Effect of domain expansion size on computational stability and efficiency of semi-resolved CFD–DEM model for fluid–particle coupling

The semi-resolved CFD–DEM model based on the domain expansion method eliminates the restriction imposed on CFD grid size and has emerged as a promising method for simulating fluid–particle coupled flow. Nevertheless, the effect of domain expansion size on the model’s computational stability and efficiency remains insufficiently understood. In this study, particle settling in a viscous fluid is first simulated using the semi-resolved CFD–DEM model over a wide range of domain expansion sizes and CFD grid resolutions. The temporal evolutions of particle settling velocity and hydrodynamic forces are analyzed to reveal the intrinsic mechanism triggering computational instability. Subsequently, interactions between fluid and particle groups are further simulated under different domain expansion sizes and parallel thread numbers. The normalized elapsed CPU time of semi-resolved simulations is used to quantitatively evaluate the computational efficiency. Results indicate that increasing the domain expansion size effectively suppresses non-physical fluctuations in hydrodynamic forces induced by the changes of discrete coupling domain, thereby significantly improving the computational stability of predicted sphere settling velocity. However, the computational cost rises exponentially with the domain expansion size under the serial computing condition. This computational bottleneck can be effectively mitigated by combining the hierarchical search method with parallel computing techniques, which optimizes computational granularity and achieves prominent improvement in overall simulation efficiency.