Watercooling help

[mainomega]

Hi, I am planning on setting up a watercooled system on my new rig. I already bought a New heater core as the radiator but now i come to my hardest questions. what Pump and block to use. I heard the mase 2 is still slightly better than the mase 3, is that true ?

Please advice me. Mainomega@mainomega.com

Thanks guys

 

[svidanag]

You've got to include what sort of rig you've got. Processor? Case? Do you want it to fit in your case?

 

[mainomega]

Sorry, i wana run a Xp processor but i want the whole water cooling system outside of the case. Also.... is a bigger reservoir better than a small one ? thanks

 

[panhead49]

the bigger reservior will have more water.......that is better because it will equate to lower temps...the more water you have ,..the longer it takes to circulate ...the longer the water stays in the radiator.......lower temps...if your fitting all your stuff outside your case.....go for the bigger reservior....

 

[mainomega]

would you recommend a block to me ? I was looking at the Maze3 or Maze2. I heard that the Maze 2 did a little better because the incoming water was right ontop of the cpu.

Oh yea.... anyone know where i can get good price on a pump ?

Thanks again guys

 

[MrThompson]

I believe the first step may be to further your education. Try reading the articles in the Watercooling section at Overclockers. Be sure to read Radiator Heat Dissipation Testing by Bill Adams. There is a lot of good info in Bill's article about tuning the flow in watercooled system.

The Maze 3 should out perform the Maze 2. I would recommend a copper top if you go with the Maze 3. Swiftech makes some fine waterblocks too.

For the pump, Danger Den has very competitive prices on Eheims.

And finally, a large reservoir will not lower your temps. At some point the water temperature is going to reach equilibrium. Also, having more water in the system does not increase the time the water stays in the radiator. Changing the length of time the water stays in the radiator would require a reduction in flow rate. The volume of water in a system has no effect on the flow rate.

Perhaps the best reason for a reservoir is as an air trap at the highest point of the system. This makes filling and bleeding easy. I also recommend a drain line at the lowest point of the system. Not only does this make it easy to drain your system, you can also release the air pressure when filling the system to prime the pump.

 

[Lizardman]

For the pump I agree with an Eheim. You might as well get the best since you will be depending on this pump. As far as a water block goes I would recommend a spiral. You can get one here.
Case-mod

 

[mainomega]

Thanks for your help. I saw the prices on danger den, and i found a Ebay store that sold them for about $10 less. I decided to go with the maze3 with the copper top too. One question, i know that adding antifreeze into the solution would lower the heat capacity of water (you know what i mean). But i've seen some water dies, that would glow in the dark... would that cause a change in the cooling capacity too ? as of now tho, H2O, watter wetter, and a bit of chlorine sound good.

 

[h2sammo]

Water molecules H-bond to each others. H-bonding is one of the strongest intermolecular types of bonds, so it takes a more heat to brake this type of bond (that would be the case when water boils). This also means that water has a very high heat capacity, which means it can "ëat up" a lot of heat before changing its properties. Adding antifreeze (ethylene glycol), or water wetter, or any other alcohol based solution, will decrease the heat capacity of water. Molecules of the alcohol will "brake the H-bonded water lattice and will lower the "strentgh" or capacity of the water molecule lattice. This will result in a decrease in heat capacity.

However, adding antifreeze will keep your reservoir water from freezing, if you go below 0 C in water temps. It will also help keeping the water clean with its toxic preperties. It will kill bacteria, fungi and the like breeding in the water, on the walls of the radiator/waterblock. You dont want these creatures thriving in your system because they will build up on the walls of the mentioned parts and lower heat transfer between the moving liquid and the hot/cold metal. Water wetter has the same use, its an antibacterial solution. IT DOES NOT MAKE WATER STAY COOLER. That is a myth, if you can explain that one to me, you could proly win a Nobel Prize in Chemistry.

hope this helped.

 

[h2sammo]

Now, about Bill's article on flow rates and radiator cooling, theres something wrong with it. The formula for determining total heat transfer is:

Q = Ww Cp (T - Ti), where:

where Q = total heat transferred, Btu/sec
Ww = coolant flow rate, lb/sec
Cp = specific heat of coolant, Btu/lb°F
T = exit temperature of coolant, °F
Ti = initial temperature of coolant, °F

Now, this formula has been determined by very intelligent physicists, and has been tested and approved by other physicists around the world. It is accepted by anybody, taught in schools, its not a subject of debate, its common knowledge.

This formula indicates that the total heat transfer is directly proportional to the coolant flow rate. This means that if you increase one of them, the other increases proportionally. If the water flow in the radiator is increased, the heat transfer should also increase.

Now, this is simply reading the formula, and needs not to be proved, or anything like that, its just fact. However, Bill's results show that in most of the radiators tried, the heat transfer is lower at higher cooling flow rates than at lower ones. This is impossible, based on the very formula he has been using to reach his conclusions.

So, since his results cannot be reinforced by accepted physics, than one of 2 things can be true:
1. His experimental technique was faulty and his results anecdotal, in other words, the experiment is not valid.
2. The experiment has been carried out correctly and it shows that 100 years of physics in flow dynamics is wrong and he should be internationally acclaimed.

Now, i do not want to sound mean to him, i am an experimentalist myself, and i understand how these things are.
What do you guys think?

 

[BillA]

new forum for me, hello all
(thanks for the suggestion Colin)

h2sammo, thanks for reading the article - and thinking about the results
some general comments:
- as with many thermodynamic 'experiments'; the results are quite dependant on the quality of the equipment used, and the procedures used to generate the actual data
- a specific problem with the rad data was the use of a digitized variable orifice flow meter which, while providing lots of resolution and having been calibrated, was later found to be inaccurate - by varying amounts at different flow rates (the worst of all possible conditions)
-> this flow rate error accounts for much, but not all, of the 'bump' seen between 0.5 and 1.0 gpm

wrt the decreased cooling seen at higher coolant flow rates, I suggest you review again that equation

when the higher coolant flow results in a decreased change in the temperature difference between the inlet and the outlet (times the flow rate, etc), then the work done by the rad decreases as well

no mystery
and can be understood by considering the convection rate difference between the relatively high coolant/copper rate and the much lower copper/air rate
('residence time' is a term sometimes used with rads indirectly addressing this effect)
- the 'solution' to this difference is to add fins, and to increase the air flow
-> but in this testing sequence the air flow was held constant while only the coolant flow was increased, to look at specifically that effect in isolation

"What do you guys think?"
I can only speak for myself h2sammo, but I think your analysis of that equation and my data was all wet
the body of knowledge of physics is unchanged

be cool

 

[Lizardman]

BillA I understand what your saying and what you mean. But I will have to agree with h2sammo
on this one. BillA you are saying there is a decreasing benefit with higher flow rates right.

This is like the law of Diminishing Returns in Economics. At first you get the most benefit out of something but the more of that thing you want the less helpful it becomes. Such as food. At first you are very hungry and eat a pizza but eventually that more pizza is worth less.

The difference here is that the rad will not care how much water it gets.
but the faster the flow rate the cooler the overall temperature of your water cooling system.

The only reason I could see for a slower water flow BillA would be if your pump could not provide a higher flow rate or that your sytems tubes and pipes could not stand the abuse of the higher flow rate for some reason.

Bill that was a very good article you wrote but no matter how you look at it the formula makes sense higher flow rate = higher heat dissapation.

 

[Bosw8er]

Try the HWFaq, section Cooling-Howto ... scroll down

 

[BillA]

over the years the 'net has taught me that there is no possible contest
against a belief system and a closed mind, but I'll give it another whirl

for starters:
described in that article are test results which have then been used in a calculation
no one is suggesting that the manner of calculation is incorrect
so the only possible disagreement can lie with the test results

here an excerpt of the raw data for the 'surplus' rad in the article:
(corrected for the flow rate measurement error)

amb T . . . in T . . . out T . . . . . delta T . . . . . . flow . . . Btus
22.91 . . 32.91 . . 29.09 . . 3.82C . 6.876F . . 0.32 . . . 1100
23.43 . . 33.42 . . 32.77 . . 0.65C . 1.170F . . 2.06 . . . 1205

as these results did not support my statement, I then recalculated the entire data set
and while the above represent the worst case (30% to no error), the results are clear

increased flow will (slightly) increase the heat dissipation by the radiator

h2sammo, you are correct - the "experimental technique was faulty"
thanks for rubbing my nose in it, as that is what it took to make me re-check the data

seems that it was my closed mind that was impeding
the correcting of an error of which I have known for 6 months
I will see how Joe Citarella wishes to handle the article's correction

as I have had numerous requests to use SI units, I'll try to address that at the same time

Bosw8er, to which specific article were you referring ?
there are 117 links on that page
I have never seen ANY quantified cooling test results - but would like to

be cool

 

[h2sammo]

Bill, i read over my initial post and i realized my tone of "voice" was maybe too sarcastic. I apologize for that right now.

I appreciate that you took my note seriously and relooked at your results. I am not a specialist in thermodynamics, and i spent all all this morning trying to interpret your first response and trying to figure out whether convection is dependent on flow or only on the metal the radiator is made of and the coolant used.

From what i figured, the convection rate should be independant of flow rate (but again, my knowledge of physics is rather basic) but dependant on the coolant and the radiator metal. If this is correct, even though heat passes faster from Cu to H2O than from Cu to air, like Lizardman said, the "rad will not care how much water it gets". No matter the flow of the liquid touching the radiator, heat will convect at the same rate, since the only thing that matters is the kinda of liquid/metal combination used. So, then, coming back to the formula, indeed, the higher the flux, the higher the work done by the radiator.

But...thx God, you posted again, and interferred with the natural increase of entropy in my head. Now things appear orderly again . Your new, corrected data shows that at the highest flow rate, as compared to the lowest, the change in "deltaT" is obvious ~83% change, and consistent. DeltaT decreases as the flux increases. In other words, the faster the water moves through the system, apparently, the less the radiator is able to cool the water passing through it.

Theoretically, if water goes faster through the radiator, a fixed volume of water will spend less time in contact with the walls of the radiator. However, it will come through the radiator more times than if it passed through it at a lower speed (a higher cyclic frequency). Now, what we have to determine is what relationshit is there between this "cyclic frequency"and flow, so that it maximizes Q(radiator).

Nonetheless, it looks like flux and deltaT are related in such a way that at a ~85% increase in flux, you get a ~83% decrease in deltaT - and that translates in the same decrease in convection rate between water and Cu, because air flow on the Cu is unchanged. Now, you did not post any intermediate points so I dont know if the relationship looks like a straight line, or more exponential, or in some other way. That will be very interesting. As of now, flux has the edgein increase, over the decrease in convection. In other words, Q will increase over all (85% > 83%). Percentage wise, it should be ~a 2% increase in Q, but from your data, Q increased ~10%. This difference in Q, I think is due to the ~30% error margin you had (pretty high).

If the relationship between flux and convection (water/Cu) is linear (even if inverse), than no matter the increase in flux, there will be a proprtional decrease in convection, and the 2% - 10% gain in Q will bethe same for all possible coolant fluxes .

If it is exponential (of some sort), (hopefully) than the gain in Q would probably increase with a increase in coolant flux through the system.

Gosh, i REALLY gotta get back to my polymers now. by the way, im trying to see if this particular molecule with a 6- charge, is able to play the role of transistors. It is able to move through a polymer multilayer pretty fast when at a certain pH, and almost not at all, in a different pH zone. It looks promising you know.

trying to "be cool"


Here, after spending somuch time thinking on this, i find a thread on Overclcockers beating this issue to death, and it seems i am so right.
Flow : ALWAYS higher flow, bigger DeltaT, higher Q (for cooling, were talking here)

 

[BillA]

h2sammo
starting with the link you posted from OC, its been trashed by a bunch of ignorant posters
please, forum popularity is NOT the measure of technical correctness
herewith an example

my last post was largely incorrect due to the generalizations made
the initial post was more accurate, here are some of the recalculated and re-plotted test results

apparently one cannot post images here ??
the complete set for the individual rads are HERE

attention is first drawn to #5, a stacked flat plate design having no attached fins
as can be seen, higher flow does NOT result in increased cooling

#6, a round continuous tube design, exhibits significant cooling variations - quite obviously related to the flow regime within the tube sections

#7, another round continuous tube design, has almost exactly the opposite characteristics of #6

yet # 11, ostensibly a doubled version of #7, seems to show just that performance one might expect of a radiator
EXCEPT it is probably the most unique of all rads tested (but not in the original article) - for it has "turbulators" in the straight sections
-> for low air (noise) flow conditions it is clearly superior

so what can be gleaned from the test results ?
it should be appreciated that since the heat is transferred AT the rad tube wall, that flow regime maximizing both contact and turbulence will transfer the most heat
AND that this condition is sometimes not at the highest (possible) flow rate

there is a trade-off between the flow rate and the rads surface area, for both the liquid and air sides, but I do not have access to the multiple sizes of rads necessary to determine this by testing

I do suggest that too much emphasis not be placed on what is 'thought' about how thermodynamics works,
fluid heat transfer is not so simple (slight understatement, eh ?)

do note the consistancy of the data sets for the individual rads, the results are not random (except #4 ?)
feel free to disagree with this post,
but
do provide real data and/or references to support what is said
(I'm not so interested in opinions as facts)

be cool