"Nothing like this will be built again" | Hacker News Reader

"Nothing like this will be built again"

antipope.org

452 pointsajdecon95 comments

Worth noting that the visitor centre for Torness recently re-opened, it looks like you need to make an appointment in advance (for security checks) but, having had a detailed tour of another AGR plant years ago, I suspect it may be worth visiting:

http://www.edfenergy.com/about-us/energy-generation/nuclear-...

Also, if you do visit Torness and you have an interest in science (which seems likely) then you might want to visit the nearby Hutton's Unconformity at Siccar Point - this is one of the places where 18th century geologist James Hutton first found evidence of the tremendous age of the Earth, as his travelling companion John Playfair put it:

"The mind seemed to grow giddy by looking so far back into the abyss of time"

http://en.wikipedia.org/wiki/Hutton%27s_Unconformity

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

Tangent thought....

Might be worth booking an appointment to see what the British government's internet slurp has on one. At least one would know if they thought one were dangerous or not. And if one finds out one is not a terrorist after all, one can celebrate by having a rather cool day out.

This wasn't a visitor centre tour.

This was a "psst, wanna come visit where I work?" tour from one of the guys who built and ran the plant.

Either pre-9/11 or very shortly post-9/11 -- with the rise of the post-9/11 security state there's no way such a tour would happen today. Indeed, the friend in question said some time later that he didn't have that kind of access-all-areas pass any more.

Mirror: http://webcache.googleusercontent.com/search?q=cache:S384WfJ...

Previous discussion: https://news.ycombinator.com/item?id=2140900

And previous previous discussion https://news.ycombinator.com/item?id=569564

Seems like quite the evergreen.

This is many, many years old - maybe even 10 years old? Maybe someone can find the original date and put it in the title.

Archive.org's earliest snapshot of it [edit: at this URL - see below] is from January 6, 2010, but have a snapshot of the same content at a different URL from December 26, 2002. Their snapshots of the site itself go back to July 6, 2005, so I'm inclined to believe the article dates only back to 2010 or late 2009 at the earliest.

Edit: and archive.org has an even earlier snapshot from December 26, 2002, under a different URL.

Here's a Metafilter discussion of it from January 2003, fwiw: http://www.metafilter.com/23068/Nothing-like-this-will-be-bu...

One often gets the "nothing like this will be built again" feeling around these kinds of engineering projects from the 20th century, since modern constructions tend to have a very different feel, more sleek and less massive. Also true with things like transit infrastructure: even though a modern subway or elevated train line, like the Copenhagen metro (which has both), is better in just about every way from the old style of construction, it subjectively doesn't have the same impressive feel of something like the NYC subway or Chicago El, with their massive steel columns and hundreds of thousands of rivets.

Agreed.

It just feels like these structures were built to last a lifetime and then some. Most of the stuff being built nowadays appears as though it already has a due date of "the not to distant future" to be torn down and replaced.

At least here in the US, the architecture just feels like everything else around me - it's disposable.

For a lot of infrastructure, the ability to more accurately model and calculate the performance of structures has meant that they don't need to be overbuilt so much. The less you overbuild a structure, the less it weighs, and the less structure you need to build - a virtuous cycle.

If you look at the last 400 years of bridge design you can see the transition from overbuilt compression structures to gossamer suspension bridges. This is a good thing, but it's easy to mistake old, overbuilt structures as being more "robust" when what they really are is an inefficient design meant to avoid failures due to a weak understanding of how the materials and structure will perform under use.

I agree that there is a vast increase in productivity and saved resources from better modelling and understanding of the materials.

But you cannot ever convince me that our steel suspension bridges are going to last for centuries like our stone bridges from the 15th century and onwards.

The thing is, their goal wasn't to last for centuries. Their goal was just to not fall down, and because people back then didn't have the modeling abilities we have now, their solution was to be very conservative in their design and overbuild everything. If we wanted to make bridges that would last a long time, we could almost certainly do it more efficiently now.

Yes, stone will last longer, simply because it doesn't get eroded by the elements as quickly as possible. But it's also simply not usable as a construction material for bridges beyond a certain size and span, because it's only good under compression and even then its own weight starts to become a limiting factor as you get larger.

By the 1940s, a stone bridge that might take 100 man-years of labor to create could be replaced by a bailey bridge that goes into place overnight. I'm sure if we wanted to create a bridge that lasts 300 years, we could dedicate more resources to it, but it would be grossly wasteful to do so - you don't know how long the current requirements are going to even be relevant. It's better to build a bridge that lasts for 50-60 years, and have the option value of replacing it with a better structure down the road.

>>>>>> It's better to build a bridge that lasts for 50-60 years, and have the option value of replacing it with a better structure down the road.

And then we get to where we are now a lot faster. . .

http://www.businessinsider.com/worst-state-road-systems-2010...

"Over 90,000 miles of roads and 71,000 bridges are in dangerous disrepair, according to a 2010 U.S. PIRG study."

"with their massive steel columns and hundreds of thousands of rivets"

We have a fantastic example of that near Edinburgh:

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

Reminded me of the first arch bridge in the world to be made of cast iron, built in the 1770s and still standing: http://en.wikipedia.org/wiki/The_Iron_Bridge

Wasn't that bridge intentionally designed to not only be robust, but to look as imposing and robust as possible, in order to reassure the public since the previous bridge in the location had collapsed?

Yes they certainly succeeded in making it look robust! You are correct in saying that this desire was based on a previous accident where a bridge collapsed - although it a bridge in a different location (over the Firth of Tay at Dundee rather than the Firth of Forth near Edinburgh).

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

One interesting thing if you like Science Fiction is that Iain (M) Banks grew up in North Queensferry, which is right under the north end of the Forth Bridge - which completely dominates the wee town. I remember hearing in an interview that Iain claimed that his fascination with megastructures (e.g. Culture GSVs and Orbitals) probably came from growing up in an environment dominated by a Victorian megastructure.

He even set one of his novels on a fictionalized mega-sized version of the Forth Bridge:

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

I guess the key difference is the sense of massive person power one gets from those older projects. Rivets and the like bear the unmistakable mark of human assembly.

That said, I still find modern day mega engineering projects similarly impressive. The Large Hadron Collider and the current construction of the ITER in France come to mind....

http://www.iter.org/construction/tkmfoundations

Nothing like it will be built again because a) successive British governments over the last 30 years have buried their heads in the sand over energy policy[1] and b) we as a country don't do actual engineering anymore. Both a and b mean that our energy future likes wholly in the hands of France!

[1] Say what you like about Thatcher, at least she had a halfway coherent energy story.

>we as a country don't do actual engineering anymore

...is that really true? The UK is currently home to Europe's largest civil engineering project (Crossrail), my hometown produces Trent jet engines and designs wings for all Airbus aircraft. And there's the cars that are both manufactured (BMW, Honda, Toyota, Nissan, etc) and designed (Jaguar, Aston Martin, Land Rover, McLaren...) here.

The engineering we do has changed, and maybe there is less of it than there was forty years ago, but it's a bit hyperbolic to say we simply "don't do" actual engineering.

I seem to recall that, measured by value in real terms, the UK does more engineering than ever. Just because it's a smaller part of the economy and doesn't support as many unskilled people doesn't mean it's not happening.

On an unrelated note, some of the social issues we complain about these days seem to be a direct consequence of the off-shoring of manual labour.

The UK does (at least before the recent recession) more manufacturing that ever - it's just that the whole process has become super efficient that it doesn't employ very many people:

"A 2009 report from PricewaterhouseCoopers, citing data from the UK Office for National Statistics, stated that manufacturing output (gross value added at 2007 prices) has increased in 35 of the 50 years between 1958 and 2007, and output in 2007 was at record levels, approximately double that in 1958."

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

Our problem is that we got too good at manufacturing!

I don't know if that is the case in the UK, but it is the case in US vs. Chinese manufacturing. The US output never went down and more things are produced in the US than ever - it just takes fewer people to do it, and it isn't the kinds of things that are currently visible to most consumers.

But not those that support a large number of jobs and how many of these engineering projects are done by non uk firms.

"we as a country don't do actual engineering anymore"

Let's see - within a few miles of Edinburgh we have being built:

- 65000 ton aircraft carriers

- A new 2700m road bridge over the Firth of Forth

- Eurofighter work

- A new railway being open to the Borders (OK - reopening an old line)

OK, I won't mention our trams, but there certainly seems to be a hell of a lot of "engineering" going on in the immediate vicinity.

Maybe Edinburgh is unusual in having such a burst of engineering activity in our vicinity?

NB Even the crane built to build the aircraft carriers is an impressive bit of engineering:

http://www.aircraftcarrieralliance.co.uk/en/media/image-libr...

I rather like the fact that the huge crane has another crane on top of it - which looks tiny but is probably quite large!

Heh, 65k ton carrier, how cute! :)

Edinburg sounds a lot like Hampton Roads, VA, where we build lots of defense stuff, including the 100k ton Gerald Ford class nuclear carriers, with the largest gantry crane in the western hemisphere - fun fact, they hoist the island of the carrier onto the flight deck. Amazing engineering work going on there!

"we as a country don't do actual engineering"

What do you mean by "actual engineering"?

- the UK makes and develops satellites - designs, develops and produces a huge number of jet engines - is home to nearly all the Formula 1 teams - the Nissan car plant in Sunderland produced 500,000 cars last year which is a huge engineering challenge and is now the home to the main production of the LEAF electric car - Teeside is still the home of chemical processing - companies like Alert Me are leading the charge in smart energy consumption - last time I checked Arm were stil based in Sheffield and Cambridge and scaring the pants off Intel - last, but not least, BAE systems is in the midst of producing some of the best and most advance submarines in the world and commissioning nuclear reactors in the process.

I do however agree with your first point - although there are a lot of wind farms around now which is nice to see

And how many of these are UK owned? Which I took to be his point, given the dig that a French company will be building the new Nuclear power plants at Hinkley Point...

I was wondering how many weren't arms industry related as every engineer I know in the UK is doing arms work. I was pleasantly surprised.

I used to work at ARM and ever since then have been invited to all sorts of weapons and defence conferences.

Does it matter who owns the company if it's British engineers?

Doesn't matter who owns the company - the engineers are British and they are solving their engineering problems in the UK. CocaCola is obviously a US company but the production plant in the UK has some very interesting manufacturing/engineering problems (as does Mars, Proctor and Gamble, Fuji (pharma division)) and if you really need examples of British companies then try Dynex, Wolfson Microelectronics, Imagination Technolgies, British Sugar and Dyson.

His dig about a Frech company was just a fashionable quip that is doing the rounds right now - yes they are better at some things than us but likewise we sell them planes etc so each nation plays to it's strengths

Um a lot of these European companies favor their own nationality it's going to be french engineers and possibly a lot of the navvies/ground workers will be imported.

Thans, that's an interesting list.

In terms of pure engineering I agree it doesn't matter who owns the company; in terms of whether engineering is having its full impact on Britain (other than by providing jobs) I'd say it does - if profits get moved to other countries and taxes paid elsewhere, we're losing out and I'd like to see more UK owned companies for that reason.

>CocaCola is obviously a US company but the production plant in the UK has some very interesting manufacturing/engineering problems

Citation needed. They are Coca Cola manufacturing plants all around the world. From Bolivia to India and from Germany to Albania, and they don't have any interesting manufacturing/engineering problems.

Nothing that wasn't already solved in the US HQs anyway.

There are, but each plant (just like in every industry) will have it's own issues, targets and geography specific problems to deal with. Manufacturing isn't a setup once and repeat around the world process - as much as the people working on it would like it to be.

My citation is a friends Dad who was a lead engineer on CocaCola's line. Your citation the Bolivia etc not having interesting problems would be?

>There are, but each plant (just like in every industry) will have it's own issues, targets and geography specific problems to deal with. Manufacturing isn't a setup once and repeat around the world process - as much as the people working on it would like it to be.*

Sure, but you wrote that (the UK one) "has some very interesting manufacturing/engineering problems". What you describe is perfectly normal engineering problems, that kind that any kind of factory has.

Nothing out of the ordinary, or anything that points to any special UK engineering in the UK CocaCola factories (which was what you provided it as an example of).

>My citation is a friends Dad who was a lead engineer on CocaCola's line. Your citation the Bolivia etc not having interesting problems would be?

Having visited two of them and seen the production process. Pretty much standard stuff all around. Back in the day (it's now closed) we had one even in my hometown in Europe (in a place of around 120K people).

Ok, I understand your comment now. The emphasis wasn't on me saying that the UK plant was the ONLY one that had interesting challenges, I was simply using that as a way to the say that any CocaCola plant (or similar, as I mention in the brackets after) has interesting engineering problems and that there IS a plant in the UK. I think people often dismiss manufacturing as "done" and easy hence that became one of my examples.

I think our definition of interesting is different as I find nearly anything in manufacturing interesting, standard or not, and making them better/faster is always a problem to me - I recently toured one of our suppliers who supplies our extruded parts and I was fascinated about how they use their machines, not one of which was less than 25 years old, to make useful shapes and multi material extrusions - on the surface everything was standard and run of the mill but below the surface it was much more interesting.

I was going to reply to the original post but this perhaps more apropos. My father started his career with Babcock & Wilcox after leaving a nuclear sub as an electrician's mate with a nuclear engineering MS; his division -- the org that designs and maintains reactors & power plants -- was later acquired by Framatome, which later formed part of Areva (https://en.wikipedia.org/wiki/Areva). It's a French company, but entirely global and the nuclear power division (Areva NP) performs nearly all of it's reactor design work in the US, as well as most of the planning/oversight for power plant support & maintenance/inspection services.

(Speaking of plant inspections, the job of inspector is one of those high-stress but über-high pay seasonal jobs you never hear about. $200k+ for 6mo/yr of work, but it's ~2000hrs of work compressed into that period.)

Anyhoo, growing up I was privileged to tour a number of different power plants in the US, and -- besides the scale of things Charlie notes in the blog post -- the thing that struck me most was that even though the designs were often very similar (all PWRs of some sort), the implementations were very different. Even though reactor design may be identical between plants, all the peripheral support infrastructure and all the wiring & plumbing is unique to each plant, and keeping track of everything is hugely complex.

As an aside, and the bring relevance to HN, because of the complexity of nuclear power plants -- particularly the plumbing and backup systems -- there is a huge market for simulation services (which is something Areva pretty much owns globally at this point), which consists of both electrical/mechanical engineering and software work. They literally build a control room for a specific power plant, identical to the actual control room but with software simulating all the inputs/outputs and allowing them to run all sorts of scenarios (and to train new operators). My dad had a few job openings for software engineers who had some nuke background, and positions requiring only a few years experience were paying $100-120k base, and this is in small city Virginia (Lynchburg) where you can buy a decent house for $150k.

"successive British governments over the last 30 years have buried their heads in the sand over energy policy"

You missed discussing the why, which was north sea natgas. Why, we'll have natgas so cheap we won't meter it, and we'll grow rich exporting it. So why not replace every electric plant with a natgas turbine?

Wot you say its 2005 and the fields have declined so much that we're now net importers and the lights are going to go out pretty soon? In fact we need to import more than we can physically be supplied with given current infrastructure?

Whoops! Its a pity really, we may be witnessing the first "first world" country to have its electrical grid shut down in a couple years. Brownouts in summer suck, but brownouts in winter can be lethal. Its not going to be pretty.

My father and I went on an equally in-depth guided tour of Hinkley Point station back in the mid '80s when I was a boy. We too got to walk on the reactor lid, view the cooling ponds, and meander underneath miles of thrumming piping. It was utterly fascinating and one of the very best memories I have of my childhood. Hinkley Point is a PWR rather than an AGR, but what the author writes about that collision of space age technology and Victorian plumbing definitely rings true.

Other highlights for me not mentioned in this article: a great video showing a high speed diesel train (not just the locomotive) crashing into a waste transport container; climbing around a whole fleet of tiny decontamination trucks with radiological protection gear for the drivers and manipulator arms for picking up material; and getting hands-on with the security system (not part of the tour; my dad made a friend) and broadcasting over the plant tannoy. No chance of any of that stuff happening for my kids, sadly.

I was happy to read this even if it is a repeat. A great write up that took me back.

Off-topic, but it's the first time I see somebody use "tannoy" in place of "PA" - really identifies you as a Brit :)

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

Guilty. I've lived in the US for a few years and my usage and spelling is hopelessly mixed up, but that kind of cultural thing is so deeply ingrained you could use it to catch spies.

Hinkley Point B is an AGR (or, more specifically, two AGRs). Hinkley Point A was an earlier Magnox station.

It's one of the leading sites for new-build nuclear in the UK, and if Hinkley Point C does happen, it'll be a PWR. Right now, the only operating civil PWR (naval reactors are a rather different matter) in the UK is at Sizewell.

I also had the Hinkley Point tour as a child and remember it fondly.

Thank you for the correction. I was relying on very old memory there.

Do you remember whether you toured A or B? I am pretty sure we only looked around one.

I'm fairly sure it was B.

"For starters, some embedded controllers in racks in the auxilliary deisel generator control rooms have EPROMs which have been known to be erased by camera flashes in the past, triggering a generator trip"

Wow, I don't know where to begin with this one. Either the engineers have completely lost all knowledge about how their controllers operate, or nobody has the right mind to think about putting opaque labels on the EPROM chips or over the controller board itself?

This kind of stuff borders on urban legend / cargo cult engineering territory.

Err, or maybe it's just a sensible precaution, because relying on a single defense against failure is a bad idea?

In 1997, someone taking a picture of an embedded controller board at the Haddam Neck nuclear plant set off the fire protection systems in the control room, causing everyone to evacuate the control room for an hour.

http://www.nrc.gov/reading-rm/doc-collections/gen-comm/info-...

And the conclusion?

"They also confirmed that the light from the Canon flash and the Polaroid flashbulb could be effectively blocked by "black bagging" the flash, or by blocking the EPROM window with "tin foil" held in place by clear cellulose tape, or by blocking the EPROM window with "standard electrical tape."

No shit, now read my post again with the knowledge that I already knew that! :)

You're absolutely right. My apologies.

Or they do put opaque labels on the EPROMs but such labels occasionally fall off or are forgotten.

Belt and suspenders.

An interesting counterpoint, in being submitted on the same day as http://www.pnas.org/content/early/2013/07/03/1300018110 "Evidence on the impact of sustained exposure to air pollution on life expectancy from China’s Huai River policy" starts getting publicity:

> This paper's findings suggest that an arbitrary Chinese policy that greatly increases total suspended particulates (TSPs) air pollution is causing the 500 million residents of Northern China to lose more than 2.5 billion life years of life expectancy. The quasi-experimental empirical approach is based on China’s Huai River policy, which provided free winter heating via the provision of coal for boilers in cities north of the Huai River but denied heat to the south. Using a regression discontinuity design based on distance from the Huai River, we find that ambient concentrations of TSPs are about 184 μg/m3 [95% confidence interval (CI): 61, 307] or 55% higher in the north. Further, the results indicate that life expectancies are about 5.5 y (95% CI: 0.8, 10.2) lower in the north...

Love your writing, but how is this a counterpoint? The article claims no more AGR reactors will be built. It's bullish on nuclear reactors in general.

He's using a different meaning of the word counterpoint than the one you're thinking of.

Don't be so sure... In cinema's this year: Pandora's Promise: http://www.youtube.com/watch?v=7fm8SVLOacQ

The claim is that the particular kind of reactor (gas cooled) is unlikely to be built again. It's (apparently) more efficient than competing designs, but also larger and more complex, and it didn't really become popular. It certainly looks like more nuclear power plants in general will be built.

Silly nitpick: this exact design appears to be finished, but some of the Gen IV designs are gas-cooled: http://en.wikipedia.org/wiki/Generation_IV_reactor

Last year I went on a tour [1] of the 1944 "B reactor" at Hanford (the first large scale reactor [2]) They don't allow you to crawl around on stuff but you can see a surprising amount and they have people who worked on it giving talks. Pretty cool and worth doing.

To me one of the coolest things was the complexity of the (pre computerization) monitoring/control room and systems.

[1]http://manhattanprojectbreactor.hanford.gov/ [2]http://en.wikipedia.org/wiki/B_Reactor

Fantastic. I live within an hour drive to the plant so I just sent them an email asking if there are any tours scheduled in coming weeks. Would love to visit myself.

Looks like whoever posted this got around the repost limit by appending a '?' to the end of the url. Someone should fix that.

You can resubmit a link after a year, I think?

In any event, there's no way to block resubmissions programatically: the site owner could just change the name of the file.

Or people could append random query parameters, HN would have no way of detecting that http://example.com/?foo, http://example.com/?bar and http://example.com/?baz points to the same article.

No need to change anything on the webpage.

The page content's hash or any other blueprint (an arbitrary regex match) could be checked against for resubmissions. For convenience, if resubmission should be detected, the original post could be topped on the frontpage nevertheless, because old material may as well be worth revisiting for several reasons -- I for one had missed this article before, and enjoyed it reading greatly now.

  page content hash

Change one byte.

  regex match

Now we get into ship-of-theseus arguments. What counts as the "same" article? 99 percent? 51 percent?

What if you resubmit a blog post that's picked up 150 more comments since it was last posted? How would HN tell the comments apart from the article text?

What if someone submits a blog post that heavily quotes another article that's been submitted, in the process of rebutting it, and ends up being a 80% match?

What if you include the full text of moby dick, in a display:none <div>? What if the page loads the text of the article with javascript? Would the HN submissionbot have to render CSS and javascript?

> Change one byte.

Most resubmissions are submitted by people other than their authors, who are typically not in a position to change any bytes of the content.

In the present case, speaking as the last person to submit this one (more than two years ago), I don't see that there's any harm in it at all. If this is still interesting two years since it was last discussed on HN -- which clearly it is -- then good for the submitter, who has provided the HN audience with something interesting to read.

The slightest spec of dynamic content makes this approach brittle at best, author or not.

You probably meant "speck", not "spec".

Looks like whoever posted this got around the repost limit by appending a '?' to the end of the url. Someone should fix that.

Yeah, because we should never, ever be subjected to the same content twice. After all, this is Hacker News, not Hacker Olds, amirite?

(Sorry for the snark. I have been reading Hacker News for several years, and I don't recollect having read this article before. I enjoyed it, I upvoted it, and I'm glad it was reposted.)

I have vague recollection of reading it on HN before, but meh, it's not like we're wading through a swamp of reposts. My scrollwheel still works.

And sometimes the page changes but the url stays the same

Y'know, if I'd known it was an evergreen I'd have posted it myself, for the link karma.

Humph.

Come on, surely you already have far more internet points than any one man needs. Share the wealth a little!

Whew, that was a close call. Still shaking. I can't believe I almost read the same thing twice.

I disagree. I — and others — have used query parameters to submit links to Humble Bundles, for example. New Bundles are launched regularly, and the broader HN community tends to have useful things to say about the titles in any given Bundle, and to appreciate the heads-up to their existence.

Heck, half the things in my house will never be built again. Like my Minox camera from the 1950's. It's beautifully machined, marvelously engineered, and utterly obsolete. And, back when I could still get film for it, the pictures were crappy even by disposable camera standards.

"Anti-rabbit defenses"?

Torness has been shut down due to jellyfish, maybe they expect suicide rabbit squads to invade and nibble vital cables?

http://en.wikipedia.org/wiki/Torness_Nuclear_Power_Station#S...

[NB I carefully avoided making any reference to The Wasp Factory]

Think radiation leaks. In event of a major incident, you do not want rabbits on site burrowing around and then spreading cesium-137 and other pernicious decay products.

So protection from radioactive rabbits? :-)

You'd be surprised where those critters can get and the kind of damage they can cause in their search for some place warm to sleep. A single rabbit making its way into a building's transformer can short the transformer, causing a surge that trips the breakers of every PDU in that building, but not before the building's UPS gets fried.

Didn't realize nuclear reactors were that inefficient. But I get most of my power from hydro, so that's perhaps an unfair comparison.

There are hard limmits on the efficiency of any heat engine; the plant described here is actually doing pretty well. And, hey, fuel is cheap; the energy density of uranium is pretty spectacular.

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

> smelling the thing (mostly machine oil and steam, and a hint of ozone near the transformers)

Possibly just metal oxides or dimethylsulphide, as ozone is odourless.

Next time choose your sources more carefully. Not only ozone has smell, its name means that.

He's technically correct in that the "smell of ozone" is just burned/oxidized respiratory system. That's why all the strong oxidizers smell pretty much the same unless they've got "something interesting" attached to them other than boring ole O2 like ozone. On a small scale like parts per billion its a unique stinkyness and mostly harmless, so they claim. Its more of a flash burn of the respiratory system, not a slow cooked fat rendering, which in that case does kinda smell like burned meat. Smell of oxidized proteins, not so much slowly rendered fat.

Can someone explain this? "generators spin in a sealed atmosphere of hydrogen gas", hydrogen, really?

Hydrogen offers low drag and good thermal conductivity: http://en.wikipedia.org/wiki/Hydrogen-cooled_turbo_generator

Why wouldn't they spin instead in a near-vacuum, instead of an environment of hydrogen?

Because near vacuum is a terrible heat conductor.

Vacuum has no drag but also no thermal conductivity, so you'd have to cool the generators some other way.

If they're sealed and there's no oxygen in there, then there's no boom.

Since hydrogen is so light it makes for less aerodynamic drag on the moving parts.

Ah ok makes sense, thanks!

Reply on news.ycombinator.com