A hybrid solver based on efficient BEM-potential and LBM-NS models: Recent BEM developments and applications to naval hydrodynamics

We report on recent developments of a 3D hybrid model for naval hydrodynamics based on a perturbation method, in which velocity and pressure are decomposed as the sum of an inviscid flow and a viscous perturbation. The far-to near-field inviscid flows are solved with a Boundary Element Method (BEM), based on fully nonlinear potential flow theory, accelerated with a fast multipole method (FMM), and the near-field perturbation flow is solved with a Navier-Stokes (NS) model based on a Lattice Boltzmann Method (LBM) with a LES modeling of turbulent properties. The BEM model is efficiently parallelized on CPU clusters and the LBM model on massively parallel GPGPU co-processors. The hybrid model formulation and its latest developments and implementation, in particular, regarding the improvement and validation of the model for naval hydrodynamics applications, are presented in a companion paper by O'Reilly et. al (2017), in this conference. In this paper, we concentrate on the BEM model aspects and show that the BEM-FMM can accurately solve a variety of problems while providing a nearly linear scaling with the number of unknowns (up to millions of nodes) and a speed-up with the number of processors of 35-50%, for small (e.g., 24 cores) to large (e.g., hundreds of cores) CPU clusters.