Improved System Models for Building-Integrated Hybrid Renewable Energy Systems with Advanced Storage: A Combined Experimental and Simulation Approach

The domestic sector will play an important role in the decarbonisation and decentralisation of the energy sector in the future. Installation numbers of building-integrated small-scale energy systems such as photovoltaics (PV), wind turbines and micro-combined heat and power (CHP) have significantly increased. However, the power output of PV and wind turbines is inherently linked to weather conditions; thus, the injected power into the public grid can be highly intermittent. With the increasing share of renewable energy at all voltage levels challenges arise in terms of power stability and quality. To overcome the volatility of such energy sources, storage technologies can be applied to temporarily decouple power generation from power consumption. Two emerging storage technologies which can be applied at residential level are hydrogen systems and vanadium-redox-flow-batteries (VRFB). In addition, the building-integrated energy sources and storage system can be combined to form a hybrid renewable energy system (HRES) to manage the energy flow more efficiently. The main focus of this thesis is to investigate the dynamic performance of two emerging energy storage technologies, a hydrogen loop composed of alkaline electrolyser, gas storage and proton exchange membrane (PEM) fuel cell, and a VRFB. In addition, the application of building-integrated HRES at customer level to increase the self-consumption of the onsite generated electricity and to lower the grid interaction of the building has been analysed. The first part deals with the development of a research test-bed known as the Hybrid Renewable Energy Park (HREP). The HREP is a residential-scale distributed energy system that comprises photovoltaic, wind turbine, CHP, lead acid batteries, PEM fuel cell, alkaline electrolyser and VRFB. In addition, it is equipped with programmable electronic loads to emulate different energy consumption patterns and a charging point for electric vehicles. Because of its modular structure different combinations of energy systems can be investigated and it can be easily extended. A unified communication channel based on the local operating network (LON) has been established to coordinate and control the HREP. Information from the energy systems is gathered with a temporal resolution of one second. Integration issues encountered during the integration process have been addressed. The second part presents an experimental methodology to assess the steady state and dynamic performance of the electrolyser, the fuel cell and the VRFB. Operational constrains such as minimum input/output power or start-up times were extracted from the experiments. The response of the energy systems to single and multiple dynamic events was analysed, too. The results show that there are temporal limits for each energy system, which affect its response to a sudden load change or the ability to follow a load profile. Obstacles arise in terms of temporal delays mainly caused by the distributed communication system and should be considered when operating or simulating a HRES at system level. The third part shows how improved system models of each component can be developed using the findings from the experiments. System models presented in the literature have the shortcoming that operational aspects are not adequately addressed. For example, it is commonly assumed that energy systems at system level can respond to load variations almost instantaneously. Thus, component models were developed in an integrated manner to combine theoretical and operational aspects. A generic model layout was defined containing several subsystems, which enables an easy implementation into an overall simulation model in MATLAB®/Simulink®. Experimental methods were explained to extract the new parameters of the semi-empirical models and discrete operational aspects were modelled using Stateflow®, a graphical tool to formulate statechart diagrams. All system models were validated using measured data from the experimental analysis. The results show a low mean-absolute-percentage-error (