Situationally Aware Rolling Horizon Multi-Tier Load Restoration Considering Behind-The-Meter DER
Wenlong Shi, Junyuan Zheng, Zhaoyu Wang
TL;DR
This paper tackles the challenge of restoring power distribution systems when adjacent feeders lack capacity, by introducing a situationally aware, multi-tier restoration framework that enables cascaded load transfers across feeders and transformers. It develops MILP-based models for multi-tier load transfer, transformer/feeder protection, and AMI-driven situational awareness, all embedded in a rolling-horizon optimization with safeguarded segment constraints to prevent myopic decisions. The approach is validated on a modified IEEE 123-node feeder, showing substantial improvements over one-shot MILP and traditional rolling-horizon methods in total restored load and reduction of switching actions, especially under multiple fault scenarios. The work demonstrates that integrating real-time net-load observations and DER reconnection dynamics enables more flexible, resilient restoration with better transformer utilization and balanced feeder operation, offering a practical path toward more resilient distribution networks.
Abstract
Restoration in power distribution systems (PDSs) is well studied, however, most existing research focuses on network partition and microgrid formation, where load transfer is limited to adjacent feeders. This focus is not practical, as when adjacent feeders lack sufficient capacity, utilities may request support from more distant feeders in practice. Such a hirarchical restoration is complex, especially when involving changing system conditions due to cold load pickup and delayed reconnection of behind-the-meter DERs. To fill this research gap, a situationally aware multi-tier load restoration framework is proposed. Specifically, models are proposed to describe the multi-tier load restoration, including the multi-tier load transfer and substation transformer and feeder protection models. By introducing binary actional switching variables and load block transfer variables, the models effectively captures the dynamics of switches and multi-tier transfer process. To integrate situational awareness of evolving system conditions, the problem is formulated as a mixed-integer linear program (MILP) and then embedded within a rolling horizon optimization. Particularly, a set of safeguarded constraints are developed based on segment-level restoration reward bounds to mitigate the myopia of traditional rolling horizon optimization. The proposed safeguarded rolling strategy guarantees that each time step is lower bounded by a $(1-\varepsilon)$-fraction of its optimal restoration potential, thereby balancing short-term switching decisions with long-term restoration goals. Finally, cases studies on the modified IEEE 123-node test feeder validate the proposed multi-tier restoration framework.