIMD parallelism for GPU architectures and detailed plans for development of a fully performant GPU implementation will be established. During Phase II, a fast-running and mature GPU version of the Loci/LBM solver will be developed and deployed for auto-piloted simulations of shipboard launch and recovery operations including single and multiple main rotor aircraft using the Navy CASTLE flight simulation software. Verification and validation of the Loci/LBM prediction accuracy will be made by comparison with experiment and high-fidelity Navier-Stokes CFD simulations.