A steam driven feedwater pump at a 966 megawatt South Carolina reactor tripped itself on overspeed during a routine test, while the linkage that bound had gone 8 years without the maintenance its own procedures called for

A machinery room in a nuclear plant, a pump the size of a small car, a scheduled test on an ordinary Wednesday.

The pump is supposed to spin up, hold its speed and settle.

It spins up, keeps accelerating past where it should stop, and trips itself offline.

The inspection ran from November of 2025 into March of this year, and the final determination was issued at the end of June, as the trade report set out.

Nothing is damaged. Nothing is released. The test simply fails.

What made it fail was a worn linkage, not a design fault.

And the linkage had been waiting a while.

Why a plant keeps a pump that runs on steam

A reactor that has just shut down is still making heat, and that heat has to go into the steam generators or the core gets into trouble.

Motor driven pumps do that job while power is available. The question is what happens when it is not.

So plants keep one emergency feedwater pump that needs no electricity at all. It is driven by the plant’s own steam, and it will run when the switchgear is dark.

That is the machine at the center of this. It exists for the hour when everything else has already failed.

Its speed is held by a governor valve, which meters steam to the turbine through a mechanical linkage of rods, pins and a cam.

Let that linkage bind and the valve stops metering, which sends the turbine past its limit and into the overspeed trip.

What the inspection found in the linkage

The parts came out worn in several places at once, which is what a slow problem looks like.

There was a burr raised along the valve stem. The stem connector had worn shoulders and worn pins.

The cam crank carried gouges where it had been running out of alignment, and an anti rotation slot had worn oval.

None of that happens in a week. It accumulates joint by joint over years of ordinary cycling.

Together the worn parts produced enough binding during valve travel to make the pump inoperable.

Which is the uncomfortable part. The failure showed up on a test day rather than on a bad day.

How the significance was actually calculated

Regulators do not grade a failure by how alarming it sounds. They estimate how much it raised the chance of core damage.

The estimate here came to an increase of about 1.8 in a million per year, over an exposure period of just over 27 days.

That arithmetic is what put the finding in the low significance band rather than a higher one.

The interesting detail is where the risk actually sat. Fire sequences made up roughly 92 percent of it.

The dominant scenarios involved a control building fire together with loss of power to the 7.2 kilovolt buses and evacuation of the control room, with the steam driven pump unavailable.

In other words the pump matters most in exactly the situation where everything electrical is already out of the picture.

The gap that turned a bad test into a finding

The last maintenance appropriate to the linkage was done in May of 2017.

Procedures did call for examining linkage wear during governor valve work, but nothing required an inspection outside that work.

A scheduled task in 2023 was pushed out to 2027, which left the assembly with no examination for eight years.

Nothing in that sequence was hidden. The task existed, it was scheduled, and it was moved.

The inspection ran from November of 2025 into March of this year, and the final determination was issued at the end of June, as the trade report set out.

It is also the second such finding on the same pump inside a year, the earlier one covering preventive maintenance on the overspeed trip device itself.

The full reasoning, including the fire sequences, is laid out in the inspection report, and the pattern in it is the real finding.

What the color actually changes

White is the second of four bands and means low safety significance. It is not a shutdown, a fine or a public hazard.

What it does is move the plant into a column that brings additional inspection until the corrective actions are verified.

The practical requirement is a documented maintenance interval for the linkage, which is the thing that did not exist.

Two findings in one year on the same pump is what draws that column, rather than either one on its own.

Aging equipment is the quiet constraint across the fleet, and it shows up elsewhere as lost megawatts rather than as findings.

It is also the argument the new build case leans on, which is why federal money keeps flowing toward smaller reactors with fewer such assemblies.

None of that helps a 1982 machine with a worn cam, where the fix is a written interval and somebody keeping to it.