Investigation of the filtration performance for fibrous media: Coupling of a semi-analytical model with CFD on Voronoi-based microstructure

3D micro-scale model of fibrous media with controllable structural parameters (such as fiber diameter, solid volume fraction and cell configuration) was constructed by using Voronoi algorithm in this study. Compared with the fibrous media model having layer-structured topology, the Voronoi-based model can well characterize the penetration behavior between fibers and layer connections. The gas-solid flow characteristics in fibrous media under different structural and operating parameters were examined numerically and semi-analytically. The results show that the pressure drop of fibrous media between numerical results and experimental values in literature is not always consistent. A new pressure drop correlation for fibrous media is proposed. Additionally, numerical results of the filtration efficiency show similar tendencies to the results that are calculated by the semi-analytical model. For submicron particles, the Stokes-Cunningham drag law combined with Saffman's lift force and Brownian force can accurately predict the capture action dominated by the combination of diffusion and inertia. The deviation of filtration efficiencies between numerical results and theoretical predictions increases with increase of the Peclet number. For micron particles, the semi-analytical model-determined filtration efficiency is generally higher than numerical results which adopting the spherical drag law and considering the gravity effect. Moreover, a threshold value of the Stokes number that determines the deviation between numerical and semi-analytical approaches is presented. When the Stokes number is greater than the threshold, a relatively small deviation can be achieved. It is concluded that the new pressure drop correlation can provide guidance for macro-scale model to study the filtration performance. As well, the correlation calculating of filtration efficiency between numerical simulation and semi-analytical model is obtained by using the Peclet number and the Stokes number as criteria.