Contributions to smart grids based on renewable energy sources with hidrogen as backup system. Energy management systemdesign, modeling and physical implementation based on model predictive control theory

Abstract

Attending to the concept of Smart Grid, these systems are closely related to the use of renewable generation systems. Despite the benefits of this technology, its dependence on environmental resources makes it impo ssible to guarantee the balance of energy between generation and demand at all times. Far this, the hybridization of systems, as well as the use of hydrogen-basedsystems, is shown as a viable technical solution to salve or mitigate the probel ms associated with this type of technologies. The use of this type of hybrid systems poses a greater compel xity in terms of managementdue to the high number of parameters and factors to be taken into account in arder to guarantee an optimal energy distribution dependingon the application and the energy status of the system. In this sense, certain aspects associated with the actual operation of the systems, such as the topology, the operating and maintenance costs, the need far a charge voltage control far batterie s, the degradation of equipment, dynamics of each system, the lossesassociated with the working point, or parameters related to the quality of the electricity supply.In the light of the above, it is necessary to use energy management strategies to determine the energy distribution between devices, in arder to optimize the response of the system from a technicaland economic point of view, thereforeposing a multi-objective optimization problem. In arder to respond to the proposed multiobjective optimization problem, in this Thesis, a distributed control architecture is used, composed of local controllers at th e first level, and at the top level, the use of a supervisory controlel r based on predictive control techniques (MPC). The main function of the proposed controlel r is to det ermine the operating setpoint of each of the equipment that makes up the Smart Grid, responding to the proposed objective function accordingto the system design criteria. The advantages of using predictive control techniques over other types of techniques are clear; allows the use of multivariable control techniques, allowing multiobjective optimization in constrained problems; as well as implementing a control strategy based on a prediction horizon, which allows the system to adapt the response of the controller based on future events, improvingthe response of the system against merely passive control techniques. As a knowledge base of the proposed controller, this Thesis presents a general discrete linear model of the plant, calculated in each sampling period, based on a recursive linearization, which allows to increase the quality of the model with respectto solutions based on lathe to a singlepoint of linearization. The model includes all the necessary parameters far the control of a real plant, including the terms associated with the energy status of the system, battery operating voltage, as well as technical and economic parameters, such as degradation, losses or operating cost, with the objective of defining a system cost function that allows its generality far any type of application or design objective. Based on the design of the proposed controller, and with the objective of guaranteeing the generality required throuqhout the orocess. in this Thesis a desian methodoloav basedon the orooosed model and a cost function that includes ali the necessary technical and economic parameters are proposed to solve the proposed multiobjective optimization problem, regardless of the application and system topology. This objective function allows to establish a tracking problem according to the instantaneous power balance of the system, while the technical and economic parameters associated with the system response are considered, see equipment degradation and performance, limits and operating dynamics, operation and maintenance costs, battery charging criteria, etc. To guarantee the generality of the proposed controller, thus promoting its use, regardless of the application and topology of the system, this Thesis proposes a design and tuning methodology of the controller parameters, according to the proposed objective function and the design criteria in terms of priority of use and energy distribution. The methodological proposal is based on the cause-effect relationships between the different parameters, which allow defining the behavior of the system according to the energy management strategy and proposed design objectives. Similarly, in order to consider the short and long-term optimization of the system, limited by the concept of the sliding horizon typical of predictive control techniques, additional control techniques are used, which act directly on the process of adjustment of the parameters of the controller. In this sense, based on the history of the system, the parameters of the controller are recalculated, if necessary, acting directly on the weighting parameters, in such a way that it allows adapting the dynamic response or energy distribution according to the controller design criteria. Finally, the design methodology and the proposed controller were validated on the experimental micro grid of the TEP-192 research group. For this, it was necessary to design, develop and implement ali the control, acquisition and power electronics for the correct operation and integration of the equipment.