36
$\begingroup$

Spaceships are a peculiar thing. We've got them in all forms, sizes & colours. They have vast computers needed to do all the real-time calculations required for astrogation.

Assuming our computers are not as perfect as we designed them to be. There have been some miscalculations and we've been hit by a rather large junk of something that should not have been there!
We managed to seal-off the crew area and life-support seems to be still-running. But there's an issue:
 Our server-room has been breached and atmosphere is venting at an alarming rate. Environment-control for that section has shut down and we cannot fix the breach, as the section of our ship containing our spacewalk equipment has been ripped off and is currently on its way into the local star...

We need to keep our computers from dying, otherwise we'll stand little chance of being able to send a distress signal, and no chance of staying alive until someone arrives to rescue us.

Q: How can we keep our computers at working temperature, so they neither freeze nor combust while we wait for our rescuers?

We do have unlimited power. But everything is managed by these computers.

$\endgroup$
  • $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$ – Monica Cellio Dec 26 '16 at 3:14

19 Answers 19

65
$\begingroup$

Actually you have just the opposite problem.

Computers inside a pressurized vessel can dump heat into the atmosphere via convection (the reason the heat sinks on the CPU in your computer have fins, and why there is a fan inside your tower or laptop.

enter image description here

Current CPU with heatsink and fan

Once the atmosphere is dumped, the computers are now encased in a vacuum, which is a very efficient insulator. The fans will be useless, and the heat energy being released from the CPU's and other computer components will have to be radiated into the vacuum, a far less efficient process. This is more difficult because the radiating fins on the CPU heatsinks are sized for rejecting heat to the atmosphere. To operate efficiently in vacuum, they would need to be much larger, and not buried inside the server racks.

If your computers are liquid cooled, and the cooling loop is still functional, the problem is going to be a bit less severe, since the coolant is presumably being circulated to an external radiator which can reject heat more efficiently. If the radiator is facing the sun, then you might actually be receiving more heat energy than the computers are producing, while when the ship rolls so the radiators face away from the sun, the radiators can now dump heat to the 3K background of the universe. (Apollo spacecraft did "barbecue rolls" to ensure the surface of the spaceship was evenly exposed to the sun and prevent heating and cooling problems).

So once the computers are exposed to the vacuum, your real problem might actually be that they overheat and shut down.

$\endgroup$
  • 19
    $\begingroup$ Computers which were not designed for a vacuum are known to melt the solder connecting the chips to the board, and literally shedding parts due to these overheating issues. When designing hardware for vacuum, there's an entire additional set of requirements put on them to ensure this doesn't happen! $\endgroup$ – Cort Ammon - Reinstate Monica Dec 20 '16 at 17:52
  • 7
    $\begingroup$ Excellent answer, which also contains built-in solution: air-cooling adds unnecessary workload to life support, and requires more space for air to circulate. Using liquid coolant saves on space and makes entire system more efficient. $\endgroup$ – M i ech Dec 20 '16 at 18:44
  • 2
    $\begingroup$ Its worth mentioning that every current terrestrial liquid cooling setup also relies on convection to remove heat from the liquid, it just happens somewhere else. You can design a radiative cooling system for the liquid, But would likely need to be much larger that any convection setup (due to the loss in efficiency), and potentially open more issues (are we facing the sun?). $\endgroup$ – shenles Dec 20 '16 at 20:10
  • 10
    $\begingroup$ @shenles that needs to be done anyway to cool the ship in general. Space is a stupidly good insulator(one of, if not the best known), so the ship needs to be designed to dump heat to vacuum efficiently. Losing air to the computers just stops air cooled computers dumping heat to this already existing cooling system. Liquid cooled computers can still of course be taken out by breaching the coolant lines, but they're a much smaller target. $\endgroup$ – Leliel Dec 20 '16 at 21:36
  • 2
    $\begingroup$ You'll also need vacuum-safe grease between the chip and the heat-sink. And while the CPU has its conspicuous heat sink and fan, a lot of other components do also rely on air cooling. These will require attention as well. $\endgroup$ – Dave Dec 21 '16 at 2:33
18
$\begingroup$

Cold won't be an issue. Computers work great in the cold so long as it doesn't cause condensation which could short out the circuits. And if there is no atmosphere, that isn't an issue at all!

As other answers have stated, the real problem is heat dissipation. Your typical home computer you are familiar with uses heat sinks that dissipate heat into the air to keep everything within operational temperatures. Of course, without an atmosphere, systems like that are going to fail. In a vacuum there is nothing to absorb the heat, so you are stuck with the radiant abilities of your heat sinks, which isn't nearly as efficient.

Hopefully, though, on a spaceship where a decompression event is always a danger and with a crew that relies on the computers, the ship's designers took this into account and designed cooling systems that could work in a vacuum. That could be a closed-loop liquid cooling system (which hopefully has some place to dump the heat; again even a liquid cooling system on your average gaming PC that has one ultimately dumps the heat into the air around it), or a heat sink designed to radiate effectively enough into vacuum. It would be utter negligence on the part of the engineers not to design for this contingency. It should also be noted that the atmosphere on a space ship is a closed system anyway and would have to have a way to get rid of the heat if you are going to dump heat into it, or eventually you are going to cook your crew!

It could be, though, that the computer cooling systems don't work as efficiently as normal without atmosphere. Or maybe the breach also damaged this system in some way so it isn't as effective. The computer might have to be slowed down to a lower clock speed so as not to generate as much heat. Or maybe the crew has to turn off non-essential functions so the computer doesn't cook itself.

Update: Thinking about this a bit more, if the engineers who designed the computer did their job, loss of cabin pressure by itself should not affect the ship computer at all. This would been an easily-considered contingency. Relying on the life support system for heat regulation of the computer would have been a bad design decision anyway. It violates the principle of separation of concerns any competent engineer would have followed. Current space craft use radiator fins to dissipate heat, and if your computer is now in danger of overheating, you probably lost some of those in the impact. To keep it running you could either reduce its workload so it doesn't generate as much heat (which may or may not be an option depending on how much work it needs to do to get you out of the crisis), or maybe stage a space walk into the decompressed room to rig up a new heat-dissipation system.

$\endgroup$
  • 2
    $\begingroup$ Good point in this answer is - why should space comp relay on air cooling, it is not efficient nor a good design for a space ship, so if system is not disrupted then there may be no problem at all. $\endgroup$ – MolbOrg Dec 20 '16 at 20:21
  • $\begingroup$ I think you leave out that even for a liquid cooling system the radiator for that system would have to be in a separate area with atmosphere otherwise, it won't vent the heat at all and you're still left with an off computer from overheating eventually as it'll still overheat. You also neglect moisture as that's going to be the only killer as it get cold and condenses on metal. $\endgroup$ – rangerike1363 Dec 20 '16 at 21:43
  • $\begingroup$ No, I mention that you still need atmosphere for a liquid-cooled system. And moisture condensation isn't a concern if there is no atmosphere, which I cover in the first paragraph. $\endgroup$ – Seth R Dec 20 '16 at 21:52
  • 2
    $\begingroup$ Normal computers actually don't work that well at very low temperatures. The capacity and dissipation factor of capacitors depends on temperature, and at -30°C your computer might not function anymore. $\endgroup$ – Aaganrmu Dec 21 '16 at 10:15
  • 1
    $\begingroup$ "And if there is no atmosphere, that isn't an issue at all!" -- Oh yes it is! Going from 1 atmosphere to zero will place mechanical stress on everything that has air/gas pockets in it. If you don't have vacuum-qualified hardware, things will go pop! This is from a satellite engineer. $\endgroup$ – Jens Dec 22 '16 at 12:12
11
$\begingroup$

Do not make a problem where you do not need one

Summary: this is not a problem, because either you — as the author — should not make it one in the story, or it has already been dealt with because heat management is a vital system on a space craft anyway.

Take one step back here and apply Chekhov's Gun: do you actually need this plot element for your story?

If the story element in question is not something that is relevant to the plot, then you should ignore the problem. Adding a superfluous story element just for the heck of it never helps the story; it is just a waste of time and effort, both for you as the author and for the consumer.

Simply assume that future development of computers have made computers so energy efficient that you simply do not have a problem of cooling (freezing — as other have pointed out — is not a problem). Already today there are moves towards not just making computers faster, but also to consume less energy per computation. Think about it: would it not be quite silly to assume that mankind has gotten advanced enough to go star-hopping, but have not solved the problem of our computers slurping up stupidly large amounts of energy?

If — on the other hand — this is something you need to deal with in your story, for whatever reasons, then the solution for you is that this was never a problem in the first place, because the designers of the space ship already considered it.

Heat and energy management aboard a spacecraft is a very serious thing. Us mud-stompers (people living planet-side) — that are spoiled with having a near infinitely capable heat sink: the atmosphere — never think about these things. But in space, this is a much different matter. Already today, heat management is a very big issue in the design of space craft, space stations and even space suits.

If you mean to say that the shipboard computers in your story are sources of heat, then the cooling of these will — of course — not be done willy nilly with you carelessly venting the waste heat into the ship's internal atmosphere.

The computers will be hooked up to the ship's internal heat management system

Just as these computers are hooked up to the network and the power supply, so will they also be hooked up to heat-transfer devices that transport away the excess heat, either doing so conductively with heat pipes, or using fluids as a heat carrying medium. The waste heat is then dealt with in a centralized manner by the ship and in the end the heat will most likely be routed to external heat radiators.

A secondary problem with using consumer grade cooling with heat sinks and fans is the risk of fouling. In your own home, this is not a big issue. You just clean it out, or the computer breaks and you get a new one. In a space ship, this is not something you want to have to worry about, because there is no handy little computer-shop around for many light years.

So if you are going for realism, then the proper thing to do is to assume that heat management, computers included, is already dealt with. This is because already today heat management is a very critical issue when it comes to designing space ships.

$\endgroup$
  • $\begingroup$ Wouldn't a reasonable assumption also be that computers aboard a star(space)ship are vacuum-hardened anyway? Your comments regarding being hooked into the ship's heat-dissipation system would certainly be part of that, but not the entirety. Unless the OP is hypothesizing organic or semi-organic computers. $\endgroup$ – Doug R. Dec 21 '16 at 20:16
  • 1
    $\begingroup$ @DougR. I was planning on writing that earlier: why waste breathable air on computers? Now granted this can go both ways: if the outer hull is the seal towards the vacuum of space it would make little sense to make a vacuum inside the spaceship just for the computers. But if the hull is partitioned, the computers could be residing in a "No/Low pressure" area. $\endgroup$ – MichaelK Dec 21 '16 at 20:21
4
$\begingroup$

It would very much surprise me, if a spaceships computers would have any problem operating under vaccum conditions.

Our current desktop and servers are designed for use in an atmosphere, mostly because thats the cheapest way to make them and the only environment they have to operate in. That said, computers can be readily made to withstand harsh environments - think of avionics systems in aircraft (especially military), space probes, communication satelites.

Thermal management by airflow is just the cheapest way to do it, heatpipes and radiators work just as well - but they cost more, so those are only used where needed.

Also, considering how much computing power we can already fit into a smartphone, I doubt a spaceship would still have a "server room". It seems much more likely there would be multiple, redundant racks distributed over the ship (again, look at military aircraft). A stock PC mainboard is pretty much not repairable anyway (costs more than replacing), so if need be, they could be encased in epoxy resin (of course the heat management has to be designed in), making them pretty tolerant to environmental conditions (Mainboards for south-east asia come with an extra moisture protection layer for example).

A spaceship where the computers can't stand vaccum and all the computer power concentrated in a single point of failure seems unrealistic. Even small business server solutions today can be built distributed and redundant (and above a certain company size its practically standard). Redundancy is standard and mandated for (civil) aircraft.

It has been argued that harddrives require atmospheric pressure to work properly. Again, thats the cheapest way to make them, not an absolute necessity. The only harddrives I know of that have been to space were in laptops on the ISS. Everywhere else either solid state memories or good old tape drives (out of fashion by now) have been used.

Also, you do not need an awful lot of computing power for everything, even if there was a supercomputer needed aboard the ship, most basic functions (like environment control, doors, stationkeeping, attitude control etc.) can be easily managed by small embedded systems with power consumption in single watts (tops). Again, look at smartphones today - they already provide thousands of time the computing power needed. We landed on the moon with a tiny fraction of that computing power, flew to jupiter, landed on mars, all with the computing power of a cheap $5 keyboard microcontroller and so on.

A spaceship with a computer not working in space breaks any suspension of disbelief for me personally. Even students can build cubesats as university projects nowadays.

$\endgroup$
  • $\begingroup$ Best and most solid answer so far, by my opinion. $\endgroup$ – MolbOrg Dec 22 '16 at 2:06
3
$\begingroup$

You describe a scenario where it is impossible to get into the room. I can think of only one solution:

Let them freeze.

If you spray water (which will freeze due to decompression) onto your computers (which I hope are electrically insulated), the resulting snow will act as evaporative cooling. Basically, when you put water in a vacuum all of the highest-energy molecules fly off and what remains is so cold it freezes solid. This will absorb heat from the electronics, cooling them at the cost of sublimating the ice until it's gone. Space suits are cooled by this effect in real life.

Of course, any water pipes that you happened by some miracle to have in the computer room would freeze solid for the same reason, preventing you from getting any significant cooling. But just think for a moment, what similar system is there in a computer room? Your spacecraft might have a CO2-based fire extinguisher system. Open it fully and let the dry ice spray onto the computers and it should cool them for a while. They're already designed for the CO2 to freeze by decompression so it should work, albeit not for long.

You can also vent nitrogen from the life support systems onto the computers. Solid oxygen results in Bad Things and should be left to the adventurous.

$\endgroup$
  • $\begingroup$ I was going to write similar answer, assuming the atmosphere is not instantenously lost. If the leak is small enough, the computer could release water every now and then, keeping pressure relatively low, but not complete void. The cool water steam would fill up the room, hottest steam (heated by the CPU) rising on top, some of it escaping into space. If the leak is small enough, the water might momentarily even seal it by freezing - but I guess that would be just lucky incident. $\endgroup$ – diynevala Dec 22 '16 at 10:38
  • $\begingroup$ @diynevala: There is no convection in microgravity, so saying that the hottest steam would rise to the top is false. In microgravity, heating the gas in one part of a gas-filled room results only in local heating until the heat eventually conducts away. Spacecraft must use forced convection. The convective cooling you want to introduce is something completely different from evaporative cooling. $\endgroup$ – user25972 Dec 22 '16 at 10:56
  • $\begingroup$ Assuming they have no rotation to create artificial gravity, there is still one more force moving the cool water around: the cpu fan, if they got one. Most probably, if the cooling system is not very modern, they got lots of fans. Anyway, this is just a race against time. $\endgroup$ – diynevala Dec 22 '16 at 12:35
  • 1
    $\begingroup$ True but at the high rate at which the question says the computer room is leaking, the pressure would drop too quickly for the fans to be much use for cooling. I'd rather wait for the pressure to drop sufficiently that the computers can be cooled with CO2 & nitrogen snow than trying to keep the pressure up. $\endgroup$ – user25972 Dec 22 '16 at 14:06
  • $\begingroup$ We clearly interpreted the "alarming rate the room is venting" differently. I thought that the leak would take minutes, during which they'll need to decide how to fix the situation. If all the air is gone in seconds, they are pretty much screwed. In any case, they might be better off wearing suits, and attempting a maintenance trip to the computer room, at least sealing off the leak. Also, turning computers off (or as low a load as possible) might be a wise move until the cooling is guaranteed. $\endgroup$ – diynevala Dec 23 '16 at 7:12
3
$\begingroup$

Space is quite an alien environment. Being so close to a perfect vacuum makes conduction almost nonexistant. Zero-G erases convection even inside the spacecraft's atmosphere. Radiation is the weakest of the heat transfer methods, but it's the only method left.

Others have already pointed that you cannot rely on air cooling at all, and mentioned the black body radiators placed on the exterior of the spacecraft. About that, I'm just gonna drop a link: https://science.nasa.gov/science-news/science-at-nasa/2001/ast21mar_1

Silicon semiconductors have very useful properties, but they don't need a lot of heat to lose them. That's bad.
But if your story is set some decades in the future, maybe the obstacles in the way of carbon semiconductors (diamond, graphene, etc) have been worked out.
Carbon semiconductors cool faster and withstand higher temperatures than silicon ones. Very convenient in our predicament.

Also, if you can, roll the spacecraft so the breached area doesn't face the local star.
Stars irradiate a huge amount of heat. They will cook anything unshielded if it's somewhat near them, and you just lost your shielding. Oh, they also spew a lot of stuff that wrecks unshielded electronics faster than heat.
So, yeah, turn the breach away, even if special shielded transistors have been a thing for decades.

$\endgroup$
  • $\begingroup$ I would recommend you to cite or make small annotation for your link, to much to read about obvious stuff and essence is not guaranteed. $\endgroup$ – MolbOrg Dec 21 '16 at 0:17
  • $\begingroup$ Diamond semiconductors are already a thing, just not cost effective for mass market use at this point. But they would definitely be an option for critical shipboard components. wired.com/insights/2015/01/the-rise-of-diamond-technology $\endgroup$ – Rozwel Dec 22 '16 at 20:08
3
$\begingroup$

You may find that due to lack of fans and atmospheric cooling the computers overheat.

Modern day computers can throttle back activity when hot, so lets assume yours do the same.

Likely, they'll keep running, but only at a few 100Hz rather than the few GHz you'r used to.

So your hapless astronauts will experience serious lag in anything they ask the computer to do.

The computer will stop making on the fly calculations - no route planning needed at the moment so that is ok.

Maybe voice activation and holographic interaction (if your computer did such things) will stop, and the only control will be via command line.

But since you said that power was unlimited - where is you generator venting its heat? If the heat sinks are still working, you may be able to get computers in that part of the ship to become the main computers and run at full speed.

$\endgroup$
  • $\begingroup$ Yes, exactly. Also your computers won't be THAT centralized, you'd have distributed systems. I mean your car does, so certainly a spaceship would. $\endgroup$ – Harper - Reinstate Monica Dec 22 '16 at 17:31
2
$\begingroup$

If this room didn't have any/minimal moisture, your computers would probably run regardless of having atmosphere or not because "cold" doesn't really harm a computer (except when approaching absolute 0 when matter starts to slow/stop) where the actual problem arises is that if there is moisture, as the room cools from the lack of atmosphere, it'll condense on the PCB traces and short something whereas if there is little to no moisture this won't happen and the computers will run just fine for awhile.

However, the real problem is the fact that once there is no atmosphere, the processors can no longer conduct heat away from anything at all because there are no molecules to move heat away from it unless your computers are liquid cooled and then it's still likely to eventually go because the motors of the liquid cooling system wouldn't be able to cool themselves unless they're in a different part of the ship that still has atmosphere.

$\endgroup$
  • $\begingroup$ You're thinking in terms of heat transfer planetside where the atmosphere has a huge volume and thus almost unlimited heat capacity. In a sealed vessel, the room air will never be essential to cooling equipment, because transferring heat from the equipment to the air only results in overheated air. $\endgroup$ – Ben Voigt Dec 22 '16 at 19:24
  • $\begingroup$ @BenVoigt - Actually, I'm just thinking in terms of basic heat transfer that you need to have either a gas or liquid to conduct heat and a vacuum isn't going to do it. As well, regardless of the heat capacity of the air in that compartment, it will still have more than a vacuum which is essentially 0 meaning that if these computers had a cooling system that involved the compartment itself, it would eventually burn itself out. $\endgroup$ – rangerike1363 Dec 22 '16 at 19:36
  • $\begingroup$ You're thinking that the air is a sink for thermal energy, but the reality is that it can only conduct the heat to somewhere else, which ultimately has to result in transferring the heat to outside the sealed environment. It's generally better to use heatpipes, etc, to get the heat to that "somewhere else" where it actually leaves the vehicle. $\endgroup$ – Ben Voigt Dec 22 '16 at 19:49
  • $\begingroup$ @BenVoigt - Air is a sink, just like nearly every material is capable of being a sink, it's just a poor one. I mean, you could even run plain old air through a closed loop "liquid" cooling system and it would cool because it would transfer heat away but it's better to use something like water that has a very high specific heat capacity. The point is you don't want the semiconductors holding the heat. $\endgroup$ – rangerike1363 Dec 22 '16 at 19:58
  • $\begingroup$ What is cooling the air? It's not a sink (actually, nothing is a sink, only a conduit). You say "transfer heat away" but never mention "where to?". Once you figure out where to, you find that the spacecraft designers got the heat from semiconductors to "there" without using air in between. (In a spacecraft, the "there" found at the end of all heat conduction is the outer wall of the spacecraft, which loses the heat energy via blackbody radiation not further conduction) $\endgroup$ – Ben Voigt Dec 22 '16 at 20:05
2
$\begingroup$

These astrogation computers are quantum computers. They need to be cooled to near zero K anyway. You can't do that with air, so it makes no difference if there is air in the server room.

$\endgroup$
1
$\begingroup$

Freezing is probably not as much of an issue as you might think. If there's sufficient moisture in the air to cause ice crystals to form, your servers are screwed anyway. If there's not that much moisture, you can try Alexander's suggestion of running everything to keep the systems "warm."

If your leak is big enough that there's no atmosphere in the room, then traditional hard disk drives will cease to function. They typically rely on a layer of air to support their heads at the appropriate height (I'm not certain if NASA has a design that works differently, but Solid State Drives do not have this failure point, they are just more expensive currently).

If the hole is very large, you might have to deal with radiation impacting your equipment. This could cause all kinds of fun failures (this is assuming that your servers weren't built with some kind of shielding, but the room itself was, which makes more sense if people were ever supposed to be in this room).

$\endgroup$
  • $\begingroup$ as others have noted, freezing is not the problem. Vacuum is a superb thermal insulator. Overheating is the problem! $\endgroup$ – nigel222 Dec 21 '16 at 9:48
  • $\begingroup$ It depends on how cold it gets. Anything with mechanical parts suffers from extremely low temperatures as components become more brittle and wear out faster, if not break outright. $\endgroup$ – Marshall Tigerus Dec 21 '16 at 14:46
1
$\begingroup$

As you mentioned, spaceships are a peculiar thing. To answer your question, "How can I keep the computers on my spaceship from dying after a hull breach": design it well while still on the ground. NASA has already done this for their satellites.

First, temperature control.

I have never warned my clients(I am a computer technician), "Now don't let that computer get too cold." In fact, freezing a failed hard drive can sometimes be useful. Heat, however, is an issue. As many have mentioned general computers use air to siphon off their heat, without oxygen that won't work. It is a good thing NASA does not use general computers. Spaceships use radiators to siphon heat off into space, these vary in design but they all reject heat by infrared (IR) radiation from their surfaces. https://en.wikipedia.org/wiki/Spacecraft_thermal_control#Radiators

When your spaceship is designed just make sure they vent the computer's heat through the radiator.

Second problem, storage.

Standard hard drives need atmosphere to work, so they currently use magnetic tape as a backup. SSDs could also be a solution, they are not currently used due to heat and cost, but your scientists may have worked out this issue.

Alternative. You seem to indicate this is some point in the future after space travel advances somewhat, so perhaps at this point computers are 100% light based. Gravity and heat should have little to no effect.

$\endgroup$
1
$\begingroup$

While many answers correctly pointed out that computers don't mind the cold much (although even transistors tend to stop working at a few degrees above absolute zero - but that's not an issue for anything producing even the slightest bit of heat). The real issue is going to be if the computers are designed with fan cooled heatsinks or not.

See, fan cooled heatsinks become quite worthless in a vacuum and if the important electronics is inside an enclosure (almost certainly) then the only method for heat to escape is through radiative effects and that's not nearly as good and conductive or convective heat transfer (think about why thermoses work so well at keeping their contents hot... the amount of heat radiated from shiny metal through a vacuum to some more shiny metal is tiny) Shiny objects don't radiate heat well (to get around that, paint everything black, it can be 10-20 times better at radiating heat than shiny metals) and a vacuum is a really good insulator so your laptop is most likely toast in a few seconds.

However, in a mainframe scenario, there is no reason why the server racks need to be fan cooled, the heat still needs to be radiated away somewhere outside the ship on the side pointing towards deep space. So why not have the computers directly connected to the ship's external radiators? That way you save the hassle of needing to dump heat from the computers into the air and then move the air to the heat exchanges then pull the heat from the exchanges and then dump into space. You skip several stages (and all the mass and complexity that comes with it) although your laptop is still toast.

$\endgroup$
  • $\begingroup$ "although even transistors tend to stop working at a few degrees above absolute zero" - industry grade components are not guaranteed to work below -55C , altera page, interesting link in this answer - they say Silicon based should freeze about 40K, Ge at about 20K, but as device I guarantee they will fail earlier, however amd they had used nitrogen(?) liquid to cool their overclocked processor(they had video) $\endgroup$ – MolbOrg Dec 21 '16 at 0:14
  • $\begingroup$ "Not guaranteed" is very different to "won't work". It just means if you do crazy things, the manufacturer won't give you a refund if the chip fails (most probably from the thermal stress of rapid cooling or heating). The mil-spec CPU with the much higher price and wider temperature tolerance is probably exactly the same chip with more extreme testing and (maybe) better packaging. There are crazy people who try to build maximally over-clocked commercial systems using liquid Nitrogen cooling. The Physics is that CMOS will switch faster at lower temperatures. $\endgroup$ – nigel222 Dec 21 '16 at 9:47
1
$\begingroup$

Supplementary answer, about passively cooled computers in atmosphere. Building completely silent (fan-less) computers has been a hobby of mine for quite some years. It's getting a lot easier these days. Some years back you needed enormous heavy cases with heat-pipe assemblies that attached to the CPU, to transfer heat to the (huge, finned) aluminium case from where it was shed to the atmosphere by convection. And if you got your calculations wrong, the CPU emitted smoke and died.

Today, you can get the power of a modest laptop PC by consuming just ten watts, and the convection cooling requirement has reduced to a finned heatsink about 8cm cubed. (That's with the baroque intel architecture. The ARM chip in your phone does much better still, and is always passively cooled).

Convection cooling involves a high power law of the difference in temperature between the heatsink and the surrounding air. When you are used to passive-cooled systems, you get to know that they will run distinctly warm, even hot to the touch. The CPU is quite happy - commercial ones are rated up to 80 or 90C, and mil-spec is probably well above boiling point (with a degraded clock speed). Also when you test one, you need to leave it running under load for several tens of minutes, monitoring the temperature. It takes that long for the temperature to stabilize with heat input = convected output. In the days of Athlons, it was a bit nerve-wracking, these CPUs could self-destruct if they got too hot. Today's CPUs just throttle their clock speed down if they are getting too hot, and are pretty much proof against damage. They just quit if they get too hot.

OK, in vacuum there's no air, so the system is going to overheat sooner or later, unless it was designed with some vacuum-proof heat transfer plant to carry its heat to massive radiator fins on the outside of the spacecraft. But perhaps, the question is how to make it "later" rather than "sooner"? From my experience, the answer to that is good thermal conductive coupling from the computer to more massive chunks of metalwork. It won't overheat until it has heated all the metalwork up to a "too hot to touch" temperature (60-70C). Which could easily be hours: long enough to rescue the situation, or realize that it's nor survivable.

One fairly low-tech way to increase thermal inertia and coupling is to put the computer in a sealed box filled with oil (low tech, flammable, higher vapour pressure) or liquid silicone "oil" (not flammable, very low vapour pressure). You could do this to a spacecraft system, but there's an obvious weight penalty. Note, this is also a solution to cooling all the other chips on the board, which normally relies on air-convection. Here on the ground, it's a way to build a completely outdoors-proof or even underwater-proof computer: the oil keeps water and water vapour out.

$\endgroup$
1
$\begingroup$

I suggest researching what NASA has already learned about operating computers in harsh environments. If your story is set in the future, your computers will most likely be designed better than what NASA already has. A spaceship, designed when space travel is a norm, will have redundancies in place to deal with these situations, so your server room would be built to run under both atmosphere and vaccuum. You should narrow your problem to a specific failure such as a failure in your communication computers cooling system. Maybe a leak of some sort.

You could then have a scenario where your crew has to disable some important systems to keep the communications system running. They have to run the now empty coolant line through the environmental system to pump air over the communication system, but this causes the temperature of the living area to become dangerously hot.

$\endgroup$
1
$\begingroup$

If you want to imagine a scenario where a hull breach would pose a danger to a futuristic computer system, I would look at oil submersion cooling. It's a cooling method that replaces air with mineral oil (or other non-conductive liquid), and moves the warm oil to a heat exchanger to cool it before returning to the computer. The upside of the submersion cooling method is that the enclosure that holds the oil & computer system can be designed to withstand not only the rigors of space travel, but also hardened enough to survive heavy damage to the computer room. You could then construct scenarios where the hull damage has pierced one of the cooling lines, and just needs a patch. The other upsides you could construct could be that the flow of the coolant can be stopped for a certain amount of time (to allow repairs or create time-sensitive tension) before the computers fail. I think it gives you the flexibility and believability you're looking for.

$\endgroup$
1
$\begingroup$

Huston, if you use regular hard-drives a decompression is likely to damage or explode them. In void melting and vaporization temperatures are different, and you will have no heat dissipation.

Prevent computers from dying in vacum in advance (better computer architecture)

  • Make them cooled with many indipendent liquid based copper dissipators.
  • Allow some failure, maybe some computers can fail without affecting the ship
  • Do not use Hard-drives, or at least not hard-drives filled with air
  • Keep computers in vacum in advance (the reason for this is that air decompression cause a instant drop in temperatur that may break equipment that is under heat stress like processors)
  • Do not keep all computers in same place.
  • Ideally each computer have its own purpose and is loosely connected to other computers
  • Use 16-bit systems instead of 64 bit systems (much more resistent to space noise).

Prevent computers from dying after the accident

  • Someone should go out and fix the hole before something enter in computer room
  • Prevent air from going inside computers' room (eventually doing vacum in nearby sectors)
  • Turn off all unnecessary systems, with the shielding of the computers' room damaged every extra radiation or electromagnetic noise can cause malfunctioning
  • Prepare to lose some computers, move important programs to computers that are doing useless functions in order to reduce damage in case of disaster.
  • Maybe the explosion disconnected some cables (you have to go there and reconnect/replace, or to hack the lan to route data on different paths, maybe sacrificing some other functionality because bandwith is limited)

Ehi basically you have to do nothing more nothing less what a datacenter is already doing. Good luck!

$\endgroup$
0
$\begingroup$

I really feel I should point out some stuff I know that most people here apparently do not. Space contains radiation and that radiation can actually cause bits in a computer to flip! In essence, your danger is not just heat exchange. Even if the coolant continues to work perfectly, your computer's hardware is now un-shielded. Expect the data on the computer to corrupt rapidly. The hard drive might be safe. I'm not real knowledgeable in this radiation stuff. I just read a paragraph mentioning it in the textbook. However, running program's memory will corrupt so expect systems to go down all over the ship whereas your digital library of Klogang languages stay just fine.

$\endgroup$
  • $\begingroup$ It's all about the size of individual transistors. For the geometries used in new consumer processors, like 14nm, individual high energy particles will cause a substantial jump in the voltage where they hit, changing the result. Older larger geometries have more capacitance on the gate of each MOSFET, making them considerable more tolerant of radiation, so they tend to still be used for computers intended for space (as well as near X-ray machines, etc). Making transistors larger works for logic circuits and memories alike. $\endgroup$ – Ben Voigt Dec 22 '16 at 19:55
  • $\begingroup$ @BenVoigt how is that at all relevant? An advanced spaceship needs a more complicated computer than that. $\endgroup$ – The Great Duck Dec 22 '16 at 22:48
  • $\begingroup$ You can make very complicated computers with 70nm production process. You'll need an order of magnitude more silicon than the same computer power using 14nm, but the mass/power/volume increase is not so great as to make it prohibitive. Alternatively, you can use an in-between amount of silicon, and put redundant copies of the 14nm circuitry and use majority-vote and other error-correction techniques (the voting logic had better be the large geometry, but it's a very small fraction of the total). $\endgroup$ – Ben Voigt Dec 22 '16 at 22:52
  • $\begingroup$ @BenVoigt fair enough, but by the same reasoning the computer shouldn't fail from heat as I can just wrap everything in coolant. It's still worth mentioning to keep in mind. There might be less-critical systems using microprocessors like for scientific databases used as reference material. :) $\endgroup$ – The Great Duck Dec 22 '16 at 22:59
  • $\begingroup$ I see in your comments on other answers, you consider the problem to be the loss of integrity of shielding, thus increased radiation. But the fact is, it's much cheaper (in the metrics that matter, mass and volume) to make computers work in space via redundancy and radiation-hardening (larger transistor geometry) than shielding. The only reason the hypothetical spacecraft has any shielding at all is for the benefit of the crew. $\endgroup$ – Ben Voigt Dec 22 '16 at 23:04
0
$\begingroup$

Redundancy.

You have a second server room, in another part of the ship, unnaffected by the issue. Maybe even a third.

It's what we do on fly-by-wire (i.e. fly-by-computer) aircraft nowadays, though you wouldn't call their computers "server rooms".

$\endgroup$
  • $\begingroup$ That's a smart idea. A distributed system could handle everything room by room with certain info transferred via wired networks. I like this idea. $\endgroup$ – The Great Duck Dec 23 '16 at 1:55
0
$\begingroup$

I initially added part of this as a comment to Michael Karnerfors' excellent response, but on further reflection, I think it's worth its own response.

  • The logical place for the computer room of a spaceship/starship is in the core of the vessel, due to its critical nature.

    • Such a ship is so dependent on its computers, that an issue (whether external or a malfunction) could be catastrophic to the ship and its crew.

    • This provides maximum physical protection to the computer.

    • Depending on the type of vessel, the crew may or may not be able to repair the computer or even have access to it.

    • On most civilian vessels, the computer would likely be sealed and would need to be repaired or pulled and replaced by qualified technicians while in port.

    • On a ship designed to be away from port for an extended period of time (i.e., exploratory or military), a drunk-but-brilliant teenager might be able to go into the computer core, replace the isolinear chip while the ship is under fire by the Romulans, and save the day...well, you know what I mean.

    • Conclusion: Unlikely to be an issue, because if the ship is this badly damaged, the crew is likely to be dead or dying anyway.

  • A ship designed to be away from its base for an extended period of time will likely have multiple, redundant computers located in different areas of the vessel and will also use a distributed network.

    • A quick Google search shows that many current NASA missions use dual- or triple-redundant computer systems.

    • This makes it statistically unlikely that a single event or malfunction would impact all computers.

    • Most likely such systems would also cross-check one another periodically to ensure that there are no malfunctions creating issues.

    • In an emergency, you can load the astrogation software on the galley computer and still be able to find your way home. Or make a chicken pot pie. Your choice.

    • Conclusion: Not an issue, because the crew can still get home or wait for rescue.

  • Computers aboard a spaceship/starship would likely be designed to be vacuum-tolerant, to guard against the possibility of loss of pressure.

    • This means they would have other means to dissipate heat, as Michael's post noted.

    • Conclusion: Not an issue.

All this being said, you don't need to have a valid explanation for how you do this. If this is crucial to your plot, you just need a logical explanation for why this is an issue (i.e., organic or semi-organic computer core in a sealed environment). Then get on with your hand-waving and make it happen.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.