EAGLE-I Outage Data 2014-2022

EAGLE-I Outage Data 2014-2022

Team: Supriya Chinthavali, Aaron Myers, Sarah Tennille, Nils Stenvig, Matt Lee, Melissa Allen-Dumas

Overview: Restoring power after an outage event is a critical piece of response and recovery for those in the emergency response sector. When analyzing data on power outages, we seek to understand spatial and temporal patterns, restoration times, correlations with other relevant datasets, future outage predictions, expected outage behavior, and how we might fill in data gaps. Analysis at the county level is recommended when possible.

Source: The Environment for the Analysis of Geo Located Energy Information (EAGLE-ITM) is maintained at Oak Ridge National Laboratory (ORNL) for the Department of Energy (DOE). EAGLE-I is DOE’s operational and scalable data and information platform for real-time wide-area situational awareness of the energy sector, providing a centralized platform for monitoring power distribution outage for over 146 million customers; just over 92% coverage of US and Territories.

Figure 1: A screenshot of the EAGLE-I platform with current power outage numbers colored by county, HIFLD transmission lines, and NWS radar overlayed.

Dataset description: The provided dataset includes eight years of power outage information at the county level from 2014 to 2022 at 15-minute intervals collected by the EAGLE-I program at ORNL. This dataset has been collected from utility’s public outage maps using an ETL process. Note that the number of “customers” does not necessarily equate to the number of people affected, as a “customer” reported by a utility could be one meter, one building, etc. Also included is the coverage of each state for each year included in the dataset.

Challenge Questions:

  1. What restoration times and patterns can be derived from the power outage data using detailed time series analysis? We are interested in restoration analysis at the county, state, or FEMA region level.
  2. Using other datasets of interest (examples listed below) what sort of potential correlations between these datasets and power outage information might be discovered and analyzed?
    Weather events: https://mesonet.agron.iastate.edu/request/gis/watchwarn.phtml 
    Social vulnerability index data: https://screeningtool.geoplatform.gov/en/#3/33.47/-97.5
    Medically dependent customers: https://empowerprogram.hhs.gov/empowermap
  3. We are interested in power outage events that are different from “normal” for a given area. How should “normal” be derived for each county, and what metrics should play a part? Consider number of customers without power, power outage length, length of time between outages, escalation of outages, etc.
  4. There exist gaps in the power outage dataset due to technical errors, planned maintenance, etc. What are reasonable ways to fill in these gaps for a more continuous string of outages?
  5. Based on the 8 years worth of historic data, what sort of outages might we predict in the near future? (Next 1-2 years)
  6. How have power outage events in the USA evolved over time? What sort of patterns over space and time can be observed?
  7. Extreme events such as hurricanes cause long-duration restoration efforts and trigger resource sharing (personnel and equipment) with neighbors and other areas of the country. During these time periods, is there evidence of abnormally long restoration times in areas of the country not impacted by the event (ie, does resource sharing cause reduced performance)?

Dataset Download (DOI):

EAGLE-I Umbrella DOI: https://doi.ccs.ornl.gov/ui/doi/436

2014 – 2022 DOI: https://doi.ccs.ornl.gov/ui/doi/435