Hacker News new | past | comments | ask | show | jobs | submit login
Open source Molten salt nuclear reactor design (github.com)
317 points by iso-8859-1 65 days ago | hide | past | web | favorite | 88 comments

I'm a reactor designer and I love the idea of open-source reactor design. It's a wonderful way to add some longevity beyond typical congressional funding cycles to these long-term reactor projects. The national labs may have 2 years of funding on gas-cooled reactors, but then mothball those and move to sodium-cooled reactors, and so forth. The lack of continuity is leads to lack of momentum which leads to stagnation. People even forget stuff and have to rediscover it years later.

Also the nuclear industry is small and struggling. Open-source may enable collaboration on a new level that could be really exciting.

Th problem so far is that the nuclear industry is really secretive. We like to think we have the best design, and all other designs are goofy. We like to patent things and protect secrets. The nuclear industry started in secrecy, developed in secrecy, and still has a lot of it.

But it's struggling so much with cheap natural gas and popular intermittent renewables that it's kind of a do or die situation. Open source is a brilliant step.

All that said, reactor projects don't run themselves. Transatomic shut down and open-sourced their design. For it to go anywhere, it needs dozens of well-funded or otherwise highly engaged people managing the project. Nuclear builds are about as complex as projects come and simply open sourcing it won't go far. So the next step here is for the nukes to agree on a particular open source design, fill in a business plan, and start going. In any case it's exciting times.

The problem so far is that the nuclear industry is really secretive. We like to think we have the best design, and all other designs are goofy. We like to patent things and protect secrets. The nuclear industry started in secrecy, developed in secrecy, and still has a lot of it.

This is precisely the condition much of the software industry was in, when Stallman came up with Free Software. You could s/nuclear/software/ and the above would have been mostly true. So maybe Open Source is what's needed here.

...the following decade used to be known as the „Golden age of Open Source reactor design“ with tens of thousands of international contributors coming up with clever, proliferation resistant designs. This time lasted until Amazon Reactor Services and Google Nuclear Cloud scooped up the best ideas of those designs (as well as hiring all important contributors) into their proprietary reactor designs, enabling anyone to easily spawn a cloud nuclear reactor worldwide, leaving 90% margin to these companies and leaving early dreams and ambitions in the dust...

> Amazon Reactor Services and Google Nuclear Cloud

The fools. The mad fools.

Near the end of the story, someone figures out a device that uses actual atmospheric clouds as material for fusion chain reactions.

Not totally. The share / guide and even the Ibm os was open source up to a point you can still run Ibm mainframe on pc for those 1970 era.

No one care about software. It is bill gate who start the commericalism and the availability of cheap pc that start it all.

What stallman did was to point to the importance of licensing scheme for the cooperation to work. Linus chose the license to avoid unix and minix issue.

Is this really true? I thought it had more to do with the four freedoms Stallman talks up so much.

Using open source software is nothing like building a reactor based on an open source design...

No but the idea of scientific journals for example is that there's a wider science community which collaboratively iterates.

It's the same thing which has happened for engineered artifacts such as 3d printers.

A world where the nuclear industry has a common best Practice design for each reactor, overseen by a steering committee of experts ala the Linux kernel, could actually lead to some progress and allow building some coherent and verifiable narratives about safety.

This is true by definition - no 2 things are the exact same (otherwise its the same thing), but the same priciples might apply here as well.

> We like to patent things and protect secrets.

This actually seems like a good opportunity to use patents. Something like "we integrate your patented idea in our open-source design as long as you agree in writing to give anyone interested a licence at reasonable terms as specified below". Of course this has to be done carefully but has the potential to finance research and promote the exchange of ideas (exactly what the patent system was originally conceived for).

You should get Involved! I bet they need nuclear designers.

> Th problem so far is that the nuclear industry is really secretive.

Why is that a problem? Do you want to make it easier for countries to figure out how to build nuclear bombs? Generally I'm a big fan of open source and so on, but when it comes to nuclear or dangerous biological agents, I prefer having these things locked away really well.

Reactors are VERY different from weapons. And not all reactors result in weapons grade materials.

Reactors are not bombs, bombs are not reactors.

Reactors can be used to breed nuclear bomb grade material. Building a dirty bomb is trivial if you can build normal bombs. For upgrading that to a nuke the main difficulty is finding nuke grade material.

Nope, this isn't true. First of all there are many different kinds of nuclear reactors, and only certain ones can be used for making weapons-grade material. Second, the isotope separation process for U238 or the implosion mechanism for plutonium are either massive or crazy difficult to pull off, and totally unrelated to reactor design.

A reactor design doesn't get you anywhere meaningfully closer to building a bomb.

A reactor gets you experience handling nuclear components, and if chosen with a bomb in mind also plutonium or other useful isotopes.

There is a real reason Israel bombed Syria as soon as they got convincing evidence Assad was building a secret reactor.

Okay, great. It’s free and out in the open, for people to study and maybe expand and iterate with.

But a directory tree full of PDF files is still less than compelling.

It’s something akin to a public domain screenplay.

Okay, we have a script. Now what?

Wow, I remember lots of buzz on HN about Transatomic a few years ago and did not realize how much they had to backtrack on their initial claims:

- From 75X electricity per ton of uranium all the way down to 2X

- No longer re-using spent fuel

What happened? And are the competing startups using molten salt doing any better?

Transatomic claimed to get as much energy from uranium as a fast reactor, without actually being a fast reactor. I never understood how that could be possible and wasn't surprised when it turned out it wasn't.

Other MSR startups are doing fine. Elysium, Moltex, and Terrapower's MSR actually are fast reactors. Terrestrial Energy and Thorcon aren't making those claims in the first place. Terrestrial Energy especially is making great progress with regulators.

I chuckled at that thought of `git clone energy`

Could someone look up what material is being used for the molten salt plumbing itself please? I believe corrosion of these is one of the key issues that's so far prevented their viability.

There are a number of molten salt solar concentrators in operation. I think materials science has overcome the hurdle.

This could be a good start, but it also seems to be lacking in some critical areas. Existing molten salt reactors have found the molten salt to be more corrosive than expected. This means that components in contact with the flow of molten salt may need extensive engineering for safe operations. Given the history of unexpected leaks it seems there also need to be precautions to avoid and limit leaks and also to respond to leaks when they do occur. On top of all of this one of the subtle problems with nuclear power is the sheer amount of precautions that need to be taken for reasonably safe operation begin to eat away at the efficiency of the whole operation. Just how efficient such a reactor might be over its operational lifetime is an extremely important question that is also not addressed yet.

Looks quite expensive for me, what it does better than basic BWR on small side, and basic PWR on large side?

I'm not an expert in this at all, but a friend of mine from college is; he's working at Argonne National Lab doing liquid sodium research for nuclear reactors--so I know enough as you would talking to a guy over beers about his job.

As said above, the chance of a meltdown is slim as sodium is a solid at room temp, therefore you don't have to worry about it ever boiling off. It would have prevented a Fukushima type meltdown, but sodium is extremely reactive and had the sodium touched water, you would have had huge explosions. Their test loop in Argonne had to go through extreme welding checks to prevent possible humidity from getting in. Salt also has better heat transfer properties than water. Interestingly, research in this area really stagnated, and its only recently gotten attention. The US went all in on water due to the nuclear navy.

That Argonne reactor uses molten sodium, which explodes when it touches water.

The open-sourced design uses molten salt, which is the very stable stuff you probably have in a shaker on your kitchen table.

I just checked and my table salt is solid, but then again it's winter time here.

So is sodium but it'll burn rapidly if you drop it in water. Salt won't even if it's molten (though it can cause a steam explosion just from the heat).

Molten salt is extremely corrosive, though.

That depends on the salt and the metal. I saw a presentation by someone from Elysium, who said that chloride salt (which they're using) corrodes stainless steel less than water does. A mix of salt and water is more corrosive, but molten salt by itself is fine.

FLiBe is more corrosive to steel but there are advanced alloys that withstand it fairly well.

Another approach is to use a modular design that makes it easy to replace parts before they corrode too much. Thorcon and Terrestrial Energy use small reactor cores that get replaced every few years.

Not an expert, but one of the big benefits of molten salt reactors is that they aren't really under pressure - no big poof if containment is breached.

It's also really easy to halt meltdowns - since the fuel and coolant are a big liquid glob, you can have thermal plugs that will melt above a certain temperature and drain the coolant+fuel into tubes small enough that, even when the small tubes are full, the coolant+fuel in the tubes is below critical mass and the reaction stops. It fails safe and is immune to mechanical failure, unlike control rods in current reactor designs that have to be moved in and out.

It’s a good system, but the main problem comes from the same source, the fact that you’re dealing with a pool of molten salt. Maintenance on a reactor vessel is never without challenges, but the problems with servicing a molten salt (or something like a Pb-Bi eutectic) are still a financial and practical roadblock to widespread adoption.

This more than anything else is likely the biggest reason these reactors never took off, they're certainly safer from meltdowns, but now instead of piping water we have to pipe an incredibly caustic molten salt! The challenge hasn't gone away, it's just shifted.

I remember reading about issues with storage of the waste from a molten salt reactor. A problem discovered was the radiation was producing fluorine gas from the metal fluorides.

Not all MSRs use fluoride salts. At least three companies are working on fast reactors using chloride salts.

I'd prefer superheated chlorine gas to superheated fluorine gas, in the same sense that I'd rather have an angry weasel in my pants than a grizzly bear, but...

It's not superheated if it's just slow emissions from waste, as mentioned above.

A pipe leak in the reactor itself doesn't create a lot of gas. It's molten salt at atmospheric pressure. It drips out and solidifies as it cools. An advantage of MSRs is that the troublesome fission products (like iodine, cesium, and strontium) don't leak out as gases, like they do from conventional reactor cores; instead they are chemically bound in the salt.

That’s both an upside (as you’ve described), and a downside because the already challenging environment of molten salt becomes increasingly radioactive making servicing even harder than the chemical and thermal environment of the salt make it. You also have to carefully monitor the salt, and a lot of the probes and other means of monitoring it tend to rapidly degrade in the molten salt.

It’s another issue that feels like the answer involves new materials that have a much longer life in situ, so that the inevitable maintence is infrequent. It’s a very promising technology, but it isn’t mature yet.

Reactor design can also mitigate these problems.

Terrestrial Energy and Thorcon use small sealed reactor cores that get replaced every few years.

Moltex uses a pool design, where everything is immersed from above in a pool of coolant salt, and can be pulled out and replaced as necessary. The actual fuel is isolated in vertical rods.

At least three companies use chloride salts. According to a presenter from Elysium, in the absence of water the chloride salt is less corrosive to stainless steel than water is.

I certainly agree that the technology isn't mature, since we don't have any production reactors yet. But we're making good progress, especially in Canada where regulation of new nuclear technology is more rational than in the U.S.

Canada is such an amazing contrast to the US in that area, true. I wish people here had a better understand of just how necessary nuclear power is if we want to survive to ever reach the hoped-for “clean energy future.”

When your coolant can go up to over 700 °C without gassing then you might also be able to cool the reactor using convection only in emergencies, which would do away with the necessity of emergency coolant pumps.

On the other hand, working with a big pool of radioactive liquid that will literally explode and catch fire if it comes in contact with water and still catches fire if it comes in contact with air may also be ... challenging.

You're thinking of sodium. This reactor uses salt, which is very stable, just like the salt on your kitchen table.

The lack of any sort of driver for chemical explosions is one of the advantages of molten salt reactors over light water reactors, in which the water can split into oxygen and hydrogen and cause explosions (as we saw at Fukushima).

Are you sure that molten salt in water is safe? This guy has beautiful molten salt/water explosions: https://www.youtube.com/watch?v=PDRWQUUUCF0

That's pretty cool but the guy's conclusion is that it's not a chemical explosion, it's just from the water turning to steam. But I'll concede that we shouldn't drop a molten salt reactor in a lake.

I was indeed thinking about the NaK (sodium+potassium) alloy (which is used in some molten salt reactors, but not this one). And indeed this reactor design uses LiF (Lithiumfluoride). However, LiF-based salts have the obvious drawback of their high melting points (pure LiF 840 °C, FLiBe 460 °C).

I looked up NaK and only found solid-fueled fast reactors: https://en.wikipedia.org/wiki/Sodium-potassium_alloy#Coolant

Do you have any examples of liquid-fueled reactors using NaK?

Some MSR companies are working on fast reactors using chloride salts, including Moltex, Elysium, and Terrapower (in a project separate from their better-known sodium-cooled fast reactor).

High temperature is a property of all MSRs I know of and does have some advantages, including better thermodynamic efficiency and usefulness for process heat.

Do they accept pull requests?

Only if your changes are atomic.

I imagine this isn't compatible with this open-source license. https://spdx.org/licenses/BSD-3-Clause-No-Nuclear-License.ht...

That's not an open source license. The 6th defining criteria of an open source license is "No Discrimination Against Fields of Endeavor." (https://opensource.org/osd)

You're absolutely correct. Looks like the SPDX license list contains nonfree licenses as well.

i'm gonna get a used Mac Pro on Ebay, gut it and build one of these inside

from https://en.wikipedia.org/wiki/Transatomic_Power

In 2018 the company announced that it would be winding down and open source its intellectual property.[5] The company discovered that in 2016 it had made errors in its early analysis and realised that the design couldn't consume nuclear waste. [6]

Brilliant idea, I hope your concept flies. Surely its time for real open science to prosper and the pseudo nonsense science to get restricted to political narratives

how liquid is molten salt?

If I’m reading the charts right in https://arxiv.org/pdf/1307.7343.pdf the the salts they are documenting are in the range of whole milk, cold water, and hot water. But there is no better than a 50-50 chance that I’m interpreting that correctly.

Can you send me this as an image file as I am unable to open PDF for some reason. You can use this converter: https://www.coolutils.com/TotalPDFConverter

Don't let the North Koreans or Iranians or any other bad actors see this... It will be nukes for every madman in the world

So it's open source but patent encumbered? How well does that go together?

A basic intent of patents is the "open sourcing" of design. You get a limited-time monopoly on your invention in exchange for publicly publishing how it works so others can use it when your patent expires (or before then others can use it by coming to a licensing agreement).

There could be a movement to give patents a much better name and work as intended for the public good, unlikely as it sounds.

This is based on a casual misunderstanding of the meaning of the term "open source". It does not merely mean, as the name suggests, that the source code is available; rather, it means that fairly broad rights to study, use, copy, and modify the source code are granted to the general public. You can correct your misunderstanding by reading the Open Source Definition, written by the group of people who coined the term "open source," immediately after they coined it, to clarify what they meant: https://opensource.org/osd

Patents, on the contrary, specifically deprive the general public of rights they otherwise enjoy, granting a monopoly on them to the "inventor" (or rather the assignee), in exchange, as you note, for publication. They do, as you say, have one thing in common with open source, but in their principal characteristics, they are precisely the opposite of open source, both in theory and in practice.

The earliest known use of the term open source was literally along these lines. I don't see that the OSI has any right to co-opt a term that was already in use, and has a fairly plain english meaning, to mean something else.

> Caldera Announces Open Source for DOS.

> ...

> Individuals can use OpenDOS source for personal use at no cost.

> Individuals and organizations desiring to commercially redistribute

> Caldera OpenDOS must acquire a license with an associated small fee.

[0] Usage w/ the quoted text http://www.xent.com/FoRK-archive/fall96/0269.html

[1] Source for it being the earliest known use https://hyperlogos.org/article/Who-Invented-Term-Open-Source

[2] Strictly speaking I mean earliest known use related to software, the term open source goes way back to at least 1685 https://books.google.ca/books?id=eY5mAAAAcAAJ&printsec=front...

There are occasional applications of the adjective "open" to the noun "source" or "source code" in Usenet articles going back to the 1980s, but it wasn't a "term", even after Adam posted this article to FoRK with the abbreviated subject line. Reading in 2019, you might be tempted to parse the headline "CALDERA. ANNOUNCES OPEN SOURCE CODE MODEL FOR DOS" as containing the subtree "((OPEN SOURCE) (CODE MODEL))" but it seems clear to me that it was intended as "((OPEN (SOURCE CODE)) MODEL)".

So, no, I don't agree that it was "a term that was already in use". When Eric circulated the mail to get support from free-software leaders to change the name from "free software" to "open source", I read it, I read the generally positive responses, and I read Stallman's furious response, which I believe was the next day. I don't recall anyone claiming the term was "already in use". And I was an administrator of FoRK a few years after that message was posted; my first post was probably http://www.xent.com/FoRK-archive/nov98/0351.html. So I'm not just trying to reconstruct the history based on fragmentary documents. I also have my own memories.

We also have this graph, which shows pretty conclusively that the term appeared suddenly in 1998 after Chris Peterson invented it: https://books.google.com/ngrams/graph?content=open+source+so...

You can see from https://www.google.com/search?q=%22open%20source%20software%... that the supposed uses of the term around 1980 that appear in the graph are entirely due to documents Google has misdated.

The OSI doesn't get to own the definition of open source regardless if they were involved in the initial usage of it. It is clear we mean different things as we use term and arguing about what it should mean is pointless.

Patents and "open source" (whatever your definition) share some goals. Specifically, to get design details published in a way accessible to the public. Both patents and OSI Open Source restrict freedoms, but in different ways.

> It is clear we mean different things as we use term and arguing about what it should mean is pointless.

I strongly disagree with this. Open source is a commonly used term; if there is disagreement in it's meaning then that issue has to be resolved first, otherwise what are you even discussing?

If you wish to take your attitude IMO you either have to first demonstrate that your definition of "Open Source" is either the one that should be used, or avoid using the term entirely and talk on neutral terms.

You are attempting to claim that it's possible to have a neutral position on "open source" when your own position is in fact highly opinionated. Commenters are absolutely right to question whether your definition is appropriate or not.

Source is obvious, what does the adjective open mean?

>Allowing access, passage, or a view through an empty space; not closed or blocked.

>exposed to the air or to view; not covered. "an open fire burned in the grate"

>having no enclosing or confining barrier : accessible on all or nearly all sides

>not restricted to a particular group or category of participants

"open source" taken literally only means that the source code is available to view. The adjective open doesn't seem to lend itself to anything more than that.

Getting back to the original point I was trying to make instead of unproductive defining, an original purpose of patents was to OPEN design specifications for inventions to the public instead of everything being kept as closely protected corporate secrets.

> Source is obvious, what does the adjective open mean?

> "open source" taken literally only means that the source code is available to view

Why is the literal meaning relevant? Language doesn't work that way. Do shitstorms have to refer to literal fecal weather patterns?

Hackernews is frequented by professional software engineers and "Open Source" is a commonly used term in that industry. Using the dictionary definition of the words composing that terms is irrelevant. What is relevant is how the term is commonly employed in industry. Your approach only makes sense if we are talking about some sort of newly coined phrase. Open Source software has existed in industry for decades and has an understood meaning.

Generally speaking, a professional software engineer will refer to something as Open Source if the source code can be both read and reused in some context. If you believe that in the tech industry it is commonly used by professionals in a different way I'd like to hear your evidence.

I'm just not interested in discussing what you think I should have meant with the words I used. The idea I was originally trying to convey should be clear.

kragen 64 days ago [flagged]

The OSI has had to deal with charlatans attempting to co-opt the term "open source" since its inception, using dishonest arguments like the one you are deploying here; this is no different from trying to sell water with chalk dust as "milk" or gold-plated tungsten as "gold". It is possible that what you are doing falls within the ambit of US federal fraud statutes, depending on whether you are in a position to receive money for it. However, the Hacker News readershp is savvy enough to understand the difference; here, you have failed and you will continue to fail.

Aside from the issue of passing off whatever sad scam you're wanting to pass off as "open source" in order to take advantage of the goodwill earned by genuine open source, it is entirely and transparently false to claim that "OSI Open Source restrict[s] freedoms." Copyright, patent, and trade secret laws restrict freedoms; open source just gives some of them back.

> Copyright, patent, and trade secret laws restrict freedoms; open source just gives some of them back.

I can't say this is 100% true.

OSI licenses divide into two camps, usually called "permissive" and "copyleft". Even permissive licenses usually restrict the freedom to publish the source without the author's name, which is, while trivial, a restriction nonetheless.

But copyleft licenses restrict the ability to modify the source without publishing your modifications. This is indeed a significant restriction of freedom, and why the BSD/MIT/etc camp are called "permissive", because copyleft is less permissive.

I prefer the permissive licenses for this very reason, although I'm not opposed to using or contributing to GPL-licensed code.

Now, you may say that the freedom hereby restricted is justly restricted, as we justly restrict the freedom to, say, put chalk in milk. I find it more consistent to reject the restrictions of copyright entirely rather than exploit them to force sharing; after all, the only way to use the clause in GPL requiring source sharing is to pursue satisfaction through the court system.

One thing on which I'm sure we can agree: both of these camps are producing open source software, and 'source available' is an entirely separate state of affairs which cannot be included.

Copyright laws are what restrict the freedom to publish the source without the author's name or to copy or modify the source. Open-source licenses give some of those freedoms back, but, in some cases, not all of them.

You're quibbling.

Everything following the word "provided" in GPLv3 limits user freedoms.

Maybe you should care less about being "right" and more about understanding your interlocutors. In this instance you haven't achieved either goal.

Sam, knock it off.

Assuming your interpretation of the law is right (and I do not think it is even close to being so), this comment comes very close to accusing the person you are replying to of committing a crime. Hopefully you didn't mean it that way, and the phrasing was an accident, but I honestly can't tell.


kragen’s comment went over the line, but he’s correct in saying that “open source” has a well-established meaning: https://en.wikipedia.org/wiki/Open-source_software

Open sourcing your design gives other people a chance to work on your project, improve it. If you design gains traction then you will earn way more from patents.

Open-source licenses usually just deal with copyright. Patents are another obstacle to overcome if your company/organization/hobby has a need for building salt reactors, but overcoming copyright is a push in some direction.

APL explicitly deals with patents and contains a patent grant clause. Same EPL, AFL and a few others. See https://en.wikipedia.org/wiki/Comparison_of_free_and_open-so... for an overview

Yeah, I'm confused about that as well. Are these defensive filings? Are they going to do what Tesla did with their patents?! So many questions!

I guess pretty good if you live somewhere where intellectual property laws are only loosely enforced.

If someone gets inspired and find alternative ways to do things, do patents apply ?

Patents are national, so I guess there are many countries where they do not apply.

I totally read the title as "Morton Salt".

My brain read “Malden salt reactor”

I think this a great example of why the “great filter” is ahead of us. Our quest for knowledge seems to grow faster than the dampening of our urge for violence.

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact