The AP1000 was supposed to be a "modular" design, where most of the difficult welding could be done off site and delivered complete, with paperwork. And failure in this modularity is what caused the project to be such a flop, and essentially kill all new large nuclear in the US (people are trying "small" modular, but their target costs are still far too high to be economically feasible).

See, for example, this 2017 report on just one aspect of what went wrong:

https://www.enr.com/articles/43325-witness-to-the-origins-of...

> To build the first new nuclear reactors in the U.S. in three decades—South Carolina’s V.C. Summer Units 2 and 3 and Georgia’s Plant Vogtle Units 3 and 4—the design and construction team would face a steep learning curve. However, says Hartz, learning wasn’t much of a priority in the rush to start work at Lake Charles. “They were clueless” about the complex geometry of nuclear welds, the nuclear supply chain and the need for a nuclear safety culture, he notes, adding, “I wasn’t a whistle-blower. I was just a senior procurement manager who was concerned.”

> Westinghouse would issue drawings to Shaw Nuclear in Charlotte. When Shaw reviewed the drawings and asked Westinghouse to correct a detail, problems ensued. The work processes were unnecessarily complicated by the separation of the team members. Giving an example of how the process got out of hand, Hartz says that, if a design called for a 3⁄8-in.-wide, 12-in.-long fillet weld, the welder might make it 14 in. long. “Instead of having Westinghouse right there saying, ‘That’s no problem,’ ” recalls Hartz, “we had to write a nonconformance report that was processed and reviewed by Shaw and then sent to Westinghouse for disposition. It was insane. From Lake Charles to Pittsburgh to Charlotte then back to Shaw Modular before the red nonconformance tag could be taken off, saying it’s OK now.” He adds, “Each change went through the same tortuous path, taking months and months.”