The Chance of Blackouts

Kevin Kilty

“What I know about electricity is that when I flip the switch the lights come on.” Some variant of this statement is what I hear from people when a discussion involves electric energy service. People do not understand much about their electric service – generally can’t even fathom their electric service bill. One friend of mine, a scientifically trained person, responded to me when I referred to Kilowatt-hours by saying, “I don’t understand those units at all.”

What is truly misunderstood here is that the lights coming on is only an expectation that has become a certainty in people’s minds because our electrical generation and distribution system, or systems, have become exceptionally reliable. Utilities often strive to reach a reliability often stated as less than 0.1 days of outage per year.[1]  A mere 0.027% of unserved demand.

Numerous places now have planned outages organized to prevent wildfires. This occurs in California, Colorado, and even Wyoming.[2] If analysis by the North American Electrical Reliability Council is to be taken seriously, unplanned outages may soon become common in much of the U.S. Figure 1 is a map, taken from the NERC Long Term Reliability Assessment, published in January 2026. It shows nearly half of the continental U.S. being at high risk of blackouts, beginning  variously from 2028 to 2030 and continuing for five years or thereabouts.

Figure 1. From the North American Electric Reliability Council 2025 Long Term Reliability Assessment.

Reasons for these elevated blackout risks are commonly that demand in each region is rising faster than new supplies can be procured.[3] These are risks related to energy adequacy and are not the only risks that the electric service system faces, but are the easiest to quantify.[4]

I live in the region shown in Figure 1 as WECC Basin. It is effectively a region served by two utilities – Idaho Power Company and PacifiCorp’s subsidiary, Rocky Mountain Power (RMP). The two face very different adequacy risks. Idaho Power serves a territory in which there has been a surprising surge in demand from an unforeseen surge in migration largely from California. It faces future risks as well from people in Oregon who’d feel virtuous if they could get the Federal government to breach dams on the lower snake river – dams which supply some 70% of Idaho’s hydropower.

Rocky Mountain Power’s problems stem from population growth in Utah and concomitant A/C demand in summer, along with increased industrial demand in Wyoming. But unmentioned is its long term commitment to replacing coal and gas with wind and solar.[5]

How does renewable power perform?

In March 2025 our Public Service Commission (PSC) had scheduled a hearing on the latest general rate case of RMP. The requested rates would raise power bills by something like 17% atop rates that had already risen by nearly 21% in the previous rate case in 2023. Even though RMP and all intervening parties had reached a stipulated settlement beforehand, I attended the meeting armed with pertinent data.

When a Rocky Mountain Power official was asked about why they were building the Rock Creek I and II wind power facilities, which were a large part of the reason for this rate increase, her answer was “to guarantee safe and reliable power to our customers.” This pat answer, meant only to reiterate what everyone knows to be the goal of utility regulation, caused me to show data undermining the “reliable” part of the claim.

During February 2025 there were multiple periods in which wind power was almost non-existent.  In particular, during a 15 hour stretch of low wind on February 11 wind power sagged to 46MW during the dark hours with zero solar energy. This represents a capacity factor of 0.8%!

I further showed that the Western Area Power Administration, from which RMP was importing power during these low-wind periods, had no wind power available either. The bulk of the PacifiCorp East Balancing Area (PACE) was being carried by coal plants, many of which, I emphasized, were soon to be shuttered.  I concluded by saying “Someone needs to address this issue, and soon.”[6]

One of the commissioners asked if the situation I just had shown to the meeting occurred also in other times of the year. I knew that it was even worse in summer and that early autumn, October for instance, could be very difficult for generating wind energy.

In preparing for a radio interview on this topic of reliability last autumn, I decided to quantify how often energy from wind and solar vanishes. I took 2025 data from January 1 to September 30 and determined how often wind energy would dip below 10% capacity factor during the dark hours – hours when solar was zero.[7] Table 1 nearby shows the result.

Table 1. Number of periods of less than 10% wind energy capacity factor during dark hours in the first 9 months of 2025.[7]

Month	Jan	Feb	Mar	April	May	June	July	Aug	Sept
Occurences	4	5	5	8	11	13	13	19	17

I was actually surprised to see a total of 95 such periods, and to see how concentrated they were toward high summer. This is exactly the time period when A/C demands are greatest in Utah (and eventually in southwestern Wyoming when planned data centers become functional).

Why use 10% capacity factor?

It is very difficult to get utility spokespersons in public hearings to state clearly what they are thinking. Their applications to the PSC for adjusted rates and permission to build are huge documents and not necessarily clear either. The IRPs, which should inform us clearly of their future intentions, do not fulfill that expectation if they aren’t vetted thoroughly. The IRPs of both Idaho Power and PacifiCorp are not vetted seriously by the PSC, and as a result are just political documents meant to show that they are towing the politically correct lines – abandoning fossil fuels.  

In January 2023 I attended a hearing at the PSC where PacifiCorp was asking for permission to build the Rock Creek I and II wind energy facilities.[8] The discussion was focussed on the need for this facility being that PacifiCorp faced a looming 1,800MW capacity shortfall. This wind plant was said to be needed to close that capacity gap. The discussion persistently referred to the nameplate rating of the plant (590MW) and several times I made public remarks that nameplate rating is the wrong metric for making decisions; that the actual output of the plant could be only 200MW as annual average and at times would surely be zero.

When one commissioner asked the PacifiCorp representatives how much of the 1,800MW capacity gap would they estimate the 590MW plant would cover, PacifiCorp answered as low as 60MW. In other words, PacifiCorp views wind energy without battery backup as being worth

about 10% of its nameplate; a value so close to the Effective Load Carrying Capacity (ELCC) of a wind or solar power system that, in the absence of other guidance from the utility, we should just assume ELCC is what they intend. I should also point out that 10% capacity factor of wind in the dark periods of a day, and considering the mix of wind+solar in PACE, amounts to a capacity factor for renewable energy in total of less than 6%.

There are several important implications to draw from this. First, the stated 60MW useful capacity of this wind plant should have drawn additional questions from the PSC, but did not. The implication of this is that to close an 1,800 MW capacity gap with wind plants like Rock Creek I and II would require 30 of them; costing $27 billion and occupying 4,000 square miles of land. Clearly a bizarre conclusion.

A further implication is that if we view this 10% capacity factor in reliability terms, there are periods on one-third of days, mainly concentrated in summer, where an adequacy shortfall would occur and have to be covered by some other means – lean on your neighbor, batteries or just shut off the power.[9]

Finally, the responsibility of a PSC is not only to ensure safe and reliable power, but to do so at rates just and reasonable. To approve investment of huge sums of money into renewable energy with the knowledge aforehand that the actual utility of the invested capital to customers in 95 periods running from an hour up to 15 hours or more (7% of annual hours) is only 6%, seems imprudent. The additional cost of using batteries to make the utility of this investment behave more like 85-95%, like a thermal plant,  is ruinous.

Planning to Meet Demand

Figure 2. From PacifiCorp’s 2025 IRP.

Figure 2 shows how PacifiCorp intends to address growing demand though the 2025 IRP planning period. This plan encompasses the entirety of PACE and PacifiCorp west (PACW), not just PACE alone.

Consider 2030 which is right in the middle of the reliability shortfall the NERC assessment considers. Total of demand and obligations to deliver power in 2030, according to the 2025 IRP, is 13,000MW (without any plus or minus figure at all). The total of resources dedicated to meeting this amounts to about 27,000MW according to Figure 2. However, note what the mix of resources to cover this consists of. There is still 7,000MW of dispatchable coal and gas. Of the remaining 20,000MW of resource, though 2,000MW is speculative efficiency (that is load that never has to be supplied) plus response management; 18,000MW is wind+solar which I have argued is going to be at 10% capacity factor or less possibly 7% of the year; and, some additional generation is from batteries, which are not generation at all but rather represent an additional load to keep charged and might (probability unknown) be a true resource for some unspecified number of hours.

Just arithmetic suggests we are well below covering obligations and load. There is no reserve.

Conclusion

The looming problems with reliability are not just a matter of acquiring new resources too slowly, as the NERC report implies, but also a more daunting problem with procuring new resources that are difficult to quantify and expensive to make reliable. In my mind, I see lots of wishful thinking. A likelihood of blackouts well above 0.027%.

Notes:

1- As a measure of reliability this leaves much to be desired. 0.1 per year works out numerically equal to 1.0 days per ten years. However the costs of a day of unplanned outage once in ten years are likely to be very different than 2.4 hours once per year depending on location, customer and time of year. What a person would like to quantify is the cost of an outage. This was the topic of a recent episode of Energy Bad Boys, which was the inspiration for my essay, though I am writing here only about the likelihood of blackouts, not their costs.

2-Outage in the service area west of high plains west of Cheyenne, outages in the region of the disastrous fire in Superior and Louisville, Colorado in December 2022, and fires in northern California, etc.

3- Table 1 in the NERC report lists the factors.

4- The April 2025 outage in Spain, for example, occurred despite ample adequacy because of a lack of system mechanical inertia; that is, too much solar energy attached through inverters and not enough contributed by spinning turbines and generators which resist and damp electrical disturbances.

5- The most recent IRP from PacifiCorp has tempered this plan by planning to run most of its existing coal fleet out to the planning horizon in 2045 or so, but there is no mention of replacing thermal assets nearing their 50 year mark and aging out.

6 – Luckily someone did address this issue. The Federal Energy Regulatory Commission (FERC) stepped in, late in 2025, to demand that these coal plants remain operating.

7 – After September 2025 the Rock Creek I and II wind plants were being brought online in an unknown manner, making 2025 October through December aggregate wind energy deliveries possibly not comparable to these nine months.

8 – Known as a Convenience and Necessity hearing. 

9 – Demand response, shown in Figure 2 by year in PacifiCorp’s 2025 IRP, is roughly 1,000MW; a value roughly equal to total system load in the Wyoming portion of PACE. To play upon the quotation beginning this essay, I’d define it as “When the switch is flipped the lights don’t go on.”