Aplicación de métodos heurísticos para optimizar la distribución de agua: un estudio de caso para la cuenca del río Machángara, Ecuador

Abstract

Water can be represented by a nexus called WEF-Nexus that includes water supply, wastewater treatment and hydroelectric power generation in a reservoir water system (Liu et al., 2018). The WEF-Nexus factor considers: avoiding floods, meeting water demands and maintaining the water level in reservoirs and river segments. The optimization of water distribution considering this factor can be approached with linear, non-linear, dynamicdiscrete and heuristic programming. In (Veintimilla-Reyes et al., 2019), the author applied linear programming to optimize the distribution of water in the Machángara river basin, and this study seeks optimization in that same context, but applying heuristic methods. The implemented heuristic model is PSO (particle swarm optimization), which was selected after a systematic literature review. Three phases are considered: calibration, validation and application. The first phase calibrates variables necessary for the model to reproduce reality with data from 1998-2001. It is validated by comparing the output of the parameterized model with the expected values in the 2002-2003 period. Finally, the model is applied to optimally distribute water in the 2004-2005 period. The data used were provided by the Program for Water and Soil Management of the University of Cuenca (Promas, 2022). From the results it stands out that PSO and the Pymoo package (used for the implementation) present difficulties in finding a solution that can satisfy all the constraints. PSO is also used to determine the optimal number of reservoirs, but the model suffers from the same handicap in satisfying constraints. For future work, the implementation of hybrid algorithms is proposed, and reduce the number of restrictions and variables