Designing groundwater supply systems using the mesh adaptive basin hopping algorithm

Designing groundwater systems is a challenging problem in industrial engineering, where pumping wells have to be located in an optimal location to minimize the cost of installation and maintenance. Groundwater flows are studied using simulators, which makes difficult to standard optimization methods to find satisfactory results, since approximating the gradient is not accurate and computationally expensive. We tackle the problem using the Mesh Adaptive Basin Hopping approach, which combines a heuristic search step with a derivative-free local optimizer. We apply our method to two design problems in the ground-water supply field; the method is able to outperform the state-of-the-Art algorithms, providing better solutions with a tight budget of objective function evaluations.