Without any fanfare, a few days ago, the Nuclear Regulatory Commission moved on to the next phase of one of the most difficult and important tasks it has ever undertaken: making a new rule to license future reactor designs. For two years prior, it had been accepting public comments on draft rules, but with that door now closed, the rest will be up to the NRC.
The NRC staff has released a couple of versions of a partial draft rule, called Part 53. While opinions on the details differ in some places, there is broad stakeholder consensus that the framework is not pointed in the right direction. To the NRC staff’s credit, they often ask for stakeholders to identify specific issues to fix. The problem is that stakeholders generally don’t want to identify minor issues with the draft rule, they want an entirely new approach—from a blank sheet of paper—and have offered to help the NRC make it happen.
Starting from a blank sheet of paper may be necessary.
The regulations on the kinds of reactors we have now just don’t fit tomorrow’s designs. So the NRC has to develop an entirely new way to license new kinds of reactors. Today’s rules assume huge machines, cooled by water at high pressure, each reactor staffed 24/7 by teams of licensed operators, and kept safe by complicated safety systems. But the designs for the advanced reactors now moving towards commercialization don’t look like that. There are lots of variations, but most are smaller, many don’t use water, some use hardly any pressure at all, and most don’t need the extensive system of emergency pumps, valves, pipes, and water tanks needed to keep today’s reactors safe. Instead, the new designs rely on gravity, natural heat circulation, and other “passive” principles. Many don’t require operators to take action in case of malfunction. In some cases, they don’t even need to have operators on site.
The mismatch between the current licensing system and new reactors is something Congress recognized and has been grappling with since at least 2019 when it passed the Nuclear Energy Innovation and Modernization Act, and told the NRC to come up with rules for advanced reactors that were “technology-inclusive,” “risk-informed,” and “performance-based.” Those terms are a bit off-putting, but it’s worth unpacking them, to see why the NRC’s task is so difficult and why early indications are that it’s not going well.
“Technology inclusive” means that whatever the reactor uses to move heat out of the core to where it can be used—water under high pressure in today’s models, but inert gas or liquefied salt in some advanced designs–one set of rules will set the safety, security, and inspection standards that all the reactors must meet.
“Risk informed” is a doctrine that the NRC has been trying for decades to incorporate into its rules for current-generation reactors; it means figuring out which systems are important for preventing damage and which aren’t, and concentrating surveillance, analysis, and maintenance on those parts, rather than enforcing cookbook style rules across all systems regarding inspection and maintenance intervals.
Finally, “performance-based” means allowing designers to make decisions on how to meet safety objectives, rather than specifying the method.
One of the reasons coming up with performance-based rules that meet Congress’ mandate is hard for the NRC is that, while staffed with smart, diligent technical experts, the commission has inculcated a culture over the last 50 years of specifying everything—a culture of writing the book and then running a by-the-book operation.
In turn, NRC engineers are used to rules that prescribe certain details, but Congress called for setting performance objectives that don’t prescribe how to meet those objectives. A prescriptive rule for a car, for example, might say that it has to have four wheels, turn signals of a minimum size on each corner, sit a certain height above the ground, and have airbags that deploy at a certain kind of impact. A performance-based standard would define the objectives that a car has to go down the road straight, stop within a safe distance, and keep the people inside it alive in the event of an accident. Then the designer figures out how to do that.
NRC’s first licensing rules (called Part 50) were onerous enough for the previous generation of nuclear reactors that the commission made a second rule (called Part 52) in 1989 in an attempt to streamline that process. But Part 52 was still aimed at the previous generation of nuclear reactors. The problem is that the new designs are so different from the old ones. With new designs diverging significantly from the previous generation, a change in approach is needed—it makes little sense to continue prescriptively mandating wheels for all modes of transportation when boats exist, and when next-generation cars could be hovercraft (who knows).
There are many examples of where technology-specific regulations limit innovation. For example, government regulations mandate inspection of airplane wings, for signs of corrosion. Aluminum corrodes, but the wings are increasingly made of carbon fiber, which doesn’t. Airplane developers have to figure out if they are allowed to use carbon fiber at all. If they can use this innovative material, how can they either show they are complying with irrelevant maintenance requirements intended for the previous generation of aircraft or, better yet, get relief from that regulatory burden altogether?
Cutting and pasting from the old rules will not meet the mandate Congress gave the NRC. Starting from a blank sheet of paper is the only option—a conclusion the NRC staff and Advisory Committee on Reactor Safeguards have also stated on multiple occasions, even though the current draft of Part 53 doesn’t reflect this need.
But even starting with a blank sheet of paper wouldn’t by itself be enough. Filling in the blank sheet of paper wrong would have perverse effects. Today’s reactors were designed in the 1960s, but are safe and reliable. The advanced reactors, because of passive safety features made possible by a half-century of engineering progress, are even safer; they are simply not capable of sustaining some of the accident scenarios postulated for current-generation reactors. (How, for example, do you get a core melt accident in a molten salt reactor, which is already molten in its ordinary operation?) But if the regulatory structure is clumsy enough, new reactors simply won’t get built. Or, perhaps they will be built, under the old, inappropriate system, which will mean that the federal government will spend tens of millions of dollars on developing an alternative system that no one will use.
It isn’t even clear that using the older systems would work. If we’re going to decarbonize the electric system, and use electricity to replace gasoline and diesel on the roads, and fossil gas in home heating and industry, we’re going to have to build reactors by the hundreds, in addition to other clean energy sources, just to meet electricity demand. That means licensing dozens of reactors per year. But it’s hard to imagine the current system handling more than a handful over that span. Near-term applicants already have to use the existing frameworks, with mixed results, because Part 53 is not available yet.
Now that the door is closed to comments on the existing draft, the NRC staff will start the work of evaluating public written and in-person comments (although the commenters, including the Breakthrough Institute, did not have a full draft to comment on). In the Spring, it will publish a formal draft in the Federal Register. At that point, despite its draft status, the general direction will be mostly set.
The NRC is not everyone’s favorite agency. But its staff are smart, dedicated engineers. The trick now is for those engineers to realize that they are facing a wide range of new technology and that, as a result, they need a fresh approach.