Energy Management Systems (EMS) for Microgrids

  Claudio CaƱizares
  University of Waterloo

Microgrids are not new to the power systems community, since these local and small grids have been widely deployed and utilized for electricity supply in remote and isolated communities such as islands and remote villages throughout the world. However, there is nowadays a rapid development and deployment of microgrids in the context of smart and resilient power networks, in good part motivated by the need to integrate distributed generation, especially those powered by renewable resources, to reduce operational costs and the environmental impact of these grids, particularly for diesel-depended isolated microgrids.
In this talk, an overview of the main characteristics and challenges of EMS for microgrids will be discussed. An EMS architecture together with deterministic, probabilistic, and robust methodologies developed at Waterloo and being implemented by an industry partner will be described in some detail. Thus, a two-step architecture based on a model predictive control plus receding horizon approach, which allows to divide the Mixed Integer Nonlinear Programing (MINLP) EMS problem into a simplified MILP unit commitment (UC) problem and a very detailed NLP dispatch problem, will be presented. Novel stochastic programing and a robust optimization implementations of the UC problem, to better deal with uncertainty of particularly wind and solar power sources, will be also discussed. The practical feasibility and benefits of the proposed EMS approaches will be demonstrated and compared based on results obtained from tests carried out on a realistic microgrid.