Reconfiguration and Real-Time Operation of Networked Microgrids Under Load Uncertainty

TL;DR

The paper tackles reconfiguration and real-time operation of dynamic networked microgrids under load uncertainty by developing a two-stage robust partitioning framework (RPOP) and a model-free real-time OPF (MFRT-OPF). It integrates a novel maximum-phase connection model and a coloring scheme to cohere grid-forming inverter roles with topology, while employing LinDist3Flow to keep the optimization tractable in unbalanced, multi-phase networks. A cutting-plane algorithm solves the two-stage RPOP, and a model-free primal-dual method enables fast real-time responses between reconfigurations. Case studies on a realistic SMART-DS network demonstrate improved solve times, AC-feasibility, and resilient operation under clustered load uncertainty, with MFRT-OPF effectively maintaining voltage limits during contingencies. Overall, the work advances scalable, robust DNMG deployment for resilience in modern distribution systems.

Abstract

Distribution networks are increasingly exposed to threats such as extreme weather, aging infrastructure, and cyber risks--resulting in more frequent contingencies and outages, a trend likely to persist. Microgrids, particularly dynamic networked microgrids (DNMGs), offer a promising solution to mitigate the impacts of such contingencies and enhance resiliency. However, distribution networks present unique challenges due to their unbalanced nature and the inherent uncertainty in both loads and generation. This paper builds upon our prior work on the two-stage mixed-integer robust optimization problem for configuring DNMGs, improving the solve time and scalability. Furthermore, we present a model-free, real-time optimal power flow algorithm to manage DNMG operations in the time between reconfigurations. A case study on a realistic network based on part of the San Francisco Bay Area demonstrates the scalability of both approaches. The case study also illustrates the ability to maintain power flow feasibility as loads vary and operating conditions change when the methods are used in tandem.

Figures (6)

• Figure 1: Dynamic reconfiguration of networked microgrids via switches. Blocks are connected by switches to form connected components (CCs). Each energized CC, which functions as an independent microgrid, contains uncertain loads and DGs, with one DG serving as a voltage source (GFM-DG).
• Figure 2: Diagram illustrating the integration of the faster time scale MFRT-OPF algorithm with the RPOP.
• Figure 3: Illustration of the load blocks and the switches that connect them. Load clusters A-G are indicated by the blue and orange backgrounds grouping blocks together.
• Figure 4: Optimal partitioning of the SMART-DS network under varying uncertain loads. Black blocks indicate de-energized blocks, yellow blocks without outlines indicate the presence of a GFM-DG or substation voltage source, and black lines represent closed switches. Load clusters A-G are indicated by the blue and orange backgrounds grouping blocks together.
• Figure 5: Optimal partitioning of the SMART-DS network assuming $20$% load uncertainty. Black blocks indicate de-energized blocks, yellow blocks without outlines indicate the presence of a GFM-DG or substation voltage source, and black lines represent closed switches. Load clusters A-G are indicated by the blue and orange backgrounds grouping blocks together.