Titre : Microgrid control with a high penetration of renewable generation
Date : jeudi le 19 mai 2016
Heure : 14:00
Lieu : Université McGill, Pavillon McConnell (génie)
Salle : 603
Directeurs : Geza Jos et Francois Bouffard
As the penetration of renewable energy-based distributed generation becomes more prevalent on the electric power system, its limited controllability and intermittent nature brings adverse effects to the system. A multi-objective optimization is proposed for the Microgrid’s energy management system as an appropriate means to mitigate the volatility of these resources while optimizing other benefits of implementing a Microgrid, instead of the status quo of a single objective optimization that exacerbates the other objectives. The proposed approach attains a Pareto optimal solution by directly comparing the quantified objectives and directly solving the economic dispatch through its scalarized valuation functions.
Furthermore, the implementation of the Microgrid controller is formulated as a hierarchical multi-agent system to reduce computational burden and to provide distributed intelligence. The central controller algorithm directly computes optimal regions of the unit commitment through virtual resources, while the economic dispatch is performed locally with modified valuation function parameters. In order to comply with the online implementation architecture, the energy storage system’s controller employs a backcasting algorithm to estimate the net stored energy value, against which the current cost of energy is compared to determine how the storage system should be used to perform arbitrage. In islanded mode, the storage system employs a novel primary isochronous control strategy to maintain a desired state of charge while maintaining power balance by requesting power from other resources through their droop curves when the storage limits are being approached. Results show that the proposed controller can mitigate the negative impacts of volatile generation to levels below that of the system load.