Gestión de energía en sistemas con fuentes renovables y almacenamiento de energía basado en hidrógeno mediante control predictivo

Our civilization, dependent on fossil fuels, has a peculiarity, namely a centralized and deterministic character of the energy system. The highly centralized nature means the overall energy control in a small number of institutions, of which depends the welfare of much of humankind. The commitment of many scientists and engineers and many other people and institutions in the world, is by the use of hydrogen as an energy carrier (Raimund Bleischwitz, 2009), resulting in what has been termed as The Hydrogen Economy. Hydrogen is a fuel that does not pollute and can be found potentially everywhere. However, it is rarely found in nature Free State, so it must be produced. The various forms of producing, which although may involve the use of energy from oil, lean towards the use of renewable energies, such as photovoltaic, wind, hydro and geothermal. These energies can generate the electricity consumed to split water into hydrogen and oxygen in the electrolysis process. But the most interesting aspect of hydrogen is a new economy less centralized, more self-sufficient, and linked to the consumer (Suzanne Shaw, 2009). Distributed generation refers to a set of small power plants, located close to the end user, or just in the same location, and it may well be integrated into a network or operate autonomously. Users can be factories, commercial enterprises, public buildings or private residences neighbourhoods. These small power plants have been called "Microgrids". Microgrids have gained great reputation recently, nations are making decisions towards a new electrical system while the old centralized system is being abandoned in favour of a new one, more sophisticated, efficient and environmentally friendly. Therefore, the energy system is facing a major transformation. This document reflects the initiative of contribution to the development of hydrogen technology in the field of hydrogen-based microgrid control systems. The implementation of such complex systems such as microgrids, drove to find out improved control systems for efficiently and safely handle these microgrids. This thesis is about microgrid control system, covering both theoretical and experimental study from simple strategies to the more complex. In addition, this thesis presents a new approach in microgrid control applying the Model Predictive Control (MPC) methodology. In order to reach the mentioned objectives, in this thesis we have started from a detailed study of these systems. The first section, Chapters 1 to 3, review the state of the art of energy storage systems, integrated system types, configurations and existing control strategies. One chapter was devoted to the description of the experimental test bench, including tests performed for the component characterization. For the control system study and design it was necessary to obtain suitable models. Modelling and validation tasks are framed in a second block (Chapter 4 and 7) which develops a complete control oriented model. In this model, most relevant dynamics were implemented, covering the gap in the literature. The models were validated experimentally with good results. Additionally, chapter 7 attends to apply the proposed modelling methodology to a real operational plant: "The Hydrogen Office" located in UK. A plant model was obtained using the same methodology applied to laboratory scale. The models were experimentally validated with data gathered from real operating conditions. The validation results show a very good fit with model predictions. This chapter demonstrated that the modelling methodology, which is the basis of the theoretical and experimental study of the MPC and operation modes, is a valid approach. The last block of the thesis is dedicated to the theoretical and experimental study of operation modes (Chapter 5) and MPC strategies (Chapter 6). The design, simulation and implementation is showed along with experimental results. The plant operation obtained in the laboratory showed a good system performance, depending on the control strategy used, in concordance with the theoretical results obtained formerly. MPC control achieved smooth operation along high efficiency. In contrast, operation modes got different results in terms of efficiency and cost depending on the operation mode used. Finally a benchmark between classical control (heuristic hysteresis band) and MPC is presented. The MPC achieved outstanding results, especially in economic cost, in comparison with non-optimization strategies despite slightly neglecting the system efficiency. This contrast is discussed extensively in the thesis and finally some future work is proposed.