Modeling and Analysis of Flows About Changing Geometries Using the Lattice Boltzmann Method
| The lattice Boltzmann method (LBM) is an increasingly popular approach for simulating fluid flows, valued for its conceptual simplicity, strong theoretical foundation, and excellent scalability on parallel computing hardware. By operating at the mesoscopic scale and discretizing the Boltzmann equation, LBM bridges the gap between particle-based and continuum descriptions of fluids, naturally recovering the Navier-Stokes equations as a special case. This flexibility allows LBM to efficiently model complex, unsteady flows, including flows that involve moving or changing boundaries, that are often challenging for conventional methods.
This work is centered on the simulation of unsteady flows around changing geometries, such as those encountered in fluid–structure interaction and maneuvering configurations. The main objective is to systematically compare a range of boundary condition schemes, including the traditional bounce-back and advanced immersed boundary methods, with respect to their accuracy, stability, and practical implementation. To facilitate this, a flexible and highly efficient multi-GPU LBM solver was developed, allowing for the large-scale simulations that are required for meaningful benchmarking. The outcome of this research includes a detailed evaluation of the strengths and weaknesses of various boundary treatments for unsteady, moving-boundary flows, as well as new insights into best practices for LBM-based simulations on high-performance hardware. These findings contribute to the broader effort of enabling reliable, accurate, and efficient CFD simulations for complex, realistic engineering problems. |
