Multiple Regression Analysis-Based Reveals the Effect of Complex Parameters of Porous Media on Preferential Flow

Air-gap diffusion distillation (AGDD) is a promising new technology that utilizes superhydrophilic porous media both as thermal channels and as evaporation surfaces to achieve high-efficiency desalination. However, the preferential flow phenomenon within superhydrophilic porous media significantly impairs AGDD desalination efficiency, so a rationally structured porous media must be designed to minimize preferential flow. Nevertheless, previous studies on preferential flow rarely addressed porous media structural parameters. In order to overcome the experimental limitation of distinguishing individual contributions of various porous media parameters to preferential flow under correlated conditions, we conducted extensive preferential flow experiments based on AGDD and introduced statistical methodologies. A stepwise multiple regression analysis was conducted to reduce the dimensionality of independent variables and construct regression equations (with coefficients of determination, R², reaching 82.1-98.4%). According to the experiments, apparent skeleton density, mean pore diameter, permeability, and tortuosity of superhydrophilic porous media are the key parameters determining preferential flow. However, only apparent skeleton density suppresses preferential flow, while pore diameter, permeability, and tortuosity promote it. There are two distinct patterns of the generation and development of preferential flow: Heterogeneous Channel Pattern (under low saturation conditions), where preferential flow is primarily influenced by the inherent heterogeneity of the porous material, as represented by the mean pore diameter. Larger pore diameters directly induce instability at the wetting front. There is also the Seepage Instability Pattern (under high saturation conditions), which is primarily influenced by the instability of the seepage flow field, represented by permeability, resulting in preferential flow. Because unsaturated hydraulic conductivity is highly sensitive to saturation, it intensifies hydraulic conductivity heterogeneity rapidly as saturation increases, resulting in velocity differentials that produce preferential flow when saturation occurs.