Intelligent optimization of a hybrid renewable energy system-powered water desalination unit

Abstract

The use of renewable energies for pumping and desalination of seawater and/or brackish water can be a viable solution to reach a more sustainable freshwater production and reduce environmental impacts. In this way, a new approach based on the design of experiment method for optimal sizing a stand-alone solar–wind-reverse osmosis desalination system is investigated. For this, system modelling is presented with the development of a single sizing parameter between desalination motor pump and reverse osmosis unit and a dynamic simulator of the proposed energy–water system with its energy management loop is developed using climatology year data of southern Tunisia. In optimization loop, the methodology with one-year dynamic simulation necessarily leads to very long convergence times of several days. To try to reduce this time, a track has been proposed which consists in using meta-models instead of dynamic simulators. In order to apply this track, a meta-model (hybrid spline) that represents the system constraints and objectives is investigated based on design of experiments tool and a bi-objective genetic algorithm is applied in the optimization process via the developed meta-model. Two objective functions are used: loss of power supply probability (reliability indicator) which is used to present the dissatisfaction of the water production and embodied energy (environmental indicator) which is used to compute energetic cost (MJ) and to evaluate the environmental impacts potential (across the whole life cycle). Optimal sizing of the system via meta-models instead of dynamic simulator led to encouraging results with significantly reduced CPU times (from several days to 13 min).