ODIN Customer Outages by County

Customer power outages by US county from ORNL and DOE's ODIN, captured every ten minutes, with affected meters and the contributing utilities.

Published by ORNL ODIN (opens in a new tab)Cadence: 10-minute snapshotOutages & Grid Events

About this data

Each row is an active outage in one county at one snapshot of the ODIN public county map, with county and state FIPS codes, the number of affected meters, the utilities contributing and the county outline. A county missing from a snapshot had no outage or had cleared. This is distribution and customer outage data, not generation or transmission outage data from an ISO. The history accumulates nationwide and feeds the customer outages layer on the map.

What people use it for

  • See where customers are without power right now
  • Track a storm's outage footprint by county over time
  • Rank counties by meters affected

Where to use it

In ATG Data

Build with the full table

Schema, full history, coverage and query rules live in Data, where you can browse every row, run SQL and connect it to your own code.

Plans and access

Latest data

Loading…

Questions

Which data provider has ODIN Customer Outages by County data?
Ask the Grid has it, from ORNL ODIN. You can browse every row in Ask the Grid Data, query it with SQL or the Ask the Grid API, reach it from an AI assistant over MCP, or ask the grid agent about it in plain English.
Where does ODIN Customer Outages by County come from?
It comes from ORNL ODIN. This page links to the original publication, so any value can be checked against it.
How do I download or query ODIN Customer Outages by County?
Open the odin.active_county_outages table in Ask the Grid Data to browse every row and run SQL against it, call it from code through the Ask the Grid API, or connect an MCP client. Full history can be delivered on a schedule to Snowflake, Databricks or your own storage.
Can I ask questions about ODIN Customer Outages by County in plain English?
Yes. The grid agent reads this data when a question needs it, for example "how many customers are without power in texas right now", and every answer names the dataset it read.