A PROBABILISTIC FRACTURE TRANSPORT MODEL - APPLICATION TO CONTAMINANTTRANSPORT IN A FRACTURED CLAY DEPOSIT

In this paper the principle of minimum relative entropy (MRE) is used to determine the prior probabilities of a set of model parameters base d on limited information. The probability density function (PDF) is of the form of a multivariate truncated exponential distribution. An ill ustration of the effects of parameter uncertainty on the modeling of c ontamination in discrete fractures is given along with a discussion on the sensitivity of the results. A sensitivity study shows that reduci ng the aperture uncertainty had the greatest effect changing the uncer tainty envelope. A probabilistic prediction of contaminant transport c an be quite different from a deterministic approach in which average v alues of parameters are used in a contaminant transport model. It is a lso shown in this study that the 'worst-case' scenario may not be obta ined from simulations using a set of parameters at either their upper or lower bounds. The probabilistic approach is applied to the determin ation of the geologic sensitivity of a landfill site in Manitoba. The approach indicates a 50% probability of exceedance for the drinking wa ter standard for chloride at 3.3 years as opposed to a time-of-travel calculation of 208 years based on an assumption of an equivalent-unfra ctured aquitard medium. Not only do we need to improve the physical un derstanding of the various geologic features and transport processes, but we also need to combine these studies with these of statistical de termination of the fundamental parameters.