An effective multi-objective optimization approach for groundwater remediation considering the coexisting uncertainties of aquifer parameters

An accurate groundwater simulation model and an efficient optimization algorithm are critical for building a reliable remediation scheme using simulation-optimization method. The hydraulic conductivity (K) and the porosity (n) of aquifers are two main controlling parameters for groundwater flow and contaminant transport, whose uncertainties are high and often coexists. Such coexisting uncertainties have not been fully considered in previous studies on the optimization of groundwater remediation system. In this work, a new multi-objective optimization algorithm, namely the probabilistic multi-objective fast harmony search and genetic algorithm (PMOFHS-GA), is developed, capable of finding Pareto optimal solutions with high reliability. The algorithm is applied on a case study in designing a groundwater remediation system based on pumping and treating (PAT) technology considering different scenarios with varying K and n realizations. Furthermore, a novel decision-making strategy based on the reliability of different objective functions is proposed to help designers deter-mining the final best remediation scheme from Pareto optimal solutions. Results show that the coexisting un-certainties of K and n will cause significant impacts on final optimization results. In addition, if there is a correlation between uncertainties of K and n, the reliability and variability of optimization results will be different from that under no correlation condition. Thus, accurate information of aquifer parameters is critical for the final optimization results. Furthermore, the proposed decision-making strategy can help decision-makers to identify the final contaminant remediation scheme more reasonably. The PMOFHS-GA and decision-making strategy developed in this study are promising methods which can help decision-makers to design a ground-water contaminant remediation system effectively.