ASSESSING UNCERTAINTY IN SUBSURFACE SOLUTE TRANSPORT - EFFICIENT FIRST-ORDER RELIABILITY METHODS

Due to the heterogeneity of natural groundwater systems, any quantitat ive description of aquifer hydraulic properties is subject to uncertai nty. Consequently, prediction of groundwater contaminant transport is also subject to uncertainty. Stochastic approaches to transport simula tion quantify this uncertainty in terms of random variables and proces ses. An important practical consideration in the application of such m ethods is their large computational cost. In recent years the first-or der reliability method (FORM) has been introduced as a possible techni que for obtaining stochastic results with low computational expense. S pecifically, the implementation of FORM known as advanced FORM (AFORM) has been shown to produce reasonably accurate results when applied to simple problems. However, recently published results indicate that th e computational burden of AFORM can equal, or even exceed, that of Mon te Carlo simulation when applied to groundwater contamination problems with a large number of variables. If FORM is to be a viable alternati ve, the computational costs of the method must be lowered. In this wor k we propose two alternative implementations of FORM that have a highe r computational efficiency. The primary numerical difficulty that aris es in AFORM is locating the linearization point, a procedure that requ ires the solution of a non-linearly constrained optimization problem. We minimize this difficulty by zoning spatially variable aquifer param eters and by defining a new linearization point that can be found more easily. The new approaches are shown to produce results that are comp arable to those obtained with AFORM when applied to a one-dimensional transport problem. Future work will be aimed at generalizing the proce dures described herein. Copyright (C) 1996 Elsevier Science Ltd.