jump to navigation

Is nuclear energy actually cheap(er)? September 20, 2022

Posted by Maury Markowitz in Uncategorized.

Supporters of nuclear power often make the claim that nuclear is not being built out for some nefarious reason or another. “Nefarious” is being used loosely here, in the sense that the reasons are unfair or undeserved and even a little under-handed. I’m sure this comes into play at some level, but I don’t believe it is even remotely close to being anything like the primary reason, or even secondary.

I believe the reason is money.

The reason is always money.

If you’ll hurry aboard…

I recently came across an article that started with the claim:

For many years, we’ve heard nuclear waste is terrible and we should forget about building low-carbon nuclear power plants.

7 Things The Simpsons Got Wrong About Nuclear | Department of Energy
It so unfair!

This is the sort of thing that really gets my goat. I’m sure everyone reading this has heard this claim, or some variation – replace “waste” with “meltdown” for instance.

Sure, lots of people complain about waste, but does that actually have an effect? After all, lots of people complain about the weather, and nobody ever does anything about it!

I don’t think the public’s concerns have very much bottom-line effect, I think it’s all about the money. And I found this particularly ironic, given the article claims:

It’s a form of denialism to focus on the non-problem (nuclear waste) at the expense of the real problem (climate breakdown).

I’d say it’s also a form of denialism to focus on the on the non-problem (nuclear waste) at the expense of the real problem (money). So that got my back up.

The article led me, roundabout, to a European group, RePlanet. So I reached out to the RePlanet people inquiring about writing something like a rebuttal. Rob De Schutter suggested I drop into the Discord instead. I posted my concerns about this line of argument, saying that the article is ignoring the real problems and going after non-issues. “It’s all about the money”, says I.

Rob replied back with a link to an article by Joris van Dorp. Back in 2019, he wrote an article for Medium that suggested money is not really an issue. Sure, it’s expensive, but that’s looking at it the wrong way.

NOTE: the article adjusted after he wrote and explained what is not explained in the original article: all future and present value is being deliberately ignored. As such the entire argument really boils down to something much simpler, and the article has been edited to account for this.

That’s no moon…

… it’s more like a disk

The article uses the example of the new Hinkley Point C (HPC) reactors being built in the UK. This is the latest of several European examples of the new “EPR” design from France. In the section “And what are the benefits?”, he outlines why it’s so inexpensive.

First statement in the section:

On an annual basis they will together supply 26 billion kWh to the British electricity grid… HPC has a design life of 60 years, with the possibility of extending it to at least 80 years… This yields a construction cost per kWh of 1.6 cents.

At the time he wrote the article, Hinkley was still expected to be around 25 billion Euro. That’s gone up a bit to 26 billion since, but that’s small potatoes. So basically the equation is:

25 billion Euro / 26 billion kWh per year x 60 years = 0.016 Euro/kWh.

Now this is very cheap, wholesale prices here in Ontario are currently hovering around 4 cents, or about 3 Euro, so if we could build a reactor at this price it would be very competitive.

But only if you ignore time.

… that’s inflation!

File:"Help Your OPA Fight Inflation" - NARA - 514468.jpg

If I give you a dollar and you keep it in your pocket for ten years, is it worth more or less at the end? Less, because inflation.

So, keeping this in mind, let’s say I would like to take a loan from you, I will take $100 and pay you $10 a year for the next ten years. Sound like a deal? No? Because inflation, you say?

Well here’s the rub; that’s exactly how we pay for power plants. We build them using today’s dollars, and pay that off using future dollars. This is a common calculation known as “present value”, or PV.

And here is the really crazy part, the value of electricity has actually fallen over time. That’s not wholly unexpected, as electricity is primarily a tech issue and tech gets better and cheaper over time. Since 1971, for instance, the inflation-adjusted value of electricity has actually fallen 2% in the US, and since 1915, the average rate is about 1.5%, compared to the average inflationary rate of 3.15%.

In other words, if you calculate your profitability using flat values, you’ll go bankrupt, because those future payments are worth less money. For instance, if we take the UK average inflation rate of 2.5% since 1990 and apply that to the 60 years of output?

25 billion Euro / 26 billion kWh per year x 60 years [FV @ 2.5] = 0.032 Euro/kWh.

The price just doubled simply by accounting for value vs flat pricing. Note, this is not the discount rate, that’s something different. Nor do I have better numbers for the UKs rates, so I can’t say is this is the correct curve to use in this case or not. 1.5% would be the rate to use here in Canada.

But this isn’t a killer problem, 3.2 is still competitive.

It’s worse

Of course the capital cost of the plant isn’t the only thing that goes into the final price of power coming out. That, in industry lingo, is the “levelized cost of electricity”, or LCOE.

The LCOE is simple in concept, it’s all the money that went into the plant divided by all the electricity that came out. We generally break that “all the money” into two piles, the capital cost of the plant, or “CAPEX”, and the operational costs of running it, the “OPEX”. Sometimes fuel will be considered separate from OPEX for clarity, sometimes not because it is part of operations.

We’ve covered the CAPEX side above, so now let’s see about the OPEX. Back to the original article:

The operating costs after commissioning of the plant are estimated at between 1.5 and 2.5 cents / kWh. This includes all running costs during operation, including personnel costs, fuel costs, permits, insurance, maintenance, taxes and premiums to the decommissioning fund and the processing and, importantly, the ultimate disposal of nuclear waste.

Now here is a place where inflation absolutely does apply; the cost of running a nuclear plant, where fuel costs are pretty tiny, is mostly in salary – both direct and indirect in things like licensing. So the OPEX goes up with inflation. This is one of the main reasons people shut down old plants (not just nuclear), they just become so expensive to operate they’re simply not worth it.

That’s precisely what happened to Palisades. The price of operating the plant kept going up, and the price on the grid kept going down. Eventually it simply produced power for more than it was worth, and even federal intervention didn’t help.

But OK, once again, let’s just ignore the time axis.

That’s not true! That’s impossible!

The I am your father cinema audience reaction video from 1980 is a fake and was doctored and was not actually a real recording from Star Wars Episode V The Empire Strikes Back.
If you only knew the true power of compounding inflation!

So, based on these calculations, the article states:

The construction costs and the operating costs together are therefore at most 4.1 cents / kWh.

Basically, if we could build the plant, poof, in one day, and we could buy our operational expenses forever without inflation, this is the price you get.

Now the HPC strike is 11.3 cents. Some of that is the cost of financing, about 3 cents, and the rest is profit and similar. Note that the 11.3 cents will be adjusted for inflation.

Well, that’s the trick, isn’t it?

Now here we come to the part that I just totally disagree with. We are supposed to simply ignore the difference between the 4.1 financing-free LCOE and the 11.3 financing-in strike price because the later has no “societal cost”. And what is this?

the societal cost of HPC is low because the costs that society experiences are exclusively the costs for building and operating the plant, not the interest and dividends, because those are returned to society… Most of the money… goes directly back to society in the form of pensions, public spending and so on

So basically, the argument here is that in spite of the fact that your electricity bill will go up because of these financing costs and profit taking, we shouldn’t care, because that cash will come back to us. So, for instance, after a couple of decades that money might pay for an old age home through the profits returned to the shareholders.

But it won’t. HPC is being financed primarily by France and the China General Nuclear Power Group. So those payments are going to foreign countries and/or banks. I’m pretty sure the UK public already takes a dim view of this, and would disagree that these payments benefit them at all.

And if the idea is to maximize societal benefit, why not just built a cheaper power plant and use the savings to build those things right now that you would instead have built in the future after amassing those payments?

File:Poll Tax Riot 31st Mar 1990 Trafalger Square - Police Pinned down.jpg
We love societal costs!

To take the argument to an extreme, why not just double the price of power in the UK today and use the extra income to give to the shareholders? According to the definition, this is a zero-sum game, and this version costs nothing at all to implement. But the UK already tried this, it didn’t end well

Roll credits

The weird thing is that everyone in the financials world and on the power plant operational side is perfectly aware of the actual problems in terms of financing. Even the nuclear industry itself admits cost and timeline are the main problems. At the recent World Nuclear Symposium, the industry was all-in on SMRs because they would help:

reducing costs and build times by constructing smaller, more advanced and less disaster-prone reactors

They’re not necessarily any cheaper (so far), but that’s beside the point, they’re faster and that helps remove the time uncertainty and makes the risk/reward ratio improve. This potentially eliminates the reason plants aren’t being built today: the bankers aren’t putting up the money because they’re scared of the risk.

It has nothing to do with nuclear waste. These are the same people that happily hand out billions to oil companies every year to blow up the planet, you think they care about a couple of gallons of waste? They do not. They have their reasons, they are their reasons, and they are perfectly valid reasons. Until those reasons are addressed, nothing is going to change.

The industry simply has to deliver on its promises. That is something we should all demand. Until they can demonstrate reasonable costs (which absolutely can be higher than other sources, higher prices are not a problem of itself) combined with predictable deliveries and construction, no one is going to buy this stuff. You wouldn’t buy a car from companies with this sort of track record, why would you buy a power plant from them?

Can SMRs actually fix this? Well I’ve watched SpaceX compete with Boeing and the outcome is a joke. I never thought this would happen, I thought it was rich-man-fixes-world hubris and was doomed to fail. Now it’s clear the management structure of these older companies is just not competitive. The companies making SMRs (in the US at least) are a whole lot more like SpaceX than Boeing. So I’m going to give them the benefit of the doubt.

The Sequel

NOTE: this has also been updated to remove FV and PV.

Let us consider a recent install that I’m a little bit familiar with, the new 500 MW Oberon project in the US. They are also adding 200 MWh of storage – practically every project in the US has storage now – but I’m not going to consider that in the cost calculation (for now).

Construction began in July 2022 and the commission date is September 2023. But because of the way PV works, they were able to bring it online in blocks, the first one in late August 2022. This, right here, is why renewables are killing it. There’s practically zero timeline risk. They had the system up and running, to a small degree, in one month. Whatever numbers the money people are pitching the bank are pretty likely to be true next month. Ten years? Hmmm. This is the schedule risk that is killing practically every large infrastructure project in “the west”, it’s not just nuclear.

So now, as I have done in the past, let’s use PVWatts to estimate the yearly production. I chose Twentynine Palms, which is close enough, a 1 kW system size, one-axis with backtracking, premium panels and 10% losses. That gets me ~2300 kWh/year. The system is 500 MW, or 500,000 kW, so the yearly output is 500,000 x 2300 = 1,150,000,000 kWh, or 1.15 TWh. Now PVWatts is notorious for underestimating, but in this case we can cross-check that using the SDCP’s numbers; they contracted for 150 MW of the station and expect that to produce 460,000 MWh a year, so multiplying by (500/150) gets us 1.53 TWh, so we’re in the ballpark. We’ll go with the PVWatts number as the worse-case scenario.

Ok, now the CAPEX bit. Right now the average price for a one-axis system in the US today is about 95 cents/Wp, so in this case something on the order of $475 million. Here too we see why renewables are doing so well, they’re so scalable that you can easily arrange financing from smaller firms. This lets you shop your rates in a way that the larger projects simply can’t do, and that’s why they end up paying such huge interest rates.

PPAs in the US are generally 15 or 20 years, and in this case the one SDCP one is 15 years. This is also a worst-case because it means we have less time to pay off. So using the “simple calc”, the LCOE is is 2.7 cents. If we use the 35-year average system timeline instead of the PPA, which is what we did above with the 60 year term for HPC, then the number drops to 1.5.

There is something I am ignoring here; unlike a reactor, panels lose about 0.25 to 0.5% of their output each year. This is precisely the same as having a discount rate on the output. Again, taking the worst case at 0.5%, that brings us to 2.8 cents for the 15-year, or about 1.7 for the 35 year case.

OPEX is not so easy. Because PV doesn’t have moving parts, there’s no component of the OPEX that is based on the production – contrast with a coal plant for instance, where if you want to produce more power you need to burn more fuel and do more maintenance on the turbines and such. In 2019 NREL estimated all OPEX at $17/kW capacity, so in this case about $7.5 million a year for Oberon. Using the non-inflation number that’s $113 million over the lifetime. So let’s add it, and that gets 3.2 cents for 15 years or 2.9 for the 35.

So even if we ignore all the financing costs and inflation, the bottom line is still below the nuclear option. If we are to consider this in terms of societal benefit, where is the benefit to paying more for the same electrons – don’t get me wrong, there’s all sorts of grid benefits, but that’s nothing like the usage above.

If SMRs can address this, then I think they’ll be competitive, but I simply don’t see how new large designs like EPR are going to be built without significant government intervention. If you use in-market financing rates, about 8% in France of 12.5% in the US for instance, they simply cost too much to attract attention. There appears to be some appetite for this in Europe and the UK, where they continue to explore another reactor, but Canada is done (AECL’s design division is long gone) and the US has instead directed their efforts to keeping old plants going. I suspect older plants will continue to be refurbed until something much faster comes along, and that appears to be what is happening in the market.


1. Joris van Dorp - September 20, 2022

Thanks for reading my article.

My nuclear costs numbers are correct and easy to verify by using the IEA LCOE calculator.


Select nuclear, set the discount rate to whatever inflation rate you think is right and look at the LCOE of GENIII projects.

Maury Markowitz - September 21, 2022

The number on that page is 7.1 euro/kWh for French Gen III. That number is well within the 6.6 to 8.9 cent range using my calculation. How does this verify your 4.1 cent number for the UK?

Joris van Dorp - September 21, 2022

You didn’t adjust the discount rate. The 4,1 cent in my article assumes a discount rate of zero. The IEA calculator won’t let users set the rate to zero, but if you set it to 1%, the number for France is 3.7, which is even lower than my 4.1 cent number for the UK. At 2% discount, the IEA puts French nuclear at 4.1 cent.

Additionally my article contains a screenshot of the UK National Accounting Office cost curve against interest, which also aligns with my numbers.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s