Abstract
Accurate topology estimation is crucial for effectively operating modern distribution networks. Line outages in a distribution network change the network topology by disconnecting some parts of the network from the main grid. In this paper, an outage detection (or topology estimation) algorithm for radial distribution networks is presented. The algorithm utilizes noisy power flow measurements collected from a subset of lines in the network, and statistical information characterizing errors in forecasting load demands. Additionally, a sensor placement scheme is presented. The sensor placement provides critical sensing for the outage detection algorithm so that any number of possible outages in the network can be detected. The performance of the proposed outage detection algorithm using the proposed sensor placement is demonstrated through several numerical results on the IEEE 123-node test feeder.
Introduction
As modern electrical infrastructure integrates more distributed energy resources (DERs) and microgrids, maintaining an accurate map of the network topology—knowing exactly which lines are connected—becomes critical. However, distribution networks, the "last mile" of the grid, have historically remained "dark." Unlike high-voltage transmission lines, which are heavily instrumented, distribution networks are often radial (tree-shaped) and lack extensive sensing.
Traditionally, utilities relied on customer "trouble calls" to identify outages. While automated methods exist, they often struggle with sparse data: pinpointing a failed line with only a few sensors and highly uncertain load forecasts is a significant challenge. This research addresses the dual problem of optimal sensor placement and robust data interpretation to detect outages reliably.
Methods and Network Modeling
The study employs a two-part solution focused on radial distribution networks. These networks are modeled as tree structures where power flows from a root node (the substation) to various leaf nodes.
1. Optimal Sensor Placement
The authors developed a recursive, depth-first search algorithm to determine the minimum number of sensors required to make every possible outage "identifiable." The algorithm traverses the network tree, analyzing expected power flows. If two different outage scenarios produce identical flows at a parent node, the algorithm flags that node for sensor placement.
2. Decoupled Outage Detection
Detecting multiple simultaneous outages is computationally intensive (NP-hard). To overcome this, the authors exploit the grid's radial structure by decoupling the large network into smaller, independent sub-trees rooted at sensor locations. Within these sub-trees, a Maximum Likelihood (ML) detector compares real-time noisy measurements against statistical load forecasts using linearized DistFlow equations.
Key Insights: The Identifiability Problem
A core challenge in outage detection is the "numerically indistinguishable" problem. Some outages cannot be differentiated by sensors alone—for instance, if a node has zero load, losing the line before it looks identical to losing the line after it.
The research formalizes these indistinguishable sets, ensuring the algorithm identifies these groups rather than making arbitrary guesses. Furthermore, the study highlights that power flow measurements are superior to voltage measurements for detection, as outages cause discrete, significant shifts in flow that are easier to distinguish from background noise.
Results and Evaluation
The framework was evaluated using the IEEE 123-node test feeder, a standard complex distribution system model. Key findings include:
- Minimal Sensing Requirements: For the 123-node system, only 20 strategically placed sensors were needed to ensure full identifiability of all detectable outages.
- Enhanced Accuracy with Reactive Power: Including both real and reactive power measurements significantly boosted the probability of correct detection, particularly under high load forecast uncertainty.
- Robustness: The system maintained high detection probability even with a standard deviation of 2.0 in load forecasts (where the smallest load is 20).
- Efficiency: Detection probability saturates quickly with a small number of measurement samples, allowing for rapid response.
Conclusion and Future Work
This research provides a scalable framework for monitoring "dark" distribution networks. While currently optimized for radial topologies with a single power source, future work aims to extend these methods to "mesh" networks and grids with bi-directional flow from solar and battery storage. Integrating these algorithms into active microgrids represents the next step toward a more resilient, self-healing power grid.
Summary updated