Assessment of coarse grained CFD–DEM modelling strategies in horizontal plug flow pneumatic conveying

The computational expense of CFD–DEM simulations can become infeasible at an industrial scale, where the number of particles may reach billions. Using scaled-up particle sizes in coarse grained (CG) CFD–DEM simulations could potentially reduce this computational expense substantially while preserving the key physics, provided that appropriate scaling methods are employed in the simulations. This study examines six contact force and three drag force scaling methods in plug flow pneumatic conveying, aiming to develop a clear methodology for similar operations. Comparisons identified cubic drag force scaling as the most accurate, while linear and quadratic scaling both resulted in unrealistic air velocities. Contact force scaling methods which involved scaling of the sliding friction coefficient led to inaccuracies; other contact force scaling methods produced consistent results. Higher CG ratios increased plug porosity and affected the plug velocity and particle exchange rates between the plug and stationary layer. Spatio-temporally averaged continuum fields indicated that the CG ratio affected the solid fraction and momentum flux within the plug. The constant relative overlap model preserved contact pressures better than the constant absolute overlap model, especially at higher CG ratios. Smoothing exchange fields improved stability and accuracy when particle sizes exceeded the CFD cell size at high CG ratios. Overall, the combination of cubic drag force scaling and constant relative overlap contact force scaling produced the most reliable results, preserving contact pressures and allowing larger DEM time steps. An analysis of computational costs suggested an achievable balance between computational expense and spatial resolution. • Three drag and six contact force scaling methods were compared. • Cubic drag force scaling yielded results closest to the original case simulation. • Scaling the sliding friction coefficient led to poor results. • Quadratic contact force scaling preserved contact stresses and sped up simulations. • A pipe-to-parcel diameter ratio of approximately 17 ensured optimal results.