Optimization with a simulated annealing algorithm of a hybrid system for renewable energy including battery and hydrogen storage

Abstract

Abstract Wind and solar energy based hybrid systems incorporating energy storage can often provide cost effective and reliable energy alternatives to the conventional systems commonly used by remote consumers. To integrate a high level of variable wind and solar energy, energy storage is important. The primary contribution made by the present article is the development of a new efficient methodology for modeling and optimally sizing a hybrid system for renewable energy considering two energy storage devices: hydrogen (as a form of chemical storage) and batteries (as a form of electrochemical storage). To optimize the decision variable values, modified versions of the simulated annealing algorithm-based chaotic search and harmony search are developed. This is dedicated to the optimization of the supply of residential electrical load via stand-alone hybrid energy systems, so as to achieve the minimum life cycle cost of the system by continuous and integer decision variables. The proposed modified approach is used to size optimally the components of six schemes for a remote area in Iran: wind/hydrogen, solar/hydrogen, solar/wind/hydrogen, wind/battery, solar/battery, and solar/wind/battery. To determine the methodology quality, the performance of the proposed hybrid algorithm is contrasted with that for simulated annealing and hybrid harmony search and simulated annealing algorithms. The optimization results demonstrate that a wind and solar energy based hybrid system with electrochemical storage offers more cost effective and reliable energy than a hybrid system for renewable energy with chemical storage. Also, among hybrid systems, the wind/battery system is clearly advantageous economically for supplying power. The portions of life cycle cost of the wind turbine, batteries, and converter/inverter are 67%, 5%, and 28%, respectively. The relative errors between the Mean index of are shown to be at most 11%. Finally, a comparison of the Min., Max., Mean, and Std. values, in the six hybrid systems, shows that the proposed hybrid algorithm is more robust than the others considered since it has lower index values.