Microgrid load forecasting and optimization for future remote island electric vehicle mass penetration: The case of Okinawa Island

Electric vehicles (EV) are becoming increasingly popular, with new models that offers driving customers away from gas-powered vehicles. Given the expected growth of electric vehicles in the coming years, charging a large number of EVs in a short time frame would put a burden on the microgrids of remote islands. In this study, assuming 500,000 electric vehicles on the main island of Okinawa, we used a Probability Density Function (PDF) to predict the charging demand and Start-Of-Charge (SOC) distribution of 500,000 Electric Vehicle (EV) owners in various urban areas of Okinawa. The Monte Carlo method (MC) was used to simulate the electrical load required to charge these electric vehicle. In response to the huge load on the grid produced by the uncontrolled charging of a large number of EVs, this study proposes a demand response-based Genetic Algorithm (GA) to determine the ideal charging time for EVs to bridge the gap between the peaks and valleys of the island's grid. Furthermore, to achieve better optimization results, this study also conducted a comparative study of four types of GA, which showed that the optimization succeeded in reducing the cost of electricity by 45.3% and the variance by 29.8%. Optimized sequential charging of EVs will result in a more stable urban microgrid on Okinawa Island and lower electricity costs.