Maintaining Reserves and Their Cost

Kevin Kilty

October 18, 2026 a headline in the The Denver Post read “The Feds ordered a coal-burning power plant to remain open. But it has run less than three weeks since April.”

It isn’t often that a headline alone can suggest so many problematic beliefs about energy. The Denver Post managed it, though.

Yes, the Federal Energy Regulatory Commission ordered Craig Unit 1, which was slated to discontinue operations in December 2025 or thereabouts, to remain operating. Yes, costs are piling up. Yes, the plant owner is very annoyed about being so ordered. Yes, the readers of the Denver Post are left with an impression that the Trump Federal Energy Regulatory Commission FERC is staffed by dummies just raising energy costs for consumers to aid “Big Coal” or for other reasons they might imagine. That is the mission of the Denver Post.

Let’s untangle this.

First, by whining about costs associated with this plant The Post is helping reinforce the widespread belief that closing an old power plant and replacing it with renewable resources would save ratepayers money. Money is wasted by keeping an old coal plant running.

The reality here is that renewable energy is only a least cost way of delivering power if a person narrows their focus to just the marginal cost of spinning up a wind turbine, or some small grouping of solar panels. However, the electrical energy grid doesn’t run on the margin, especially when considering contingencies that could leave it short of adequate power. The boundary between adequate power and being short thereof is another way of saying “margin”. A reliable grid should stay well away from its margin. How far back from the margin should one stay?

Traditionally utilities have sought to have something like a 20% reserve available for contingencies. Some, however, seek twice this reserve. Their experience has taught them about operations in their particular part of the grid and considering their particular vagaries of weather that requires a greater reserve. Others use a smaller reserve margin.[1]

There was a time before some people began to learn better that wind/solar generation were counted into system reserves.[2] However, as wind/solar generation has become a larger and larger portion of grid generation, it has become apparent that non-dispatchable sources can’t be counted as reliable reserves. Yet this hasn’t stopped utilities from shutting down coal- and gas-fired power plants over their marginal costs being above those of wind/solar. 

For people who can comprehend graphs, the issue is best told through a sequence of three graphs which I showed our Public Service Commission in March 2025.   Figure one shows a week’s worth of grid operation in the PacifiCorp East Balancing Authority Area (PACE) in January 2025.

Figure 1.

Beside the obvious importance of coal to provide most of demand, and the obvious ramping of coal and gas plants to balance variable solar and wind, note especially the period from midday on January 28 to late day January 30 during this winter week of low wind production. At times wind energy produces less than 300MW, but never rises above 700MW. Translated into capacity factor, wind is producing between 6% and 12% of installed nameplate in this balancing area. This is during a season in which wind is expected according to its seasonal average to exhibit a capacity factor well above 40%.

If the seasonal average capacity factor were to be included into a reserve margin, then this measly production would not only be a disappointment but could lull system operators into a false sense of adequate reserve that doesn’t exist.

Let us continue.

Figure 2, nearby, shows interchange between PACE and three neighboring balancing authority areas during this same week; Idaho Power Company (IPCO), the Western Area Power Authority (WAPA), and Los Angeles Department of Water and Power (LDWP). The transfer of power to IPCO is above 1,000MW throughout the period of low wind. IPCO, in turn, transfers about half of this power to places beyond in the Pacific Northwest. LDWP is able to send some power toward PACE at midday when there is excess solar energy in the Southwest, but draws power from PACE when the “duck curve” appears after sunset.

Note that this typical interchange continues through the period of measly wind power. The curve of interchange with WAPA shows that WAPA supplies the needed power to PACE for these transfers elsewhere. WAPA is not entirely making up the shortfall from wind, but is contributing greatly to it.

The obvious question now arises about where does WAPA derive the power to make this transfer? Figure 3 shows that during the critical period from midday on January 28 to late day January 30, WAPA had no wind power available. Wind and solar are highly correlated over large areas, so when one balancing authority area has little available wind, neighboring areas often lack wind as well. To lean on one’s neighbor often boils down to that neighbor simply leaning on another neighbor – it’s “lean on your neighbor” all the way down.

Luckily, Figure 3 also shows that WAPA possessed adequate coal powered generation to deliver to PACE.

I concluded by pointing out that the specific coal plants that made all of this possible in January 2025 were all slated for closure; some as soon as nine months ahead. I concluded in my summary to the PSC “Someone needs to address this issue, and soon.”

As it turns out FERC did just that and ordered Craig Unit 1 to remain operational for three and now, according to the Denver Post headline, a fourth successive three-month long period.

Figure 2. The 1000+ MW of power provided to IPCO is typical. So is the swing in power flow with LDWP. Power flow of 1000MW and over is higher than normal for WAPA which had enough excess coal, especially, to provide PACE.

Now, though, the usual suspects whine that keeping Craig 1 running is perhaps going to cost ratepayers lots of money, and, perhaps worse, releasing nearly 2 million tons of CO2 when it did run, all for only three weeks of energy generating activity.[3]

Let’s untangle this.

First, maintaining a reserve is expensive. A reserve doesn’t produce revenue earning service, otherwise it doesn’t serve as a reserve. Large markets like the Western Energy Imbalance Market (WEIM) exist to make better use of resources and supposedly reduce the reserves that each balancing authority area must keep. So, keeping Craig 1 open will surely cost ratepayers something, but so would other methods of maintaining a reserve. There is no way to argue around this fact.

Second, in this case the critics are admitting that the plant was run a mere three weeks over half a year, which to them sounds like it was unneeded. In fact, running for only three weeks demonstrates that Craig 1 actually was needed. How valuable was the power that Craig 1 provided during those three weeks? I don’t know and I doubt anyone else knows. [4]

However, the outages that occurred in Texas during the infamous February 2021 “Blue Norther” the Dallas FED estimated as costing 130 Billion dollars mainly in insurance claims. That’s over $4,000 per utility meter which for me is two years worth of electric service bills.

There is no power is more expensive than power that isn’t available when needed most.

Figure 3. Note that during this last week of January WAPA generation is dominated by coal. Wind vanishes during the same period that it does in PACE.

The final important lesson in this illustrates why, when full system costs are included, solar and wind adoption is very expensive, indeed. Instead of the Levelized Cost of Energy (LCOE) metric which has led many people, utility managers included, into believing that solar and wind are the cheapest sources of power. Figure 4 shows this. Robert Idel [5] has proposed an alternative Full System Levelized Cost of Energy. This methodology estimates that full reliance on solar or wind in an environment like Texas, would lead to energy costs of $291 per MWhr if one tried to supply entirely using wind, or  $413 per MWhr if attempted using only solar.  

Figure 4. Note the divergence in cost of delivering so-called renewables versus thermal power when full system costs are considered.

In the case of PACE specifically, I showed, using a hypothetical rebuild of the PACE system using approximately one-third solar and two-thirds wind, with as little battery storage as I thought possible to replace the dispatchable coal and gas powered plants, would cost rate payers very nearly the $372 per MWhr (37.2 cents per kWhr) that Figure 4 implies.

The cost of keeping an old coal-fired plant like Craig 1 in reserve is just one reason among dozens of others why Figure 4 is probably accurate. There aren’t any alternatives to maintaining a geriatric fleet of thermal plants unless one were to consider building new coal/gas plants or nuclear. New plants are far too expensive to use as reserve power (i.e. expensive assets with very low capacity factors) and batteries are even more expensive.

As more coal and gas plants are retired across the country and batteries, if it’s even possible to procure enough, are used to supply dispatchable power, more ratepayers will come to realize the warning that Figure 4 provides. Maybe not at the Denver Post.

Notes:

1- In a number of Integrated Resource Plans (IRPs) that I have read, the reserve margin is stated to three significant figures, which considering the unknowns, is excessive precision. It more likely represents just using some figure that makes resources plus reserves balance against demand.

2- In July 2020, former Texas State Representative Jason Isaac published a widely circulated commentary warning that the Electric Reliability Council of Texas (ERCOT) was relying on “phantom reserve margins” computed using seasonal capacity factors. In the case of Colorado, Xcel energy has made clear to the Colorado PUC that they cannot guarantee reliable power without keeping some old coal plants in operation and putting some others back into service.

3-As I emphasize repeatedly, China releases maybe 12 billion tonnes of CO2 per year which means 2 million tonnes is the equivalent of one and half hours of Chinese activity.

4- Recently people in Wyoming are complaining that coal and gas plants are operating at times at capacity factors of 20-30% without realizing that this is what results from shackling thermal plants to the task of balancing wind and solar. They are unwittingly recognizing the truth of Figure 4.

5– Robert Idel, Levelized Full System Costs of Electricity, Energy, Volume 259, 15 November 2022