There’s a design choice in most pressurized water reactors that I’ve always found kind of elegant, and it’s the reason this story exists. The control rods, the things you push into the core to stop the chain reaction, are held up by electricity. Electromagnets grip them from above, and as long as the magnets have power, the rods stay out of the core and the reactor runs. Take the power away, or let it sag far enough, and gravity does the rest.<\/p>\n
So when the voltage feeding the control rods at G\u00f6sgen, in northern Switzerland, dropped during a test at 6:40 on Tuesday morning, nobody had to decide anything. The rods fell. The reactor stopped. The plant’s own description of it, from Max Brugger, who runs communications for the operator, is that the rods fell in on their own and shut the reactor down “ohne aktives menschliches Zutun,” which translates as “without any active human involvement.” He said it to the CH Media newspapers that cover the area, and it’s about as plain a description of a fail-safe as you’ll get from a nuclear operator.<\/p>\n
That’s the whole event, more or less. But it’s a 1,010-megawatt reactor that in a normal year covers around 13 percent of Switzerland’s electricity, it’d been back from a 10-month shutdown for about seven months, and the operator still doesn’t know why the voltage dropped. So stick with me for a few minutes.<\/p>\n
Both the operator, Kernkraftwerk G\u00f6sgen-D\u00e4niken AG (KKG for short), and the Swiss nuclear regulator, ENSI, put out short statements the same day, and they agree on the sequence. At 6:40 a.m. on October 20, during what ENSI calls a function test<\/a>, the plant registered undervoltage on the busbars that feed the control rods. The German word is “Steuerstabschienen,” literally control-rod rails, and it means the electrical bus that supplies power to the rod drive mechanisms. Voltage on that bus fell below the limit, and the reactor protection system did what it’s built to do.<\/p>\n KKG says every safety-relevant system responded as designed, there wasn’t any additional release of radioactivity, and the plant went into a safe shutdown state. ENSI’s own radiation monitoring network around the plant, which it calls MADUK, showed no increase. The regulator was told right away, which it’s required to be, and it’s logged the event as a reportable occurrence it will analyze the way it does every automatic shutdown.<\/p>\n A function test, if you’ve never worked around one, is a scheduled check where operators exercise a piece of equipment to prove it still does its job. Plants run them constantly, on diesels, on valves, on instrumentation, and most of the time nothing happens. This time something did. ENSI hasn’t said what was being tested, and neither has KKG, so I can’t tell you whether the test was on the control-rod power supply itself or on something next to it that pulled the voltage down.<\/p>\n KKG still hasn’t explained why the voltage dropped. Brugger told CH Media the cause was being worked out on the day of the shutdown, and nothing I can find since has changed that. Which is a slightly frustrating place to leave a story, but it’s also pretty normal: the analysis of an event like this takes weeks, and the regulator publishes its findings when it has them, not before.<\/p>\n You might expect a reactor that shut itself down to sit idle for days. It didn’t. ENSI cleared the plant to restart the same day, and KKG synchronized the generator to the grid at 5:50 p.m.<\/a>, 11 hours and 10 minutes after the rods dropped. Power would be ramped back up to 100 percent from there, the company said.<\/p>\n That’s quick, and I’d read it as a sign that whatever failed was understood well enough to be isolated or worked around, even if the root cause wasn’t nailed down. A regulator doesn’t let a reactor restart if it thinks the thing that tripped it is still lurking. Still, KKG hasn’t said what it fixed, if anything, before the restart, and that’s one detail I’d like to see in the eventual report.<\/p>\n Compare that with the trip at Waterford 3 in Louisiana four days earlier. That reactor shut itself down the night before a planned refueling outage<\/a>, with a failed bus transfer and an emergency diesel starting on its own, and it was always going to stay down for weeks because the outage was already on the calendar. G\u00f6sgen had no outage waiting, so it just went back to work.<\/p>\n A one-day trip feels bigger than it is at this plant because of what came before it. G\u00f6sgen went off the grid on May 24, 2025, for what was supposed to be a routine annual outage of about a month.<\/p>\n It didn’t come back for 10 months.<\/p>\n The reason was a pipe problem, or more precisely a what-if about a pipe problem. During a modernization project, KKG’s engineers worked out that if a feedwater line broke in the non-nuclear part of the plant, the sudden slam of the check flaps in that system could put more force on the piping and its supports than they’d been designed for. Feedwater, if you’re wondering, is the water that gets pumped into the steam generators to be boiled by the reactor’s heat; it’s on the turbine side of the plant, not in the core. KKG reported the finding to ENSI in March 2025, and after the annual outage wrapped up in June, the regulator wanted a full safety case before the plant ran again.<\/p>\n KKG’s answer was to swap the old undamped check flaps for damped check valves, which close more gently and cut the forces on the pipes. Making and installing the new valves, and writing up the proof that they worked, took until early 2026. ENSI signed off on the restart on March 23, 2026<\/a>, and the plant was feeding the Swiss grid again by 1 p.m. that day. It then took its normal 2026 annual outage from May 30 to late June, about four weeks, and ENSI cleared it to run again on June 26. No fuel needed changing that time, because the core had been reloaded with 44 fresh assemblies before the March restart instead of the usual 36, enough to run through to the 2027 outage.<\/p>\n The financial side of that shutdown is in KKG’s own numbers. The plant produced 3,428 gigawatt-hours in 2025, against 7,994 in 2024<\/a>, which is five percent of Switzerland’s electricity instead of the usual 13. The normalized cost of each kilowatt-hour went from 4.44 Swiss cents to 10.10, a bit more than double, because a plant with the same fixed costs and less than half the output is an expensive plant. Alpiq and Axpo, the big Swiss utilities that own most of it, each warned shareholders of a nine-figure hit in francs back in 2025.<\/p>\n So you can see why a trip seven months after that restart gets attention in Switzerland, even though, as far as anyone has said, the two things have nothing to do with each other. The feedwater valves are on the steam side. The control-rod power supply is on the reactor side. I haven’t seen anyone connect them, and I wouldn’t.<\/p>\n Honest answer: on its own, no. An automatic shutdown is the reactor protection system doing its job, and the fact that the rods drop on low voltage rather than staying put is the entire point of hanging them on magnets. The design assumes the power will fail someday and makes failing safe the default. If anything, the clean sequence here, rods in, safety systems on, no release, regulator told right away, is the boring kind of event you’d want.<\/p>\n What makes it slightly less boring is that G\u00f6sgen doesn’t do this often. The previous automatic scram was on November 12, 2021, which the operator at the time blamed on a limit value wrongly triggered during a swap of a control-system component, and that one was back on the grid the same afternoon too. Nearly five years between unplanned automatic shutdowns is a pretty good run for a reactor that started up in 1979.<\/p>\nBack on the grid by dinner<\/h2>\n
The plant had a rough year before this<\/h2>\n
Is a scram like this a big deal?<\/h2>\n