Extracting resilience events from utility outage data based on overlapping times and locations

TL;DR

Problem: Understanding resilience requires grouping outages into coherent events driven by extreme weather. Approach: automatic event extraction using time, location, and their combination, via overlapping time intervals and proximity modeled as 3D cylinders and a graph of overlapping outages. Contributions: precise definitions for time-based, location-based, and time–location events, plus an efficient graph-based implementation. Validation: applies to Massachusetts detailed outage data and to EAGLE-I county data, with $t_{\max}$ and $d$ thresholds, and cross-checks against NOAA/DOE weather records. Significance: provides a scalable foundation for quantitative resilience analysis and extends county-level analyses to wide-region events.

Abstract

To study resilience with real data, it is necessary to group the individual outages recorded by utilities into events in which the outages bunch up and overlap due to extreme weather. We show how to automatically group utility outage data into resilience events based on their time and location. The methods work with both detailed utility outage data and EAGLE-I data.

Figures (2)

• Figure 1: Example of six outages resulting in different events in time, location, and time and location together. The distance threshold is $d=20$ km.
• Figure 2: Two different examples of events extracted by time and location grouping in A detailed outage data, and B EAGLE-I data.