Abstract: The direct measurement of unsaturated permeability coefficient is costly and the accuracy cannot be fully guaranteed. Therefore, the establishment of a simple and practical prediction model for saturated/unsaturated permeability coefficient based on Darcy’s law has important theoretical and practical significance. In this paper, a self-made unsaturated permeability device was used to test the unsaturated permeability coefficient of Yan’an compacted loess with different dry densities, and the pore size distribution curve was determined by nuclear magnetic resonance technology. Based on the characteristics of pore distribution, Darcy’s theorem is differentiated, and the relationship model between pore ratio and saturated/unsaturated permeability coefficient is established. The research results show that the parameters (D and B) in the prediction model can be determined by the slope and intercept of the cumulative pore volume at two points (peak point and half-width point) on the pore size distribution curve and the pore size in a straight line in double logarithmic coordinates; the porosity ratio and the dominant pore diameter have a linear relationship in the double logarithmic coordinates, and the model parameters can be expressed by the porosity ratio; the predicted results of the model are basically consistent with the measured values, which suggests that the proposed model is simple, reliable and practical.

Key words: compacted loess, unsaturated infiltration, permeability coefficient, pore size distribution, prediction model