MPC-Bases energy mangement system for hybrid renewable energies

Abstract

Environmental pollution and the gradual depletion of conventional energy have induced significant research in energy supply systems equipped with renewable and conventional sources. Although wind and solar energy sources emerge as the most promising renewable energies, a supply system comprising two or more sources is recommended to fulfill local loads, as the power generated by renewable sources depends on weather conditions. For this, storage devices are frequently used to store the excess energy generated by the renewable sources and conventional energy sources are used as backup when the production of renewable energies is low and the energy stored is not enough to fulfill the load. The main idea of any energy supply system is to fulfill the energy demand with the minimum cost, considering the operational constraints related to the components. As a result, some issues, such as security of supply, improvement in the combination of energy sources, efficiency, energy saving, improvement in access to isolated systems, and the development of renewable energy, should all be taken into account. Up to now, cost reduction and energy saving have been understood almost exclusively as the technological improvement of renewable sources: wind turbines, solar panels, solar collectors, etc. The misconception of “the best system is made with the best components” is still used for the design of energy supply systems. An optimum system is typically more complex than the sum of the components which integrate the system. Technological advances in renewable sources should be coupled with a sophisticated energy management system. This thesis proposes an energy management system based on control ideas. From a control point of view, the main difficulty is their dynamics, defined according to differential equations and logic rules. Therefore, in this thesis, the design of a hybrid controller based on predictions of energy, estimated from physical models and previous measurements, is considered in order to satisfy the energy supply. Model Predictive Control (MPC) has been chosen as the main control strategy since it is able to handle variations in the supply of renewable energy; while, in the energy demand, MPC includes a cost function to be minimized and adds the constraints on the manipulated and controlled variables. The cost function takes into account the value of the energy generated, the cost of storing energy locally, and the aging of the components. A process model that represents the process dynamic is used to predict the output signal accurately. It is selected to be simple because the future control actions computed by the optimizer take into account the integration of the model along the prediction horizon. Hybrid process models are then considered in the proposed MPC. Although this gives formulation problems, Mixed Logical Dynamic (MLD) involves continuous variables (involved in linear dynamic equations), discrete variables (specified through propositional logic statements), and the mutual interaction between the two. In this case, the resulting mixed integer quadratic programming (MIQP) could present problems for real time implementation, because the solution is computationally complex and depends exponentially on the number of binary manipulated variables. To simplify the MPC problem, the use of binary manipulated variables is avoided by the parameterization of the binary manipulated variables into continuous variables. This transforms the mixed integer optimization into a nonlinear optimization made up of only by continuous manipulated variables (NMPC). To illustrate the applicability and effectiveness of the proposed predictive control, three energy supply systems have been considered: a Solar and Wind based Microgrid for Desalination, a Solar Gas Air Conditioning Plant and a Hydrogen based Microgrid. The study and applicability of these energy supply systems will be carried out throughout the different chapters. The proposed algorithms are demonstrated in detail using simulations of existing systems, and are partially validated in these systems. The implementation has been done as a modular structure to facilitate changes. Additionally, a library on microsources of renewable energies has been developed in EcosimPro©, as it is a powerful modeling and simulation tool that follows an advanced methodology for modeling and dynamic simulation. The renewable energy library facilitates the training of technicians and engineers in how to operate energy supply systems, how to check different configurations and control strategies, before implementing them, and how to facilitate their design.