GPU-accelerated CFD for high-viscosity mixing

Results from this joint study by Unilever and the University of Manchester show that GPU-based Simcenter M-Star outperformed a CPU-based solver in both speed and efficiency. Mixing processes are vital for ensuring product quality in industries like food and pharmaceuticals, especially when dealing with high-viscosity fluids. This study compares two CFD methodologies—RANS-FVM and LB-LES solvers—in predicting blending times for miscible liquids with contrasting viscosities, validated using experimental ERT data.