Sizing and maintenance visits optimization of a hybrid photovoltaic-hydrogen stand-alone facility using evolutionary algorithms

Abstract

This paper tackles the optimization of a stand-alone hybrid photovoltaic-batteries-hydrogen (PV-hydrogen) system, using an evolutionary algorithm. Specifically, a stand alone power system for feeding a remote telecommunications facility is studied. The considered system is specifically designed to cover the power necessities of remote, isolated telecommunications facilities, so it must be able to work in an unattended way during a long time period. On the other hand, if maintenance visits are scheduled, it is intuitive that the cost of the stand alone system could be reduced. Thus, two different optimization problems have been considered in this work. The first one consists in the obtention of the optimal number, distribution (two different arrays of batteries must be fed) and disposition (slope and azimuth) of the PV panels in the facility, for the case of autonomous operation of the telecommunication system during at least two years. The second problem considered consists of scheduling a maintenance visit per year, where a technician is able to reconfigure the system. In this case, the problem consists of obtaining the optimal number, distribution, disposition of the PV panels, and also the time of the year where the maintenance visit should take place. An evolutionary algorithm, able to tackle both problems with very few changes, is described in this paper. The proposed evolutionary algorithm has been analyzed in a simulation of a real PV-hydrogen system sited at National Spanish Institute for Aerospace Technology (INTA), at Torrejón de Ardoz, Madrid, Spain. The well-known software TRNSYS has been used in order to simulate the behavior of this PV-hydrogen system. Several simulations of the system recreating different weather conditions of three Spanish cities (Madrid, Barcelona and La Coruña) have been carried out, and a comparative analysis of the results obtained by the evolutionary algorithm has been done. The results obtained in the first problem tackled showed that in Madrid the system was able to work in an unattended way during 23 months with 6 PV panels, whereas adding one extra panel, the system was able to work in an unattended way for more than 24 months. In the case of Barcelona and La Coruña solutions with 6 PV panels provide 21 and 20 months of unattended work of the system. In the second problem tackled, we have obtained an important reduction in the number of PV panels needed for obtaining an unattended work of the system between two maintenance visits.