Groundwater contaminant source identification using swarm intelligence-based simulation optimization models

In this study, a linked simulation optimization (SO) model is presented for identification of groundwater contaminant sources. The SO model consists of two steps namely, simulation and optimization. The simulation step entails developing a groundwater contaminant transport model in which the advection-dispersion-reaction equation (ADRE) is solved for predicting the concentration of the contaminant. The system parameters (hydraulic conductivity, dispersivity, etc.) and control variables (pumping, recharge, etc.,) are given as model inputs. A meshless technique called the meshless Local Radial Point Interpolation Method (LRPIM) is employed to solve the contaminant transport equation. The simulation model is linked with three different swarm intelligence-based optimization models namely, teaching-learning based optimization (TLBO), grey wolf optimization (GWO) and particle swarm optimization (PSO) to form three SO models namely LRPIM-PSO, LRPIM-GWO and LRPIM-TLBO. The SO model minimizes the difference between the predicted and observed concentrations to determine the unknown source locations and release histories. The applicability of the developed SO models for source identification (SI) is demonstrated with a hypothetical and real aquifer problems to identify the groundwater contaminant sources. All the 3 models are able to locate the sources and release histories satisfactorily. However, the LRPIM-TLBO has been found to be more accurate followed by LRPIM-PSO and LRPIM-GWO.