March 20, 2015

IE Questions: How Long Is Your Blackout?

Print More
There's going out, and then there's going out.

Jordan Wirfs-Brock

There's going out, and then there's going out.

In the United States, we assume that when we flip a light switch the lights will go on. Reliable electric power is a given. Except when it’s not.

Take, for example, the time I visited my parents in Oregon in 2008. Buried under 19 inches of snow — actually it was snow, then freezing rain, which formed a frosty creme brulee crust — we spent Christmas eve huddled under blankets, reading books by candlelight and eating cold pie. For more than 24 hours, we were without power.

Major events like this dominate our perception of power outages. In reality though, small outages happen constantly—but because they may affect only a few homes instead of an entire region, we rarely notice them.

So how much time do you spend without power, really?

Until recently, that question was very, very hard to answer. There is no central, standardized database for grid reliability information. Electricity operators are required to report major outages affecting 50,000 customers or more to the Department of Energy, creating a database which Inside Energy compiled. But the DOE data is under and misreported, which makes it problematic for studying long-term trends, according to an analysis by researchers at Lawrence Berkeley National Labs. What’s more, it only addresses the “bulk power system” (power plants and major transmission lines), not the lines that bring power to your house.

But someone out there is monitoring recording data on every single power outage: utility companies.

The bad news is that getting historical information about outages — the type of information that can help us understand if our grid is getting more or less reliable over time — means going directly to utilities and public service commissions. It’s a tedious and monumental task. (A task some have attempted: A study of utility-level data by Lawrence Berkeley National Laboratory suggests grid reliability decreased at a rate of two percent a year from 2000 to 2009.)

But there’s good news, too. As of 2013, utilities now include data about the reliability of their service to the Energy Information Administration, in their annual reports. Data that was released, for the first time, last month. We only have one year of comprehensive reliability data, but it’s a step in the right direction.

What does this grid reliability data tell us? (Or, what are SAIDI and SAIFI and why should you care?)

We know, by utility, how long and how many times customers were without power in 2013. Utilities report two main reliability metrics: SAIDI and SAIFI. SAIDI stands for “system average interruption duration index” and SAIFI stands for “system average interruption frequency index.” Peeling back the alphabet soup and jargon, SAIDI means how many minutes an average customer is without power in a year, and SAIFI means how many outages an average customer will experience in a year. For example, in 2013:

  • Customers of the Public Service Company of Colorado (aka Xcel Energy) experienced, on average, 114 minutes without power (SAIDI) and 1.18 power outages (SAIFI)
  • Customers of Northern States Power (also Xcel) in North Dakota experienced, on average, 74 minutes without power and 0.81 power outages
  • Customers of PacifiCorp (aka Rocky Mountain Power) in Wyoming experienced, on average, 422 minutes without power and 2.06 power outages

Or, to put it another way…

There's going out, and then there's going out.

Jordan Wirfs-Brock

“Going out” means something entirely different if you are talking about a dog and not the grid.

For now, let’s focus on SAIDI. (We’ll come back to SAIFI in a future post. To fully understand how reliable the grid is you need to look at both.)

Inside Energy compiled 2013 SAIDI data by state to come up with an average estimate. We found that state average SAIDI ranged from seven minutes in Vermont to 1,100 minutes (more than 18 hours) in South Dakota. The national average was around 200 minutes in 2013.

Toggle between the two images below to see the U.S. states ordered by SAIDI, and ordered alphabetically (so you can easily find yours):

Get the data: CSV | XLS | Google Sheets | Source and notes: Github

SAIDI has another wrinkle: Some utilities include major event days — like the one my family and I experienced during our arctic Christmas — in SAIDI calculations; others exclude them; many report both.

Here’s what SAIDI state averages look like without major event days (again, ordered by SAIDI and alphabetically):

Get the data: CSV | XLS | Google Sheets | Source and notes: Github

Now, Vermont still leads with five minutes without power per customer per year, but West Virginia trails with 418 minutes. The national SAIDI average, without major events, is about 100 minutes.

States generally have multiple utilities operating within their borders, so SAIDI numbers can vary widely within a state. For example, some customers in Minnesota could have experienced anywhere from one minute without power, if they were Willmar Municipal Utilities customers, or 3,302 minutes without power, if they were Nobles Cooperative Electric customers.

This graph shows the average SAIDI, with major event days, and the SAIDIs for the individual utilities, for six states:


Jordan Wirfs-Brock

This graph shows, on a logarithmic scale, how many minutes customers were without power in 2013. The orange dots show individual utilities and the grey squares show a state average, weighted by the number of customers served by each utility.

Get the data: CSV | XLS | Google Sheets | Source and notes: Github

What doesn’t this data tell us?

The 2013 SAIDI and SAIFI data published by the EIA is a snapshot, not a documentary. Reliability metrics like SAIDI and SAIFI vary as much as 20 percent year-to-year, so we can’t draw many conclusions from a single year. Furthermore, not every utility reports SAIDI and SAIFI in the same way. Inconsistent reporting makes it difficult to make cross-utility comparisons or to analyze broad trends.

Our aging grid infrastructure faces mounting challenges: solar storms, severe weather, cyber attacks, population growth and adapting to renewable technologies. While this is alarming, what’s even more alarming is that we don’t have the data necessary to fully asses the grid’s reliability. It’s critical that we have accurate, comprehensive grid reliability data. How are we going to address a problem we can’t even measure?

What’s next?

Inside Energy will be tackling this topic over the coming months, bringing you stories about the grid, its vulnerabilities, and what people are doing to upgrade it, as well as ongoing analysis of grid-related data.

In the meantime, we’d love to hear from you:

  • What would you like to know about the electric grid?
  • When was the last time you remember the power going out?

Send your thoughts to or Tweet at us @InsideEnergyNow.

Data notes: View the code and the methods behind the analysis; download the crunched data (SAIDI with major event days and SAIDI without major event days by state and utility); get the raw data from the EIA .