Multiobjective Design of Dynamic Monitoring Networks for Detection of Groundwater Pollution

Abstract

A methodology is developed based on the solution of optimization models for optimal design of groundwater quality monitoring networks. The optimal monitoring network is time varying as the monitoring wells are installed in stages, considering the transient pollutant transport process. The monitoring locations specified as solution to the optimization model change with time. This ensures additional economy in the installation of the network, compared to a single stage network design. The optimization model incorporates uncertainties in prediction or estimation of some of the aquifer parameters such as hydraulic conductivity and dispersivity. Advective, dispersive, and radioactive decay processes of transport in a two-dimensional groundwater system are considered. Randomly generated aquifer parameter values are used to simulate different realizations of resulting pollutant plumes incorporating uncertainties in predicting the transport process. The simulated pollutant plume realizations are subsequently utilized to obtain cumulative distribution functions (CDFs) of actual concentrations at different spatiotemporal locations. These CDFs are incorporated as an approximated distribution function in the optimization model. These CDFs are used to define chance constraints with associated reliabilities. Both single and multiple objective nonlinear optimization models are developed. Performances of these models are evaluated for different specified values of reliabilities using illustrative problems considering tritium