Search for: All records

Distribution network safety should not be compromised when distributed energy resources (DERs) provide balancing services to the grid. Often DER coordination is achieved through an aggregator. Thus, it is necessary to develop network-safe coordination schemes between the distribution network operator (i.e., the utility) and the aggregator. In this work, we introduce a framework in which the utility computes and sends a constraint set on the aggregators’ control commands to the DERs. We propose a policy to adjust the charging/discharging power of distributed batteries, which allows them to be incorporated into the framework. Also, we propose a data-driven approach for the utility to construct a constraint set with probabilistic guarantees on network safety. The proposed approach allows the DERs to provide network- safe services without heavy communication requirements or invasion of privacy. Numerical simulations with distributed batteries and thermostatically controlled loads show that the proposed approach achieves the desired level of network safety and outperforms two benchmark algorithms. 

Systems often face constraints at multiple levels. For example, in coordinating a collection of thermostatically controlled loads to provide grid services, the controller must ensure temperature constraints for each load (local constraints) and distribution network constraints (global constraints) are satisfied. In this paper, we leverage invariant sets to ensure safe coordination of systems with both local and global constraints. Specifically, we develop a method for constructing a controlled invariant set for a collection of subsystems, modeled as transition systems, to ensure they indefinitely satisfy the constraints, based on cycles in individual transition systems. Then, we develop a control algorithm that keeps the state inside the maximal controlled invariant set.We apply these algorithms to a demand response problem, specifically, the tracking of a power trajectory (e.g., a frequency regulation signal) by a population of homogeneous air conditioners. The algorithm simultaneously maintains local temperature requirements and aggregate power consumption limits, ensuring the control is nondisruptive to consumers and benign to the distribution network.