Management of coastal aquifers based on nonlinear optimization and evolutionary algorithms

A method of assessing the optimum pumping rates of coastal aquifers based on nonlinear optimization and evolutionary algorithms (EA) is developed. The objective is to maximize the total pumping rate while protecting the wells from sea water intrusion. The formulation of the constraints is based on numerical simulation of the freshwater flow equations. The simulation model is based on the sharp interface and the Ghyben–Herzberg approximation and is applicable to unconfined aquifers and steady-state flow. The single potential formulation of [Water Resour. Res. 12 (1976) 1165] is followed and the governing equations are solved numerically using finite differences. The numerical model can handle aquifers of complex shapes, nonuniform hydraulic conductivity, nonuniform distribution of surface recharge, etc. The constraints are nonlinear with respect to the decision variables resulting in a nonlinear optimization problem. Two optimization methods are investigated, specifically Sequential Quadratic Programming (SQP) and Evolutionary Algorithms (EA). SQP requires less computer time than EA but can get stuck on local optimum solutions. The simulation and optimization methodology is applied to a real unconfined coastal aquifer in the Greek island of Kalymnos for determining the optimal pumping rates while protecting the wells from sea water intrusion.