Design methodology based on multi-objective optimization with evolutionary algorithms for humidification-dehumidification desalination systems using updated approaches

Abstract

This paper develops a methodology for multi-objective optimization using Non-Dominated Sorting Genetic Algorithm II (NSGA-II). The methodology aims to address the problem of the close relation between variables based on a mathematical model that simulates design variants of humidification-dehumidification (HDH) desalination systems, where the variation of one parameter can affect another. The approach consists of an exhaustive study of the state of the art to incorporate as many results as possible, including variable and parameter constraints established by previous research. Energy efficiency and potable water production are optimized with respect to the total area of the system's active components and the ratio of seawater converted to potable water. The calculations generate 20 Pareto optimal solutions, with an average pure water production of 14.01 l/h, and an average of the total and solar heating areas of 22.43 m2 and 8.52 m2, respectively, including many other parameters with similar values compared to other publications. The methodology is flexible, updatable, and generates high efficiency values, confirmed by the comparison of equivalent parameters and the pinch analysis, also with respect to previous results. The sensitivity analysis performed, presents different trends in the relation between variables in this scenario contributing to feedback on the behavior of hyperparameters of the algorithm, indicating a very wide potential of solutions. For the water temperature entering the dehumidifier, the range between 30 and 37 °C is determined as the optimum range, however, this is not the only result from the trend analysis of these 20 solutions, compared with other references. This methodology is a contribution to the overall objective of improving the HDH technology design process; optimal solutions are obtained for systems of the configuration chosen to apply it, but it can be adapted to others.