Hacker News Reader

Comment by OrwellianChild | Hacker News Reader

As much as I'd love more nuclear power on the grid, this project cost over double its initial bid and is actually raising rates for GA utility customers... I can't tell from the story whether "nuclear is hard" or if "contractors are idiots" here. Reactor was supposed to open in 2016 - it's very hard to cheer for a 7-year delay and a 140% cost overrun.

View on HN

Basically, you can look at [the list of nuclear plants under construction](https://world-nuclear.org/information-library/current-and-fu...) and see that the Russians VVER series are still going up around the word (over 20), the Chinese are building a bunch (of many different international and domestic designs, I don't think they've built one for another country yet), the South Koreans have 4 under construction around the world, there are 4 AP1000 (two in China, two in Georgia) and 6 EPR (2 in China, 1 in France, 1 in Finland, 2 in UK). (Plus a couple of oddities like Argentina and India building a new domestic design each.)

I think an under-appreciated point is that if you do a pause on nuclear construction, say because of the Fukushima Dai Ichi or Three Mile Island or whatever, you lose the skills to deliver them on time and quickly very fast, and it actually seems to be harder to rebuild than it was to build the first time. My suspicion is that the first time your standards are lower, and so you don't design as aggressively and your workmen deliver okay workmanship. Then you pause for a while, all your experience scatters to the wind, but you need to deliver next generation performance, and you just don't have the industrial base to actually build it on time and under budget.

This doesn't seem to be a just-America problem: France alone built ~60 reactors in the 1970's and 1980's and Germany and the UK built more, but the EPR has been just as big as disaster as the AP1000, because they didn't build new reactors in the 1990's and that experience withered away, but are still trying to deliver that next generation performance.

The overall number of operating reactors doesn't go up much though: https://www.statista.com/statistics/263945/number-of-nuclear...

Supplied energy seems to have leveled out also: https://world-nuclear.org/information-library/current-and-fu...

Time to say goodbye and maybe think hard why those Russian and Chinese reactors come up so fast and cheap, and prepare for the fallout...

now do the list of plants slated for decommissioning

We lost much of our nuclear engineering capability in the US, so in many ways contractors are "idiots" here and made many costly mistakes during construction. Many articles on this and how it lead to the Westinghouse bankruptcy. Sadly, there's only a handful of western-friendly firms in the world that can build a reactor, so hoping the expertise and lessons learned will be worthwhile for future projects.

Nuclear isn't necessarily the cheapest option for rates, but it's the best for base generation AND greenhouse gas emission.

This "loss of nuclear engineering capability" doesn't hold up when you see reactors build by the French having the same problems: overdue and over budget.

The French never stopped building reactors all over the world.

"Base generation" is an outdated concept. It can be replaced by storage and grid size for example.

They did stop building them in France though, and they switched technology to EPR

French engineers were still allowed to travel abroad and work on other French projects which were overdue and over budget.

I thought "overdue and over budget" is natural state of any big project

No, it's not.

But this generalization comes always up if someone wants to sell a project which they knew will be overdue and over budget.

Do you have any examples of other energy plants that weren't nuclear, were consistently under budget and delivered before the due date?

It's entirely routine in the natural gas world;

https://www.businesswire.com/news/home/20170518005288/en/Pan...

> After the installation of 1.4 million linear feet of electrical cable, 113,161 feet of pipe, 890 tons of steel and expending more than 1.6 million man hours, Panda Power Funds’ 778-megawatt “Stonewall” generating station has achieved commercial operations. The Loudoun County, Virginia power facility was finished ahead of schedule and under budget, and initial tests show the plant is exceeding performance guarantees for both power output and efficiency. The plant is capable of continuously supplying the power needs of up to 778,000 homes in the Northern Virginia/District of Columbia metropolitan area.

Lest you believe the PR page after the fact -- They fundraised in 2013/2014, started site prep in November 2014, broke ground in August 2015 with a target commercial launch date of June 1, 2017 and Bechtel finished commissioning in May 2017.

https://web.archive.org/web/20151015014525/http://www.power-...

https://www.power-technology.com/projects/stonewall-power-pr...

The French have the same problems of loss of engineering capability. They built many reactors of the same design during the Messmer plan, but haven't had to build new reactors in a long time. The fact that nuclear plants are so long-lived is a huge strength: they are among the cheapest forms of green energy if they are allowed to serve the entirety of their service life. But because they last so long, if you build a bunch of them there's no reason to build more for half a century.

Weird that the French somehow missed all those basic facts some random person can come up on the internet when they presented their designs in recent years and started building those massively over budget and overdue projects, eh?

> they are among the cheapest forms of green energy

The only interpretation of "cheap" which makes nuclear "cheapest" in something is the one where you ignore everything related to nuclear waste. From reprocessing over storage to decommissioning. Making it a lie, basically. Just like the "green" in this greenwashing.

What facts are the French missing? The point is that building 48 reactors of the same design (like in the Messmer plan) is cheaper on a per-unit basis than building 2 reactors of a new design. The learnings from the first few builds inform subsequent iterations, and these learnings decay over time as plants haven't been built in decades. I'm not sure what you're referring to here.

Nuclear waste disposal is not a huge portion of nuclear power's cost, it accounts for about 10% [1]. Nuclear plant builders have to finance the cost of disposal upfront. Decommissioning is even less - less than one percent if the plant serves its full service life.

1. https://world-nuclear.org/information-library/economic-aspec....

They had reduced costs within each generation but the costs between generations increased wildly.

They never saw any kinds of economics of scale.

Have a look at Figure 25 here and add Flamanville 3 waaaay beyond the end of the scale at $12 000/kWe

https://www.oecd-nea.org/upload/docs/application/pdf/2020-07...

https://en.m.wikipedia.org/wiki/Flamanville_Nuclear_Power_Pl...

You just described an economy of scale: costs within generations went down. The more plants in the same generation the lower the cost.

Flamanville is the first EPR France has built, and this it's not leveraging the intra-generational learnings.

Take the 20% cost reduction the last two generations lead to. That would change Flamanville 3 from exorbitantly expensive to exorbitantly expensive.

The French nuclear program as a whole has had a huge negative learning curve. The more they've built and learned the more expensive they have become. Sure you can brush of 20% with a lot of money down the drain, but that does not make it in any way economical.

At the end of the day, the cost of reactors built when there was large scale serial production was considerably cheaper than they are now: https://www.sciencedirect.com/science/article/pii/S030142151...

The "negative learning curve" both in France and in the USA happened as the pace of reactor construction slowed. The "negative learning curve" you're referring to is actually the loss of the economies of scale.

We had economies of scale in the 50s when it was the cheapest? Or was it because we had this attitude towards nuclear engineering?

https://en.wikipedia.org/wiki/Demon_core#Second_incident

It was in the 1960s and early 70s that it was cheapest. And the Demon Core incident was part of the nuclear weapons program, not nuclear electric power - this is a common pattern in misleading anti-nuclear rhetoric.

Your paper shows a very nearly monotonic increase in costs with no dip at any pont between the peak of construction and when the reactors started during that peak were all finished. It also shows that in the US at least, the negative learning rate was much steeper between the end of the turnkey error and both TMI and the construction peak. It's also not normalised for availability factor or upgrades/reliability retrofits.

You can't just post evidence of the opposite to your claim and then repeat it.

France built most of their reactors to one design in the 1980s. That worked out well, but those reactors are nearing end of life. More recently, reactor projects in France have been taking far too long and going far over budget.[1]

You have to make multiples of something to get production efficiencies, and you have to keep building on a regular basis to keep the production line alive. Newport News Shipbuilding and Dry Dock was once so fed up with the Navy's intermittent aircraft carrier orders that they said that if Congress would order two at once, they'd throw in a third carrier for free.

[1] https://www.popularmechanics.com/science/energy/a33499619/fr...

The contractors aren't idiots. They got an extra 7 years and a 140% bonus.

We're assuming cost-plus contract terms, then, I assume? :-)

And don’t forget, this was the “revolutionary” AP1 design that was supposed to fix all the bad things about the previous revolutionary designs.

I don't see how that's related.

I doubt that much - if any - of the cost overruns and delays are related to the nuclear portion of this public works project. Governments and their contractors at all levels are not known for their ability to accurately forecast these things, and they're generally not incentivized to care during the work order, as they're not held accountable.

Jacksonville, FL’a JEA (public utility company) lost a lot of money investing in this plant as well

Are there any infrastructure projects that don't have cost overrun or finish on time in this country?

A lot of it seems to come down to very, very agreesive regulation. Eg, plant designs got approved, work started, then the regulators came back and said "actually, you need to make some major changes to the design", so everything paused and some completed work was torn up, new designs were made, approved (slowly), and then work restarted from scratch. Then it happened again. And again.

The changes weren't bad per se; in one case the regulators decided that the plant needed to be resistant to aircraft impacts; in another that it needed to be more resistant to tornados. But when you have a fantastically complex and expensive engineering project, going back and requiring retroactive changes after approval and after bg has already started is going to blow the budget to hell. And it did.

From the Federal Register notice for the aircraft impact change:

> In making these additions, the NRC is making it clear that the requirements are not meant to apply to current or future operating license applications for which construction permits were issued before the effective date of this final rule. This is because existing construction permits are likely to involve designs which are essentially complete and may involve sites where construction has already taken place. Applying the final rule to operating license applications for which there are existing construction permits could result in an unwarranted financial burden to change a design for a plant that is partially constructed. Such a financial burden is not justifiable in light of the fact that the NRC considers the events to which the aircraft impact rule is directed to be beyond-design-basis events and compliance with the rule is not needed for adequate protection to public health and safety or common defense and security.

That sounds pretty reasonable. As the US government said, applying this rule where existing construction permists exist could result in "unwarranted financial burden" which is "not justifiable" and because the rule is "not needed for adequate protection". It's just a "hey, if you haven't started building yet, why not make it extra super duper safe?" rule. And why not!

But the Georgia plant, at this point, didn't technically have "construction permits", but it did have certified designs, and signed firm contracts for engineering, procurement, and construction, had already submitted its certified costs and schedules to state regulators, and begun manufacturing long lead time components. The cost impact was enormous, but they were still required to go through it.

This regulatory environment is excellent for ensuring that nothing dangerous gets built, because nothing will get built. Is that wise? I suspect not.

See:

https://www.ans.org/news/article-1646/root-cause-of-vogtle-a...

https://www.reuters.com/article/us-toshiba-accounting-westin...

https://atomicinsights.com/nrcs-imposition-of-aircraft-impac...

Really excellent context - thank you for sharing all this detail and citation!

All of this was a great explanation of some of the delays so far...nine years ago.

https://www.ans.org/news/article-1646/root-cause-of-vogtle-a...

Reply on news.ycombinator.com