Carnegie Mellon University
February 19, 2021

Apt and Lavin publish op-ed in The Washington Post

This article was first published by The Washington Post
---

Temperatures plummeted far below freezing. Water in natural gas froze, cutting off flow from wells. More than 150 electric power generators at 60 power plants in Texas didn’t provide the juice they promised. More than three-quarters of those plants relied on natural gas. Rolling blackouts ensued.

That was Texas — but not this month. It was a decade ago, in February 2011. But still we haven’t absorbed two key lessons. What happened in Texas this week will keep happening until we do.
 
We know extreme cold snaps are not particularly rare in the United States. The January 2014 polar vortex at one point took out nearly 22 percent of the generation in the country’s largest grid, the Mid-Atlantic’s PJM Interconnection. Temperatures in Texas stayed below freezing for three days in January 2018.

In just the past decade, the United States has had five nearly identical cold snaps. The first step toward better reliability is recognizing that this sort of weather is common. Blaming frozen wind turbines, which some are doing in Texas, is very much beside the point. Texas’s grid didn’t consider most of the wind as an emergency resource, and much more conventional generation capacity was frozen.
The second step is to realize that, in hot or cold temperatures, many power plants fail at once.
 
We saw that last summer in California, and we saw it this week in Texas. Power plants cannot be assumed to fail independently of each other when it is really cold outside. The availability of a huge data set showing failures of every generator has let our team of researchers determine that natural gas plants in PJM, for example, are historically five times more likely to fail at 5 degrees Fahrenheit than at 30 degrees Fahrenheit.

But here’s the problem. Although there are three ways that power grids in the United States procure enough generation capacity to compensate for generator failures, none takes these simultaneous failures into account.

The PJM grid managed to keep the lights on in the 2014 polar vortex, but vowed never to come that close to the edge again. It redesigned its market for generation capacity to include penalties if generators or demand response do not show up during events like the vortex. PJM’s market for generation capacity is the first way that grids in the United States attempt to provide incentives for adequate generation resources.
 
Texas has chosen a second mechanism. It allows prices for electricity to be bid up at times of scarcity. Wholesale prices this week reached $9 for a kilowatt hour, far above their typical 3 to 8 cents per kwh. Texas feels that entrepreneurial suppliers will respond to the high prices by building more generation.

A third choice is prevalent in the Southeast. These states have traditional regulated power utilities, and regulators determine a given level of generation to build as reserve. They give the power companies a set rate of return on the investment required to install these reserve generators.

In each of these three methods, the operator has to figure out how much generation capacity should be incentivized. This is routinely done using data on how often generators fail on average. Then, assuming each generator that fails when needed does so independently, they compute the probability that they will have a blackout due to insufficient generation. In general, they procure enough generation to have only one blackout every 10 years due to generator failures.
 
But the math that grid planners use has the same fatal flaw — it overlooks the tendency of plants to go offline in clusters. Calculations must be changed to take into account that either more or better-performing reserve generation is needed in extremely hot and cold temperatures, when electricity demand is highest. This sort of seasonal procurement recognizing peak demand and supply changes is familiar to every airline and supermarket. Why not in power generation?

Texas’s grid failed this week due to frozen gas wells and coal piles, iced-up generator air filters and turbine blades and other specific causes that will take time to uncover. But it is clear that planning should better incorporate how likely generators are to fail in extreme weather. Quantifying that is the first step toward deciding what solutions are worth investing in.

Extreme weather is quite common and is likely to become more frequent in decades to come. We ought to recognize that 2021 isn’t a black swan event. We should use what we’ve learned about why — and when — power plants of all types fail to better prepare. It won’t prevent every weather-related blackout, but it sure will help.