Fiche individuelle

Networked microgrids is a cluster of local grids that can be connected through a weak network and can provide ancillary services. On a system point of view, it is desirable that every microgrid exhibits a behaviour at the Point of Common Coupling (PCC) which enables to share active and reactive powers with other grids. This paper proposes a Model Predictive Supervisor that encompasses internal objectives and enforces a virtual droop control at the PCC. The supervisor uses a prediction module based on a model of the microgrid and an outer loop which supplies the PCC powers with respect to the aggregated capability diagram at the PCC. For a weak network, this diagram depends on the PCC voltage that varies with time, generating a set of time varying non-linear constraints. The supervisor generates the references to control the local distributed generators which are inside the microgrid. Through the three steps developed in this paper, the need for model based control is highlighted and a Model Predictive based Controller proposed and validated from simulations.

In this paper a Power Hardware in the loop simulation has been realized to test in a safely way the performances and reliability of a device called “PowerCorner” used to supply an islanded microgrid. A real-time model has been developed in order to simulate the microgrid, batteries and photovoltaic panels. Some modeling criterions have been proposed to reduce time-step simulation and enhancing the Power Hardware in the loop simulation stability. Power Hardware in the loop simulation is used to emulate the AC and DC environments around the power inverters. On the DC side, DC power amplifier is used to emulate photovoltaic power plants and storage devices made on Lithium batteries. On the AC side, AC power amplifier is used to emulate the behavior of the microgrid. These two power amplifiers are controlled by a digital real time simulator which embeds the dynamic behavior of both DC and AC sides.

This paper proposes to optimize the real time operation of a microgrid controlled with a two-layer Model Predictive Controller supervisor. Based on the classical decomposition of control level, the proposed supervisor tracks long-term economic references from a classical economic optimization routine. The optimization problem is formulated as a Mixed Integer Quadratic Problem and uses the different power references as levers to reach this optimum and maintain the state of the microgrid within limitations. In addition, it is able to minimize the grid losses.

Ces travaux portent sur la commande d’un micro-réseau interactif pour fournir des services système à un réseau électrique faible, et plus particulièrement une réponse primaire en fréquence et en tension au point d’interconnection (PCC). Le premier objectif de cette thèse est de superviser un micro-réseau afin d’assurer un fonctionnement stable tout en respectant les objectifs économiques définis par un optimiseur externe.Dans un second temps, une nouvelle méthodologie en trois étapes a été mise au point. Premièrement, pour fournir les services auxiliaires au PCC, il est nécessaire d’estimer et de coordonner les flexibilités des différents équipements tels que les générateurs d’énergie distribués, les énergies renouvelables, les stockages, etc. Un algorithme d’optimisation est proposé pour l’agrégation de ces flexibilités afin de déterminer les flux de puissance active et réactive maximum que le micro-réseau peut fournir. La deuxième étape détermine le comportement possible du micro-réseau à son PCC. Enfin, deux nouveaux algorithmes de contrôle ont été développés pour assurer un comportement de type statisme au PCC.Une première solution, basée sur un superviseur centralisé et la commande prédictive, assure un ajustement en temps réel des points de consigne. La seconde est une solution distribuée qui détermine de nouvelles lois de contrôle locales primaires pour les différents actionneurs. L’efficacité des deux architectures de contrôle a été validée par simulation sur un modèle de micro-réseau de référence. This work focuses on the control of a grid-interactive micro-grid to provide ancillary services to a weak power system, and more particularly a primary frequency and voltage response at the point of common coupling (PCC). The first objective of this thesis is to supervise a micro-grid in order to ensure stable operation while enforcing the economic objectives defined by an external optimizer.Then, a novel three-step methodology has been developed. First, to provide the ancillary services at the PCC, it is necessary to estimate and coordinate the flexibility of heterogeneous equipment such as distributed generators, renewables, storages, etc. An optimization algorithm is proposed for the aggregation of these flexibilities to deduce the maximum active and reactive power flows that the micro-grid can provide. The second step determines the possible behavior of the micro-grid at its PCC. Finally, two new control algorithms have been developed to ensure a droop-like behavior at the PCC. A first solution, based on a centralized Model Predictive Control based supervisor, ensures a real-time adjustment of the set-points. The second one is a distributed solution that determines new primary local control laws for DERs. The effectiveness of the two control architectures has been validated by simulation with a benchmark micro-grid model.