Valve parts at 5 nuclear plants in five states were traced to steel bar stock whose own test certificates fell below the strength minimum, and the analysis that followed cleared every one of them

A length of stainless bar sat on a bench, half an inch across, cut from a single melt.

The paperwork that came with it said the metal met every requirement.

The numbers printed on that paperwork did not.

Parts from those orders had shipped to Salem in New Jersey, Brunswick in North Carolina, Sequoyah in Tennessee, Catawba in South Carolina and Palo Verde in Arizona.

By the time anyone noticed, parts machined from that bar were already installed in safety valves at five power plants.

Nobody knew yet whether they would hold.

So how does a sheet of paper become a safety component?

Why the certificate is doing structural work

A valve actuator turns a control signal into physical movement, driving a stem against a seat to shut off a flow path.

In a safety related application that valve may be asked to close at full line pressure during the worst hour the plant was ever designed to survive.

The stem carries every pound of that load.

If it yields or fractures at that moment, the valve cannot do its job, and the stem is the single mechanical link between an operator’s instruction and the physical world inside the containment building.

So designers do not size it by feel. They calculate a margin against a stated minimum strength.

That stated minimum comes off the material certificate.

Which means the certificate is not documentation of the design. It is an input to it, and a wrong number there propagates forward into every part cut from that steel.

What the standard actually requires

The parts in question are plug and stem assemblies, machined from cold finished bar to a common stainless specification.

That standard fixes floors rather than targets: a minimum tensile strength, the force per unit area at which the metal finally breaks, and a minimum yield strength, the lower point at which it deforms without springing back.

It also sets minimums for elongation and for hardness, and it requires mechanical testing to confirm each of them.

The traceable unit underneath all of it is the heat.

One melt of steel gets one number, and every bar, every part and every certificate downstream inherits that number.

Trace a stem back far enough and you land on a single heat and a single test report.

If that report is wrong, everything below it is wrong at the same instant.

What was found and when

The deviation surfaced on February 17 of 2026, when tensile and yield values on vendor certifications came in under the specification minimums.

By March 23 the supplier had confirmed the scope: seven purchase orders in total, from a bar stock supplier in Middleburg Heights, Ohio.

Parts from those orders had shipped to Salem in New Jersey, Brunswick in North Carolina, Sequoyah in Tennessee, Catawba in South Carolina and Palo Verde in Arizona.

On April 21 the valve manufacturer, Fisher Controls, filed an interim notification with federal regulators under the rule that governs potential defects in nuclear components.

Interim is the operative word: at that stage no determination of a defect had been made.

The company was still running stem buckling and structural adequacy analyses, using actuator loads and design conditions specific to each site.

It told the regulator those would be finished by May 17.

How it actually ended

They finished early, and the answer was that the parts were fine.

On May 13 the manufacturer concluded that the supplied material continues to meet or exceed the required design strength in the actual valve and actuator configurations involved, with no adverse effect on form, fit, function, operability or safety related performance.

No defect. No substantial safety hazard.

The reason is the gap between two different numbers: the standard’s floor and the design’s actual demand.

The margin engineered into those assemblies was wide enough that steel a few percent under the specification still cleared the load it would ever see.

Which is a good outcome and a slightly uncomfortable one, because the interim notice had already put five utilities into evaluation.

The metal was never weak. The record was wrong.

What the near miss leaves behind

Certified steel bar stock failing its own paperwork is not an isolated category of problem.

Material certification failures run through the industry’s history from fasteners to pipe fittings to valve assemblies, and the common thread is the same every time: reliance on a document that is only as good as the testing behind it.

In a few cases certificates have been deliberately falsified, which draws criminal referral alongside regulatory action.

The structural fix is unglamorous and expensive: incoming verification for safety related components cannot rest on a supplier’s paperwork alone, and receipt inspection, third party testing or supplier audit each add a layer a single certificate cannot.

That matters more as the supply chain consolidates around fewer large suppliers and single points of failure multiply.

It matters more again for the reactor fleets still on paper, which will inherit whatever supply chain rules get written now.

The final report closed this one in under three months.

The next one will hinge on the same sheet of paper.