A high-performance three-dimensional lattice Boltzmann solver for water waves with free surface capturing.

The lattice Boltzmann method (LBM) is a numerical method with high parallel computational efficiency. However, currently few 3D numerical wave model based on LBM can be widely applied to water wave simulations in coastal engineering due to several problems related to numerical instability and numerical accuracy. To develop a high-performance 3D LBM solver for water wave simulations with free surface capture, a new 3D collision operator with sufficient numerical stability for wave simulation is first proposed. Then, wave generation and active wave absorption methods for the volume of fluid coupled with LBM are implemented in the solver to ensure accurate wave motion. The performances of this solver and other popular numerical wave solvers are analyzed and compared. The parameters of the numerical model are tested and determined to simulate wave transformation around structures. Through the comparison with three benchmark physical experiments, the accuracy of the presented solver proves that the present model is a promising tool for wave propagation and transformation simulation in practice. Benchmark tests and comparisons with the IHFoam and SPHysics solvers prove that the 3D LBM solver developed in this paper has a very good computational efficiency and parallel scalability. Overall, the work in this paper provides a new high-performance 3D solver for water wave simulations with free surface capturing for coastal and ocean engineering applications. • A new D3Q27 MRT collision operator with sufficient numerical stability for wave simulation is first proposed. • A 3D solver based on the new collision operator for water waves is established. • Wave generation and active absorption for the LBM-VOF model are presented. • The LBM solver has the same accuracy as the IHFOAM and the SPHysics solvers for coastal engineering applications. • The parallel computational efficiency of the presented LBM solver is 3~5 times greater than that of the RANS-based solvers. [ABSTRACT FROM AUTHOR]