Water cooling maths?

[Fat Tone]

Hi
Does anyone have a link for the maths involved in specifying a water cooling system? e.g. for a cpu that produces x Watts, how may lpm do I need going through what size rad to achieve y temperature on the cpu given an ambient temperature of z.

I've looked, but can only find hand-waving, and I can't remember enough of the physics!

Cheers

Tony.

[DuRaNgO]

This sounds like a job for "Super Greebe!"

[Novdid]

I'll give you this. 1g (0.001 kg) Water absorbs 4,4 J energi for every K (or C) when it warms up. Now you can figure it out I don't have time nor energy. Well it depends a lot on other factors to though.

[Fat Tone]

So, we use std. h/s calculations for energy input to water, flow rate to calc temperature rise of water, and some sort of reverse h/s calc for heat output by the rad?
Great, now someone with a better memory than me please remind me of the precise details! btw, I thought it was 4200J/K?

Thanks

Tony.

[Novdid]

Yes that is correct it's 4,18 J for one g, 4180J for one kg. That's how it goes when your fingers don't follow you.

[rubank]

Why can´t you guys stick to calories in your calculations so an old man can follow.

I know nothing of Joule, but I remember Yule. We had one last year, and it was a lot of snow.
I remember it took a lot of calories to shovel my way to the post box, and then some to melt the snow from my boots into a nice little pond inside my door.

rubank

[Greebe]

Well to be honest unless you can test every component yourself, calulating this in the design stage would be a waste of time.

Reason: No radiator, pump, water block (including line (hose)/reservior losses, etc) specs are ever stated.

It's much easier to build and then to test after the fact.

Some rules of thumb do apply tho... you don't need high pumping volume. Typically ~ 1-1.5 gallons/min (~3.79-5.68l/min) is sufficient. You can use a submersible pump tho they do raise the water temp much more than an external pump... the size above would raise the temps by ~3-4c typically. The waterjacket (water cooled HS) is very important in how it's designed and what material it's comprized of. Copper being virtually twice as good as Aluminum, but also how the internal layout is also greatly affects thermal transfer. Greater the surface area the better.

Radiator size and design also greatly effects getting rid the dreaded heat build up... aluminum thin fin designs (your typical auto oil cooler) aren't as good as they first appear, but they're cheap ($). Auto heater cores are much better typically cause they're made from copper.

With my setup the objective went for lowest cost (bang for the buck), lower the sound level and utilizing as much materials I had on hand. Which is what mine does very well. Remember I'm running an Athlon500@850 (producing 85.4watts peak!), it's thermal headroom is extremely limited. Thus if room ambient temps prior to water cooling exceeded 77f (25c) it would crap out if really stressed (ie gaming Rendering 3D, converting vid files, etc). Now I have an additional 7c of room ambient temperature headroom, which suits me fine, especially here in Hot/sunny Florida!

The other major issue with air cooling was noise the 120mm 136cfm fan I was using for air cooling produced... it's spec'd at 50dBA... tho I know this is lower than what it actually was because it's a sleave design (tho an industrial AC fan with cast aluminum frame) which is old and moldy (worn out)... if mounted it induced a great amount of vibration into the case (or HS) which augmented the noise levels greatly.

With the water cooling setup, I'm utilizing 2 el cheapo 80mm case fans... much much quieter

First thing to be replaced here shortly will be the water jacket. The one I currently have is a modified P2 stud pin aluminum jobby... cost me just under $3 to make. Of course I will mod some copper HS to make it instead of buying a prebuilt for alot more money. (can't get making as much as possible out of me blood

I haven't given a spec on the performance of what I'm running now because extreme effeciency wasn't my goal. Make it work better than any air cooled HS and keep the sound level to a minimum (within reason) for as cheap as possible was. Total cost ~$75

[Paddy]

Thanks for the very informative reply!

[Fat Tone]

Thanks Greebe, very useful. Now lets see what damage we can do to Paddy's bank balance (and hopefully not his rig!)

Cheers,

T.

[Greebe]

wouldn't it be easier to just install a sprinkler?

[Fat Tone]

nah - there's always the risk he might twig that one!

[Novdid]

Tony, I've always wondered. What's that avatar of yours meant to resemble?

Waiting eagerly for an answer.

[rubank]

It´s his first attempt at digital photography.

[Novdid]

AAAAHH!!!

Englishmen should stay away from difficult tasks like that.

[lenl]

Just to add my 2cents, I work in the refrigeration and heating field and can add the following.
1 watt = 3.42 BTUs
a BTU is a " British Thermal Unit" and is a measurement of heat
- the amount of heat to raise one lb of water 1 degree farenheit-
BTU,s are a constant used measurement in regards to heat loss or heat gain in North America on all heating and cooling equipment. As greebe had stated it depends on your objective.
If you are looking to cool the water in your process with just the ambient air temperature in the space (house, apt, etc) then you have to look at that as the extreme temperature you are working with. That will be the warmest air that will be entering the heat dissipating rad / coil. So lets say if it is 75 deg. F air going into your rad, the closer you want your aproach temperature, the larger the coil and fan have to be. So if your processor can easily tolerate 85 deg. Farenheit water in the heat sink then you would be working with a 10 deg. F temperature difference that you want to achieve through the heat dissipating rad / coil.
If the processor generates 85 watts, then that would be 290 BTU,s per hour we would be looking to get rid of. To transfer this through the water at a 10 deg. F differential you would have to pump approximately .06 GPMs U.S. or 3.6 gallons (U.S.) per hour, of which a 50 to75 watt pump should do easily. Now for the heat expelling coil, Coils are given BTU rating dependent on many factors - The size of tubing used, how many feet of tubing in the coil, the type of finn material and finns per inch and of course air flow and size. If you want a quiet system you go with a larger fan blade, slower rpm motor and a large coil with wide finn spacing, but then it may work against you if you are cramped for space. If you want your processor to run cooler lets say 5 deg. F T.D. , then you double the water pumped and the air going through the coil. Its all relavent. The whole idea behind water cooling is to try and maintain a cool and constant temperature on the processor,
helps keep the cabinet temperature down, frees up what usually becomes a very cramped area, and sound. I think TomsHR may have some info on this subject and components avaliable, but if you are looking to do the "handyman" thing Ill see if I can get you an Idea of sizes, coil and air.

Hope this helps

Lenl