Contaminant detection probability in heterogeneous aquifers and corrected risk analysis for remedial response delay

Abstract

A Monte Carlo stochastic model was developed to simulate contaminant transport from an instantaneous source into heterogeneous, two‐dimensional aquifers. Probabilities of detection Pd and contaminated groundwater areas were calculated for different arrangements of monitoring wells. It was shown that detection decreases as heterogeneity increases. Monitoring with 20 wells provides high detection, while 3 wells result in four out of five contamination cases to remain undetected. For fixed heterogeneity, for each well arrangement, Pd increases up to a certain value with increasing transverse dispersion coefficient aT and then it decreases. The frequency of sampling is critical in heterogeneous aquifers with bi‐annual or monthly sampling improving Pd by 40%, and 70%, respectively, relative to that by annual sampling. It appears that at a minimum sampling should take place twice a year, with the once‐in‐a‐month sampling appearing the optimum choice considering the effort involved and the improvements in detection. In heterogeneous aquifers a large number of monitoring wells sampled infrequently does not perform any better in terms of detection than a lower number of wells sampled regularly. Finally, we propose an expression that accounts for the delay between detection and remedial action in order to provide a correction to decision analyses that evaluate the economic worth of well monitoring. Our expression illustrates that delays of over 3 years are equivalent to reducing the economic performance of 12 wells to that of a lower number of wells, or equally, having to consider higher failure costs than those assumed in current risk analyses.