Load Dependence of Power Outage Statistics

TL;DR

This study reveals that power outage sizes and frequencies depend on grid load levels, with higher loads increasing the likelihood of larger failures and regional differences, supported by data and modeling.

Contribution

It demonstrates load-dependent outage statistics using real data and models, highlighting regional variations and potential for vulnerability assessment.

Findings

Higher load correlates with larger outage events.

Regional differences influence outage size distributions.

Modeling confirms load dependence of outage biases.

Abstract

The size distributions of power outages are shown to depend on the stress, or the proximity of the load of an electrical grid to complete breakdown. Using the data for the U.S. between 2002-2017, we show that the outage statistics are dependent on the usage levels during different hours of the day and months of the year. At higher load, not only are more failures likely, but the distribution of failure sizes shifts, to favor larger events. At a finer spatial scale, different regions within the U.S. can be shown to respond differently in terms of the outage statistics to variations in the usage (load). The response, in turn, corresponds to the respective bias towards larger or smaller failures in those regions. We provide a simple model, using realistic grid topologies, which can nonetheless demonstrate biases as a function of the applied load, as in the data. Given sufficient data of small scale events, the method can be used to identify vulnerable regions in power grids prior to major blackouts.