Optimal location of pumping wells by a mesh-free numerical method

Abstract

Abstract In the present study, the optimal place to excavate extraction wells as the drawdown is minimized was investigated in a real aquifer. Meshless local Petrov–Galerkin (MLPG) method was used as the simulation method. The closeness of its results to the observational data compared to the finite difference solution showed the higher accuracy of this method with a root mean square error for MLPG of 0.757 m compared to the finite difference which equaled 1.197 m. Particle warm algorithm was used as the optimization model. The objective function was defined as the summation of the absolute values of difference between the groundwater level before abstraction and the groundwater level after abstraction from wells. In the Birjand aquifer, which was investigated in transient state, the value of objective function before applying the optimization model was 2.808 m, while in the optimal condition it reached 1.329 m (47% reduction in drawdown). This was investigated and observed in three piezometers. In the first piezometer, the drawdown before and after model enforcement was 0.007 m and 0.003 m, respectively. This reduction occurred in other piezometers as well.