Hacker News Reader

Comment by alex_young | Hacker News Reader

From the article:

  > The only country where the costs of nuclear plant construction seem to have steadily decreased is South Korea:

  > The fact that South Korea is the only country to exhibit this trend has led some experts to speculate that the cost data (which comes directly from the utility and hasn’t been independently audited) has been manipulated and we shouldn’t draw conclusions from it.

View on HN

Koreans excel in cost-efficient construction projects. Korean companies are often considered for best bang-for-buck value when developing countries are interested in big infrastructure projects nowadays.

https://en.wikipedia.org/wiki/1915_%C3%87anakkale_Bridge

https://en.wikipedia.org/wiki/Burj_Khalifa

https://en.wikipedia.org/wiki/Bataan%E2%80%93Cavite_Interlin...

In absolute terms, Japan and India have costs similar to South Korea's.

https://www.sciencedirect.com/science/article/pii/S030142151...

See figure 12 for a quick overview. And from the introduction: "In contrast to the rapid cost escalation that characterized nuclear construction in the United States, we find evidence of much milder cost escalation in many countries, including absolute cost declines in some countries and specific eras. Our new findings suggest that there is no inherent cost escalation trend associated with nuclear technology."

> In absolute terms, Japan and India have costs similar to South Korea

That countries like Japan and Korea have similar absolute costs as India which has a 4-5x lower PPP adjusted GDP per capita suggests that the price for labor for nuclear power plant construction is globally set (relatively few qualified engineers who can demand a high price), or India uses a lot more labor, or a combination of both.

India probably uses a lot of foreign expertise. It’s either the French or the Russians that supply the reactor. That would change in the future I suppose and cost would drop. Also, corruption.

India was under nuclear sanctions until recently and had to build and design everything from scratch. Every component is made in India for PHWR which adds a lot of cost, due to r&d and economies of scale. Also, India is still not a member of NSG, due to repeatedly being blocked by China.

Japan ceased all construction of nuclear reactors in response to Fukushima daiichi, and since that was a decade ago I'm betting that all the competence they built up has disappeared.

In other words, very similar to what happened in the US in response to Three Mile Island: after a scary nuclear incident there was a lengthy pause in nuclear construction which meant that all of the skills and learning-by-doing that had accumulated up to that point went away, and starting again would be significantly more expensive and subject to massive schedule and cost overruns.

The government that was in power from 2017 to 2022 put a moratorium on new reactor construction there and promised a full phase-out, and although the newly-elected government promised to reverse this, it likely has done some damage to S. Korea's civilian nuclear capabilities.

If anything this seems to support the position that the only way to reduce costs is to increase volume. A classic economies of scale example. Instead the experts want to disregard the data for vague reasons.

South Korea, Japan, and India all have similar costs which suggests South Korea isn’t benefiting significantly from continuous construction.

Economies of scale generally exist, but it’s not magic. A large fraction of construction costs for nuclear power plants is very similar to other structures. A high pressure steam pipe is a high pressure steam pipe and people are constantly building structures using them.

also, quantity is the (primary) independent variable in economies of scale, and at quantities of dozens for nuclear plants, you can't get much economies, as opposed to when quantities are in the many thousands/millions.

The cost of flying has come down by 50% since 1980, and while an airplane is a simpler machine than a nuclear plant, the two industries also have a lot in common (such as the perception of risk not being in their favour).

By doing international standardization and coordination in ways similar to the aircraft industry, the same should be possible for the nuclear power industry.

It should be possible to consolidate most of global production down to a handful of companies (like Boing and Airbus), with a forest of subcontractors in the same way that was done for airliners, and achieve similar economies of scale.

Successful designs could be re-used over a period of 20 years or more, with only minor modernizations of things like electronics, like the Airbus A320 or Boing 747.

Ideally, we should have done this in 1980, but even if we start today, nuclear can provide a lot of energy at very competitive prices in the next 60-100 years. By then, we should have fusion or the ability to build energy storage cheeply enough to make renewables (probably solar) competitive.

By the way, this blog (Construction Physics) is about why construction in general (not nuclear power construction in particular) is expensive. One big part is that construction is done on site, and site-to-site variation hurts standardization and economies of scale.

Finished airplanes can transport itself by flying. This advantage is particular to aircraft industry and probably can't be copied by other industries. Finished buildings can't transport itself.

According to the original article, only 16.5% of the costs of a nuclear plant is from the buildings, though.

Most are things like reactor equipment, turbine equipment and electrical equipment. I'd bet most of that could be built in a central factory, and then shipped to the site.

In fact, when comparing to air travel costs, the reactor buildings can be compared to the airports.

Most costs are actually wholly-legal corruption, charged to various other accounts.

Ideally, we could have started building out solar in 1989, and would have had today's PV prices by the '90s, and would be well along to mitigating the climate catastrophe now. But we chose otherwise.

Starting nuke construction now would just be kicking the can down the road again.

How much of the development in PV tech is linked to developments in other technologies, and how much is due to dedicated R&D on PV's? My personal understanding is that basic science is the main long-term driver, while directed R&D into a specific field is important for short-term gains in efficiency, but will tend to reach diminishing returns if it goes beyond what is enabled by basic science.

Economies of scale is more of a fixed saving on top. This may have helped take the cost per kwh of solar from $10 to $3, but the scientific development was what made it possible to go to $0.1.

In any case, PV prices are already very low. We are reaching a point where installation and maintaince are becoming limiting factors, and where storage is becoming the primary bottleneck.

Development in storage (batteries and hydrogen) already has so much momentum that building out some nuclear power would not threaten it.

Solar panels have barely improved since the 80s. What drove down cost, and continues driving down cost, is purely manufacturing volume.

New PV tech, like perovskites, might carry prices down further once the learning curve for Si levels off, but we are not there yet.

There is no "storage bottleneck". We just don't have anywhere near enough renewable generating capacity to justify diverting capital from expanding it to building storage. Diverting capital to boutique nukes, and to coal while we wait a decade or two for the nukes, would be far, far worse than that.

When we do start building out storage, vanishingly little of it will be batteries.

>handful of companies (like Boing and Airbus)

Why yes everyone should consolidate global production of _nuclear power plants_ to a handful of companies probably based in the Western world.

Imagine if there were 200 companies producing airliners instead of a handful? How would the airports know which were safe enough to land?

Imagine if there were 200 operating systems for phones, instead of 2? How would corporations know which were safe enough to install 2-factor auth on?

Imagine if there were 200 suppliers of x86 cpu's, all with small differences in features? How would we build safe software for those?

Consolidation to a few vendors (but more than 1) is a huge benefit for both costs and safety.

And there is absolutely no reason these companies need to be "Western", unless you include Korea, Japan, Taiwan and India in the "Western World".

China would probably have one such company, but would perhaps only be trusted by their allies, just like US companies would probably not be allowed such infra in China or Russia.

https://en.wikipedia.org/wiki/South_Korean_nuclear_scandal

Reply on news.ycombinator.com