Complete design space exploration of isolated hybrid renewable energy system via dynamic programming

Abstract

The electrical demand of isolated regions without connection to the main grid is usually satisfied by a diesel generator. By replacing this conventional system with a hybrid renewable energy system (HRES) including renewable power generators and battery system, the initial cost increases due to additional components. The operating cost, in contrast, can be reduced by minimizing fuel consumption. The economic viability of HRES, therefore, is heavily dependent on both component sizing and energy dispatch strategy. In order to calculate the minimal operating cost, the control trajectory has to be optimized for each design. Within the common publications, the optimal design is not determined based on optimal control trajectory, but using simple rule-based (RB) or advanced RB control. Furthermore, the correlation between component sizing and economic performance is not identified clearly. In this paper, the complete design space for HRES is explored by dynamic programming (DP) based optimal control. In this way, it is guaranteed that the economic performance of each design is deduced from its full potential and the fair comparison among different designs is enabled. The economic performance is evaluated based on four economic parameters including initial cost, operating & maintenance cost, life cycle cost (LCC), and the payback time (PBT). Based on simulation results using DP, the influence of component sizing on each economic parameter is systematically studied. The simulation results show the overall economic performance is determined by the size of renewable power generator and battery storage, whereby diesel generator plays a minor role. Since the LCC and PBT are dominated by initial cost rather than operating cost, small component sizes are preferred minimizing the wasted dump power. Finally, the impact of energy dispatch strategy is investigated by comparing DP based optimal control with simple RB and advanced RB control. Although the loss of optimality using advanced RB control is negligible within 3%, the application of simple RB control suffers from an optimality loss of 7–16%.