Optimization of concentration control by evolution strategies: Formulation, application, and assessment of remedial solutions

The uniqueness and mathematical complexity of typical groundwater remediation or control problems involving numerical models necessitate appropriate solvers that find optimal solutions reliably and within reasonable computational time. The aim of this paper is to introduce an innovative evolutionary algorithm, the evolution strategies with covariance matrix adaptation and rank mu update (CMA-ES), used as an external solver in combination with groundwater transport models. A broad range of hypothetical pump-and-treat design problems is set up to derive recommendations for a robust CMA-ES configuration. One- and five-well cases are distinguished, for which both extraction rates and well positions have to be optimized in order to partly capture a contaminant plume. Intrinsic natural attenuation of the contaminant that emanates from a continuous source is considered. During this study, it is revealed that when pumping rates are minimized, well positions should be focused close to the source. In contrast, when contaminant mass extraction is minimized, positions distant from the source close to the plume fringes prove to be preferable.