Probabilistic approach to unsaturated seepage analysis

The McWhorter and Nelson model for unsaturated seepage beneath a tailings pile has been widely used to estimate the time-dependent movement of the wetting front. This paper presents a probabilistic approach to the first stage seepage using a Monte Carlo approach. Water content and porosity are taken as normal variables while saturated hydraulic conductivity is taken as a log-normal variable. Pore size distribution index is assumed to follow a truncated normal distribution. The probabilistic approach is possible because of the large amount of laboratory data available. The model is mathematically simple enough that very small (8 bit, 64k) micro-computers can perform the computations cheaply and efficiently. Large variances in seepage time are found as a result of the large variances in saturated hydraulic conductivity. The resulting seepage times follow a log-normal distribution