Safe Coordination of a Collection of Switched Systems, Applied to Coordination of Distributed Energy Resources

Abstract

As global awareness of climate change increases, renewable energy sources are being rapidly integrated into power grids. However, this integration introduces significant uncertainty and variability in electricity supply due to intermittent wind and solar power generation. Consequently, we cannot solely rely on the conventional balancing methods leveraging traditional generators. Distributed energy resources (DERs) are emerging as a promising solution to manage this variability as they can collectively provide significant flexibility and enhance grid reliability with diverse grid services.

While DERs are providing grid services, certain operational constraints must always be maintained. First, the end-users of the DERs should not experience any disruption from DERs' involvement in grid services. Also, DERs must not cause any issues within the distribution network while providing these services. One challenge is that third-party aggregators, responsible for coordinating DERs in the U.S., lack access to private network and DER information, complicating their ability to evaluate the impacts of the DERs' actions on the network. This lack of information makes it difficult for aggregators to coordinate DERs while ensuring that all operational constraints on individual DERs and the network are satisfied.

This dissertation addresses these challenges by developing control algorithms for third-party aggregators that ensure constraint satisfaction. We adopt a specific framework wherein the Distributed System Operator (DSO) sends constraints on either the aggregator's action or network-level behavior of DERs for the safe operation of the distribution network. Within this framework, we devise algorithms that provide formal guarantees of constraint satisfaction, utilizing tools from control engineering and formal methods. While our approaches are not limited to specific types of DERs, this dissertation primarily focuses on algorithms for a collection of thermostatically controlled loads (TCLs).

First, we develop a control algorithm for a collection of TCLs providing balancing services, with formal guarantees on the satisfaction of both individual TCL constraints and network-level constraints. We present a scalable method that derives an implicit representation of a controlled invariant set for a collection of switched subsystems that model TCLs. Subsequently, we propose a model predictive control (MPC) algorithm with formal guarantees on safety and recursive feasibility by incorporating the obtained implicit representation. Second, we provide another MPC algorithm that is applicable to a collection of TCLs with unknown temperature dynamics. This algorithm uses a data-driven approach to identify modes of the TCLs in which temperature constraints are not violated. Third, we develop a coordination framework that enables distribution network-safe control of a large collection of TCLs and batteries. Under the framework in which the utility sends a constraint set on the controller's input, we develop a sampling-based method to construct bounds on the input. We show that abiding by these bounds guarantees the satisfaction of the chance constraint on network safety. Lastly, we perform a comparative analysis between our proposed approach and those utilizing nodal operating envelopes for network-safe coordination. Our findings indicate that our input constraint approach achieves a good balance between flexibility and fairness.

Case studies for all the developed approaches demonstrate that the proposed control algorithms and framework ensure the satisfaction of all operational constraints. Overall, the proposed control algorithms and frameworks enable third-party aggregators to control DERs while ensuring the safe operation of the distribution network without disturbing the end-users.