What's a "fogbank"? Nobody knows, but nukes need them, so we reinvented them. | Hacker News Reader

What's a "fogbank"? Nobody knows, but nukes need them, so we reinvented them.

thewaythefutureblogs.com

77 pointspavel_lishin43 comments

Wiki has more information (ok, more like speculation) on what a Fogbank is: http://en.wikipedia.org/wiki/FOGBANK

"Many arms experts believe that FOGBANK is an aerogel material which acts as an interstage material in a nuclear warhead, i.e. a material designed to become a superheated plasma following the detonation of the weapon's fission stage, the plasma then triggering the fusion-stage detonation."

Also, just because someone's retired doesn't mean you can't ask for help...

The wikipedia article also mentions,

    [..] the NNSA eventually decided to invest $23 million
    to attempt to find an alternative to FOGBANK.

and,

    In 2008, following the expenditure of a further 
    $69 million, the NNSA finally managed to 
    manufacture FOGBANK [..]

So not counting the initial cost (which isn't mentioned), we're already up to 92 million USD.

All things considered (this being the military, government, and science all rolled into one), I sorry to say that I'm a little impressed by 92 million USD.

I would have expected it to run a few orders of magnitude higher.

Yes. My guess is we had some information on them and the money was spent to fill in the gaps.

Indeed. The blog post seems to assume no reverse-engineering occurred. That means they assume that reverse-engineering some highly complex thirty year old technology in a top-secret military context would somehow cost significantly less than that number.

The US no longer tests Nukes right?

So how do they know the new stuff works I wonder . . .

My understanding is that they simulate them via supercomputing. I imagine this fact has caused a few Pentagon generals to go bald prematurely.

A friend of mine from college works at Livermore labs doing exactly this (simulating, not going bald.) He can't really talk about work, but it certainly seems like a pretty good gig for a physicist.

As far as I can tell the US hasn't had a nuclear test that went far different from expectations since the 60s. And modern nuclear warheads haven't used dramatically different designs for quite some time. Given that it's a pretty good bet that all US nuclear weapons in service are reliable.

4 of the top 20 super computers in the world, and another 2 in the top 100[1], are operated by the US National Nuclear Security Administration. What do they do?

"One of the primary missions of NNSA is to maintain and enhance the safety, security and reliability of the U.S. nuclear weapons stockpile. NNSA, through its Office of Defense Programs, ensures that the U.S. nuclear arsenal meets the country’s national security requirements and continues to serve its essential deterrence role."

[1]: http://www.top500.org/list/2010/11/100 [2]: http://nnsa.energy.gov/aboutus/ourprograms

Inspections, subcritical testing, and computer simulation.

The Russian solution is much simpler: when their bombs reach their shelf life, they're recycled into new weapons. Like the Soyuz versus the Space Shuttle, it's a lot easier to just rebuild a complex system then to predict how it will age. Of course, it does have the tiny drawback of generating enormous amounts of highly toxic waste, but in Mother Russia you can simply dump the waste in a lake.

Recycling a nuke doesn't create much waste. The main components of a nuke are the fissionable material = reused as is (plutonium has a rather long shelf-life), the chemical explosive = reused ( a lot of the explosive used in gulf-war-1 was WW11 surplus stuff, baritol and compB explosives can be remelted and recast continually), and possibly a neutron trigger which does decay - but into a less radioactive isotope.

The only waste is irradiated casings, but since Pu isn't really that radioactive it's a lot less of a problem than a used cancer therapy machine.

Also, just because someone's retired doesn't mean you can't ask for help...

That's the case for most problems, but it might not have been possible when dealing with top-secret classified materials like "FOGBANK". The people who retired may have given up their security clearances, which means that you have to go through the whole process from scratch to get them re-cleared.

Versus redeveloping it for $80-100 million, that still sounds fairly cost-effective.

(Caveat: I've never worked on anything that required a clearance, so I know nothing at all about how they work. This might be a stupid question.)

It seems to me that asking such a person what they remember about some work years ago, would require only that the person hearing the information have the right clearance. As long as no information was provided to the person telling the story, why would that require that retiree to get a new clearance?

Probably a blanket clause that says that you can't disclose classified information at all after you've lost clearance. It's pretty plausible that in an attempt to keep security tight, they're not allowed to discuss it at all to anyone --- situations like FOGBANK are probably the exception.

This is a substance that apparently costs millions of dollars to make. It's probably complicated enough that someone is unlikely to remember every detail about it without some hints and prompting.

I can tell you all about code fixes I made years ago if you describe the code and problem, but asking me "what code did you write 3 years ago?" gets less certain results.

There is a cost to maintaining the clearance. For Top Secret, I believe every 3 years it needs to be reauthorized. Though I agree with the point that it would have been cheaper to bring back the people who knew the first time - if they remembered. I'm sure a lot of people involved with the project know in general what it is, but the details are the missing problem. If there are no official files, the retirees don't know any better. Who remembers chemical formulations of 1 thing you designed out of hundreds, a couple decades ago?

That explanation would also seem to match the name, if only slightly, aerogel and fog share some resemblances.

The Wikipedia page links to this issue of Los Alamos Nuclear Weapons Journal, which has some interesting additional reporting (page 20): http://www.lanl.gov/orgs/padwp/pdfs/nwj2_09.pdf

Apparently they did have documentation for how to manufacture FOGBANK, but when they rebuilt the manufacturing pipeline the end product didn't pass the tests.

After some debugging they realized that back in the 70s, one of the raw materials had been purified to a lower standard than today, and one of the contaminants actually turned out to be crucial to the process. Once they knew that, they could start intentionally adding that contaminant and track the concentration of it.

At least they had unit tests for the material.

Imagine if they only way to test was to run a whole system test.

Some background: in a thermonuclear weapon a fission warhead "primary" explodes inside a radiation channel which also contains the fusion/fission "secondary". The explosion of the primary provides the energy for imploding the secondary.

Quite a lot of unclassified publications describe the mechanism for this implosion as plasma pressure from the vaporized foam that had filled the radiation channel. In truth the overwhelming force causing the implosion is ablation of the secondary's container via absorption of x-rays. I'm not sure why this error continues to be so widely disseminated, I'd half suspect some government plot to spread misinformation.

FREDERIK POHL is still alive??? And has a blog?? My mind is blown.

(Non science-fiction readers: this is like hearing that not only is Elvis still alive, but is performing an acoustic set nearby)

I know, right?!

Gateway remains my favourite hard-sci novel of all time.

http://en.wikipedia.org/wiki/Gateway_%28novel%29

RANT to follow, feel free to ignore:

It feels like this story is being spun to raise "serious questions about the federal government's nuclear weapons management" by Mother Jones and others, as well as to demean hackers.

From my three minutes of research, it looks like Fogbank is a material that is used in just three models of nuclear warheads, one of which, W76, is used on a the Ohio class of nuclear submarine. More to the point this is a story about refurbishing a device that is 20-30 years old. One weapon in the nuclear arsenal, the W76 nuclear warhead, is due to be refurbished and one component is not readily available since they shut the manufacturing plant down 21 years ago. And now that they tried to make the component again, they have run into some problems since no one has manufactured this stuff for 20+ years. The stuff is a nightmare to make--they had to evacuate the manufacturing facility three times in one year alone. (Has your office ever been evacuated three times in a year so you wouldn't DIE?)

This story angle(and its related links) is really about pundits who NEVER DID ANYTHING HARD mocking nuclear engineers who are trying to remanufacture something from the past, essentially from scratch.

"Tee hee, they're so dumb they forgot how to make the atom bomb." No. They are trying to make component to a discontinued 30-year-old nuclear warhead. I would love to have one of this guys fix my grandmother's 600lb. RCA Color TV, the one in her living room in the gigantic wooden furniture case. All they have to do is order a new cathode ray tube...oh wait...they don't make them anymore for that, now what? These critics are guys whose knowledge of electronics extends to (possibly) being able to use a light switch, and who have never MADE anything their entire lives that involves science, math, or reality. So what's my point? (Following is NOT directed at F. Pohl, I know nothing about him:) Maybe in Science Fiction Land, old crap is easy to fix, maybe your robot girlfriend could even do it for you---but in the real world, complex old junk and old codebases and anything that comes from intelligent, creative lifeforms does not get fixed by squirting stuff on it that you can get out of a toothpaste tube you can buy at Wal-mart. And frankly, my guess that if they spent $60-100 million dollars to create this new facility and make juicy new Fogbank, it is LESS than the cost would have been to keep the old facility open for 20-30 years until now. A top secret nuclear weapons manufacturing facility? What do you think that costs to run per year? (The Microsoft Ultra Top Secret Software licenses alone probably run into the millions per year.)

They won't reveal what Obama's last junket to India cost (and/or Bush before him), but if it was in the $5-10 million dollar per day range (that was the latest figure, I saw quoted, and verified by snopes) times 10 days and there goes the Fogbank budget. Pffft.

Hasn't Obama heard of Skype? Sheesh.

  > This story angle(and its related links) is really about pundits who
  > NEVER DID ANYTHING HARD mocking nuclear engineers who are trying to
  > remanufacture something from the past, essentially from scratch.

Seems to me that the 'moral' of the story is that the government should have kept better records (of the design, and the manufacture process). So that they didn't have to start from scratch. $69M is the price tag for our lack of foresight/incompetance.

The only other smaller moral is that we need to find a way for classified things to be found when they need to be found, rather than ending up in a bottomless pit because no one that has classification knows that it exists or how to find it.

They learn that lesson periodically. At one point nearly 50% of the money they spent to have systems built by my people was spent on documentation no one ever read. ever.[1]

If it only cost $70M to develop a process and replace them that is a bargain compared to the wasted documentation on the 9999[2] contemporary items that were never needed again.

To be more concrete, the guys building the bigger sister system to ours probably blew through that $70M on write-only documentation every 12 months for years.

[1] It piled up under desks until they asked us to "only" submit one paper copy because they didn't have enough desks at the program management office. We did eventually talk them down to a reasonable level of documentation, but it took years and the program level staff had to fight hard to get there.

[2] I made this number up.

Actually, I am glad they didn't write a step-by-step "Making H Bombs for Dummies" guide like you suggest. Because NOTHING is ever kept secret if it can be passed along like papers or computer files. Witness the long and ongoing history of nuclear espionage breaches. Having the full recipe broken up and distributed in the brains of a trusted group of people is significantly more secure.

Then they should at least have a plan for knowledge transfer rather than letting those people die/retire without passing on the knowledge, no?

I'd be surprised if the specs were simply lost.

More likely, they said, "Load <archaic file format> into <obsolete software> on the <ancient machine you'd have to get from a computing history museum> to control <manufacturing process nobody uses anymore>."

And after fiddling with the design for a while, someone said, "Well, we can approximate that with <modern manufacturing process>, but the <technical characteristic> is going to be 10 instead of 5. Is that okay?"

And there was nobody around who knew.

> Following is NOT directed at F. Pohl, I know nothing about him

It might be interesting to note that Pohl's Heechee books are exactly about repairing and reusing complex old junk left over by highly intelligent aliens in way that is far from scientific or systematic, more based on random experimentation and pure luck.

I'm with you until the last two paragraphs. Most of the figures thrown about on the India story were utter, utter bullshit.

http://www.snopes.com/politics/obama/india.asp

My understanding of the situation is that most of the documentation on manufacturing fogbank was destroyed to evade litigation by workers poisoned by acetonitrile (which your body metabolizes into hydrogen cyanide) which was used in the manufacture of the substance.

One should remember that not long prior to the shutdown of the facility making fogbank, the Rocky Flats facility (which made plutonium bits for nuclear weapons) was the first nuclear facility raided by the FBI and EPA for environmental violations (among them was burning plu contaminated waste in open barrels and losing track of hundreds of pounds of plu).

I'm sure this is true for any sufficiently complex engineering project that wasn't wound down with an eye towards restarting it later. It's practically impossible externalize all the practical details that the individual engineers hold in their own heads and pass down organically when the project is still active. I'm sure at the time they simply assumed that this generation of weapons would be long obsolete and replaced before the inability to make more fog banks became an issue.

This is a good example of the importance of institutional knowledge. IIRC this is why Boeing has to design a new plane every decade or so -- otherwise it 'forgets' how to when enough people leave, retire, etc. I am sure there are many more examples of this out there.

edit: a more relevant example to HN -- the supposed run on cobol hackers. One might also be able to argue this about mainframes, too.

This is the sole reason we are still designing fighter planes, submarines etc. Nobody else in the world can match the US at the moment, but if we stop now we'll be starting from scratch in 30 years when we want to again. The way to make it cost effective is to slow down the pace and lengthen the projects.

Am I the only one who sees the face of a giant clown in that nuclear cloud?

It is from that game twisted metal

http://playstationlifestyle.net/wp-content/uploads/2010/01/t... (must be a new one if it has ps3 but the old one was for the original playstation)

Edit: maybe not http://ps3.ign.com/articles/805/805953p1.html but twisted metal was the first thing I remembered

It was created by http://www.thomasmangold.com/ for a Playstation 3 ad some years ago.

I saw it too. My first thought was that this was the creepiest case of pareidolia ever. http://en.wikipedia.org/wiki/Pareidolia

I'm glad you saw it and that others have indicated it's not my mind playing tricks on me.

There is always a trade off between security and usefulness.

The best secret is one that no one knows. It is also completely useless. This trade off is one that startups/hackers deal with all all time and there is no right answer, just worries.

Are we too insecure? Is our security costing us users? How much security is too much/little?

It is also a problem you can never finish just continually review and refine.

The concept of having to reinvent stuff because we lost all of the tacit knowledge associated with it is very interesting. Donald MacKenzie wrote about this in the context of nuclear weapons - Tacit knowledge and the uninvention of nuclear weapons ( http://books.google.com/books?id=c5YaHkcP6DEC&lpg=PP1... )

It's highly recommended if you're wondering how things can get uninvented.

P.S. The author of the post is Frederik Pohl, a pretty awesome science fiction writer.

http://en.wikipedia.org/wiki/Frederik_Pohl

Reply on news.ycombinator.com