The effect of wall roughness on particle dynamics in gas–particle flows has been studied both experimentally and numerically, but mainly for channel, pipe, and boundary-layer flows. The present authors did numerical studies and detailed phase-Doppler particle anemometry (PDPA) measurements on the effect of wall roughness for particle dynamics in separating (sudden-expansion and swirling) gas–particle flows, as well as channel flows. The simulated gas–particle flow showed that the prediction results accounting for the wall roughness agree well with the measurement results. The PDPA measurements of backward-facing step gas–particle flows showed that as the wall roughness increases, the longitudinal and the transverse time-averaged particle velocities decrease, but the longitudinal and transverse particle fluctuation velocities increase.

Swirling gas–particle (droplet) flows are commonly encountered in gas-turbine combustors, cyclone combustors, and furnace burners. To better understand the flow behavior, many investigators did measurements, RANS (Reynolds-averaged Navier–Stokes) simulation, and LES (large-eddy simulation) of these types of flows. Most studies were done for weakly swirling gas–particle flows with swirl numbers less than unity. Experimental and numerical studies were done by the present author and his colleagues for PDPA measurements of swirling gas–particle flows with swirl numbers greater than unity, Reynolds-averaged two-fluid (Eulerian–Eulerian) simulation using k–ε–k_p and USM (unified second-order moment) two-phase turbulence models, and two-fluid LES using a two-phase sub-grid stress model. The measurement and simulation results give the two-phase time-averaged and RMS fluctuation velocities, and particle concentration distribution, showing the complex recirculation structures in the two-phase axial velocities and the Rankine-vortex structures in the two-phase tangential velocities, the anisotropic two-phase turbulence properties, and the effect of swirl number on the two-phase flow behavior. The simulation results show that the two-fluid approach using the k–ε–k_p and USM two-phase turbulence models are better than the Eulerian–Lagrangian approach for simulating swirling gas–particle flows