Exactly why does high flow perform better than low flow?

[skane]

Im wondering why high flow performs better than low flow. I already know it does, and read the stickies, but cant seem to find it anywhere.
i have heard a few times that low flow would perform better because the water would spend more time in the blocks, but i know thats BS because its a closed loop. But if the water is an equal amount of time in every piece of the system, no matter if its high or low flow, why does high flow perform better?
it might be a stupid question for you guys, but im really puzzled by this. I hope you understand what i mean, because i got the feeling i cant explain it properly.

[einCe]

i know what you mean, id like to know too, mabe theres some sort of balance between the water removing heat from the blocks and the radiator removing the heat fom the water. like a median of performance.

[[XC]Atomicpineapple]

It may be to do with the higher flow rate increasing turbulence in the waterblocks, which aids heat removal by having more water molecules of a lower temperature contact the base/sides of the block.

[Fairydust]

There are 2 reasons, the primary reason is what Atomicpineapple said. The higher flow rate will increase coolant velocity inside the waterblock, to a lesser degree also in radiators, and thus reducing laminar flow which acts as a low flowing layer on the waterblock's surfaces and thus hindering thermal transfer.

In (very) low-flowing setups the temperature difference between coolant and waterblock will become smaller as it heats up inside the block and hinders heat transfer that is directly proportional to the temperature difference. With more flow the thermal energy is spread among a bigger amount of water and thus the coolant will stay colder inside the block, at "normal" flowrates the delta of in/out watertemperature of a waterblock is sub 1/10 Kelvin.

[skane]

ok, thanks people.

[Torin]

Most modern waterblocks use some form of turbulence in order to aid heat transfer. More flow = more turbulence. The faster you can get the cooler water to the surface where heat is transfered, the faster you can carry more heat away, and thus the more heat you can transfer to the water. Warmer water transfers less heat.

Some blocks benefit more from higher flow than others, and some blocks can deal with low flow better than others. Take the old Swiftech MCW-6000/6002 blocks... they work well with low flow, and have minimal (but still noticable) benefit from high flow setups. Compare that the the Swiftech Storm blocks that benefit a lot from high flow setups, and perform very noticable worse with low-flow setups. It's mainly because of the method uses to transfer the heat, where the Storm block uses a jet impengement method that relies on strong flow for heat dissipation, whereas the MCW-6000/6002 just moves water over the heat dissipation area.

Blocks that rely more on high flow usually have better cooling potential than low flow blocks. Thus, a high flow loop with a high flow block will offer noticably better cooling capabilities.

[Stigma]

There are 2 "major" factor

Increased velocity of the cooling fulid gives more turbulence and thus promotes trasnfer of heat between water and metal.

also, more velocity simply means that the heat is more quickly moved from the hot spots to the rad. it can be hard to visualize, but try to think of it this way:

If you loop has no flow at all, what happens? The water in the cooling blocks heat up, your temp increase because the heat isnt moved away. Now what happens if you have just a a tiiiiny flow? the heat is moved, but because the flow is so low, the difference in heat between the water in the rad and the waterblock is so high, the thermal transference between the CPU and the water is less than optimal.

In the best case scenario, your flow is so high that the water temp is aproximately equal no matter where you measure.

The point to keep in mind though is that flow gives very quickly diminishing returns. In other way a little flow gives you a lot of benefit, but having a crazy pump dosn't do much (if anything) over having a normal capacity pump.

Thats why flow-rate in general is HIGHLY over-rated. Thats what happens when people start buying equipment based on "bigger is better" and assuming that it just works better. in general, the difference between a low-capacity pump and the most powerful pumps sold for use with watercooling is allmsot none - usually no more than 1 to 1,5 celcius. Of coruse, if you have an insane load on your system (like using several heavy TECs), then this differance will be slightly higher.

here is actually a test made to measure this in a real-world application - made by the resident guru "Skrue" over on hardware.no . Sorry that its in norwegian, but I'm sure you will understand the result-data anyway. Pm me if you want more info and il translate:

http://www.hardware.no/guider/kjoling/betydningen_av_flow_i_et_vannkjolingssytem/23408

-Stigma

[Smalltimer]

Im wondering why high flow performs better than low flow. I already know it does, and read the stickies, but cant seem to find it anywhere.
i have heard a few times that low flow would perform better because the water would spend more time in the blocks, but i know thats BS because its a closed loop. But if the water is an equal amount of time in every piece of the system, no matter if its high or low flow, why does high flow perform better?
it might be a stupid question for you guys, but im really puzzled by this. I hope you understand what i mean, because i got the feeling i cant explain it properly.
It's quite simple really, water absorbs heat, once a water volume has reached a heat load, the absorption rate continues to decrease as the fluid nears the heat source temperatures. The maximum heat absorption will take place at peak contrast rating and diminishes from there. If you placed none moving water above a heat source the temperatures will stabilize evenly between them (or to ebullition point).

Having said this, once water has reached its maximum effect on the heat source it can no longer cool, in fact it will actually act as an insulator extending the heat load. In a circulatory heat exchanger, water must be moved out of the way to keep the cooling fluid as efficient as possible in heat exchange process. If flow rates are to high you may not get maximum absorption and if you go to slow you may add heat to the loop.

It's safe to say though, one thing you can never get enough of is heat exchange - dumping all the heat from a RAD or otherwise would be the ultimate scenario. :)

[skane]

its getting clearer and clearer with every post.
thanks for the link Stigma, the norwegian in no problem for me, i have lived a year in sweden, so i can understand some norwegian.

[creidiki]

If flow rates are to high you may not get maximum absorption and if you go to slow you may add heat to the loop.
Thats actually not true... for a given loop lenght, the amount of time spent by a water molecule in any 1 part of the loop over a unit time is the same irrsepective of the flow - faster flow = shorter passes but more passes so time in block is the same - but temperature gradient is larger beause of the increased heat transfer thanks to turbulence, hence more flow is ALWAYS better, though some blocks may not scale well enough to justify the added pumpage...

but if youre going to use 2 pumps on an apogee you deserve to waste your cash.

[Smalltimer]

I guess I should have been more specific. Since I was not speaking on a PC water cooling basis I guess the notion of higher flow rates = better cooling holds true.

But since I was speaking on general heat exchange terms I thought it would be interesting to put up the entire spectrum into the mix. There are a few precautions to be observed in increasing coolant flow rate. Going too far could result in aeration and cavitation of the loop, due to excessive pressure drops through the heat exchanger phase. The ideal coolant flow rate is one that will provide optimum coolant flow velocity through the radiator tubes in the range of 6 to 8 feet per second. Flow velocities above 10 feet per second should be avoided.

However this is far from a defined condition due to the infinate variables and possibilities in the cooling process. So the answer is simply yes, better flow rates = better cooling unless your moving into extreme flowrate which I doubt would apply here :)

[creidiki]

Yeah not really, considering an RD-30 will struggle to break 3gpm in a loop.

[Torin]

The point to keep in mind though is that flow gives very quickly diminishing returns. In other way a little flow gives you a lot of benefit, but having a crazy pump dosn't do much (if anything) over having a normal capacity pump.

Thats why flow-rate in general is HIGHLY over-rated. Thats what happens when people start buying equipment based on "bigger is better" and assuming that it just works better. in general, the difference between a low-capacity pump and the most powerful pumps sold for use with watercooling is allmsot none - usually no more than 1 to 1,5 celcius. Of coruse, if you have an insane load on your system (like using several heavy TECs), then this differance will be slightly higher.
There is a fine line to be drawn before you can say that flow-rate is "highly over-rated". Yes, when going to the extremes, something like an RD-30 and it's flow rates can be a bit unnecessary. But, to compare for example, a Laing DDC and any of the "silent" Eheim pumps, flow-rate is absolutely not overrated. The flow rate offered by many "completely silent" pumps even for a single block loop is not ample to achieve near-optimal performance of a block like the Storm. The "average" medium flow pump like a DDC does not offer unnecessary flow, but nears the point where additional flow vs. the diminishing returns you get in performance makes it not worth it to get a bigger pump.

This is all of course in the context of the loop. The more blocks you have, the bigger the rag you have, the more restrictive the blocks you have, the more flow you need. In a setup with 2 or more blocks and a good sized rad (120.2 or bigger), having a pump with at LEAST DDC-level flows is a necessity, and absolutely not over-rated nor unnecessary.

If you have a very small loop, with a 120.1 rad and a single "low flow" block such as the MCW-6000 or any of the 1U german "bling" blocks, you can get away with a low flow pump such as an Eheim. But, with a more "enthusiast configuration" such as a Storm block, Maze 4, 120.2 or 120.3 rad and enough tubing to fit it all in a big case like a Stacker, even the flow rates of a modified DDC is barely ample. Situations like this are fine examples of where flow rates are not over-rated, even in the case of an Iwaki MD-20/30 or multiple DDC/50-Zs.

To make a blanket statement that "flow rates are HIGHLY over-rated" is just silly and uninformed, as it needs to be taken in context of the loop, and we all know that every loop is different.

[Fairydust]

... having a pump with at LEAST DDC-level flows is a necessity, and absolutely not over-rated nor unnecessary...

To make a blanket statement that "flow rates are HIGHLY over-rated" is just silly and uninformed, as it needs to be taken in context of the loop, and we all know that every loop is different.

Well, it being over-rated or a necessity are both arbitrary judgements. The differences in comparison to phase change are pretty small, even the most aenimic eheim pump is usually within a 1 digit Kelvin range of a high end pump in the worst case scenario. The few Kelvin difference seem to justify hundreds of dollars for some, but why call people silly because they don't see the use of it?

[creidiki]

http://forums.procooling.com/vbb/showthread.php?t=10825

http://forums.procooling.com/vbb/showpost.php?p=129150&postcount=11


Let's rank it for easier browsing shall we?

+20.8C => 2 x Swiftech MCP600 @ 13.8v (series)
+20.9C => Swiftech MCP600 @ 13.8v
+21.1C => Iwaki MD-20RZ @ 60Hz
+21.1C => 2 x Swiftech MCP600 @ 12.0v (series)
+21.2C => Iwaki MD-20RZ @ 50Hz
+21.4C => Swiftech MCP600 @ 12.0v
+21.4C => 2 x Eheim 1048 (series)
+21.5C => Iwaki MD-15R @ 60Hz
+21.6C => Swiftech MCP650 @ 12.0v
+21.6C => Iwaki MD-30RZ @ 50Hz
+21.7C => Iwaki MD-30RZ @ 60Hz
+21.9C => Eheim 1250
+22.3C => Eheim 1048
+23.3C => Eheim 1046

thats 2.5c between MD-20 and 1046, but note very carefully: thats with a whitewater block, a dual 120mm rad, and nothing else. Imagining these figures for a triple-block system is an exercise left to the reader ;)

[Fairydust]

thats 2.5c between MD-20 and 1046, but note very carefully: thats with a whitewater block, a single 120mm rad, and nothing else. Imagining these figures for a triple-block system is an exercise left to the reader ;)

Cathar used a HE 120.2 not a single rad, and if my math is correct it was 2.5 K to a dual overvolted MCP600, difference to a MD-20RZ was 2.2 K.

The biggest difference i've personally seen was 7 K in a bip3, nexxos xp, DIY GPU and NB coolers loop switching from an Eheim 1046 to a dual DDC (Although cleaning the rad might have been worth 1-2K by itself :P).

[creidiki]

right you are, 120.2

[Torin]

When it comes to a 3C difference in a W/C loop, that could easily mean 100Mhz or more to the total overclock.

If performance is the point, and for most of us it is, then a temperature drop that noticable simply by increasing the flow in your loop would hardly quality it as over-rated.

Yes, super high flow pumps are over-rated in all but a few scenarios. But, for every loop that can get by with an Eheim 1046, there is a loop that needs an RD-30 to maximize performance.

Let's not turn this into a German vs. American W/C logic debate... If we're talking about performance, and that is what this thread is about, making a blanket statement that high-flow pumps are over-rated is baseless.

No matter what anyone says, flow is a concern in every loop. How much of a concern and how worthwhile it is to attain an extra 0.5gpm depends entirely on the components of the loop and the performance goal in mind.

Fairydust, I can see where you are coming from, being German and all, and knowing full well the general German credo towards W/C in reference to balancing silence, appeareance and performance. Woe be the day that I meet a German that uses a 40DB fan simply to cool his CPU by an additional 1.5C.

[creidiki]

the thing they have for silence and low flow pumps wouldnt be a problem in itself, if they at least used blocks that are appropriate, instead of coupling the most restrictive block designs known to man with the weakest pumps...

ok, ill stop now :3

[Torin]

the thing they have for silence and low flow pumps wouldnt be a problem in itself, if they at least used blocks that are appropriate, instead of coupling the most restrictive block designs known to man with the weakest pumps...

ok, ill stop now :3
Nexxos XP + Eheim 1046! :slap:

I'll stop too. :cool:

[Radical_53]

Well, Nexxxos is less restrictive than a MP-05 ;) The Cuplex that some people (like over at [H]) like to use, THAT's restrictive...

One real problem on most low-flow loops is the use of elbow fittings... as many block designs can't be fit without an elbow (like the full cover blocks, most of them).

[Fairydust]

Fairydust, I can see where you are coming from, being German and all, and knowing full well the general German credo towards W/C in reference to balancing silence, appeareance and performance. Woe be the day that I meet a German that uses a 40DB fan simply to cool his CPU by an additional 1.5C.

Well personally I have several "extreme" systems myself, dual D5s, dual DDCs and in the near future some dual DDCs+. I consider myself a performance oriented guy, , like most on xtremesystems, but I know plenty of people that prioritize other factors, be it noise or price/performance. The problem I have is people on either side telling the other what is "best".

[skane]

the problem is that both of them are right, but in their own way. the silence guys say the best system is one that performs ok, but is really silent, while you say that the best system is the one that performs the best, no matter what.

[Radical_53]

Well, with one DDC/DDC+ and something like a PA rad you can still have very good efficiency will still keeping it dead quiet...

[Smalltimer]

Dammit we should have both :D

I'm commited to high performance /low noise - I have no doubt it can be done :)
I have 500 sunk into the project thust far and I am only cooling my CPU. I think I will simply build a second loop to take care of the GPU and remaining components.

However it will be silent :p

[skane]

Well, with one DDC/DDC+ and something like a PA rad you can still have very good efficiency will still keeping it dead quiet...
thats probably why most european shops are starting to offer the ddc also.
i do wonder why they keep sticking to those tube rads though.

[Torin]

Dammit we should have both :D

I'm commited to high performance /low noise - I have no doubt it can be done :)
I have 500 sunk into the project thust far and I am only cooling my CPU. I think I will simply build a second loop to take care of the GPU and remaining components.

However it will be silent :p
The Germans have a very strict definition of silence, one you won't achieve with a high performance system. When they meant silent, they don't just mean "really quiet." :)

@Fairydust: No one is saying the other side is wrong, I was just clarifying that when on the topic of performance in a W/C system, flow rate is only as over-rated as you deem necessary for your performance. Taking the basic stance of "flow rate is over-rated" in general means to me that the person could care less about performance, regardless of what additional flow rate can do. Flow rate is an important factor to performance, and can be overkilled, but generally speaking, in most "performance" w/c system, an Iwaki MD-20/30 is not overkill, and it's flow is not over-rated if your goal is low temps.

[Radical_53]

@skane: Yes, they have the DDCs for like a year and half now. But they're selling quite some of them ;)
Tube rads, good question. Some aren't that bad, still you'd always find a heater core style with less restriction, better performance and maybe even lower price. I guess most of those people do believe in "tradition" or solely marketing :D

@torin: Well yeah, like a guy I know who uses a triple-stage decoupling for his Eheim... you're right about that, but you can't consider these people "sane".

[Torin]

@torin: Well yeah, like a guy I know who uses a triple-stage decoupling for his Eheim... you're right about that, but you can't consider these people "sane".
No arguments here. :D