Piezoelectricity for CPU Thermal Waste Management?

sunilraman

okay... so who here has a PhD in Thermodynamics??



anyway, it dawned on this (self-declared slightly autistic, slightly ADD) mind that a lot of thermal waste management (liquid cooling, bigass fans, peltier, phase change, etc...) has got to do with REMOVING the heat from a CPU and other computer components



data centers now face these challenges as more and more 24/7 equipment gets stuffed into increasingly(?) expensive real estate, alongside the current trend that performance per watt starts to hit some nasty exponential-style curves



instead of focusing so much on using more energy to remove heat energy, can we CONVERT some of that heat energy back into that useful stuff called electrical energy?



can we harness piezoelectricity?

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



for those not familiar you know those annoying kids with the shoes that light up when they stomp around all over the mall ~ not to be confused with the annoying kids with the built-in-tiny-rollerskate-thingys in their shoes.



anyway, the idea is that a force applied to a piezoelectric material generates a voltage difference IIRC and vice-versa.



what is the worst thing that a high-performance, kick ass pr0n-downloading, email checking, solitaire-playing machine do?

it generates lots of heat.

now what does heat do?

heat can cause expansion

what can expansion do?

expansion can generate a force

what can the force do?

this force can be applied to a piezoelectric material

hence

converting heat energy back to electrical energy



assuming such a device can operate at say, 30% efficiency, that means 30% of heat energy that is just vented out of a computer could technically be converted back into electricity ~ omfg powering the computer itself!



now i'm not being an idiot here fantasising about a perpetual-motion-type computer which is powered purely on its heat waste, because that's not quite possible with current available physics and there's that annoying entropy thing.



but think about if you could harness just 30% of that heat waste and turn it back into electricity.



who thinks this is possible? who thinks i'm on crack? who wants what i am smoking?

brendon

Quote:

Originally posted by sunilraman

okay... so who here has a PhD in Thermodynamics??



anyway, it dawned on this (self-declared slightly autistic, slightly ADD) mind that a lot of thermal waste management (liquid cooling, bigass fans, peltier, phase change, etc...) has got to do with REMOVING the heat from a CPU and other computer components



data centers now face these challenges as more and more 24/7 equipment gets stuffed into increasingly(?) expensive real estate, alongside the current trend that performance per watt starts to hit some nasty exponential-style curves



instead of focusing so much on using more energy to remove heat energy, can we CONVERT some of that heat energy back into that useful stuff called electrical energy?



can we harness piezoelectricity?

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



for those not familiar you know those annoying kids with the shoes that light up when they stomp around all over the mall ~ not to be confused with the annoying kids with the built-in-tiny-rollerskate-thingys in their shoes.



anyway, the idea is that a force applied to a piezoelectric material generates a voltage difference IIRC and vice-versa.



what is the worst thing that a high-performance, kick ass pr0n-downloading, email checking, solitaire-playing machine do?

it generates lots of heat.

now what does heat do?

heat can cause expansion

what can expansion do?

expansion can generate a force

what can the force do?

this force can be applied to a piezoelectric material

hence

converting heat energy back to electrical energy



assuming such a device can operate at say, 30% efficiency, that means 30% of heat energy that is just vented out of a computer could technically be converted back into electricity ~ omfg powering the computer itself!



now i'm not being an idiot here fantasising about a perpetual-motion-type computer which is powered purely on its heat waste, because that's not quite possible with current available physics and there's that annoying entropy thing.



but think about if you could harness just 30% of that heat waste and turn it back into electricity.



who thinks this is possible? who thinks i'm on crack? who wants what i am smoking?

I'll take a crack at it. Once the heated materials have expanded then what? What is the effeciency of Piezoelectric energy conversion? Guess what the by product is, heat. There's some thermodynamics for ya'. Kind of like that saying 'Don't sweat the small stuff and it is all small stuff', the thermodynamics version is a bit shorter 'It's ALL heat' Is the electricity that is being generated usabe? I think that what is happening is some of that high voltage, very low current type electrical energy, it would be like trying to power your laptop using static electricity. This is the same stuff that used to be in those cheap-ass phone receivers, your voice would send out waves they would disturb some crystals and that disturbance would be amplified, way on the other end the same crystals would be disturbed by an amplified version of that electricity and wold cause a magnet to move a diaphram, next to the other persons ear. Nice try, how about using Ammonia, that is what was used in the old gas powered refridgerators. They were cold and had no moving parts just a gas pilot light like looking flame, oh yes they all sported a check valve, ok so one moving part, the spring of the check valve. Sorry to take away from your thread, I still love the ammonia trick. Heck with all of that fancy materials making that is goin' on today it looks like some will at some time soon find a more effecient medium than ammonia, hopefully it will be inert.

wingnut

Actually, I think AMD (and maybe Intel, too) has developed something like this, but haven't taken it that far yet. In theory, it would work like a reverse peltier, turning some of the heat generated by the CPU die to power again. I think AMD came up with the idea back in the hot Tbird days, but nothing ever came of it again. It might be because it's not practical yet, so I think it's a matter of cost at this time. With AMD and Intel being so competitive, it's hard for either to justify developing a new technology without clear performance benefit. While it would be cool for a CPU to increase its energy efficiency, most users wouldn't see a real cost benefit in their energy bill, so it won't likely increase revenue. Granted, on a server/global level, it would become more significant. However, people would save far more energy by just throwing out their CRTs.



Eventually, I think this will be standard on CPUs, but at smaller nodes. Currently, process technology is good enough that exotic cooling layers in the die are unnecessary. However, as dies get smaller, thermal density goes up. Cooling a smaller CPU with a higher thermal density will be much harder, so this situation will demand something creative like thermal recycling. So yeah, I think we'll see it one of these days.

splinemodel

If there were an easy way to turn heat directly into electricity, power plants wouldn't be using steam turbines.



While you guys solve this problem, I'll work on the frequency condenser.

skipjack

My lab partner wanted to do a piezoelecticity to power a cell phone as a senior project. Our calculations showed that it wasn't practical.



Back to the subject, you would need to cycle or provide continuous expansion for the piezoelectreic device to work.



On the other hand, you might be thinking more about a thermocouple. I believe that these have been installed in nuclear reactor cores (besides the obvious application as driving temperature sensing devices).

sunilraman

Quote:

Originally posted by Splinemodel

...While you guys solve this problem, I'll work on the frequency condenser.

heh. okay. fuck this idea. i'm moving on... i hear cold fusion is back in vogue.

sunilraman

Quote:

Originally posted by Skipjack

My lab partner wanted to do a piezoelecticity to power a cell phone as a senior project. Our calculations showed that it wasn't practical.



Back to the subject, you would need to cycle or provide continuous expansion for the piezoelectreic device to work.



On the other hand, you might be thinking more about a thermocouple. I believe that these have been installed in nuclear reactor cores (besides the obvious application as driving temperature sensing devices).

cool. i was just wondering if the latest advances in material science, nanotech, 'smart fabrics'



gawd... i watch(ed) too much sci-fi. thank goodness i'm doing multimedia stuff nowadays and not physics or engineering.

telomar

The biggest problem is the heat that isn't turned into electricity. The substances they use for these are great insulators so in the end you make the whole heat situation a lot worse because it is even harder to get the heat out of the area. Small gain for a massive amount of additional pain.



It's the same when designing the heat recovery/control systems on a chemical plant. Often recovering every last bit of heat just becomes uneconomical and you are better off going with the less efficient method that solves your problems than trying to save every bit of energy.