Sensitivity of Optimal Groundwater Remediation Designs to Residual Water Quality Violations

Abstract

This work explores the sensitivity of optimal remedial design policies and their associated costs to the residual constraint violation, which is the sum of any small violations in constraints that may occur over all points of interest. To evaluate the sensitivity, a genetic algorithm is used to solve two different groundwater remediation design problems: pump-and-treat using granular activated carbon and enhanced in situ bioremediation. The sensitivity to the residual violation is tested given a range of water quality goals and for static and dynamic cases. The range of residual constraint violations tested was small, so that in all cases greater than 98% of the remediation goal was reached. Nevertheless, it was found that the cost sensitivity to these small constraint relaxations was of the same magnitude as the cost sensitivity to changes in the ultimate water quality goal. The greatest sensitivity was seen for the lowest water quality goals. This work indicates that a remediation designer using optimization tools should consider the trade-offs in cost and performance that will occur depending upon one's approach to constraint enforcement.