Decision support system based on genetic algorithms to optimize the daily management of water abstraction from multiple groundwater supply sources

Abstract

The use of irrigation water extracted from aquifers with submerged pumps is essential to ensure agricultural production mainly in water-scarce regions.. However, the use of the water source requires of a considerable energy consumption by water user associations (WUAs) being key factor to consider due to their high share of total management, operation, and maintenance costs. In this work, a new tool (MOPWE, model to optimize water extraction) to optimize the water and energy use of wells in WUAs was developed. MOPWE was applied to a real WUA located in Castilla-La Mancha region (southeast of Spain). This WUA utilizes groundwater as water source that is extracted from several different wells of different characteristics (discharges, water table levels, efficiency, variable speed drives…).. Therefore, these kind of WUAs must decide not only which well to activate at a certain time but also at what frequency the variable-speed drive should run the pump. With the aim of aiding decision-making in groundwater abstraction, a new management model (MOPWE), which is based on multi-objective genetic algorithms and is implemented in MATLAB®. This model helps determine the optimal daily management of a WUA with multiple underground supply sources and focuses on the management of wells while considering the water reservoir level. After 18,000 generations of the genetic algorithm, the pareto front was obtained with the best WUA managements achieving a water and energy savings of 25% and 54%, respectively. At the end of the irrigation season, the optimal total energy consumption per unit of water applied was 38% lower than that achieved by the current management. Results showed that a more realistic approach can be implemented when several water supplies operate jointly under a collaborative principle.