Kevin Kilty
On November 13, 2025 a network outage occurred in central and northeastern Wyoming (the Northeast portion of the Western Intertie) that affected 250,000 customers, including many industrial customers. Some of these were without power for up to 10.5 hours. The North American Electrical Reliability Council (NERC) and Western Electric Coordinating Council (WECC) have released a report on this major event, and a hearing about was scheduled to take place at the Wyoming Public Service Commission on Wednesday July 15, 2026, but has now been postposed until August 7.
I have read the 49 page report, reports that it references, and I have collected additional information, or what I could collect from EIA, in order to be informed to some degree before the meeting, which I plan to attend. The event was complex to say the least.
Timeline of the Disturbance
In what follows I often refer to voltage at a substation, but I intend this to also mean voltage on the electrical “bus” that this substation serves.
I think what can be established about this event is that it involved a long sequence of blunders some of which were latent errors in the system since the inauguration of some parts of the supervisory data network in 2020. There are a few mysterious items in its description that I hope to have cleared up by the end of the hearing in August.
The proximal cause of the outage was a scheduled disconnect of a 500kV transmission line in Central Wyoming, Aeolus near Medicine Bow, to Clover near Mona Utah, which generally carries wind generated power when such is available, or thermal power when needed. This is a new transmission line (Gateway South) which was advertised in its Wyoming PSC application for necessity and public convenience as a way to move wind energy out of Wyoming. Wyoming has far more wind generation at times than it can possibly use. The nature of this planned outage was to replace/decommission some reactors, the series compensation for Aeolus-Clover, at several electrical substations along its path.
The outage was scheduled for 8 am on November 13, but was delayed until 12:43 because needed personnel were not available at a substation in southern Wyoming (Little Snake on the Colorado/Wyoming border) and noon became a new targeted time instead. The Southwest Power Pool (SPP) as it turns out had planned with the Western Area Power Administration (WACM) to shutdown another area transmission line in the area and was informed they could begin this during the delay period of shutting down Aeolus-Clover. As the scheduled time for Aeolus-Clover shut down approached, SPP became concerned about excess grid energy from wind generation. A number of transmission lines in the area were already down since November 12 for reasons of maintenance and SPP became concerned that rising wind power and solar would present problems for grid stability.
Updated plans made at this point (late morning on November 13), involved discussion among a number of reliability coordinating entities, and it was first determined that PACE would need to curtail 820MW of wind energy to avoid excess energy in the bulk electrical system in the area.
While it would seem that adequate planning ahead of time had alerted everyone about what to expect. What actually occurred was a surprise. Schedules were constantly changing and a communications outage was preventing PACE from fully measuring the state of its network in southern Utah. Discussions among all entities resulted in RCW and PACE agreeing to curtail only 200MW of wind generation, rather than 820MW planned earlier. At this point PACE (and SPP too) felt that the system was clear to remove Aeolus-Clover, but RCW had not yet assented to the shutdown – a case of miscommunication.
In a mistake reminiscent of the 2003 blackout in the northeastern US and southern Canada, low voltage alarms in the PACE network had never been enabled since their installation, and PACE didn’t realize their substation at Aeolus was already near a low voltage limit.[1]
Moreover, even though PACE had modeled their network state with Aeolus-Clover shut down, the modeling parameters for various generators were not physically possible. In other words, PACE had a valid solution to a model giving them confidence to proceed shutting down the transmission line, but not a physically viable one.
Upon shutting down Aeolus-Clover at 12:43, there was now deficient power available to parts of the network rather than the worry of SPP that there would be too much power. According to the report, at the moment of shutting down Aeolus-Clover a cascading failure began…
“This cascading was due to a shift of flows onto the Aeolus–Anticline line, which caused the local voltage to collapse to 138 kV (~0.28 pu)…”
I am unsure if the authors of the report are using the word “onto” in the manner I would understand clearly. I would think that phrase would indicate power that had been flowing on Aeolus to Clover now switched to Aeolus to Anticline. Yet, this makes the collapse of voltage difficult to see as flows into the remaining transmission line would cause voltages too high instead.
The Aeolus-Anticline (anticline substation is near Jim Bridger Power Station east of Rock Springs, Wyoming) voltage dropped to 28% of its nominal operating value in 344mSec which opened its circuit breakers. In turn, the substation at Windstar, northwest of Glenrock, Wyoming which was already in a precarious voltage state unknown to network operators, saw its voltage decline to 23% of nominal operation in another 279mSec and it disconnected. By this time thirteen other transmission lines disconnected. Northeastern Wyoming and western South Dakota became an electrical island which then collapsed completely in another six (6) seconds.
By the end of those six seconds the grid lost 1,140MW of load, 4,800MW of generation, 33 bulk electrical system transmission lines, and 250,000 customers.
All four units of the Dave Johnson power station near Glenrock disconnected, and the chaos and guessing about network state that followed for a time allowed power at low voltage to be reconnected to Unit Four of Dave Johnson. Unfortunately this voltage was below the voltage alarm limit that would prevent the generator of Unit Four from entering a motoring mode (generator operating as a motor rather than a generator). This did quite a lot of damage and Unit Four is still offline eight months later.
Restoring power
Restoration of power proceeded slowly because operators had incomplete knowledge of their network state. In such a condition their first step should have been to assess and verify the network state before trying to restore power. Instead there was some guessing which led to mistakes. Re-energizing transmission lines without knowing the full state of the network was a major problem. I think the damage to Dave Johnson unit four was part of this.
A lesson drawn so far from this episode is that while operators had been trained in restoring a completely black grid, they were not trained for this partial blackout. There were also deficiencies in the supervisory data system, including some of SPPs data acquisition and supervisory system dropping out at 12:20 well before the outage. It wasn’t until SPP organized a conference call at 14:16 that the full state of the network was recognized, and restoration of power began to proceed productively.
Puzzling things I hope to learn about at the PSC
There are several statements made in this report that I would like to see validated.
First, I find it odd that while everyone concerned saw the potential problem as too much energy in the networks once Aeolus-Clover was disconnected, it seems the opposite occurred – voltages collapsed at first. Maybe this is explained by simulations ahead of time using incorrect parameters, or that each entity had an unclear picture of the network of others, or that communications prevented both PACE and WAMC from having a complete picture of their networks.
Second is this, while the report focuses on recovery efforts and mistakes made there, there is no real explanation of what caused the cascading failure in the first place. Was there a disturbance propagated across the grid that caused Inverter Based Resources (IBR) to fail a “ride through“ of the disturbance. Failure to ride through had been observed four times in 2024 in the Western area, one of which occurred in PACE. An example of failure to ride through is shown here in Figure 1 from the report of those earlier failures.
Figure 1.
These earlier IBR failures were related to the inverters involved not being set to maximum capability to ride-through. So, one would think that advice having gone out to all resources about this problem would have prevented it from occurring in November 2025.
Instead, PACE claimed that
“… IBRs supported a faster frequency recovery across the interconnection, driven largely by battery energy storage systems, while wind generation provided readily available capacity during system restoration…”
And made this orphaned claim
“PACE identified that the ability to use wind resources helped it restore the system…”
Almost 5,000 MW of generation was lost, and it took many hours to restore power. I am doubtful that there is enough battery storage in this area to have aided much. After all, the region is rich in coal and gas-fired power to backup wind/solar. PACE’s Integrated Resource Plan (IRP) indicates plans to add battery storage, but not that it has been installed as yet.
There is a suggestion in the report that the mix of resources available at the time provided frequency support to the tune of 2,800MW per 0.1Hz which is far greater than the Frequency Support Obligation required. I would like to know what parts of the mix were valuable for this frequency support as the report indicates that the grid was short of inertia and reactive power.[2]
Figure 2. Generation by source in PACE. The red arrow indicates time of the Wyoming event.
As far as wind resources aiding the recovery I refer to Figure 2 nearby, which shows generating data from PACE. It indicates that wind energy rose substantially in the two hours before the outage. I’d like to know the origin of this wind energy, but I’ll simply note that it occurs coincident in time with the RCs and SPP negotiating with PACE about curtailing 820MW of wind. Could it be that estimates of the needed curtailment were complicated by this sudden ramping of wind energy?
Figure 3 is a close-up of a spreadsheet I put together from EIA data from the relevant time period. Note that I have added columns that are labelled with a “Delta”. These show the change in the category of generation from the previous hour. The “Delta Wind” column shows a drop of 816MW in the hour the outage began which is suggestive of the 820MW that SPP advised would be prudent. However, the RCW and PACE were ultimately speaking about only 200MW. Maybe I’ll get an explanation of this.
The sudden loss of 816MW of wind in the final 17 minutes of hour 12pm to 1 pm, suggests a sudden loss of 3,000MW of wind at about 12:43 that persisted through the next hour. Indeed, the 1pm to 2 pm hour shows a loss of an additional 2,200MW, making a total loss of 3,000MW hours at 12:43.
Figure 3. Spreadsheet of EIA data from PACE balancing area. I have added columns labelled “Delta” to show changes from the previous hour.
Then note in the column labelled “Delta Firm PACE”, which combines the two firm power sources in PACE, coal and natural gas, that these firm sources were being scaled back to accommodate rising solar and wind, just like a normal day. Solar energy is generated in PACE largely in southwest Utah, and I might justifiably suspect the following. Was solar power from Southern Utah flowing toward Aeolus on the Gateway South transmission line at the time of the cascading failure? PACE admitted that they were having trouble getting a full picture of network state in Utah, but they assured the RCW that “they didn’t need to shift power around much.”
Something I have never had clarified about Gateway South is this.
When Gateway South was pitched originally it was to aid moving wind energy out of Wyoming, but a letter I sent to the Office of Consumer advocate pointed out that there was two-way metering intended at the Colorado/Wyoming border and this line could convey curtailed solar from Utah, complicating Wyoming hopes for selling wind and coal. Maybe solar power was flowing toward Aeolus at 12:43 and opening that line caused the sudden decline in voltage. Maybe?
Note also that wind was added back into the PACE balancing area in wimpy amounts totaling only a few hundred megawatts by 6pm on the 13th. In contrast the “Delta PACE firm” column shows 2,500MW of firm power was added back over the same period. So, while it is true that “wind power helped restore the grid”, firm power did more. It also helped to cut exporting power to Idaho Power Company by 800MW as Figure 3 also shows. In fact
Finally, wind would likely have been ineffective late in the day on November 13 anyway, as I assume it was declining in intensity toward evening. I have no direct measure of available wind in PACE because of the chaos, but Figure 4 is a plot of Public Service Company of Colorado generating sources, and one will see that wind disappears in eastern Colorado just when efforts to restore power in Wyoming and the Dakotas are in full swing. Wind energy in PACE is pretty highly correlated to wind in Eastern Colorado. PacifiCorp is a very wind/solar supportive utility. Is all the talk of wind and batteries aiding true, or just protecting the reputation of renewables?
I await August 7 to find out.
Notes: 1-The 2003 outage I refer to was the result of a network solver program that was updated early in the day, but then had failed to be restarted, giving operators at a utility in Ohio faulty network state information. When operators restarted the solver nine minutes before this cascading blackout, it reported a network state so at odds with normal experience, the operators at first doubted it.
2-This statement is reminiscent of the Spanish blackout in 2025.
Glossary of Abbreviations to aid anyone trying to read the report.
BES – Bulk Electrical System
BHE – Black Hills Energy
IBR – Inverter Based Resources (wind/solar or battery direct current (DC) resources connected through an inverter that makes AC for connection to the network)
PACE – PacifiCorp East balancing area
PU (pu) – per unit measure. Used instead of actual units because of the variety of voltages used in a network. So, voltage dropping to 0.28pu, means dropping to 28% of nominal operation.
RAS – Remedial Action Scheme
RC – reliability coordinator
RCW – Reliability coordination west
RTCA – Real Time Contingency Analysis
SPP – Southwest Power Pool
TSGT – TriState Generation and Transmission
TOP – Transmission operators
UFLS – Under Frequency Load Shed
WACM – Western Area Power Administration, Colorado Missouri balancing area (Note: Since WACM joined with SPP in April, I am no longer able to fetch EIA data for this entity. Its November data may still exist, but I haven’t found it.)
Looking at figure 2 for Pacificorp East it would appear that on Nov 13, 2025 they managed to get near maximum capacity generation for about 10 minutes with Utility Scale at 1685MW and rooftop at 464MW for a total available of 2149MW. The rest of the days…nnnnnot so much
Rocky Mountain Power has had difficulties with the growth of rooftop solar for a long time now. It makes their grid a bit difficult to control and they are trying to discourage it. They changed some of their avoided power tariff a few years ago which irritated the rooftop solar owners.
Here is something interesting that I am struggling to understand. On typical days before April 2026 we would typically see a maximum of 2,000MW of solar at midday. Since the last week of March our solar seems to have jumped by 1,000MW — suddenly. This is exactly the time period in which we joined the Extended Day Ahead Market (EDAM) with CAISO. I don’t think this power is generated within PACE, but it looks that way.
Moreover, our interchanges with neighboring balancing areas has gone to near zero. Whereas we were always sending a couple hundred MW to Los Angeles Dept. of Water and Power, Sending maybe 1,500MW to Idaho power, etc; since joining EDAM our interchange is only with Public Service of Arizona and doesn’t amount to 200MW.
It is becoming more and more difficult to figure out where power is actually generated, and this is not a good thing for people interested in keeping the utilities honest.
Even the PSCO in Colorado never seems to meet Renewable Capacity. In their chart Wind managed to operate at full capacity at the end of the day on Nov 16th while Solar never reached full capacity on any minute of any day provided.
Xcel, you may know, has told the Colorado PUC that they cannot supply reliable power without Hayden generating station, and that they need to put Comanche unit 3 back into service. Why Comanche 3 is out of service is probably an interesting story in its own right.
Wind seldom reaches capacity because it takes sustained wind over 30 mph which is rare.
I’m surprised that at roughly noon that far north there is that much roof top solar available in November.
How does that compare to a day in June or July?
I am not entire sure what you are referring to. The solar in PACE is mainly solar plants, located in southwestern Utah, and rooftop solar is a minor addition. Rocky Mountain Power encouraged rooftop solar years ago with a feed-in tariff that was too generous, and they’ve been trying rid themselves of it lately.
Among other things, rooftop solar is expensive for the utility to handle and it has a tendency to raise the energy cost of all customers which is something that public utility regulation would seek to avoid.
Now that I re-reviewed the above thread, I thought I saw 2,000 MW of roof top.
However, what I actually read was:
rooftop at 464MW
from Bryan.
Still quite a bit more than I thought there would be from middle of WY, especially in Nov.
His data/guess(?) is from figure 2, which Nov 13 looks to have been the sunniest day of the week.
Seems like a very messy grid design just looking for trouble. They found it big time.
You could almost dream of a system where you only had controllable or dispatchable generators, with even a hierarchy amongst them so they could work together in an orderly fashion to ensure no blackouts.
We can dream…
Dream, perhaps, but the reality is that the system grew in the way it needed to piecemeal over a very long time period. There is no way to bear the expense of going back and building a more rational system at this point, and because of the distribution of large loads, particularly large industrial loads, you’d end up with something similar to the current topology anyway.
Grid design and response was a major issue the entire 35 years (began 1969) I was in the industry.
See my other comment about no one understanding complex algorithms.
V&V Failure Illustrated:
I wonder if they had ever validated their contingency models. I think that is the usual practice.
It is, and one of the NERC recommendations to the utilities is doing so more surely. But as I wrote they also had alarms they were depending on, but hadn’t enabled; then they put power back on certain lines at such a low voltage that it didn’t trigger lockout on a generator. Lot’s of mistakes.
Typo
Should be 3,000MW, without the hours.
Yes, the EIA graphs are in MWhr for the hour just passed, thus they are the average power over that hour of the day. I am so use to looking at the data their way that I forget people may have unit trouble when they first encounter these graphs.
Yes, I see what you are saying. I was thinking the loss of 3,000 MW hours over each hour occuring at 12:43 which is what I think actually happened.
Just a nit, but I noticed the “electric generation by source graphs have “megawatt hours.” not megawatts. That threw me off at first when trying to understand the situation.
Yes, EIA graphs their data this way, and it is what you’d get in the data download. I’m use to looking at it, but others may have trouble at first. If you think about it, though, they don’t even collect instantaneous power, and there isn’t any utility to them doing so until…a blackout. Then people like the FERC or NERC can request such from the utility.
A case of upper/middle management telling their operators
“you don’t understand what we are doing , so be quiet and do what we tell you to do ” ??
Not an uncommon attitude ….
Fascinating and worrisome. I see where questions naturally arise. (Would be nice to have defined PACE within the text instead of having to search to the end, since it’s one of the principals.)
The acronyms proliferate until it is impossible to read these reports. PACE is a balancing area authority, that is it is an area possessing a group responsible for dispatch orders to keep generation and load balanced within that area. PACE is essentially Rocky Mountain Power Company. Other balancing area authorities, like the Southwest Power Pool (SPP), are a combination of many, many utilities.
From the article: “Then note in the column labelled “Delta Firm PACE”, which combines the two firm power sources in PACE, coal and natural gas, that these firm sources were being scaled back to accommodate rising solar and wind, just like a normal day.”
So, what are the economics of this?
How much money do coal and natural gas companies lose when required to cut back to accommodate wind and solar? If they are compensated, then how much more does this cost the consumer?
Windmills and solar are a burden on our power grids. They are the problem, not the solution. They are only viable with taxpayer money.
Before windmills and solar were added to our grids, there were no blackout warnings. After windmills and solar were added, there are blackout warnings every summer and winter.
Windmills and solar are basically a scam on taxpayers and voters.
Accounting is an impenetrable activity for most of us in the first place, but unless I would ever wish to become an intervenor in a rate case I’ll never be privy to the economic details. PACE is absolutely convinced that wind power delivers value for customers. I have real doubts. I actually had a meeting with the President of Rocky Mountain Power and some of his managers in Spring of 2024, and one of those managers told me “if I need additional power, the cheapest power available is that I can just call up wind.” I asked, “how is that possible?” No answer.
Like all large organizations there are people in positions of responsibility who have a hazy view about how the enterprise works. The marginal price of wind or solar might be very cheap, but our rates are set to cover operations of the entire network plus give the company access to capital markets. It’s a hard argument to convince me that the cheap solar for a few hours of the day leads to low rates for the customer. In fact, rates have risen very quickly here in the past three years.
Rocky Mountain Power is very quick to defend their sizable investments in wind/solar, but we all know that PACE could not operate a full day without coal and gas. So, the costs of idling coal plants particularly to make way for wind and solar is just something we have to accept at present.
About a decade ago, I noticed that socialists began raising their ugly heads in Wyoming.
I hope it doesn’t go the way of Colorado which, following the introduction of all mail voting, has been sliding into a political shithole like California.
I lived in Colorado from 1982 until 1992 and it was a fun place in those years, but Roy Romer was not a force for the good and it has gone downhill since. I cannot believe the poor quality of candidates for office. One-party rule is destructive no matter where it takes root; and so, Colorado will follow California just as Oregon and Washington followed California a generation ago.
Where there are people, there are collectivists and utopians. It’s unavoidable. One hopes to keep them in the political minority, but an influx of young people, especially from California, is difficult to manage. I observed Washington go this route while I lived there from 1996 to 2004.
“All four units of the Dave Johnson power station near Glenrock disconnected, ”
Decades ago I toured the station as part of a job interview. All four tripping and lifting safeties simultaneously. Bet they heard that in Casper. Chaos is an understatement.
Recall a cartoon that observed the power plant control room of the future would have two occupants: an operator and a dog.
The operator was there to feed the dog, the dog was there to keep the operator from touching anything.
That’s funny!
Did the November 2025 power outage cause problems in Salt Lake City?
I don’t see that it did reading the report, but there were adjustments throughout the interchanges between PACE and its neighboring balancing areas and so it’s possible there was re-routing of power.
The lead-in graph to the above article is very telling: in just one week in November of electrical energy generation accounting by PSCO:
— electricity generation from wind dropped to below 15% of its peak for several hours on several occasions but was not backed up by output from batteries during these times
— electricity generation from solar dropped to ZERO for multiple hours on several occasions but was not backed up by output from batteries during these times
— natural gas, and less so coal, were the “tried and true”, major sources of uninterrupted electrical energy generation for the full week; with natural gas doing the major amount of “level loading” to compensate for the variations in wind and solar
— just “eyeballing it”, it appears that across the full week, the battery backup for this segment of electrical energy generation—whatever nameplate capacity it is stated to have in terms of MWh—supplied less than 5% of the average daily electrical energy use from all sources.
Some might say “Well, it’s just a start”, but I ask “Is the benefit of battery backup yet proven to be worth the cost?“
Without reading the report my first guess as to the root cause was insufficient synchronous generation (i.e. traditional generators) to maintain voltage. SDG&E’s response to “renewable” sources being added was to install synchronous converters to provide voltage control as well as grid inertia.
Possibly. NERC did indicate that there was a lack of inertia and reactive power, but I interpret their comment as pertaining to the period after about 0.75 seconds into the event until the islanded grid collapse around 5.25 seconds later; and really pertains to the islanded grid.
However, arguing against this is that disconnecting Aeolus to Clover would leave a huge amount of coal/gas generation toward the side that failed. Remember, Aeolus-Clover supposedly carries wind out of state. The failure of Aeolus to Anticline would remove quite a few synchronous machines, but this occurred after the initial transient.
Maybe wind resources behind inverters saw a transient and failed to ride through. But arguing against this possibility is that NERC sent notice to all IBRs about making certain their inverters were set to maximum capability to ride through, and even double checking after software or equipment upgrades.
Possibly Aeolus to Anticline was running so close to the margin (it was running around 0.90pu) that Aeolus-clover disconnect alone set off everything that followed. But arguing against this was that everyone was more concerned with too much energy in the grid north and east of Aeolus after shut down.
There are three episodes I would like clarified. 1) The initial transient, 2) the next roughly 0.75 seconds while 13 transmission lines disconnected and 3) the following roughly 5.25 seconds while the islanded portion of the grid struggled and then collapsed. I hope the August 7 meeting can clarify all this.
Since my original post, I did download and read the report and have some changes to my understanding of what happened.
Most significant root cause was insufficient reduction in wind generation prior to disconnecting the 500kV line. Contributing to this was inadequate real time data on the state of the grid in the forms of raw data (node voltage, frequency, power flows and line status) and grid models. Another is what I refer to as “insufficient paranoia”, with respect to the number of transmission lines already out of service.
The relatively low amount of synchronous generation was still a root cause as there is a need for controllable sources and sinks of reactive power to maintain voltage control. Heavily loaded lines need a source for reactive power to keep voltage up and lightly loaded lines need a sink for reactive power. The local battery systems were quick to respond with real power, but not with reactive power.
One of my hobby horses with respect to wind farms and utility scale solar is to require these facilities need to provide local energy storage to compensate for rapid changes in wind or insolation along with providing reactive power support equivalent to a synchronous generator.
I guess the truth will never come out – all interested parties will want it protected as commercial in confidence.
November last year and still a mystery. Who would believe that?