A Simulation–Optimization Approach for the Optimal Remediation of Reactive Contaminants in Confined and Unconfined Aquifers

Abstract

The degradation of groundwater quality is a challenging concern, and the restoration of contaminated aquifers is often expensive and resource-intensive, which requires reliable solutions. The reactive groundwater contaminants, which include radioactive substances and certain hazardous species, naturally degrade over time, which impacts the restoration process and should be considered in remediation designs. Simulation–optimization (SO) models are effective tools for designing effective groundwater remediation systems. In this study, a novel SO model is proposed for the remediation of confined and unconfined aquifers that are polluted with reactive contaminants. The meshless weak–strong (MWS) method is used to simulate the coupled groundwater flow and reactive transport, which offers advantages such as high stability and low computational cost. An MWS simulator is coupled with hybrid differential evolution (DE) and particle swarm optimization (PSO) (HDEPSO) due to the properties of HDEPSO to evade local optimal convergence. The MWS–HDEPSO SO model is used for the identification of optimal well locations and extraction rates with the objective of the minimization of remediation costs. In addition, although a high number of remediation wells are considered, only the necessary number remain active during the modeling process. To evaluate the benefits of the MWS–HDEPSO, stand-alone optimizers that are based on the MWS–DE and MWS–PSO are developed. The three SO models are successfully tested on aquifers with simple and irregular geometries. In a large irregular aquifer, the MWS–HDEPSO model identifies a single location for effective remediation, compared with the four and two well locations that are identified by the MWS–DE and MWS–PSO; therefore, the well installation and remediation costs were significantly reduced. In addition, the MWS–HDEPSO model requires significantly fewer iterations than the MWS–DE and MWS–PSO to reach convergence. Therefore, the HDEPSO converges faster, and the proposed model performs better. Therefore, the MWS–HDEPSO model could be a valuable tool for developing effective remediation plans that could aid in the field remediation process of reactive contaminants.