Hacker News Reader

Comment by runesofdoom | Hacker News Reader

To quote the late Admiral Rickover, "An academic reactor or reactor plant almost always has the following basic characteristics: (1) It is simple. (2) It is small. (3) It is cheap (4) It is light. (5) It can be built very quickly. (6) It is very flexible in purpose (’omnibus reactor’). (7) Very little development is required. It will use mostly off-the-shelf components. (8) The reactor is in the study phase. It is not being built now.

“On the other hand, a practical reactor plant can be distinguished by the following characteristics: (1) It is being built now. (2) It is behind schedule. (3) It is requiring an immense amount of development on apparently trivial items. Corrosion, in particular, is a problem. (4) It is very expensive. (5) It takes a long time to build because of the engineering development problems. (6) It is large. (7) It is heavy. (8) It is complicated."

From the article, "The first module is expected to be operational by 2029 with full plant operation the following year."

View on HN

In this particular case, what is this quote supposed to imply?

> In this particular case, what is this quote supposed to imply?

I don't know that it's supposed to imply anything but rather shine a light on the different acceptance criteria for a study reactor v. one that generates power for, in his case, mission critical needs (though power for homes and businesses would be life-critical, so it's up there).

If a study reactor breaks, even though there's a possible risk to life depending on how it fails, you can endure the downtime for a bit while bringing it back up. If a practical reactor breaks, people are far more likely to die.

That's why it's important to have a multitude of energy sources on a competitive power grid. The nice thing about SMRs is you can build enough of them to make the overall output consistent as needed, even with a higher tolerance for failure of any individual reactor.

A higher tolerance for random failure, but extreme exposure to expensive repairs for systematic failure.

Anything $M you have to fix in N of them costs Nx$M.

A study reactor is also designed to break. It is meant to be easily opened up and examined after a fault. It is meant to validate the core process but also explore edge case situations. So it needs to fail gracefully. That is a more complex beast than any final product.

That small, modular, cheap reactors have been promised many times for decades and have never panned out.

We use them in nuclear submarines and aircraft carriers, so the tech seems feasible - not sure why commercial use for small reactors is such a difficult stretch.

Nuclear submarines cost on the order of $6B a pop, so I'd hesitate to describe anything about them as "cheap". We use nuclear reactors in them because there's no other power source that's as compact and self-contained.

Plus worst case for a nuclear submarine is that you flood the reactor with sea water if you have a loss of cooling.

Surely there are worse cases for a sub than this?

The idea is that you are not going to have a meltdown even in a badly malfunctioning reactor.

Is it? It's not at all difficult to stock enough seawater to flood a terrestrial plant. What's distinguishing the submarine?

Smaller size than commercial power station reactors, easier to keep cool.

This contradicts the comment I responded to, though. The scenario is that we've already failed to keep it cool.

There is less of it to cool.

A submarine (when submerged) has a large body of water for heat to radiate out into.

A (previous generation) terrestrial plant already has systems that can dump ludicrous amounts of water onto the core, but they all rely on there being power to operate pumps. A nuclear submarine that can dump in sea water does not need this - the body of water it is in is cold enough (relative to the heat in the core) that it will naturally carry heat away.

Wikipedia says a 688 costs less than $2B in 2021 dollars. And that’s all in for the submarine on top of the reactor.

Sub and ship reactors are effectively sealed units that are not intended to be refueled on a regular basis. They can be serviced and refueled eventually, but it is a major overhaul to do so.

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

The problem with a small reactor is that there's a lot of per reactor cost. Things like security, communicating with the grid, and paperwork don't get cheaper with a smaller reactor so the percent of cost going to overhead goes up.

The exception proves the rule. If we have them already for subs, but still don’t have them on land then they must be bad ideas when cost matters.

don't they normally run on highly enriched fuel?

I'm not sure there we currently have the supply chain, security and ability to deal with geopolitical fallout of using the same fuel for mass domestic use.

Naval reactors usually run on weapons grade nuclear fuel, but not always.

The French K15 reactor runs on 7% enriched fuel.

It is not built yet. So it is theoretical. ... And Academic.

In that case Rickover's observations don't apply at all:

(1) It is simple -> nobody said NuScale's reactor is simple. It took NRC about 6 years to approve it, and in the process NuScale had to produce about half a million pages of documentation

(2) It is small -> not really. It is smaller than full size reactors, but then it delivers only 50 MW, not 1GW. Per unit of electricity delivered, it is most likely somewhat larger than a full size nuclear power plant

(3) It is cheap -> relative to what? it's not that cheap. If anything, see the comments in this thread, it appears to be expensive

(4) It is light -> this does not apply here. Rickover was concerned with submarines, where weight was important. This is not a reactor designed for submarines. I don't know how light it is, but nobody cares about this

(5) It can be built very quickly -> well, if the first one is supposed to get online in 2029, that does not seem to be very quick, does it?

(6) It is very flexible in purpose (’omnibus reactor’) -> NuScale's reactor is designed to generate electricity. That's it. What is flexibility in purpose?

(7) Very little development is required -> NuScale has already worked for one decade on this. It will take until the end of this decade to see one come online. Nobody claimed "very little development is required" It will use mostly off-the-shelf components -> not really. NuScale will use a Korean manufacturer that is accredited by NRC to manufacture componenets for nuclear reactors. There's nothing off-the-shelf about this.

(8) The reactor is in the study phase. It is not being built now. -> it depends what "being built now" means. NuScale can't start building before it has all the approvals. It is working on getting these approvals, if this counts as "being built now", then it's being built now.

More to the point. Rickover was talking about a completely different context. People venturing cheap ideas, while he needed concrete reactors for his submarines. We are in a different world. NRC is extraordinarily stringent. The fact that NuScale got their approval is a phenomenal achievement. This should not be dismissed with the same tired old quote from Rickover that gets posted on HN almost every time we talk about nuclear reactors.

> What is flexibility in purpose?

Reactors can generate electricity, generate fission products (for biomedical uses, for physics uses, for weapons, etc), or both

Not really applicable here though

Ideas of the top of my head:

* Base-load vs turn on during peaks

* Simplicity vs shorter bring-up time

* Maximization of generation capacity vs efficiency, for a given volume of the plant

etc.

Reality is harder than fiction. And people are bad at maintaining pipes.

I'll answer that with another quote:

"In theory there is no difference between theory and practice, in practice there is."

Why the flippant quote?

It's not flippant at all. It is the essence of the much longer version and the real lesson contained therein, you asked a question and I answered it. If you see it as flippant then that might reflect on you. Your other comment further down in this thread is flippant, even though it is a much longer one: you reject out of hand what is contained within these words without taking the time to ponder how they do apply to the matter at hand.

> If you see it as flippant then that might reflect on you.

It does reflect on me. It reflects the fact that I can't read minds.

You gave a widely known quote, and in your mind it was crystal clear what you meant. But to other people, who can't read your mind, it just sound flippant.

It sounds like you think NuScale is a bunch of theoreticians. And that with your remark you are trying to put down their decade-long quest to achieve something.

Maybe it does not sound flippant to you, but I assure you, it is flippant.

The message, to expand on it is that they are in the earlier stages of their development and the safe bet is that by the time it is all said and done things will be much more in line with what we've come to expect from the nuclear industry, including any and all of the statements already in line with that today.

So by the time it is done it will be more expensive, likely heavier, likely way late, more complex and narrower in its possible range of applications. It's not a law, it is an observation made over many nuclear deployments and to the best of my knowledge there isn't a single project that was an exception so I expect it to be true this time around as well. Which is why you can safely ignore any of the touted advantages until the product is ready to be fielded in quantity. Assuming it ever will be fielded in quantity, plenty of designs were slated for large numbers of deployment and ended up being one-offs or at best single digit runs because of unforeseen issues with the design.

Fair enough.

But although the most dangerous words in history are "this time is different", there are reasons to believe things could be different this time:

1. the Nuclear Regulatory Commission. NuScale got the design approval from them. That means this design is as good as frozen. The NRC is an extraordinarily conservative organization. If NuScale, or anyone who licenses their technology, will try to diverge from the approved design by a bit, ..., well that's not even a possibility, why should we think of consequences. So this design, by the time is in production, won't be heavier, more complex, or narrower in scope, because it will be this exact design. It may be more expensive, late, in fewer numbers than expected, however.

2. Conservativeness of the design. This reactor is just a pressurized water reactor design. The most widely used today. It's in a small form factor and modular, but otherwise there's nothing revolutionary or radical about it.

3. History. After about 70 years of reactor operation, the NRC has seen lots and lots of failure modes. The number of new failure modes to be discovered is not zero for sure, but it asymptotically approaches zero. The reason for a lot of past construction delays was the moving of the goalposts by the NRC, but that was due to the discovery of new failure modes, not to any malicious intent. The goalposts may still move between now and 2029, but much less.

4. Politics. In the past few decades, at least in the US, the left was against nuclear power, and the right was for it. Now it appears a large number on the left have embraced nuclear power, so some form of bipartisanship has been achieved. This is quite unprecedented. Will it last? Probably not forever, but it might last a few election cycles, and this could be enough for this SMR design to achieve some escape velocity.

Sure, but having the design frozen is one reason why this may never happen in the first place, after all any kind of deviation and it is back to a recertification and that gets expensive quickly. No such thing as 'move fast and break stuff' in the nuclear power business, your design works or you're history.

And as you tacitly admit modular is revolutionary and there are a whole pile of challenges right there, having a single (large) point of failure may well be much easier to manage than several smaller ones, I never saw the modularity as the problem to solve, because existing (large) reactors are more often than not also created in sets to deal with downtime due to maintenance and overhauls.

Your third point is accurate, but also may end up being this projects downfall: by changing some core design parameters (such as size) new failure modes may well pop up. MTBF is mostly a function of the number of components, if you increase the number of components, even if the components are smaller you will probably still see more failures.

Four is true, but: here in NL we currently have the same situation and a new reactor was just greenlit (two, actually if I understood all of the material right). But this is a decade+ project and I suspect that given the fickle nature of these projects that there is a fair chance that it will eventually never see the light of day because of the dark horse of economics: a typical nuclear plant will generate power, but it will never generate any money. Without subsidies these are essentially dead ends.

Our new nuclear reactor here in Georgia (Plant Votgle unit 3), the first new one in 29 (?) years, is a great example.

Cost and time overruns galore. It's going to end up costing almost double what they projected.

The time overrun is attributed to Westinghouse's bankruptcy, but Westinghouse's bankruptcy was also tied to trying to take on this project.

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

I wonder what causes the cost projections to be so far off. Not enough time or research into what the final structure will actually cost? Overly optimistic projections to help ensure the project is approved and started, thus getting the "foot in the door" to complete it?

It's rule number one of the big physics playbook. Make outrageous claims that funders can't verify cause they ain't good at math. Then let the good times roll!

See the LHC, ITER, James Webb, F35, quantum computing etc.

I think the biggest cost is compliance. Paranoia is the maximum after the various incidents.

>"I think the biggest cost is compliance." Bingo!

The big problem is the lack of knowlage and experienced. Simply put almost nobody has experience with building reactors, and specially not that one. And outside of it being nuclear, its also an incredible large civil engineering project with incredibly high specification. The US is not exactly known for being great at executing large civil engineering projects.

The reality is if you want cheap nuclear you need to mass produce it, just like with everything else. But that requires large scale state action and planning.

Or alternatively having a competitive market for smaller nuclear. But for that to happen regulatory approval processes and many other problems are in the way.

After SpaceX showed how it is possible to build rockets much cheaper than previous ones I attribute the size and complexity of current reactors to outdated mentality by their designers

Setting goals like "maximum efficiency" and trying to bring cost per MWh too low is a self-defeating exercise

Example: designs that require active cooling.

Kind of a fraught comparison. SpaceX intentionally learned by blowing rockets up, figuring out what went wrong, fixing, and trying again.

The same learning process applied to nuclear reactors is problematic at best.

You obviously don't need to blow up reactors to learn stuff

Build, prototype and see how parts react (for example, corrosion) way before an emergency happens. Build a first version, then improve on a second version. Iterations will work better with smaller reactors than with bigger ones

And you do need some research into materials, but that's part of the process and evolution.

> You obviously don't need to blow up reactors to learn stuff

Some people do. Did, even.

If only Coal plant accidents were taken with such knee-jerk reaction

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

We have seven+ decades of experience now. We are certain of two things. (1) Price of reactor n+1 is not less than n's; (2) every number produced by the nuke industry is a lie.

What nonsense. We literally have great evidence that n+1 is n, this is well document and claiming otherwise is nonsense.

Rockets blow up because that's what rockets do. You can absolutely build reasonably safe test reactors to verify ideas. That is what they did constantly in the 60s, with multiple new reactors every year.

The whole Molten Salt reactor experiment was something like 70 million $ for an amazing project.

And dispute that somehow the world didn't blow up and there are no mutant ninja turtles anywhere.

What helped SpaceX out was the COTS and CRS program, and something like that could absolutely be world changing in nuclear, but nobody at the DoE has that kind of vision sadly. In fact the mastermind behind COTS has been advocating for something like that for many years now.

this could be said about computers back in the day too. they used to take up entire floors, now they fit on your pocket. i suspect something similar applies for tech in general and reactors also

Cars are mature tech. Can you fit one in your pocket?

Yeah. RC cars can fit in your pockets. But you can't be the passengers.

Sure, you can buy a 1:76 scale electric car for around $80.

In other words, no.

What is a car? Is it 4 wheels, a driver seat, passenger seat, three read seats, and a trunk? Or is it an efficient means for getting from point A to B on paved roadways? If a "car" is the latter, then there have been plenty of advancements. Electric scooters and bikes are two pieces of tech you can very easily take with you.

Your car analogy doesn't really work. It's like saying "desktop computers are mature tech, can I fit one in my pocket?" No, but you can fit a smartphone in your pocket, which does many of the same things a desktop computer can do.

Oof, reactor in my pocket is a tasty prospect. Personal computing and personal transport could get super exciting. Basically a grown up way of saying Iron Man suits one day...

Keeping an ionizing radiation source in one's front pocket could improve the gene pool.

The chances of that being an improvement are extremely slim at best.

Individuals willing to do it self-selecting against contributing to the next generation could be a great boon.

Nuclear reactors are almost as old as computer, yet we have seen nothing like the scaling we have in the semiconductor industry. Part of this is because they half of the power plant is essentially a steam turbine just like in most other thermal power plants. That technology is essentially centuries old. Not much scaling to be done there.

If we put a fraction of R&D into reactors as we have with semiconductors, it would be a VERY different story. Instead we have spent 50 years scared shitless of the nuclear boogyman. The worst proponents of nuclear are so called environmentalists who's feelings over facts have caused far more harm to the environment (mining for rare earths anyone?). If we would have continued research into other than light water reactors instead of abandoning it we'd be burning "nuclear waste" instead of arguing where to bury otherwise useable fuel.

Fission gets billions of dollars of free public money for research every year and had much more spent in the past.

> The worst proponents of nuclear are so called environmentalists who's feelings over facts have caused far more harm to the environment (mining for rare earths anyone?).

Modern PV involves less rare earths than a nuclear reactor for the same energy output. DFIG wind generators use none. New batteries use none.

Uranium mining on the other hand poisons thousands of km^2, and is almost never properly remediated. Tailings dams of mines that ran out decades ago still need constant maintenance (often on the public dime) to not spew heavy metal laden dust everywhere.

> If we would have continued research into other than light water reactors instead of abandoning it we'd be burning "nuclear waste" instead of arguing where to bury otherwise useable fuel.

Trillions were spent on 'burning nuclear waste' by which you actually mean creating larger quantities of Pu240 and Pu241 (the worst high level waste) than a burner reactor does and dumping fission products into the ocean. It went nowhere.

The US has tons of money saved up to handle nuclear waste, all reactor operators have been paying for 5+ decades, but because of grid lock the money is just pilling up. Instead of making constructive policy the idiots dead-locked themselves with this idiocy at Yucca Mountain.

And there has never been a real program to systematically use the civilian nuclear waste. This was of course planned but the program was stopped in the general anti-nuclear attitude of the 70s. This program would have lead to a huge amount of green energy and a reduction in the nuclear waste.

And the uranium mining would be much less of an issue if we had modern reactors. We are operating at 2% efficiency, a thermal breeder could reach high 90%s. The existing civilian nuclear waste could be enough for a long time. No more mining. And thorium is a waste product of current rare earth mining in absurd quantities. So really any more mining that we do is just because of bad policy.

And when talking about mining, the reality is solar and wind need far, far more mining and those mins have tailings too.

> Fission gets billions of dollars of free public money for research every year and had much more spent in the past.

And nuclear was excluded from almost all green energy subsidies for decades a far larger number in total. Even more important the changes in the grid pricing basically simply did not favor long term base-load investments. Not to mention all the programs that force utilities to adopt a % of renewables.

And even today, the places that have stable reactors have low electricity prices because of long term price agreements while everybody else is suffering.

France had decades of low energy prices and even today most in France have low energy prices. And the cost of that was mostly on the books of the utility and has been paid down for many decades now even with the low energy prices.

France has sadly mismanaged its fleet for the last 30 years, with the anti-nuclear state abusing the amazing systems the generation before built for them. And despite all of that abuse and lack of maintenance and investment France still has reliable, low price, green energy.

The idea that environmentalists had any influence on IFR is a complete fantasy or that pouring another half a trillion into failed attempts at separation and reprocessing programs would magically make them clean or cheap is laughable. If you want to lament the mismanagement of funding for green energy, then lament that large scale wind was abandoned in the early 40s because the very first attempt was a little bit more expensive than coal or that the first ten billion dollars moving down the learning curve for solar to make it viable for mass market took 5 decades.

Isn't there load and loads more government regulation and control on nuclear reactors compared with semiconductors?

Reply on news.ycombinator.com