ORNL's Three-Armed MedUSA Printer Just 3D-Printed a Nuclear Pressure Vessel

Why Nuclear Hardware Needs a New Manufacturing Route

The United States wants to build more reactors. That plan hits a bottleneck: the country has very few forges large enough to make the pressure vessels those reactors need. These vessels are enormous steel cans that must hold radioactive coolant under intense heat and pressure for decades. Forging them is slow, expensive, and dominated by a shrinking number of suppliers.

Oak Ridge National Laboratory (ORNL) and Idaho National Laboratory (INL) think wire arc additive manufacturing can help. Their demonstration is simple in concept and audacious in scale: use a multi-robot metal printer to build a nuclear-relevant pressure vessel in one piece, then qualify it with data captured during the build.

MedUSA: Three Arms, One Arc

The printer is called MedUSA. It uses three robotic arms working together around a single build. Each arm feeds steel wire into an electric arc, melting and depositing metal layer by layer. ORNL recently used the system to print a pressure vessel roughly three feet by five feet in a steel alloy relevant to nuclear applications.

That is not a desktop project. Wire arc additive manufacturing deposits metal quickly and uses inexpensive wire feedstock, which makes it practical for very large parts. The challenge has always been precision, repeatability, and confidence: a pressure vessel cannot have hidden flaws. Three arms add flexibility, letting the system reach complex geometries from multiple angles without re-fixturing the part.

Born-Qualified Parts

The real innovation is not the print itself. It is the idea of a born-qualified component. Instead of building the vessel, then cutting samples and running months of post-build tests, ORNL and INL want to gather enough sensor data during printing to prove the part is good as it is made.

Cameras, thermal sensors, and process monitors track arc behavior, temperature, layer geometry, and thermal history. Machine-learning models compare that stream against validated manufacturing records. If the data stays within bounds, the part may move through qualification faster than a forged equivalent. INL brings nuclear materials expertise and its Prometheus AI tools; ORNL contributes large-scale additive manufacturing and materials characterization.

This is still a research program. A single demonstration vessel does not mean reactors will soon run on printed parts. But the project has already produced neutron sensor brackets for Antares Nuclear's R1 Mark-0 microreactor, which reached criticality at INL in June. Those brackets are smaller, yet they give the team a chance to test the same data-driven verification approach on hardware tied to an operating nuclear research program.

Beyond Nuclear

If the process works, the benefits spread beyond reactors. Aerospace, defense, chemical processing, and oil and gas all need large, high-integrity metal structures. A reliable additive route would reduce dependence on scarce forging capacity and shorten lead times from years to months.

For now, the achievement is a proof point: three robotic arms, one enclosed vessel, and a path toward manufacturing components that were once impossible to print at this scale. The next question is whether the data can match the ambition.