Water Cooling Myths

[greenman100]

Originally Posted by jaydee116

Sure, wasn't knocking your system or anything. Got to do what you got to do.
My next system may have the fill tube on the rad itself. I was thinking about installing it like you have it there and drilling a hole ontop of one of the tanks and another hole through the case and solder on a copper or brass pipe with a threaded end for cap. The pipe will stick up just far enough above the top of the case to fit the cap on.

good idea, but I wanted a little more system capacity.

/threadjack

[greenman100]

http://www.overclockers.com/articles1088/

yeehaw

still in editing, but exciting nonetheless

[Mars]

Originally Posted by greenman100

you're getting a little rediculous.

here's a pic for you, no cap, pump not running.

Dude, are we talking about totally different f$%ken things here or what?

Clarification is in order.

- T-line at bottom, open, and pump running.

The OPEN END of your T-line is facing UP. And unplugged and yes, it is attached to another line that runs along the bottom of your case. If that is what you meant all along than I am wasting my time debating. Of course its not gonna leak.

Doesn't matter if the pump is running or not. Water will level out.

I was thinking that you had your T-line OPEN END at the BOTTOM of your case, with the pump running.

In realization it doesn't matter where the 'T' is in your system. Some systems don't even have a 'T' period.

I thought there was a miscommunication.

[Etacovda]

hmm some people seem confused about the antifreeze part - how much does a 15-20% antifreeze mix effect temperatures? that 18 times worse quote can be quite misleading....

[greenman100]

18x worse as far as viscosity, but not thermal transfer.

on another note, the article has been up less than an hour and I already have one email:

Hi there,

I just read your very interesting article, and I saw the section about water and flow, and your statements about the radiator. I don't agree on that one 100%. While you are right in your assumption about the time spent in the radiator (the racing car analog) I think you forget that the car will use much more fuel in order to spend the same time on the stretch.

In a closed loop water cooling system you will have the same flow of water but not necessarely the same speed. If you lower the speed in the radiator (relatively to the rest of the system) the water will stay longer in the radiator and thus become more effective. With your statement, speed doesn't matter as the water will spend the same time in the radiator anyway. While this is correct, my take is that when the water passes for instance twice per minute (and thus spend the same time as with the lower speed) it also passes the heat source (CPU) twice. As the water absorbs heat twice instead of once, it will lead to a less efficient radiator.

The conclusion is that yes if you have the SAME speed everywhere in the system, changing it up or down will not make a huge difference, but if you have a relatively lower speed in the radiator (maintaining the same speed in the CPU-block), it will have a positive impact on the radiator performance. If you take a look in this document from our 1st generation WaterChill, you will see that it takes a 10x increase in flow in order to see some impact - so we do agree on that one http://www.asetek.com/DownloadArea/M...eport_v1.1.pdf

Again your article was interesting reading, and we do have to fight the same myths on an everyday base

Best regards,
André
asetek Inc.

how do we go about handling this, when the manuf. is believing in myths?

[AngryAlpaca]

I just read your very interesting article, and I saw the section about water and flow, and your statements about the radiator. I don't agree on that one 100%. While you are right in your assumption about the time spent in the radiator (the racing car analog) I think you forget that the car will use much more fuel in order to spend the same time on the stretch.

How the hell did fuel get involved?! ROFL

how do we go about handling this, when the manuf. is believing in myths?

Suicide? Homicide? Solitude? Ownage? They're all good!

[HammerSandwich]

I read that email with the usual dejection until I realized who'd written it. Then I said, "Oh, God!" out loud.

I'd point out his strange assumption that 2 passes through the block must add more heat to the water while 2 passes through the rad must not dump as much. Also aim him at a "heat transfer is proportional to delta-T" article.

[HammerSandwich]

Needed to add something...

This really explains why we "big flow, big money" folks dismiss the Euro equipment so quickly. It's true, we ARE biased; biased against an apparent lack of relevant knowledge.

[threeputt]

I hope the guy from Asetek is not one of their engineers for Asetek's sake.

Greenman, I guess you can still edit it? On the gold myth section, there's a dash instead of period right there--> "Diamond (6-50w/cm-k, dependent on purity)"

[nikhsub1]

Oh boy, I just looked at that old ass Asetek pdf... in one graph, Asetek compares various blocks at different heat loads, twilight zone time... non linear c/w's. Then, in another grid of the Asetek blocks, the c/w's hold linear at various heat loads. It's no wonder he made the comment about needing 10x the flow, did you see those blocks? They are like flat copper on the inside, 'half moon' and 'maze'. Me thinks I should stop now.

[AngryAlpaca]

Wait... Can we post that e-mail (name removed, of course) to show that even some major companies have problems with myths?

[greenman100]

Originally Posted by threeputt

I hope the guy from Asetek is not one of their engineers for Asetek's sake.

Greenman, I guess you can still edit it? On the gold myth section, there's a dash instead of period right there--> "Diamond (6-50w/cm-k, dependent on purity)"

ah, but the dash is supposed to be there!

check out the source link, dependent on purity, the heat transfer coefficent changes.

Originally Posted by nikhsub1

It's no wonder he made the comment about needing 10x the flow, did you see those blocks? They are like flat copper on the inside, 'half moon' and 'maze'. Me thinks I should stop now.

you're not eyeball testing, are you?

Originally Posted by angryalpaca

Wait... Can we post that e-mail (name removed, of course) to show that even some major companies have problems with myths?

post it where? the OCers front page?

don't think they're looking for that kind of attention

[greenman100]

Originally Posted by HammerSandwich

I read that email with the usual dejection until I realized who'd written it. Then I said, "Oh, God!" out loud.

I'd point out his strange assumption that 2 passes through the block must add more heat to the water while 2 passes through the rad must not dump as much. Also aim him at a "heat transfer is proportional to delta-T" article.

here's what I said, note the link to the thermochill article/graph

(on a side note, my emailbox if filling up about the myths)

Originally Posted by email

Thanks for your email.

Maybe I am still a bit unclear.

A radiator WILL benefit from higher flow/speed, in a closed loop system. You see, the slower the water goes, the less turbulence there is. If there is little turbulence, then the water molecules near the metal will cool down, while the water molecules near the center of the pipe/fin will not change temperature as much. As we all know, heat transfer is dependent on deltaT, and if the warmer water molecules don't ever get to the colder metal, then efficent heat reansfer will not happen.

I am rather surprised that an employee of a major waterblock company does not have a better grasp of heat transfer and thermodynamics.

You would probably benefit from the following graphic:

Note that for a given airflow, as coolant flow is increased, heat transfer is increased.

http://www.overclockers.com/articles778/thermo2.gif

from: http://www.overclockers.com/articles778


Also, if it takes a 10x increase in flow to see a change in your waterblocks, I suggest you upgrade your test equipment. Take a look here:

http://www.procooling.com/html/pro_testing.php

and you'll see the test results of various waterblocks.
Performance improves markedly across the .5--->2.0GPM range.

Thanks,
Tim

and his reply:

Originally Posted by email

Thanks for your email.

With regard to turbulent or laminar flow, there are huge differences in radiator designs and their response to the flow velocity of the water. As an example radiators based on cylinder shaped hairpin tubes reacts differently compared to a heater core style of radiators. Furthermore some radiators use corrugated tubing design, why the water will be mixed very well independent of flow velocity.

With regard to flow rate, if you read the document that is based upon real life tests and CFD simulations, you will see that there IS a difference, but in order to see significant changes (which of course is subjective) you would need a much larger flow. For instance moving from a L20 to a L30 or moving from 3/8" to 1/2" does not have a major impact.

I'm rather surprised about your arrogant attitude. I'm sure that our success in this field must be pure luck, as we obviously don't have a clue and thermodynamics... Thanks for learning us!

Thanks,
André

WTF.....

[threeputt]

Originally Posted by greenman100

ah, but the dash is supposed to be there!

check out the source link, dependent on purity, the heat transfer coefficent changes.

Woops... I didn't relize you were referencing a range of 6 to 50... my bad

[greenman100]

Originally Posted by HammerSandwich

I read that email with the usual dejection until I realized who'd written it. Then I said, "Oh, God!" out loud.

I'd point out his strange assumption that 2 passes through the block must add more heat to the water while 2 passes through the rad must not dump as much. Also aim him at a "heat transfer is proportional to delta-T" article.

here's what I said, note the link to the thermochill article/graph

(on a side note, my emailbox if filling up about the myths)

Originally Posted by email

Thanks for your email.

Maybe I am still a bit unclear.

A radiator WILL benefit from higher flow/speed, in a closed loop system. You see, the slower the water goes, the less turbulence there is. If there is little turbulence, then the water molecules near the metal will cool down, while the water molecules near the center of the pipe/fin will not change temperature as much. As we all know, heat transfer is dependent on deltaT, and if the warmer water molecules don't ever get to the colder metal, then efficent heat reansfer will not happen.

I am rather surprised that an employee of a major waterblock company does not have a better grasp of heat transfer and thermodynamics.

You would probably benefit from the following graphic:

Note that for a given airflow, as coolant flow is increased, heat transfer is increased.

http://www.overclockers.com/articles778/thermo2.gif

from: http://www.overclockers.com/articles778


Also, if it takes a 10x increase in flow to see a change in your waterblocks, I suggest you upgrade your test equipment. Take a look here:

http://www.procooling.com/html/pro_testing.php

and you'll see the test results of various waterblocks.
Performance improves markedly across the .5--->2.0GPM range.

Thanks,
Tim

and his reply:

Originally Posted by email

Thanks for your email.

With regard to turbulent or laminar flow, there are huge differences in radiator designs and their response to the flow velocity of the water. As an example radiators based on cylinder shaped hairpin tubes reacts differently compared to a heater core style of radiators. Furthermore some radiators use corrugated tubing design, why the water will be mixed very well independent of flow velocity.

With regard to flow rate, if you read the document that is based upon real life tests and CFD simulations, you will see that there IS a difference, but in order to see significant changes (which of course is subjective) you would need a much larger flow. For instance moving from a L20 to a L30 or moving from 3/8" to 1/2" does not have a major impact.

I'm rather surprised about your arrogant attitude. I'm sure that our success in this field must be pure luck, as we obviously don't have a clue and thermodynamics... Thanks for learning us!

Thanks,
André

WTF.....

[nikhsub1]

Originally Posted by greenman100

you're not eyeball testing, are you?

No. I looked at Asetek's data. Suggests to me those particular blocks are rather poor designs. If you note, now they use a WW clone with impingement.

[rundymc]

Originally Posted by greenman100

here's what I said, note the link to the thermochill article/graph



WTF.....

"greenman100 = obnoxious ass hole" - not your week

the guy doesn't seem to offer much in the way of explanation for his point
more flow, more turbulence, a little more efficiency, sooo what's not to agree upon? why else would more flowrate through a 'pipe' yield better results? :shrug:

I'm rather surprised about your arrogant attitude. I'm sure that our success in this field must be pure luck, as we obviously don't have a clue and thermodynamics... Thanks for learning us!

come on now, I don't think its a complete secret as to how asetek rose through the watercooling industry so far
what is the greatest thing that seperates 7Up from Sprite

[greenman100]

Originally Posted by nikhsub1

No. I looked at Asetek's data. Suggests to me those particular blocks are rather poor designs. If you note, now they use a WW clone with impingement.

was just giving you a hard time

come on now, I don't think its a complete secret as to how asetek rose through the watercooling industry so far
what is the greatest thing that seperates 7Up from Sprite

not sure I follow...

copying others?

[rundymc]

other than the antarctica, not really
more like publicity

[threeputt]

Lemon lime flavor???

[nikhsub1]

Originally Posted by threeputt

Lemon lime flavor???

My answer = not much.

[threeputt]

It just hit me!!! This Asetek guy is Graystar!!!!

[HammerSandwich]

Greenman, you shouldn't have focused just on turbulence. As water travels through the radiator, it's temperature decreases. Since transfer depends on delta-T (and Andre didn't deny that point), we want to maximize delta-T by moving the cooling water out of the rad as quickly as possible. This argument is persuasive in its simplicity.

[pHaestus]

That article needed a lot more beef and supporting information in my opinion. You guys can do better. Not horrible but not enough explanation for some points and NO data to support "rules of thumb". I'm sorry I didnt get a chance to read this thread earlier; didn't realize an article for OCers was being formulated in here.

[Brians256]

Interesting discussions with Asetek. I think that the answer is more complicated than saying that higher velocity is better or worse.

You can speed the water up with a single pass thin-tubing radiator and have worse performance than a heatercore. On the other hand, you can have a large tubing multipass radiator that performs worse than a thin-tubed multipass radiator. You have examples where velocity is not the primary problem.

Instead of focusing on "slowing the water down" in the radiator, focus on keeping the water close to a heat exchanging surface for a longer proportion of it's time traversing the loop. Both elements have to be there: 1) high proportion of time and 2) quality of heat exchange.

The slow vs. fast radiator flow debate (to me) seems better answered by looking at the costs of impedance to water and the amount of surface area that the water (not just the air) is exposed to.


Blah. Feels like I'm talking in circles on this one.