Evap Waterblock

[Boli]

Here’s a random thought:

Now given that when water evaporates it cools is it possible to design a waterblock that will incorporate this useful effect.

I have NO IDEA how this could be accomplished but I thought why not let people with more practical experience think this over. (or at least point a curious mind in the direction of previous thread where such a discussion already took place and the feasibility of which was flamed out of existance)

## Start Spiel

I was thinking of a dual pass waterblock where water would pass as per normal (maybe following the Maze4 design and above separated by a semi-permeable-membrane (guess who stayed too long at a computing lecture and woke up in a biology one ;P) would be a second area where a jet of dry air would cause forced evaporation.

This will hopefully help create some sort of temperature gradient in the water helping it to draw more heat out of the copper.

Maybe even (getting ahead of myself here) heat pipes could be looped round from the base to this second air chamber to cool.

## End Spiel

~ Boli

[Cathar]

Basically you're talking about heat-pipe technology. However as you've described it, it would seem to require a constant supply of water as you're injecting "dry air" which implies that the system is not sealed.

[Aardil]

Not that I have a clue but some how I think it might work if you use a bong type system since they are not sealed anyway.

Probably way off base ( I usually am)
Aardil

[Boli]

I understand the principle of "bongs" but I wasn't thinking along those lines. Maybe more along the lines that the actual waterblock itself allows for evaporation.

So I'm just trying to open the discussion up (ignoring waterloss for the time being) is can you make a waterblock more efficient using an evap method or maybe even a hybrid between the two?

~ Boli

[Incoherent]

I think that in order to make this idea work, the necessary temperatures are too high. i.e. it would need to be boiling to transfer the amount of heat needed and at atmospheric pressure 100C is a bit too high for a CPU.
If you evacuated the WB chamber it might work, feed a small supply of water and draw off the vapour to somewhere it can be condensed and reused. Oh, a heatpipe!
If you took the vapour away via a tube to be dumped or reused then you have part of a very conventional phase change system with a very inefficient (at these temperatures) refrigerant.
It'd be very interesting getting it to work, there are very critical balances of pressure, temperature and water volume which have to be maintained in order for it to work at all, let alone be efficient. I'd be concerned that the band of efficiency would be too narrow for the load fluctuations of a CPU.
It'd be pretty neat though.

Cheers

Incoherent

[bigben2k]

There's a way to make it work. I've taken this on as a project, but unfortunately I have to keep it under wraps.

All I can say is that it was inspired by a military grade cooling solution.

[-J-]

i believe i've seen someone using the evaporation concept.

the system had the cooling circuit at a low pressure (as in heat pipes) but used a pump to have a higher flow of coolant.

the coolant evaporated with a little variation on its temp, cause of the low pressure.

basically its a phase change system without a compressor, using the radiator to return the coolant to its liquid phase.

its a complex system cause it used all copper pipes, and the low pressure, but its simplier than a phase change unit, and cheaper too.

ill try to find out the system of this guy, he was cooling 2 xeons.

[WAJ_UK]

I don't know if anyone has seen this before but hp have had a go at evaporation cooling. http://www.hpl.hp.com/news/2002/apr-jun/cooling.html

[bigben2k]

Originally posted by WAJ_UK
I don't know if anyone has seen this before but hp have had a go at evaporation cooling. http://www.hpl.hp.com/news/2002/apr-jun/cooling.html

Nice one! I wonder how they map the heatload: IR imaging?

[Volenti]

I made an evaporatively cooled heatsink ages ago, I can't remember exactly but it was on the order of 5-7c better "wet" than dry.

It only held around 50ml so it needed topping up every 15 min or so, at least it could run dry without adverse affects...

[Volenti]

.

[bigben2k]

Originally posted by bigben2k
Nice one! I wonder how they map the heatload: IR imaging?

Nice work, Volenti!

HP's solution I guess, is enclosed, I believe. Different cat...

I thought about a fine grid of thermocouples, for mapping the heatload. Makes more sense. I wouldn't do it though: seems like a waste of effort to try to optimize the flow for the heatload, by using mappings. Maybe later, when the heatloads reach outrageously high levels...