Optimal Groundwater Remediation System Design With Well Locations Selected As Decision Variables

Mustafa M. Aral and Jiabao Guan

Multimedia Environmental Simulation Laboratory

School of Civil and Environmental Engineering

Georgia Institute of Technology

Atlanta, Georgia 30332, USA


The design of a pump-and-treat groundwater remediation system can be formulated and analyzed as an optimization problem. In this study, an optimization model which minimizes the total cost of pump-and-treat remediation operation is used to analyze this problem. This analysis usually involves coupling of groundwater flow and contaminant fate and transport simulation models with optimization algorithms. Similar problems were investigated in the literature using classical optimization methods. Solution procedures used in the earlier work usually tend to be numerically complex or computationally inefficient and costly when the governing equations are nonlinear. These solutions require repeated use of the simulation models throughout the computational cycle to reevaluate the new conditions generated by the optimization algorithm. To overcome this drawback, a new computational procedure is described for the solution of the optimization model and coupling of simulation models in this solution algorithm. The optimization model we utilized includes concentration, velocity and extraction and injection equilibrium constraints along with an objective function which describes the total cost of remediation operation in terms of volumetric extraction and injection rates. In this formulation, location and extraction or injection rates of the candidate pumping wells are selected as decision variables. The location of the injection and extraction wells were searched in a continuous manner which requires the reconstruction of the idealization of the solution domain for each simulation. This degree of freedom added to the proposed procedure yielded significant improvements in the results. Based on conventional Genetic Algorithm (GA), a novel computational approach identified as the progressive genetic algorithm (PGA) is used to solve the optimal design problem. PGA is an alternative method which offers significant advantages in the solution of highly nonlinear optimization problems. Test problems investigated indicate that the proposed approach provides a feasible alternative for the solution of optimization problems in groundwater management.

Back to MESL Reports