Comparison of Commercial Finite Vol ume Method and Lattice Boltzmann Method Software for Mixing Applications

Abstract

This thesis investigates and compares two numerical methods for simulating indus trial mixing processes: the Finite Volume Method, applied through the commercial software STAR-CCM+ executed on multiple CPU cores, and the Lattice Boltzmann Method, implemented in M-Star executed on a GPU. The methods were tested on three mixing cases: two small-scale biomedical tanks and a large-scale anaerobic digester. The comparison focuses on simulation accuracy, computational efficiency, and practical usability. Results show that both methods accurately capture impor tant flow features in the biomedical tanks, with LBM offering significantly faster simulation times on GPU hardware. However, in the large-scale digester case, LBM faced challenges related to resolution, the need for local refinement, and limited flex ibility in numerical models and solver settings. These factors contributed to long simulation times and less reliable results compared to the FVM approach. These findings suggest that while LBM is a promising method for smaller systems with low tank-to-impeller aspect ratios, while FVM remains more robust for complex, large-scale mixing applications.