Hacker News Reader

Comment by rbanffy | Hacker News Reader

Right. We should be afraid of people who want to cut corners with nuclear power.

Also, building a reactor takes a long time when all the precautions are taken. Building hundreds simultaneously is begging for precautions to be dropped or mismanaged.

View on HN

At this point we have some proven designs that make things a little easier to get right because we're not building from scratch. But a big part of the problem now is lack of expertise. We don't have enough trained hands to build these sorts of specialized systems, or keep them running once built.

We need to bootstrap our way out of this problem. Appropriate education programs should be created and subsidized. The aging workforce that still exists should be paid well to put off retirement or hired out of retirement to help implement things like apprenticeships. In five years we could have a new generation of workers with an established pipeline to continue training more people.

At this point I'm skeptical of these designs. How recent they are and how many units are in operation? How long have they operated? Did we see changes in projected useful life?

CANDU reactors are doing pretty good in Ontario

CANDU reactors are a moribund technology. The attempt to sell new ones to Ontario ten years ago came in too expensive by a factor of 3. No one is buying them.

I'd expect the opposite: The more we build the better, cheaper and safer they become.

That's what happened with cars and airplanes for example. I'm not aware of counterexamples.

They didn’t build all the cars they would need concurrently; they mostly finished a small number at a time. That let them learn along the way, but there isn’t time to do that for nuclear because we need hundreds or thousands of plants and they each take years to build.

It doesn't take 20 years, billions of dollars, and thousands of specialized workers to build each car.

Counterexample: Large infrastructure projects that take >10 years to build. Big Dig, trains, et cetera. They take so long to build that people only work on a very small number during their career so the ability to learn lessons from one to apply to another is small.

Nuclear plants are in this category.

They're only in this category specifically because each one is a bespoke custom project. That's not a compelling argument against standardization and mass factory production.

Even the NuScale SMR reactors only produce the core in a factory setting. They still need a large bespoke building and containment vessel.

If you work on years 1-2 of Project 1, shouldn't that put you in a good position to work on years 1-2 of Project 2? You might not see the effects of bad decisions that start mattering in year 6, but it seems there should be lots of workers for whom that doesn't matter, and for the rest, it seems like having some people do close inspections and produce good reports (which others read when they're on Project 3+) might mostly cover that. Seems like this should be mostly solvable...

China has learned to build those projects in 6-12 months.

Even China has huge problems with nuclear. When they took the French EPR design, they thought each would take less than 4 years to build. Ended up taking 9 years.

https://en.wikipedia.org/wiki/EPR_(nuclear_reactor)#Taishan_...

Western countries aim for a decade of construction, but will probably take 2x as long as the initial schedule.

Nuclear is itself a counterexample. It didn't get cheaper as we built more.

My take is we never built a lot.

The industry died from bad PR before it got to mature much.

It was grossly overhyped by a government trying to make trillions (current dollars) spent on nuclear weaponry not seem such an egregious drain on humanity. That it would crash back to Earth as the hype and groupthink met reality was inevitable.

That’s not my understanding. That’s precisely the benefit France enjoyed when it decided to scale up its nuclear power. Right now very few new reactors are built so the benefit and the expertise are gone. But if a country decides to scale up rapidly in the same way you should see that benefit.

France didn't see good learning effects either.

Uninformed conjecture here, but, should nuclear power be a private industry? Where the incentive is profit?

Could it be a Government Utility, where the incentive can be safety above all else?

Will it be cost effective then? It's already pretty marginal.

How expensive are highways?

I thought that the costs come mostly from insurance? (Which the public is already on the hook for to some degree)

There are an lot of industry where they are not cost effective without heavy government involvement, and even then the public barely get to reap the rewards.

(Though I supposed if this happened in the USA it would be the death knell of the fossil fuel industry)

What about the opportunity cost of having a planet to live on?

You can have that with renewables, at a much lower cost.

Different incentive structure in Chernobyl; still fucked up.

Given a large enough number of chances, people will fuck up. The fallout from nuclear fuckups is just too big to risk it.

And the number of opportunities to fuck up is mindblowingly high. A lot of nuclear accidents start with accidents on support systems that cascade into other systems up to the point the reactor can't be safely operated.

Interesting question, how many nuclear plants can be built at once in the world. I imagine there are not that many outfits with the experience to take on that kind of job.

> Building hundreds simultaneously is begging for precautions to be dropped or mismanaged.

So playing with people's fears again?

That's not some kind of abstract fear, even without doing it at such a mass-scale there have been and are more than enough examples of cutting corners [0]

The nuclear industry, just like every other industry on this planet, is run by flawed and corruptible humans. The problem is that their flaws and corruption can do damage for many generations to come, so being a bit careful and skeptical is only due diligence.

[0] https://www.powermag.com/frances-nuclear-storm-many-power-pl...

Reply on news.ycombinator.com