Comparison of point-particle models and direct particle-fluid simulations for non-spherical particles

Coupled particle-laden pipe and free jet simulations are performed at a Reynolds number of R e D = 15 000 based on the nozzle diameter. The flow is laden with heavy rigid spherical and prolate particles, with aspect ratios in the range of 1 ≤ β ≤ 8 typically found in the context of bio-mass oxy-fuel combustion. A fully developed turbulent periodic pipe flow with a volume loading of ϕ v = 6.67 ⋅ 10 − 4 supplies the inflow for the free jet through a slicing technique, thereby ensuring consistent inflow conditions for the fluid and dispersed phases. An Euler-Lagrange approach is used for the simulations, where a large-eddy simulation is performed for the carrier flow and individual particles, represented by point masses, are tracked in a Lagrangian frame. Three different point-particle models are used. A conventional spherical model, an ellipsoidal model derived for creeping flow conditions, and an ellipsoidal model that includes orientation-dependent correlations for drag, lift, and torque valid for finite particle Reynolds numbers. In this study, results from the point-particle models are compared with interface-resolved direct particle-fluid simulations (DPFS) and the predictive capability and accuracy of each model are assessed. Discrepancies in the wall-bounded pipe flow are discussed and improvements concerning, e.g., the collision model and the limitations of the correlations derived under the assumption of uniform flow are outlined. Within the jet, the extended ellipsoidal model shows good agreement with the DPFS data for the translational particle motion, radial spreading, and preferential orientation. • Pipe and free jet simulations laden with heavy rigid non-spherical particles. • Spherical and ellipsoidal particles with aspect ratios 1 ≤ β ≤ 8. • Large eddy simulations at a bulk Reynolds number of R e D = 15 000. • Spherical and ellipsoidal point-particle models are investigated. • Point-particle simulations are compared to direct particle-fluid simulations.