Probabilistic model to predict soil-water characteristics on unsaturated soils

Soil-water characteristics are fundamental to predict the mechanical behavior of unsaturated materials. The latter comes from the fact that unsaturated materials show a hydro-mechanical coupling behavior. In this paper, we present a porous model that can be used to simulate the soil-water characteristic curves. This model is developed under probabilistic considerations and is based on the principles of the percolation theory. The model is based on the probability of a site or bond of certain size to be saturated or drained under certain suction during a wetting or drying process, respectively. With these equations and the pore-size distribution of the material, it is possible to define the volume of water invading the pores and therefore plot the main soil-water characteristic curves at wetting and drying. Also the scanning curves can be obtained during wetting-drying cycles. This model represents an alternative to computational network simulations and imposes on them with several major advantages as it only requires a minimum of conventional computer memory size and is able to simulate wetting-drying processes in few seconds. Additionally, it allows neglecting boundary conditions since it is based on an infinite size network. Validation of the model is based on comparisons made with experimental soil-water retention curves and porosimetry curves for different soils.