Coordinated Energy Management in a Network of Microgrids

Abstract

This thesis is concerned with the problem of coordinated energy management in a network of grid-connected microgrids. In particular, the research investigates methods for optimal coordinated control of storage units and sharing of electricity costs among the microgrids. New multi-objective optimization models are proposed for efficient integration of the microgrids local energy storage and renewable energy resources into the power grid. In these models, individual microgrids can exchange power locally with each other as well as with the utility grid. Components of the objective function are the electricity costs of the individual microgrids over a rolling window of time, e.g, a 24-hour prediction horizon. A pricing regime is introduced in which differences in the local and grid buy and sell time-of-use rates of electricity incentivize local inter-microgrid exchanges of power over power exchange with the grid. Optimization problems are formulated and solved on a rolling horizon basis to allow for on-line management of energy resources, using up-to-date forecast of microgrid electricity demand, renewable generation, and electricity rates.

In Chapter 4, a novel formulation of a multiple-objective constrained optimization is presented for solving the microgrids energy management problem under the proposed electricity pricing regime using the concept of compromise programming. This approach minimizes l1 or l2 distances of the microgrids cost vector to a utopia point in the solution space. Components of the utopia point are defined as the minimum cost achievable by the corresponding microgrid when it always uses the favourable local buy/sell rates. The proposed optimization models are in the form of convex linear/quadratic programs without any binary or integer variables for l1/ l2 norms, respectively. In Chapter 5, the multiple-objective optimization is converted to a single-objective optimization by adding up the costs of the individual microgrids. An equivalent linear program free of binary/integer variables is derived from the original nonlinear optimization model, which can be effectively solved using existing solvers. The total cost saving and computational complexity are significantly improved in this method compared to the compromise programming technique in Chapter 4. In Chapter 6, the multi-objective optimization is formulated as a lexicographic program, to allow for preferential treatment of groups of microgrids based on pre-assigned priorities. This, for example, allows for giving greater incentives to microgrids that bring lager storage and renewable energy capacity into the network. Finally, in Chapter 7, the optimization model is extended to enable dispatch of reactive power in addition to real power. Simulation results with real data demonstrate that the proposed coordinated energy management strategies can yield substantial cost savings, in some cases in excess of 70 %, for the microgrids in the network compared to a case in which they manage their resources individually. Moreover, the solution to the convex binary/integer free optimization models can be obtained in real-time for a fairly large network, making the proposed models suitable for on-line energy management.