Development of efficient simulation- optimization methodology for identification of groundwater pollution sources using meta-heuristic hybrid optimization methods
No Thumbnail Available
The alarming increase in the rate of groundwater contamination has motivated the hydro-geologists to work on identification of groundwater pollution sources. The identification of groundwater pollution sources is the initial step for sustainable management of a groundwater aquifer. The groundwater pollution sources can be identified by using the inverse optimization technique. In this technique, an error function is formulated which minimizes the absolute difference between the observed and the simulated contaminant concentration at observation locations. The observed concentration can be obtained from the field whereas, the simulated concentration is obtained by using groundwater simulation model. Hence, the groundwater simulation model is required to be incorporated to the optimization model. As groundwater simulated model is linked to the optimization model, the technique is called the simulation-optimization model. The model is computationally expensive as the simulation model is repeatedly used by the optimization model. Therefore, the performance of the groundwater source identification model is related to the efficiency of the groundwater simulation model. To overcome this computational burden involved, the artificial neural network (ANN) model can be used as an approximate groundwater simulator. It has been reported that for a large aquifer system, a single ANN model is not sufficient to simulate the flow and transport processes of the aquifer and separate ANN model is required for each of the observation wells to simulate the process.
Supervisor: Rajib Kumar Bhattacharjya