Understanding power system resilience requires analyzing how individual outages relate to major weather events and grid disruptions. Utilities record thousands of outages annually, but distinguishing between isolated incidents and widespread weather-driven events has proven challenging. Researchers have now developed a sophisticated algorithm that automatically groups outage records into coherent resilience events based on temporal and spatial clustering patterns.
The methodology represents each outage as a three-dimensional cylinder, with its base positioned at the geographic location and vertical extent representing the outage duration. Two outages are considered part of the same event when their cylinders intersect in both time and space. This geometric approach prevents the common problem of merging geographically distant outages that happen to occur within similar timeframes, which traditional time-only clustering methods often produce.
The grouping algorithm constructs a graph where outages become nodes and overlapping cylinders create edges. Connected components in this graph represent individual resilience events. The researchers propose optimization metrics to calibrate algorithm parameters, minimizing false event groupings while maintaining accuracy.
Validation against NOAA storm records and DOE-417 outage reports demonstrates strong correlation between automatically extracted events and documented major weather phenomena. The method functions effectively with both detailed utility datasets and web-scraped EAGLE-I outage information, providing flexibility for utilities with varying data availability.
This advancement has significant implications for grid planning and resilience assessment. Utilities can now systematically characterize how weather events translate into cascading outages across their service territories. Energy planners gain better tools for identifying vulnerable infrastructure corridors and designing targeted hardening strategies. The approach also enables more accurate resilience metrics and improved emergency response coordination during extreme weather events.
As climate change intensifies weather extremes, automated event extraction becomes increasingly valuable for power system operators seeking to quantify and mitigate weather-related vulnerability.
