A surrogate model for simulation–optimization of aquifer systems subjected to seawater intrusion

摘要: Summary This study presents the application of Evolutionary Polynomial Regression (EPR) as a pattern recognition system to predicate behavior nonlinear and computationally complex aquifer systems subjected seawater intrusion (SWI). The developed EPR models are integrated with multi objective genetic algorithm examine efficiency different arrangements hydraulic barriers in controlling SWI. optimization is minimize economic environmental costs. model trained tested for control scenarios, on sets data including pumping patterns inputs corresponding set numerically calculated outputs. results compared those obtained by direct linking numerical simulation tool. two above-mentioned simulation–optimization (S/O) strategies excellent agreement. Three management scenarios considered involving simultaneous use abstraction recharge Minimization cost process salinity levels functions used evaluating each scenario. By considering effects unsaturated zone, subsurface pond collect water artificially aquifer. distinguished feature emerges its metamodel S/O where it significantly reduces overall computational complexity time. also suggest that other sources such treated waste (TWW) and/or storm water, coupled continuous brackish desalination most effective method A sensitivity analysis conducted investigate external recovery ratios plant optimal results.