Modeling droplet-laden flows in moisture separators using k-d trees

Moisture separators, such as cyclone and swirl-vane separators, are crucial devices that are used for eliminating droplets in flows, and supply dry-saturation steam for turbines. In this paper, droplet-laden flows were simulated in cyclone and swirl-vane separators based on Lagranian-Eulerian approach. With respect to details, the actual droplets are represented by some characteristic droplets, and the carrier phase is described by Reynolds-Averaged Navier-Stokes (RANS) equations. Our solving routine of the model was coupled with the commercial CFD (computational fluid dynamic) softwares, which were used to calculate the fluid properties of the carrier phase and store them on the mesh grids. However, the CFD softwares could hardly export the grids in an explicit data structure. Therefore, we carried out a data structure called k-d trees to rearrange the grids. Firstly, the results show that the executing speed of our new solver using k-d trees is thousands times larger than the old approach. Secondly, we computed the separation efficiencies of cyclone and swirl-vane separators. Thirdly, our modeling outcomes are verified against experimental data. Finally, the moisture distributions of the separators were analyzed.