Optimal coordination of distributed energy resources in smart grids enabled by distributed optimization and transactive energy

Thesis: S.M., Massachusetts Institute of Technology, Department of Mechanical Engineering, February, 2020Cataloged from the PDF version of thesis. "Some pages in the original document contain text that runs off the edge of the page. See Appendix C - MATLAB Code p.213 - 301"--Disclaimer page.Includes bibliographical references (pages 303-310).Modern active distribution grids are characterized by the increasing penetration of distributed energy resources (DERs). The proper coordination and scheduling of a large numbers of these DERs can only be achieved at the nexus of new technological approaches and policies, primarily distributed computation and transactive energy. Transactive energy is a control mechanism which uses economic incentives, such as time-of-use or real-time electricity prices, to influence the behaviour of independent agents (i.e. DERs in the grid) as needed by the grid operator. This thesis tackles the problem of DER coordination by considering the role of distributed optimization algorithms in solving the optimal power flow (OPF) problem, when a large number of small scale DERs are present. The OPF problem minimizes costs to operate the grid, while subject to network constraints. The distributed implementation makes large-scale problems computationally tractable, while also maintaining privacy of local information. First, we utilize a new convex formulation of the power grid based on current injection (CI) and McCormick Envelopes, to model distribution grids of meshed topology and unbalanced structure. The OPF problem is then solved for such grid structures, using the distributed proximal atomic coordination (PAC) algorithm, which has several advantages over other distributed algorithms. These advantages include reduced network communication requirements, reduced local computational effort, and improved privacy. The DER coordination problem is then extended to consider storage devices. This requires a multi-period OPF formulation, which is also solved using the PAC algorithm. Results from active configurations of the IEEE 123-bus network corroborate the need for a multi-period optimization approach under high penetration of renewable resources. Finally, we propose a retail market mechanism which can be viewed as a transactive energy scheme within the distribution grid. Through the retail market, DERs are optimally scheduled in an energy market, and leveraged in alert system cases in an ancillary services market. The transactions of the energy market are carried out at each primacy feeder through bilateral agreements between the Distribution System Operator (DSO) and agents representing DERs at feeder buses, leveraging the PAC algorithm. These interactions determine the optimal real-time resource schedules and distributed Locational Marginal Price. The PAC algorithm and proposed market are extensively validated on a real distribution grid in Tokyo, a balanced IEEE 123-bus distribution grid, and a three feeder model using the IEEE 13-bus. The energy market is shown to result in an overall increase in revenue for the DSO.by Rabab Haider.S.M.S.M. Massachusetts Institute of Technology, Department of Mechanical Engineerin