By Shaina Cole | Contributing Writer, Rocky Mountain Voice
Xcel Energy’s Comanche 3 power plant in Pueblo went offline on August 12, 2025, after elevated vibrations damaged the turbine. Nearly a year later, it still hasn’t produced a single watt.
According to Xcel’s most recent monthly report to state regulators in Proceeding 25V-0480E, filed July 22, the plant’s turbine rotor had to be shipped off-site for repair by Mitsubishi. The parts came back this summer. Workers are now in the middle of reassembling the unit. The company’s current estimate for return to service is mid-August 2026.
That’s twelve months offline for Colorado’s newest, largest coal plant, a 750-megawatt facility that cost $1.3 billion to build and runs about $34.5 million a year to operate.
The outage has done more than knock out one plant. A neighboring coal unit that was supposed to close at the end of 2025 is still running as a backup. Xcel is also a co-owner of a third Colorado coal unit, at Craig Station in Moffat County, that the federal Department of Energy has ordered kept available past its own retirement date. And even with Comanche 3 back in service, Xcel projects a 2027 capacity shortfall that would leave the probability of an electricity shortage at nearly 14 times the utility planning standard.
That’s the story Colorado media has mostly told. What hasn’t been told is what a national reliability report, published last month, says about why plants like Comanche 3 keep breaking, and what state regulators put in writing after it did.
What the national data shows
The North American Electric Reliability Corporation, known as NERC, is the federally designated body that monitors the stability of the U.S. power grid. Its job is to track whether generators are available when they need to be.
In its 2026 State of Reliability report, released June 24, NERC found that coal plants failed significantly more often in 2025 than the year before. The forced outage rate, the share of time a plant is unavailable due to a breakdown, rose from 11.2 percent for coal in 2024 to 14.1 percent in 2025. Across the full generation fleet, the outage rate climbed to 9.2 percent, above the historical norm of 7 to 8 percent.
NERC surveyed the owners of generators that had seen the biggest increases in outage time. Of the 26 that responded, 16 reported equipment destruction serious enough to significantly extend their outages. Eight cited supply chain constraints, the same issue Xcel cited for Comanche 3’s delayed repair. Six reported turbine blade release or vibration — the same failure mode that took Comanche 3 offline.
NERC does not name individual plants and explicitly states that its findings describe industry-wide patterns, not specific companies.
“Most large coal units are over 40 years old and are expected to be retired in the relatively near future,” the report states. “These large coal and combined-cycle units were not designed for regular cycling, which has become an operational requirement in some areas and a necessity to maintain economic viability in others. These factors compound to increase the likelihood of failure over time and act to reduce the incentive to prioritize expensive restoration.”
NERC adds that the results of its analysis “align with, but do not prove, industry assertions concerning coal and combined-cycle units.”
In other words, NERC is saying the pattern exists and the logic is plausible, but it will not state that retirement timelines caused any specific plant to be underinvested.
Comanche 3 was built in 2010 and is scheduled to retire no later than January 1, 2031. It is Colorado’s newest coal plant. PUC staff wrote in 2021 that a plant like Comanche 3 should show high and predictable availability once past its shakedown years. It had the lowest availability of any of the company’s coal and gas thermal units from 2010 through October 2020, despite being the youngest. The same 2021 PUC staff report found the plant had racked up more than 700 days of unplanned shutdowns in its first decade — 736 days, against 270 days of planned outages. It has been shut down at least 17 more times since then, for a total of at least another 293 days, according to a Colorado Sun review of notice letters Xcel filed with the PUC and the Colorado Office of the Utility Consumer Advocate.
What Colorado’s own regulators said
When the Colorado Public Utilities Commission voted on December 3 to keep Comanche 2 running, Commissioner Megan Gilman said, “We are in the dark for what any of this costs. We are just in a real reliability pickle because once again Unit 3 has broken in a catastrophic way.”
In Decision C25-0892, issued December 2025, the Commission approved the one-year extension but noted it was granting it with no guarantees attached. The Commission found that Comanche 3’s outage was “the single justification” for keeping the older plant running.
“There is no presumption of prudence at this time for the repair of Unit 3 or operation of Unit 2,” the Commission wrote. “Public Service cannot assume that repairing Unit 3 is a prudent approach if better alternatives are available.”
In a footnote to that passage, the Commission added that Xcel “may be at risk for disallowance if the Commission finds that the Company’s imprudent operation of the plant led to its outage.”
The Commission did not find that Xcel operated the plant imprudently. But it put the possibility in writing.
The cost of one broken plant
Xcel says the failure was not its fault.
A preliminary root cause analysis completed in early 2026 largely attributed the damage to fabrication and design defects by third parties, meaning the turbine components that failed were built wrong before they ever arrived at the plant. Xcel says its own operation of the unit was within acceptable parameters.
The company told state regulators it has no viable alternative to repairing and running Comanche 3. Replacing its capacity with new generation would cost billions and could not arrive before 2029.
As for the repair itself, Xcel says the total cost will fall between $15 million and $26 million, with the company’s share, after insurance and cost-sharing with co-owners, coming to approximately $4.6 million. Xcel told RMV this month that it stressed in its pending rate case settlement that no repair costs are part of customer bills. That settlement is still before the Commission, and Xcel must file a standalone prudence review of the outage by June 1, 2027.
That statement covers repair costs. It does not cover operating costs for Comanche 2, the backup coal plant now running in Comanche 3’s place. Comanche 2 was built in 1975 and was scheduled to shut down permanently at the end of 2025. It is still running. Xcel’s own July regulatory filing in Proceeding 25V-0480E projects the company’s share of keeping it operational through the end of 2026 at between $28 million and $33 million. Xcel now proposes to run it until March 31, 2028, subject to a check-in in late 2027. That extension would cost a further $64.7 million in capital, operating and fuel costs, plus $1.6 million in state air quality fees.
Separately, Xcel has proposed a $77 million net-cost portfolio of additional resources to address a projected 527-megawatt shortfall in the summer of 2027. That shortfall is not a consequence of the outage. Xcel’s own modeling assumes Comanche 3 is back in service by then. The Colorado PUC, in Decision C26-0478-I issued July 17, set a hearing for the week of November 2.
So one plant’s failure, the state’s newest and largest coal unit, has set off a chain of costs projected between $33 million and $38 million — the company’s share of the repair, plus a year of running the backup unit it kept from retiring. What Xcel has proposed since is larger, and rests on a capacity shortfall the outage did not create.
The question Xcel hasn’t answered
RMV asked Xcel this month whether Comanche 3’s scheduled retirement in 2031 had affected maintenance or capital investment decisions in the years before the August 2025 failure.
Xcel did not address the question. The company said it is committed to “balancing reliability with cost effectiveness” but did not speak to whether the plant’s approaching retirement influenced how it was maintained.
NERC, for its part, says it “does not prove” the industry pattern it describes. The Colorado PUC raised the possibility of imprudent operation in a conditional, not a conclusion.
The question is open. The hearing is in November.