System design and operation methodologies for hybrid renewable energy systems

Abstract

The energy system in the future will gradually transit from fossil fuel based system to the renewable energy based system. Moreover, the state-of-art centralized energy supply will be replaced by the distributed energy supply. During this transformation of conventional energy system, the hybrid renewable energy system (HRES) will play a crucial role enabling a smooth transition. HRES consists of renewable power generators and energy storage systems supplemented by conventional energy supply such as diesel generator or electrical grid. Although the installation of HRES is linked with high investment cost due to additional components, it has the potential to minimize the operating cost and emissions of detrimental gases. In order to realize this potential, the system design and operation strategies have to be optimized. In this dissertation, three aspects related to design and control of HRES are investigated. First, the correlation between operation strategy and component sizing is studied. For this purpose, wide range of design space is constructed and each design’s control is optimized by using dynamic programming (DP) based optimal control. In this way, the resulting economic performance is deduced from the full potential of given system design. The simulation results are used to study the link between component sizes and selected economic parameters, to deliver a clear guideline for system design. Moreover, the same design space is operated with simpler control algorithms to quantify the loss of optimality versus DP. It turns out that the genetic algorithm based rule-based control delivers near-optimum solutions with lower computation cost. Next, the stochastic DP based optimal controller is developed for real-time operation in the presence of environmental uncertainties. Sine the environmental data such as weather or loads cannot be perfectly predicted, a stochastic approach is applied to minimize the expected operating cost. Thereby, the stochastic DP based controller is combined with predictive algorithm. Since the prediction error can vary from site to site, the robustness of proposed optimizer is tested by varying the accuracy of prediction models. The simulation results show, the stochastic DP based predictive controller can handle both the environmental uncertainties and prediction errors by guaranteeing near-optimal operating cost. Lastly, a comprehensive design framework for second-life battery pack is proposed. Since the high investment cost acts as the barrier for HRES installation, the application of second-life battery pack in HRES could lower the initial anxiety for investment. The battery cells after first usage have diverging performance characteristics and this can lead to severe problems of safety and power reliability. The proposed screening algorithm, therefore, selects the battery cells with homogeneous properties customized to the system requirements. This framework includes cell modeling, cell testing, parameter identification, and cell screening algorithm. The case study shows the developed framework can reduce the maximum cell-to-cell variation by up to 70%.