In-Situ Bioremediation of Contaminated Groundwater Using Artificial Neural Network

A methodology has been developed in this study to optimize pumping (extraction/injection) rates in in-situ bioremediation system design with the aid of Artificial Neural Network (ANN). The equation governing in-situ bioremediation system consists of the flow equation and the equation of solute transport coupled with biodegradation kinetics, that results into a highly non linear system of equations. Consequently, the computational effort involved is a significant problem in the design of bioremediation systems. In this study, the computational burden is managed within practical time frame by replacing the BIOPLUME III simulation model with an ANN model. Artificial Neural Network has been considered to be a universal function approximator that generates fast solutions in real time. Besides this, the search for the optimal pumping pattern is directed by a simple genetic algorithm in this study. The results show that the costs involved in the proposed management solution is consistent with those resulting from other non traditional optimization techniques which use embedded bioremediation models. However, it has been observed that one can achieve considerable decrease in the number of simulations involved using the ANN approach.