Optimal In Situ Bioremediation System Design for Contaminated Groundwater Using Meshless EFGM Simulation and PSO

In situ bioremediation is the widely accepted technology for the remediation of groundwater sites contaminated with organic contaminants. However, designing an optimal in situ bioremediation system for the contaminated sites is a major challenge to the groundwater scientists. In the past, the various simulation-optimization (S/O) models have been proposed for the optimization of in situ bioremediation system. An S/O model couples the groundwater flow, contaminant, and oxygen transports simulations with an optimization algorithm. The well locations and their pumping rates are the important entities in determining the optimal in situ bioremediation cost. However, the simulation models utilized in the previous in situ bioremediation studies were mostly based on the finite difference method (FDM) and finite element method (FEM), and hence, they are only capable of locating the optimal well locations from the discretized nodes in the computational domain. In this study, we propose an S/O model where the simulator called BIOEFGM is based on the meshless element-free Galerkin method (EFGM) and it is integrated with the particle swarm optimization (PSO). BIOEFGM uses the moving least squares (MLS) approximation for computing the shape functions and hence, it provides stability in approximating the unknown variables in governing equations. In the developed S/O model, meshless BIOEFGM allows PSO to test all the locations i.e., coordinates of the discretized nodes as well of the remaining portion for the optimality. In this study, BIOEFGM-PSO model is used for the optimal in situ bioremediation of the two-dimensional aquifer site. BIOEFGM model is first verified with FEM and then used with the PSO. The selected problem is also solved with FEM-PSO model for comparing the bioremediation costs. The optimal bioremediation cost from the BIOEFGM-PSO is found to be less than that from the FEM-PSO. This study shows the advantages of using the proposed model and therefore, can be applied to the large field problems.