Sign in with
Sign up | Sign in
Your question

Smart People: Can anyone come up with any formulas for water cooling?

Last response: in Overclocking
Share
August 12, 2009 10:47:52 PM

Just a random, bored side question...

Can anyone think of any Mathematical Formulas that an be used for water cooling? Ex: Calculating heat transfer in the loop, Calculating the curves in the water cooling loop and using that to calculate water flow resistance, etc.

I just thought it'd be cool while I'm waiting for my last part to build my pc, to come up with a general idea where to place all my cooling stuff, what order to put them in, etc, to get the best out of my water cooling setup.

Only one I can think of is E=Cm(DT), E is the energy, C is the heat capacity of the material (copper, etc), m is the mass, and DT is the increase in heat. It can be used to calculate how much warmer a material gets after heat is transfered to it. Ex: Use this formula to disprove the theory that putting a cpu block before a video card in the loop is bad.

Finally putting Calculus and Chemistry to good use...
August 12, 2009 11:20:42 PM

I thought I'ld ask to above, just because its interesting. But I'm sure I can still get help from people that dont know any formulas.

How about the placement of the rad, res, and blocks? Should one be higher than the other? I have my triple rad at the bottom, res at the top, and blocks in between. The fans on the res are drawing air away, should it be opposite? Anyone see anything wrong with this setup?
m
0
l
a b K Overclocking
August 12, 2009 11:32:37 PM

You doing a science fair project or somewthing?

I know shadow has alot of formulas about surface area etc.

I have one about heat output if your interested?
m
0
l
August 12, 2009 11:43:21 PM

Yeah, heat output would be awesome. Thanks

But nah, I'm not doing any fair. Although maybe the prof would give me a couple of extra bonus points if I showed him that the math was being put to use in a real-world situation.
m
0
l
a c 86 K Overclocking
August 13, 2009 3:50:07 AM

Umm what watercooling setup?

And just do the coolest air to the rad is better. Most of us strive for outside case air to the rad, meaning air temp to the rad.

Ican link you to MANY places where temps, fan CFM, Flow rates matter.

Linky time. Get an A. Tell us your watercooling setup. Odd you dodn't mention it.

Guides
http://forums.extremeoverclocking.com/showthread.php?t=... Pretty up to date info and buying guide
 verclocking-and-cooling&Itemid=86" rel="nofollow" target="_blank">http://gilgameshreviews.com/index.php?option=com_conten... Another good guide
http://forums.extremeoverclocking.com/showthread.php?t=... What to do once all the stuff is in the door
http://www.xtremesystems.org/forums/showthread.php?t=22... Many build logs on MANY cases, great learning tool.

My latest rig:
http://www.ocforums.com/showthread.php?t=604016

Forums
http://www.xtremesystems.org/forums/index.php? Not a noob site, but great stickies
http://www.ocforums.com/ My fav, good peeps, know their stuff, less hardcore
http://www.over-clock.com/ivb/inde [...] opic=20277 A GREAT Europe site
http://www.overclock.net/water-cooling/ Decent site

Tests on equipment, not reviews, truly scientific tests
http://translate.google.com/transl [...] n&ie=UTF-8 Info on rad testing
http://www.xtremesystems.org/forums/showthread.php?t=22... More rad testing

Read here and copy/paste for a A+++++


http://skinneelabs.com/ Host for Martins lab and some newer tests
http://www.skinneelabs.com/MartinsLiquidLab/ Test results, very technical


Stores
http://www.dangerden.com
http://www.petrastechshop.com/
http://www.sidewindercomputers.com/
http://www.jab-tech.com/
http://www.performance-pcs.com
http://www.frozencpu.com/
m
0
l
a b K Overclocking
August 13, 2009 5:49:23 AM

Here is the heat output formula...

TDP_new = TDP * (MHz_new/MHz_stock) * (V_new/V_stock)^2
m
0
l
a c 86 K Overclocking
August 13, 2009 7:23:06 AM

Your cliff note pic looks fine. But what are you cooling? A 486 100 mhz CPU and two 256 meg GPUs?

Or an i7 and two GTX 285's? That's what matters.
m
0
l
a b K Overclocking
August 13, 2009 3:53:36 PM

^NOOOO!!!!! That WILL fail. You CAN'T get a clean enough area for it to work. Oil will work.
m
0
l
August 13, 2009 4:58:01 PM

In order to come up with a specific formula for a specific set-up, one would need:
1.Pump (head, flow rate)
2.Tubing (diameter, thickness, material, length)
3.Waterblocks (how many, size, material, internal lay-out ((the number of tubes, shape, size, height, material, position in relation to the input and output of the waterblocks)))
4.Room temp
5.Radiator size or effectiveness. (if effectiveness unknown, dimensions, number of fins, material, spacing, and flow rate of fan necessary)
6. Flow profile of room (natural vs forced convection, etc. )

With this information, and possibly a small schematic of the layout for natural convection calculations purposes, $180/hr will get any of us Mech. E.s to calculate anything you want to know.

Personally, from my preliminary calculations, I am kind of disappointed in the way these water blocks are formed. Only 73% of the area of the block is actually above 50% effective in removing heat, most are designed with triangular tubing (boo, gimme that surface area, baby!), and there is no profile change in the block to the flowrate. Meaning, it is the same shape from entrance to exit, which is great if the change in temp is low, but since we are trying to get these as cool as possible from as hot as possible, raising some surface area near the exit of the blocks would help get the heat out evenly across the block, instead of having warmer water (from entering the block, heating up initially, and traveling towards the exit) try to cool the block near the exit.
Ultimately, these are all really really small concerns when dealing with such a small area to cool, rendering a relatively small heat removal potential. BUT, in the world of OCing, when ultimately one degree C can do wonders or blunders, and everyone is trying to squeeze the last little .01 ghz from their rig, I would feel like these factors might have been taken into consideration.

That is why I am just going to wait to get my own blocks water-jetted once I have the money.

I know this probably didn't answer your question, so I am going to say this. Go by tests like the links if you really want to get some information. To calculate this properly would take ages and tons of information that is just not available or tedious to come by and use. If you really care, get some college to do a senior project on it and have them optimize the cooling for ya.

My calculations might be off, I kind of half-assed the conversions to three sig figs.

Um.....yeah, I think that covers most of it. I will probably take my time and do a more in-depth view in flow visualization software (if I can torrent). If I do come up with something concrete justifying a different design of a water block, I will probably post it if it costs too much to properly manufacture for commercial use. Otherwise, look for it on shelves within 6 mos.
cheers.
m
0
l
a b K Overclocking
August 13, 2009 5:16:48 PM

^You are not way off man, your observations are pretty much on the money as seen from the sims I have run for the past month or so on aftermarket blocks. Hence why I decided to design my own blocks that use high turbulent transfer and >20k mm^2 area. (The GTZ SA is ~19472.25 millimeters^2).

And yes, cost is very huge. It's ~$150-300+ depending on the design to CNC on Copper C110. A bit less on Aluminum ($100-250). To rapid prototype a block top it's any where from $50-300 (This is the best bet to do jet type designs on, forget CNC for jets).

Check these guys for Rapid proto: http://www.metrorp.com/
I used/use them and they have good prices.

For your sims you should take a look at CF Design, Thermal Desktop, and FloWorks (now part of standard SolidWorks distro).
m
0
l
August 13, 2009 5:40:55 PM

Ah the costs of CNC. They might be falling if you find a shop that has no work (shouldn't be hard in this economy, lol). I know if you say it is for a school project, sometimes they will do it for free, but if you lie, it could be bad news.
I am in love with water jetting after seeing it in action this summer, but meh.

I am planning on getting a personal CNC eventually, since I get off on that type of stuff. Some universities sell them cheap after they use them for 4 years and get another, depending on any contracts they have.

as for the software; FloWorks! I couldn't think of the name. Thanks for that.


I just assumed that the blocks nowadays don't use turbulent flow as much as they should, but thanks for the reassurance. If somehow you managed to cavitate a block in your water cooling system, please let me know, that is just awesomely over the top, and I would like details.

Power to the people!
m
0
l
!