Abstract
•Validation with experimental data, DNS and literature data.•Best performance with flow attack angle θ=45°, especially in laminar flow.•As the pitch-to-height P/H decreases, Nu and Sh increase, but f increases more.•Nu and Sh higher than for non-woven spacers, but f higher as well.•The complex effects of Re, P/H and θ require interpolations from a large database.
Flow and heat or mass transfer in channels provided with woven spacers made up of mutually orthogonal filaments were studied by Computational Fluid Dynamics. The problem addressed was the combined effect of the parameters that characterize the process: pitch to height ratio P/H (2, 3 and 4), flow attack angle θ (0, 7, 15, 20, 30, 40 and 45°) and Reynolds number Re (from ~1 to ~4000). The Prandtl number was 4.33, representative of water at ~40°C, while the Schmidt number was 600, representative of NaCl solutions. Simulations were performed by the finite volume code Ansys CFX™ 18.1 using very fine grids of ~6 to ~14 million volumes. For Re > ~400, the SST turbulence model was used to predict flow and heat transfer; no simulations of mass transfer were performed in the turbulent regime because the resolution of the diffusive sublayer would have required a prohibitive number of grid points. Results were validated against experimental data, including results obtained by Liquid Crystal Thermography and Digital Image Processing. The flow attack angle θ = 45° was the most effective for mixing (higher Nusselt number, Nu, and Sherwood number, Sh) and caused lower values of friction coefficient (f). In the range investigated, increasing the pitch to height ratio P/H caused Nu, Sh and f all to decrease. Therefore, the highest values of Sh and Nu were provided by the configuration P/H = 2, θ = 45°. Compared with non-woven spacers, woven spacers provided a better mixing, especially at intermediate values of Re, but at the expenses of higher pressure drops.
[Display omitted]