On the distribution side of a power system, there exist many distributed energy resources (DERs) that can be potentially used to provide ancillary services to the grid they are connected to. An example is the utilization of power electronics grid interfaces commonly used in distributed generation to provide reactive power support. While the primary function of these power electronics-based systems is to control active power flow, when properly controlled, they can also be used to provide reactive power support. Another example is the utilization of plug-inhybrid vehicles (PHEV) for providing active power for up and down regulation. For instance, such resources could be utilized for energy peak-shaving during peak hours and load-leveling at night. Proper coordination and control of DERs is key for enabling their utilization for ancillary services provision.
One solution to this problem can be achieved through a centralized control strategy where each DER is commanded from a central decision maker located, for example, at the substation that interconnects the distribution network and the transmission/subtransmission network. An alternative is to distribute the decision-making process among the DERs. In order to achieve so, the DERs need to exchange information with a number of other “close-by” DERs, and subsequently making a local decision based on this available information. Collectively, the local control decisions made by the DERs should have the same effect as the centralized control strategy. Such a solution could rely on inexpensive and simple communication protocols, e.g., ZigBee technology, that would provide the required local exchange of information for the distributed control approach to work.
In this talk, we discuss distributed algorithms for decentralized DER coordination that rely on linear iterations, and that adhere to a network communication model between DERs that is described by a directed graph. Given a total amount of resource that the DERs must collectively provide, we address three different problems: i) there are constraints on DER capacity but the objective function is identically zero—constrained fair-splitting dispatch problem; ii) there are no constraints on DER capacity, and there is some quadratic cost associated to each DER— unconstrained optimal dispatch problem; and iii) there are constraints on DER upper and lower capacity, and there is some quadratic cost associated to each DER—capacity-constrained optimal dispatch problem.