Optimization of monitoring well installation time and location during aquifer decontamination

Abstract

An important question in the systematic and objective design of general-purpose ground-water quality monitoring networks is how to evaluate quantitatively the information they provide. However, many applications require the design of a groundwater monitoring network in conjunction with remedial action at a subsurface contamination site. In such a case, it is conceptually clear what is a successful network: One that reduces the net cost of meeting the objectives of cleanup. Uncertainty entails a cost because the natural management response to uncertainty is overdesign for the sake of conservatism (‘safety factor’). The additional information that the network provides must lead to cost reductions that outweigh its cost. This paper presents a method to determine the installation time and location of an additional monitoring well while the aquifer is being cleaned up. While rates of pumping and treatment are determined by the dual control method (a method for optimization with incomplete information) candidate well locations are ranked according to a ‘cost-to-go’ index that measures the costs expected until the goals of remediation are met. This index accounts for the cost associated with uncertainty about the system and thus is useful in appraising the value of information from new measurements in the context of the specific cleanup effort. The usefulness of the method is illustrated through application to a hypothetical two-dimensional aquifer with uncertain initial estimates of the system parameters and variables. Monte Carlo simulations demonstrate the cost effectiveness of solution obtained through this method.