Refueling outages are the boring part of running a nuclear plant, and that’s on purpose. You shut the reactor down, pull out a third of the fuel, run a few thousand inspections, fix whatever needs fixing, and then switch everything back on in the right order. It’s that last step, the switching back on, that went wrong at a French plant on the Luxembourg border on Tuesday afternoon.<\/p>\n
At about 5:20 p.m. on October 20, a transformer serving Unit 3 of the Cattenom plant caught fire. Cattenom is EDF’s four-reactor station on the Moselle River in northeastern France, with each unit rated at 1,300 megawatts, and Unit 3 had been off the grid since the night of September 25 for a refueling outage. According to EDF’s statement<\/a> that evening, the fire broke out while crews were in the middle of re-energizing the transformer, which is the plant’s way of saying they were switching it back on.<\/p>\n Then the plant’s own fire system kicked in. EDF says the automatic protection triggered immediately and had the fire stopped before the Moselle fire department arrived. The plant activated its internal emergency plan at 5:50 p.m., the firefighters went over the transformer with a thermal camera, and at 7:56 p.m. they confirmed it was out and drove home.<\/p>\n No one was hurt. EDF says the event had no consequence for the safety of the plant, its staff or the environment, and that it informed France’s nuclear regulator, the ASNR, along with the local authorities. That’s about as clean an incident report as you’ll get from a nuclear operator, and I’d argue it’s still a pretty good story, because of what this transformer does and what put the fire out.<\/p>\n You may be thinking a shut-down reactor doesn’t need much electricity. The opposite is true.<\/p>\n A 1,300-megawatt unit in an outage is still running cooling pumps, ventilation, cranes, instrumentation and a few hundred people’s worth of lighting, and none of that power is coming from its own generator, because the generator is off. So the unit pulls power in from the grid instead of pushing it out.<\/p>\n That’s the job of the transformer that burned. EDF calls it a “transformateur de soutirage,” which translates literally as a draw-off transformer, and in American plant-speak it’s an auxiliary or station service transformer. It sits outside, next to the big main transformer that normally pushes the unit’s output onto the grid, and it feeds the unit’s own equipment.<\/p>\n It also sits well away from the reactor building. EDF was careful to say the fire happened outside the controlled zone, which is the area of a nuclear plant where you’d need to wear a dosimeter.<\/p>\n Why switch it back on near the end of the outage?<\/p>\n Because the unit has to be re-powered in stages before it can make its own electricity again, and the auxiliary transformer is one of the first things that comes back. As far as I can tell, energizing a big oil-filled transformer after it’s sat cold for weeks is one of the moments electrical engineers watch most closely.<\/p>\n I don’t know whether that’s what went wrong here, and officially neither does EDF. The plant told L’essentiel<\/a> on Wednesday that it had no information yet on the cause or on how much of the transformer is damaged.<\/p>\n EDF’s statement on Tuesday night said only that automatic fire protection did the job, which could mean a lot of things. L’essentiel asked, and the answer is my favorite bit of this story.<\/p>\n The plant said it has a water-spray system on certain pieces of equipment, that it reacts very quickly, and that it worked well on Tuesday. The plant’s own response team was first on the scene, ahead of the fire department.<\/p>\n A transformer this size is basically a steel tank full of insulating oil wrapped around copper windings, and that oil burns. So the big ones at power plants sit under a ring of spray nozzles, plumbed to a water main, that open on their own when a detector picks up heat or flame. The idea is to cool the tank and smother the fire in the first minute, before the oil really gets going and before any human has to decide anything.<\/p>\n From the looks of things, that’s exactly how it played out. By the time the fire department rolled up, the spray had done the work, and the firefighters’ job was to make sure it stayed done.<\/p>\n Compare that with the German wind turbine that burned 276 feet up<\/a> on Monday night while 60 firefighters stood on the ground and watched, because nothing they owned could reach it. Here the hardware was built to get wet, and it got wet.<\/p>\n How many firefighters came? EDF didn’t say.<\/p>\n ICI Lorraine, the local Radio France station, reported six fire trucks from the Thionville, Hayange and Sierck-l\u00e8s-Bains stations, confirming an earlier report in Le R\u00e9publicain Lorrain. I couldn’t find a headcount of the crews anywhere, so I won’t guess at one.<\/p>\n The firefighters didn’t just take EDF’s word that the fire was out. They went over the transformer with a thermal camera, which shows you hot spots the eye can’t see, and, according to L’essentiel, they also put up a drone to look at it from above. That’s a pretty sensible move with a tall steel box full of hot oil, since the top is where you’d expect a flare-up and also where you’d least like to send someone up a ladder.<\/p>\n Count the clock and you get the shape of the evening.<\/p>\n About two and a half hours, start to finish, and most of that was checking. If you read our piece on the Louisiana reactor that tripped itself the night before its own refueling outage<\/a>, you’ll recognize the rhythm: the automatic systems act in seconds, and the humans spend the next few hours confirming what they did.<\/p>\n The transformer fire isn’t the first thing to go sideways on Unit 3 since it came off the grid. EDF’s outage note from September 26<\/a> described it as a simple refueling stop: swap out a third of the fuel, run close to 4,500 inspection and maintenance jobs, including valve work and camera inspections of the reactor vessel and steam generators, and come back.<\/p>\n Then on October 7, EDF reported a Level 1 event on the seven-step INES scale, which is the bottom rung of the scale and the kind of thing plants declare fairly often. During tests on September 27, a calculation error left four relief valves on the shutdown cooling circuit set to open a few tenths of a bar higher than they should have.<\/p>\n The teams caught the mistake on September 29 and reset the valves. The valves stayed functional the whole time, EDF says, and the worst case would have been a slightly delayed opening. Under the plant’s operating rules, though, they counted as unavailable for those two days, so Cattenom had to file it with the regulator on October 5.<\/p>\n And now the transformer. L’essentiel reported that Unit 3’s return to the grid could come later than planned, and I’d treat that as the newspaper’s reading rather than an EDF announcement, because EDF hasn’t put out a restart date one way or the other.<\/p>\n A transformer that’s been on fire usually needs its windings and oil inspected before anyone energizes it again. If it needs replacing, a unit that size isn’t something you pick up at the hardware store, and the wait can run into months. That’s a guess on my part, not anything EDF has said.<\/p>\n Cattenom sits a short drive from the Luxembourg border, and the plant has been a sore point on the other side of that border for about as long as it’s existed. On the same Tuesday as the fire, Greenpeace Luxembourg published a study on Unit 1, the oldest reactor on the site, which this month reached the 40 years of operation it was originally designed for. The group says the report finds significant safety shortcomings in the reactor’s original design, including how its safety systems are separated from each other and how they’d hold up in an earthquake, and that upgrades can’t fully close the gap with a plant built today.<\/p>\n Greenpeace’s nuclear campaigner, Roger Spautz, told L’essentiel the fire “shows there are regular problems at Cattenom, almost every month.” EDF, for its part, told Le R\u00e9publicain Lorrain that the fire wasn’t linked to nuclear production and that the transformer sits outside in the open air. I’d say both statements are accurate, and they’re talking past each other. On Friday, October 23, Greenpeace staged a photo op outside the plant to mark Unit 1’s 40th birthday.<\/p>\n For what it’s worth, French reactors don’t have a hard 40-year legal limit. Unit 1 is due to go offline in early 2027 for its fourth ten-year inspection, about six months of work by Greenpeace’s own description of the process, and that inspection is what decides whether the reactor runs for another decade.<\/p>\n Meanwhile the plant is running light. On Saturday, October 24, EDF said<\/a> Unit 2 had been shut down overnight for a short repair to a feedwater component in the turbine building, the non-nuclear side of the plant. With Units 3 and 4 both in their annual outages, that leaves Unit 1, the 40-year-old, as the only one of Cattenom’s four reactors feeding the French grid this weekend.<\/p>\n","protected":false},"excerpt":{"rendered":" Refueling outages are the boring part of running a nuclear plant, and that’s on purpose. You shut the reactor down, … <\/p>\nSo what’s an auxiliary transformer, and why was anyone switching it on?<\/h2>\n
The plant put the fire out before anyone with a hose showed up<\/h2>\n
A drone, a thermal camera and two and a half hours<\/h2>\n
Unit 3’s outage was already having a rough month<\/h2>\n
Greenpeace Luxembourg picked the same day to publish a report<\/h2>\n