Anyone who’s driven through a Michigan winter knows how a car battery dies. The engine cranks slow one morning, then slower the next, and eventually you’re standing in a parking lot holding jumper cables while one cell inside that plastic box quietly drags down the rest. It turns out the safety batteries inside a nuclear plant can fail the exact same way, and one of them just did.<\/p>\n
On September 17, the Cook Nuclear Plant in Bridgman, Michigan, filed a defect notice with the Nuclear Regulatory Commission<\/a> over cell 50 of 1-BATT-CD, one of Unit 1’s safety-related station batteries. A routine quarterly test had caught the cell below the voltage limit its technical specifications set. Crews tried to charge it back up, it wouldn’t hold, and the cell came out a week later.<\/p>\n C&D Technologies, which built the cell back in February 2016, ran the failure analysis and found an internal short. A lump of paste had pushed through the separator and touched the negative plate next door, and that short kept the cell from holding a stable current while charging. No amount of topping it up was ever going to save it.<\/p>\n If a lump of paste inside a battery sounds odd, it isn’t. A lead-acid cell is basically a stack of metal plates coated in lead-oxide paste, with thin separators keeping the positive and negative plates apart, and all it takes is one blob squeezing past a separator to give you a dead short inside a sealed box. This one sat quiet for more than ten years.<\/p>\n Not for making power. A plant like Cook keeps banks of safety batteries around for the day everything else quits: if the site loses its outside power lines and its emergency diesel generators on top of that, a mess the industry calls a station blackout, the batteries are what keeps instruments and controls alive while crews claw power back. If that sounds theoretical, it’s the scenario that wrecked Fukushima in 2011.<\/p>\n Why does one bad cell matter?<\/p>\n Because a station battery isn’t one big block. It’s a long string of cells wired in series, so the string is only as strong as its weakest link. The filing skips the cell count and the exact voltage numbers, so I’m not going to guess at either, but it does spell out the worst case: the same flaw sitting in several cells at once could cost the battery the capacity it needs, or leave it short of the run time it has to deliver in a blackout.<\/p>\n So how do you catch it? You don’t, not by looking. Per Cook’s filing the flaw stays invisible right up until the short actually happens, which leaves boring old data collection as the only defense, and that’s why the plant now reads the battery’s terminal voltage every week instead of leaning on the quarterly check alone. Honestly, that’s about as unglamorous as nuclear safety gets, and I’d argue it’s exactly the kind of fix you want.<\/p>\n The battery wasn’t the only piece of Cook’s backup-power gear generating paperwork this month. On September 18, Paragon Energy Solutions, a supplier out of Fort Worth, filed its own defect notice with the NRC after Cook reported that four starting-air compressors for the plant’s emergency diesel generators had failed.<\/p>\n Emergency diesels this size don’t start with a key. You spin them up with a blast of compressed air, and the compressors in question are the machines that keep those air tanks topped off. When the compressors can’t make pressure, the diesels are living off whatever air is already stored, which isn’t where anyone wants to be.<\/p>\n Paragon’s teardown traced it to a gasket. The gasket on the second-stage valve plate had broken apart, pieces of it lodged in a valve seat, and the resulting back-pressure lifted the first-stage relief valve and dragged discharge pressure below what the diesels require. Every compressor that failed this way, part numbers H7100X10-MOD and H7100X15, was built in 2022 or later with an updated valve plate design.<\/p>\n The fix is what your mechanic reaches for when a redesigned part keeps failing: go back to the old part. Paragon is shipping Cook replacement compressors built with the original valve plate design from before the update, and it’s telling the plant to watch the units still in service while the root-cause work with Ingersoll Rand wraps up. A new design losing out to the design it replaced is a pretty humbling result, but frankly I’ll take the old boring gasket that holds.<\/p>\n And if you’re wondering why anyone loses sleep over starting air, look at what happened when Europe’s largest nuclear plant lost its last outside power line: seven emergency diesel generators kept six shut-down reactors cool<\/a>, with crews stretching the fuel over 18 days. Diesels that start are the whole ballgame.<\/p>\n None of this touched the reactor’s calendar. Unit 1 shut down at 3 a.m. on Saturday, September 19, for its planned refueling outage, Unit 2 stayed at 100 percent power, and Indiana Michigan Power<\/a> says customers won’t notice a thing. The bad cell was already out well before the outage, swapped a week after that quarterly test flagged it.<\/p>\n Both defect notices went in under Part 21, the federal rule that makes plants and suppliers report flaws in safety-related parts. It’s essentially a recall system for nuclear hardware, and it catches impressively small stuff: earlier this month that trail turned up acrylic paint inside another plant’s breaker<\/a>. This time it’s a paste lump and a crumbled gasket. As far as I can tell, neither defect ever put the plant in danger; two flawed parts got caught by routine testing and written up in public, which is pretty much the system doing its job.<\/p>\n Cook owes the NRC its full written battery report within 30 days of the September 17 notice, Paragon is still finishing the compressor root cause with Ingersoll Rand, and the terminal voltage on 1-BATT-CD now gets read every single week.<\/p>\n","protected":false},"excerpt":{"rendered":" Anyone who’s driven through a Michigan winter knows how a car battery dies. The engine cranks slow one morning, then … <\/p>\nSo what’s this battery actually for?<\/h2>\n
Four starting-air compressors failed the same way<\/h2>\n
The refueling outage started on time anyway<\/h2>\n