Scaling Example #1: Small Modular Nuclear Reactors

Now, let’s used the tools we’ve learned, to took at some examples from the effort to decarbonize our economy.

The first example to take a swing at with our new understanding of vertical and horizontal scaling is the small modular nuclear reactor, or SMNR for short. SMR means something else to me- steam methane reformer- a technology which pre-dates the entire nuclear industry.

First, a disclaimer. I am not a nuclear engineer. I have no nuclear design experience, and I claim no special expertise in relation to nuclear power. I am however a chemical engineer with decades of experience helping people scale up- and down- chemical process technology. I have also spent decades designing and building small modular chemical plants in a factory environment, and have more than a passing familiarity with their engineering economics.

Let’s also be clear: I think nuclear power can be made adequately safe, and that its use in the 1970s and 80s undoubtedly saved countless premature deaths in my home province of Ontario relative to its true market competitor at the time- burning coal to make electricity. Nuclear is also very clearly a dispatchable, or at least reliably available, source of electricity which also has very low GHG emissions. While I acknowledge that the both the decommissioning of nuclear plants and the storage of nuclear waste are major political and public relations problems, I think that they both have quite practical technical solutions- though the cost of those solutions is unclear to me. That’s a long list of pluses for nuclear power- rather big ones. There is some argument as to whether or not that list of pluses is worth whatever nuclear might cost us, but I’m not of that opinion. Nuclear is just one option- and there are others. We need a societal discussion about what those options are, and what their costs and impacts are.

Nuclear Power’s Present: Giant Vertical Scale

The modern nuclear reactor power plant is the epitome of “go big or go home”- of maximizing vertical scale in an (often seemingly vain) effort to keep the cost per kWh low for consumers. Initially, nuclear reactors were small as we learned how to use them. Engineers understood the economy of vertical scale just as well in the 1950s as they do today, and so it was quite clear that if nuclear power were to become cheap, it would do so by building nuclear reactors at considerable (vertical) scale. And as the industry grew, and project experience was gained, reactors got bigger- not smaller.

It is important to realize what a nuclear reactor is, at its essence. It is a steam power plant with a nuclear fission heatsource, and associated safety and controls equipment to operate and keep that heatsource safe. Given that there’s nothing magical about a steam power plant, surely the steam power plant portion of the project should benefit from ordinary economy of vertical scale.

What about the fission reactor itself?

There have been plenty of examples in the recent past of cost and schedule over-runs in nuclear power projects in numerous locations around the world- in fact, it’s much harder to find examples of nuclear power projects which are anything close to on time and under budget (though nuclear apologists always have a few of those to trot out as examples). The preponderance of cost/schedule over-runs has led people to conclude, perhaps not unreasonably, that nuclear went too far into a giant, megaproject vertical scale that was larger than truly practical.

The current scale of reference is, round numbers, 3 GW of thermal output, or about 1 GW of electrical output per reactor. The cost/schedule overruns would imply that an optimal scale for nuclear power deployment might be at some scale smaller than the current scale- a scale requiring fewer “heroics”.

A nuclear power plant generally has several reactors of around 1 GW electrical capacity, sited together. An example is the Darlington nuclear power station located east of Toronto. It consists of four units with a total generation capacity of 3.5 GW of electricity- roughly 20% of Ontario’s electricity demand. The units are operated independently but share common infrastructure, again to save capital cost.

Darlington’s construction started in 1982 with unit 1, and ended in 1993 when unit 4 was completed. The original budget of $4 billion was exceeded, considerably- the plant cost $14.4 billion, or roughly $23 billion in 2020 dollars. Some of that overage resulted from financing costs arising from project delays caused by government “interference” etc. – the story is a long and sordid one, which likely is an object lesson in why projects like this should not be allowed to become political footballs. In 2021, the 30 yr refit project for the plant was started, at a cost of another $13 billion. If we’re honest, that refurbishment cost was “baked in” when we decided to build the project in the first place.

In 2013, Ontario went out to tender on a “twinning” of Darlington. Prices came back so eye-wateringly high that nuclear ambitions in Ontario came to a screeching halt.

Until 2021, that is…

A new project, referred to as “Darlington New Nuclear”, hit the headlines. The plan is for a new 300 MW (and it looks like that means 300 MW electrical capacity) Hitachi boiling water reactor for the Darlington site.

The moniker “small modular nuclear reactor” has been attached to the project- but the plant, at 1/3 the size of the existing units, is not small in objective terms- it’s in fact bigger than the 200 MWe Douglas Point plant, the prototype for the CANDU reactors later built at Darlington and elsewhere, constructed in the 1960s.

It’s also not modular in the sense that most people understand. It will not be completely assembled in a factory and delivered in sections that are easy to put back together, on the back of a truck or trucks. The plant, if ever built, will be substantially site constructed, just as the larger Darlington units were.

But I digress…

Here’s the SMNR pitch, being made by many firms today:

  1. The reason nuclear power plants are so expensive is that they’re always a brand new, 1st of kind design. There’s no steady crew of people with the specialist skills to efficiently build them, because we never build the same design twice nor do we build them one after another. No common design means people spend more time engineering, and less time building.
  2. A “new” nuclear fission reactor technology will be used. Sometimes, that’s just a new twist on the existing “boiling water” reactor. Sometimes, they’re talking about a totally different technology, such as a molten salt fuel cycle.
  3. The units will be built at much, much smaller scale than the existing state of the art. NuScale, for instance, the project which seems to be farthest along in the USA, has a capacity per unit of about 77 MW electrical per reactor. Twelve (12) units operating in physical parallel would be required to replace a single Darlington-scale unit.
  4. The smaller scale is claimed to be small enough to be “intrinsically safe”, or something near enough to that, so the hope is they will be simpler to build, and easier and quicker to permit.
  5. The units are small enough that they can be built (and apparently, also fuelled) in a factory, and shipped to site “largely assembled”- in NuScale’s case, in three truck-shippable pieces per reactor, totalling 700 tonnes per reactor. The claim is that will make projects faster to begin producing some power, and hence much cheaper.
  6. Because the factory will make the same unit again and again, what the unit lacks in economy of vertical scaling, will be more than compensated for by a) factory fabrication by a trained team b) “mass production” and c) “simplicity” arising from the smaller scale
  7. The small units will be perfect for use on remote sites like mines, small remote communities etc.

Let’s examine the claims one by one, using NuScale as a reference case because its information is quite widely available to the public, not because it is especially worthy of either praise or criticism:

  1. To me, this is a popular myth, not a reflection of the real reason nuclear power plants are expensive. Like all myths, there’s a grain of truth there: nuclear is a specialist industry with a heavy certification burden. The real reason they’re expensive is that they are massive capital projects with extremely long design life, a high risk profile, and accordingly a long permitting, approval and construction process. Projects which must be done at positively massive scale to deliver sufficient economy of vertical scale to make each kWh seem cheap enough for ratepayers to afford, for reasons made obvious in my 1st article in this series. And the regulatory attention is inescapable, because the risk profile means that only the public has deep enough pockets to insure these projects against accidents.
  2. To my non-nuclear eye, NuScale isn’t really a new nuclear technology- it’s just a small boiling water reactor with a different cooling scheme involving a thermosiphon and water immersion rather than active pumping. And, shockingly, from a brief review of NuScale’s website, it seems that the plan is to connect each unit to its own (tiny) “skid mounted” (modular) steam plant, such that even the steam plant part of the job will lack economy of vertical scale.
  3. At 77 MW electrical output per unit, the unit definitely qualifies as “small”. Therefore, numerous individual units will need to be installed, either in physical parallel on the same site with common infrastructure, or on numerous sites, to supply equivalent amounts of power to the units they seek to replace. Which of these two options will be cheaper? Obviously the former, and by a lot!
  4. “Intrinsic safety” is obviously something which is easy to claim, but quite hard to demonstrate to the satisfaction of a regulatory body who knows that the public, not a private entity, will be providing insurance against an accident. By “hard”, I mean “will cost a lot and take a long time to achieve”
  5. Certainly the pieces of each NuScale reactor look to be small enough to be shipped by a number of different means, including by heavy logistical truck/trailer units- but by no means would those be “routine” shipments given the diameter and weight, even if they didn’t also contain active nuclear material. The project is, however, still modular in the way people typically understand that term in the industry.
  6. Until orders of such reactors are so common that maintaining a dedicated factory full time for their fabrication is a practical option, each unit will be built more or less by hand, albeit in a factory environment. Subcomponents and sub-assemblies will be made in other dedicated factories, just as all plants are built today, whether they’re modular or “stick built”. But the project would have access to the benefits of modular fabrication. Calling that “mass production”, however, is more than a small stretch of the definition of that term! Such a factory would have almost nothing in common with, for instance, a factory making cars. People, not robots, will be doing most of the work. The notion that sufficient savings in labour and schedule would be possible to overcome the rather obvious lack of economy of vertical scale of each unit is therefore very questionable.
  7. It is clear that the lack of economy of vertical scale will not be compensated for adequately by modular fabrication even if the units are ganged in parallel on a common site with common infrastructure. The notion that putting tiny units alone or in small groups on numerous different sites could yield affordable kWh for consumers is just preposterous.

Let’s look closely at claim 6) – that mass production in a factory environment would overcome the conventional economy of vertical scale.

Let’s take the unit cost of one NuScale 77 MW unit as x units of capital cost. What would we expect 12 such units, factory modular, to cost if they were all ordered at the same time? I’d guess 12^0.9 x at best, to be generous, or about 9.3x. It could easily be higher.

What should, in comparison, one unit of 12*77 = 924 MW electrical output, cost? About x * 12^0.6, or 4.4x

For a project which hinges on capital cost per unit of value production, that’s a death sentence. On the basis of decades of experience doing it for a living, there’s no way that factory modular fabrication is going to drop the price per unit sufficiently to make up for that ocean of a difference.

Now let’s look at a few other issues which seem obvious even to me as someone who absolutely makes no claim to be a nuclear power expert:

  1. From a nuclear proliferation, security, terrorism etc perspective, distributing nuclear reactors on numerous sites, particularly remote/rural ones, is far riskier than larger centralized sites which can be better planned and protected. Power distribution costs won’t be reduced unless we also decide to site these numerous little nukers much closer to population centres- something that is unlikely to go over well with the people who would be living next door. You can claim that such fears are unjustified, but that doesn’t mean they won’t present themselves with pitchforks and torches at every public meeting.
  2. Because fuelling costs are low, and capital costs are (very) high, nuclear power is generally operated as close to 100% capacity factor as physically possible, generally being given preferential access to serve loads on the grid. The issue isn’t that nuclear power plants can’t be “turned down” in output- the issue is that you can’t afford to operate them that way. And that fact means that nuclear doesn’t play well with intermittent wind and solar power, which are cheaper when they are available and simply not available when they aren’t. Making the plants smaller won’t change that, at all. The putative benefit of having 12 units you can individually control, really adds not very much to that economic equation.
  3. Could Wright’s Law really be counted on to make each subsequent reactor cheaper than the last? Can SMNRs become like solar panels or Li ion batteries? That depends on how applicable you think Wright’s law is to fairly conventional equipment- heat exchangers, welded pipe systems, steam power plants etc. My bet is that it’s not very applicable because manufacturing processes for such equipment are already very well understood- we make enormous numbers of pieces of such equipment in the world yearly already. The potential for Wright’s Law “doublings” which lead to learning-based cost reductions, seems small, though I don’t doubt there would be a learning rate if there were sufficient doublings.
  4. For Wright’s Law to kick in, we’d need to have a single design which is the obvious favourite, and to build that one only. Does such a design exist? No- rather, there are many designs being proposed, for both conventional and new fuel cycles, with no clear winner.
  5. If a particular future fuel cycle (molten salt, thorium, what have you!) is somehow limited to a small maximum scale due to its nuclear physics, to me that’s a flaw, not a feature. It means that the technology will have challenges to achieve a low cost per unit of production, ie. per kWh it makes for consumers. It also means that each technology will make its cheapest kWh when built out to its largest practical scale- just like all other technologies which produce commodity products.
  6. Can the smaller units be refurbished? What’s their design life, and how would one extend that to maximize the number of kWh each unit generates before it becomes a pile of (low level) radioactive waste? I sincerely don’t know the answer to these questions, but I’m sure others might.
  7. Is a nuclear reactor really a great tool to site at a remote location like a mine or remote community? Are such locations ideal in terms of emergency response, skilled and trained maintenance staff etc. Etc.? And does the power use of such sites, and the resulting GHG emissions, really make a big difference to total world GHG emissions? Is this a “hard to decarbonize sector” or just an excuse to sell units to places already accustomed to paying high prices for power from diesel generators and the like?
  8. It is sometimes claimed that SMNRs provide greater possibility to provide combined heat and power than conventional nuclear power plants, given that heat can’t be shipped over distances as great as electricity can in economic terms. However, that’s only true if we put them on numerous sites which are each closer to populated centres, and then we’re willing to spend the money to build district heating etc.. That is, as already noted, not a recipe for low capital cost.

From this analysis, and based on long discussions with nuclear advocates and nuclear critics, I can say that I consider the small modular nuclear reactor to be basically nearly pure nuclear #hopium. It’s a concept that fails a basic economic “sniff test”- a proposed solution that seems incapable of solving nuclear’s really big problem, which is its enormous capital intensity- not its tendency to draw out “no nukes” protesters.

It also seems to run quite contrary to the learnings of the past. And you know what they say about that: doing the same thing over again and expecting a different outcome is a fairly accurate definition of delusion.

Why are SMNRs So Popular, Then?

Lots of smart people, and entire companies with world class pedigrees such as Rolls Royce, are lined up in opposition to what I’m telling you in this piece. Why am I so sure that they’re wrong and I’m not?

Simple. It’s a concept known as “moral hazard”.

When I was in the business of designing pilot plants for new processes for clients, on occasion the client or their investors might ask me whether I thought the process was “worth piloting”, i.e. Did it have a likelihood of economic success? I would (rightly) refuse to answer such questions, and if pressed, I would simply repeat the client’s own claims to them and say, “If A, then B”. Why did I give such a cagey answer? Because, as a designer/builder of pilot plants, I benefited financially from designing and building the pilot plant, whether the process had any chance of economic success or not! Any opinion I offered to such questions was therefore offered from a position of an actual conflict of interest- and I was in a position of “moral hazard”. As an aside, I love the fact that as an independent consultant, I can now tell clients straight up about every strength and every weakness I see in their plans- with no moral hazard.

Clients who can’t take the truth as I see it, I’m quite happy to part ways with- I even advertise this as a feature of my consultancy on my website.

Now put yourself in the shoes of a nuclear engineer: you’re coming to the end of your career in what is basically otherwise a dying industry. Very few at-scale nuclear projects are being built, so maybe you’re working on a refurbishment project- the last one on that plant. To you, the chance to work on a SMNR project, probably one lasting many years, especially one funded by governments or by people who will pay your salary whether the project achieves its goals or not, is likely a very pleasant one relative to trying to find a new industry to work in late in your career. You’re an expert in nuclear power, certainly- but do I give your opinion about its potential for success of SMNRs, any real weight? Or do I consider that opinion to be one offered from a position of moral hazard?

That’s certainly not an accurate description of everyone who supports SMNRs, by a long shot. There are many people of such high personal integrity that they will tell the truth, when asked, and the whole truth, even if that truth is contrary to their personal economic interest. But it describes a lot of them, especially many of the ones advocating the concept most loudly in public.

There are some people who are absolutely not in a position of moral hazard who also think SMNRs are the bee’s knees. They may genuinely believe that SMNRs have a real chance to make cheap kWh one day, if we only finally standardize on one design and then build them by the thousand every year. I just don’t think those people are thinking clearly. I think they’ve smoked a little too much #hopium for their own good. Doesn’t make them bad people- doesn’t make them right, either.

Finally, there’s the even more cynical group of people. People like Doug Ford, recently re-elected premier of the province of Ontario. Doug is many things- aside from being the older brother of the imfamous late crack-smoking former mayor of Toronto Rob Ford, Doug is quite likely also a closeted climate change denialist, although he is far too cagey to ever admit that publicly.

Let’s say you’re our dear leader DoFo. You have lots of people clamouring at you about climate change, and since you fancy yourself to be a populist, you don’t want to appear to be doing nothing. But you don’t believe in it, so you don’t want to spend any real money dealing with it. Especially not on “green power” projects which you ran against as costly boondoggles and a blot on your rural voters’ landscape- projects which you cancelled, then passed legislation preventing the project’s proponents from seeking the cancellation fees former governments had agreed to. You also know that the Pickering nuclear power plant is scheduled to close down in 2024, for good, because it’s finally too many years past its best-before date to extend any further.

So: what do you do?

How about planning a 300 MW “small modular nuclear reactor”? Are you troubled that it’s not really small, nor modular? Nope. You talk about all those Ontario jobs. And you know that it will be quite happily studied- by people largely in a moral hazard position- until you’ve retired from office. The money spent on this gives you plausible deniability about the whole AGW issue as you see the province build more fossil gas-fired power plants to replace Pickering. “Just wait”, you say- “the SMNRs are coming to save the day!” Doug can add #hopium dealer to his long resume…

Recommended Reading: you can’t go far wrong in reading @Michael Barnard ‘s treatment of the same topic, which was informed in part by discussions we had about the topic but which also contains Michael’s usual, top notch research and analysis.