Repetition is the key to cutting nuclear costs, not innovation

In the early 1950s, when he was building the world’s first nuclear submarine for a sceptical US Navy, Admiral Hyman Rickover had a problem. He needed to fix on a working atomic reactor that could power his proposed experimental vessel. Yet everywhere he was besieged by engineers who offered tweaks or modifications, or just a new and better plan.

Determined to move things forward, Rickover wrote a memo for his team intended to silence the technical importuning. In it he distinguished between what he called an “academic” and a practical reactor.

An academic reactor, Rickover wrote, had the following characteristics: “(1) it is simple; (2) it is small; (3) it is cheap; (4) it is light; (5) it can be built very quickly; (6) it is flexible in purpose; and (7) very little development is required [as] it will use off-the-shelf components”. In almost every respect, it was superior to the practical reactor, which was the one actually under development. This sad specimen was, after all, behind schedule, expensive, large, heavy and hard to perfect.

But the academic reactor had one big drawback, which meant Rickover did not want it distracting from the pursuit of its practical alternative. It was “in the study phase” and “not being built now”.

Rickover would no doubt crack a wintry smile at some recent debates over nuclear’s role in decarbonisation. These again pit today’s “practical” large reactors, such as EDF’s EPR (European Pressurised Water design), against such “academic” brainchilds as the small modular or thorium reactor — neither of which is yet in commercial development. 

Despite embarking on the construction of a giant EPR at Hinkley Point in Somerset, the UK government is said to be tempted by the idea of leapfrogging on a generation to the SMR (small modular reactor), perhaps influenced by the possibility of showcasing homegrown Rolls-Royce technology.

More generally, the case for a leap forward has been made by Michael Liebreich, an energy expert and renewable advocate. He argues the existing generation of reactors is just too costly, citing cases such as Flamanville in France ($9bn over the original $3.8bn cost and eight years overdue) or Vogtle in Georgia, US (almost double the cost and five years late). It has been “tested to economic destruction”, he says.

Instead, Mr Liebreich calls, rather like Rickover’s engineers, for what we do not have today: in this case the SMRs or thorium reactors that will be: (1) simple; (2) small; (3) cheap, etc.

What is the problem with this vision? Well, call it nuclear experience. In a 2015 paper, two French scientists, Michel Berthélemy and Lina Escobar Rangel, tried to explain one of the paradoxes of nuclear development. Whereas, with most industries, innovation drove down cost over time, it seemed as if nuclear had a negative learning curve.

What they discovered was that innovation was indeed the enemy. Only by sticking firmly to the same specification, engineers and builders could you drive down construction costs. This “learning by doing” could lead to meaningful reductions. Take the Shin-Kori plant in South Korea, now building its fifth and sixth unit. The time taken to build the second set was 25 per cent less than the six years to build reactors one and two, according to the pro-nuclear campaigner Michael Shellenberger. And all this while increasing the size of the reactors in output terms by 40 per cent.

There is a lesson in this straight from Rickover’s memo. It is to fix on a design and drive out cost through incremental improvements. Having swallowed the hefty “first of a kind” (FOAK) costs — see Hinkley and Flamanville — and invested in the supply chain to build the latest big “practical” reactors (after a 30-year nuclear hiatus), countries such as the UK should build large “cookie-cutter” fleets.

Twice weekly newsletter

Energy is the world’s indispensable business and Energy Source is its newsletter. Every Tuesday and Thursday, direct to your inbox, Energy Source brings you essential news, forward-thinking analysis and insider intelligence. Sign up here.

The last thing needed is a multiyear pause while the next generation is brought to its FOAK phase, turning that costly investment rapidly into dust.

Of course, this does not exclude research into next generation systems. The SMR might be a winner in terms of cost or flexibility. And if it is, it should be developed in its turn. But Rickover was right about the virtues of continuity. South Korea has basically built the same sort of light water reactor since the 1980s. That is why its capex costs are close to $2,000-$4,000 per kilowatt as against $8,000 for the UK’s last completed FOAK project, at Sizewell B in the 1990s, according to research by Energy Technologies Institute.

Nor is it a lesson that just applies to nuclear. Take solar power: as Mr Shellenberger points out, the fundamental technology has not changed much since the 1950s. “Solar panels became cheap not through radical changes in design but rather decades of incremental process-based improvements,” he writes. Nearly seven decades after Rickover’s memo, it is a lesson the nuclear industry should learn.