Westinghouse Wants to Build Ten AP1000 Reactors in the United States by 2030

Inside the forgings, pumps, and standardization that decide whether ten reactors become a fleet 

Frontier Winston 

 

 

 

The physics of a pressurized water reactor fits comfortably into a senior-year nuclear engineering course. Enrich uranium to about five per cent, load it into fuel assemblies, immerse the assemblies in water pressurized to 155 bar so it cannot boil, and let the fission heat that water to 325 degrees Celsius. Run the hot water through a steam generator, where it boils a separate secondary loop at lower pressure, and send that steam to a turbine. The AP1000 does this at 1,110 megawatts net, with passive safety systems that need no operator action and no emergency power for 72 hours after a station blackout. The design is certified. Two units are running at Vogtle in Georgi

 

Manufacturing capacity is not yet established. Westinghouse plans to build ten reactors in the United States, with construction starting by 2030, totaling approximately 11.2 gigawatts if completed. The main challenge is a small set of components that are extremely difficult to produce and for which few suppliers are qualified. These components caused significant bottlenecks during initial builds. Standardizing their production is essential to reducing costs for subsequent reactors.

The forgings almost nobody can make

When China constructed the first four AP1000 units at Sanmen and Haiyang, heavy primary components were sourced from three continents, as no single supplier could provide the complete set. Doosan in South Korea forged the reactor vessels and steam generators, Curtiss-Wright in Pennsylvania built the coolant pumps, and Mangiarotti in Italy fabricated other pressure components. Delays in any of these items affected the entire construction schedule, as they determined the project's critical path.

 

 

The reactor vessel is the most constrained component. It is approximately twelve metres tall, weighs several hundred tonnes, and is assembled from forged steel rings welded into a single sealed shell. The upper shell is a single ring forging with four inlet and two outlet nozzles integrated into the body, reducing the number of welds in the nozzle region. This is important over a sixty-year service life, as the vessel wall is continuously exposed to neutron flux, which increases the steel's ductile-to-brittle transition temperature. Fewer welds mean fewer sites to monitor for embrittlement over decades. The alloy must be low in impurities, as elements like copper and phosphorus accelerate embrittlement. Forging and post-weld heat treatment are optimized to maintain steel toughness long after the vessel cannot be replaced.

The circumferential welds joining the rings are complex and time-consuming. Each weld is a deep groove filled gradually over several weeks, with the ring pre-heated to prevent cracking. Every weld undergoes multiple inspections and must meet nuclear standards, a process that can take years to certify before production begins. Only a few forging houses can press ingots of the required size, and only a few fabrication shops have the necessary welding and inspection qualifications. As a result, the wait time for a finished vessel can extend for years.

Steam generators present similar challenges, but on a larger scale. Each is a heat exchanger taller than the reactor vessel, containing thousands of thin-walled Inconel tubes that separate the radioactive primary water from the secondary steam loop. The integrity of these tubes determines the plant's operating life. Steam generators also start as heavy forgings from the same limited group of qualified suppliers. Early procurement of both vessels and steam generators is considered the main constraint on the construction timeline.

The pump that had never been built at this size

The AP1000 uses four reactor coolant pumps, two per steam generator loop, which present a different type of bottleneck. These are canned-motor pumps, with the entire motor located inside the reactor coolant pressure boundary. The casing is welded directly to the steam generator channel head, creating a single hermetically sealed assembly. This design eliminates the shaft seal, removing the most common leak path found in conventional reactor coolant pumps. Both the stator winding and rotor squirrel cage are hermetically sealed to ensure a dry winding rated for the plant's sixty-year design life.

While the design eliminates the shaft seal, it had not previously been manufactured at the AP1000 scale. Curtiss-Wright's Electro-Mechanical Division in Cheswick, Pennsylvania, designed and built the pumps. First-of-a-kind engineering challenges with these pumps, along with the explosive-actuated squib valves in the passive safety system, had to be addressed during the Chinese construction program, delaying the startup of Sanmen 1. These issues were not failures of the certified design, but rather the result of building at this scale for the first time without prior production experience.

Everything that has to be identical

The third constraint is not physical, but it has historically been as costly as the steel itself.

Almost every large reactor built before this program was partially redesigned during construction. Each project referenced the certified design but accumulated unique engineering changes, extensive licensing documentation, and separate negotiations with regulators regarding compliance. The main cost was not a single forging or weld, but the repeated engineering and licensing work for plants intended to be identical. Building ten reactors this way results in ten distinct engineering projects with little in common beyond their name.

The two units now operating at Vogtle in Georgia were completed years behind schedule and significantly over budget, contributing to Westinghouse's bankruptcy in 2017. They were the first reactors built in the United States in thirty years. There was no active supply chain, little recent construction experience, and a workforce that had not built a large reactor in a generation. Rebuilding this workforce takes years, as welders, pipefitters, and construction crews cannot be trained on the same timeline as procurement. The fleet approach ensures that the cost of relearning these skills, both on the shop floor and in engineering, is incurred only once.