Water Cooling 3 CPU's

[MrP]

easy question.

can it be done easily???

have this (http://docs.euro.dell.com/docs/dta/pexe53/00000003.htm) case. As i have said before to people, i want to get 3 motherboard in it. now at this moment in time, the idea of having 6 CPU fans (i run taisols with dual ys-tech fans glued to them) 2 80 mm fans, 1 120mm fan and 3 PSU's with fans all in one case fills me with dread at how loud it will be!!

so the thought was this, could i water cool the lot, 3 water blocks, 1 rad, decent fan (or couple of fans) to push air over the rad.

now i dont want amazingly low CPU temps, if i could keep them all below 45 degrees i would be very happy, i would settle for less than 50 though!

lemme know your thoughts

toodles :xsofa:

[Player0]

Yes, easy. Will need dual 1/2" ID rads (black ice extreme or similar). An Eheim 1250 and 3 waterblocks, if they fit on the motherboards. No sweat.

[MrP]

oh ok:)

errrrrr, wired up how??

toodles :xsofa:

[NorthernYankee]

I would do it in one long series, maybe with the two rads 1 in between CPU 1 and CPU 2 and one between CPU 2 and CPU 3 and then have it return to the pump and resovoir before it hits CPU 1 again.

--NY

[MrP]

ok,

question.

i have a rough idea of how much heat a CPU produces, but that does it mean in terms of water tempreture rising??

i mean if the water temp is at 20 degrees bedfore it hits the CPU, what will it be after, and the other way around, how much heatwill the rad get rid off??

toodles:xsofa:

[NorthernYankee]

Well that is actually a really hard question to answer because it depends on a ton of variables. CPU speed and voltage will determine how much heat the CPU will put off then alot will depend on how efficient the waterblock is and then cooling them back down will depend on how efficient the RAD is and how much water you have in the sys. But my water temps usually only go up a few degrees C 3-4 from Idle to load.

--NY

[Player0]

As NY sais, its really hard to predict.

My hot waterloop deals with over 600w of heat, using two BIX radiators in series, my water temperatures hover around 38-42c with ambient at 20ish degrees.

Three CPUs, even overclocked will barely be 300w, so using two radiators should really keep your CPUs cool. Depending on fannage, ambient temps and humidity, expect full load CPU temperatures maybe 15-20c above ambient. 40-45c maybe.

Hard to say.

[MrP]

hmmmmmmmm,

ok, sorry if you hear the cogs wirring here!!

BTW, making a large number of assumptions here

lets say that my water temps are 30 degrees with an ambiant of 20

the first CPU will be 35 degrees full load, the second 40, thelast 45 . is that fair to say, or would i expect different results?

toodles :sofa:

[NorthernYankee]

Well thats not neccessarily true, the reason I said put the rads between the cpus is to cool the water before it hits the next in the series, and as it goes from #3to the resovoir before it hits #1 again the time it sits in the resovoir will cool it too. So you may see temps between cpu's differing only 2 degrees or so.

--NY

[Player0]

If 2 degrees. Usually I find in water cooling systems with good water flow that the temperature is pretty even through out the entire system. It balances out so quickly that you really dont have hot spots and cold spots to worry about.

But, at the same time, ive found it best to design the cooling system around the idea that there will be hotspots and cold spots because, why not. So i would put the waterblocks and radiators in alternating series like NY said, as long as it WONT require too much more tubing to do so.

Keeping the tubing shorter and wiring the 3 cpu blocks and then the two radiators (not alternated) might offer an advantage because it should allow for more water flow. The benefits of less tubing may outway the benefits of alternating cpu blocks with the radiators. But I dont know.

Perhaps alternate the blocks and radiators and use a larger pump than the 1250 (is the 1080 the next step up?)

[kms]

i'd agree with da boys and put your rads between alternate cpu's as there very efficiency is dependant upon the water / ambient dT so raising the water temp again after the 1st chip / rad is the best wtg all be it a small amount as P0 says.

That said (i know dicki will jump in here :D) rads in parallel will out perform rads in series though for a system as you propose series is prolly the best wtg.

[Player0]

Sorry KMS, I have to fully disagree with you there.

Rads in series perform better than rads in parallel. There's another thread around here somewhere where I proved this mathematically :)

[kms]

Hi Tom, sorry bud we'll have to agree to disagree, unfortunately the art of w/c still has no governing body of standards to back up the many web claims, how / where to measure temperature & flow etc, how to validate results, etc. That said the one man that comes close to being the authorative body within w/c due to his extensive testing and indepth w/c articles is Bill Adams, i'm sure you'll have seen his extensive rad testing results. Well to quote another article (http://overclockers.com/tips915/) that always gets a good debate going too, i.e. which flow pattern is better pump > block > rad > pump or pump > rad > block > pump ???

Many forum discussions debate their use in series or in parallel, but Bill pointed out that radiators work best when the delta-temp is highest and that the parallel arrangement was superior.

[kms]

Originally posted by kms
Many forum discussions debate their use in series or in parallel, but Bill pointed out that radiators work best when the delta-temp is highest and that the parallel arrangement was superior


which should ring true to those with an understanding of basic energy transfer, entropy, enthalpy etc :)

[kms]

p.s got a parse error on those last 2 posts, something to do with badwords ?

[Player0]

I wouldn't be so quick to say that I have less experience than Bill Adams, or anyone else water cooling right now for that matter. But Im not going to use credentials, Im going to use some cold hard numbers. I have done many tests with different radiatior configurations to find out which one is best. Series has always worked best for me in practice.

But lets look at the mathematical theory. Thermodynamics is fun because it sort of emulates electronics, voltage and current. Voltage being the temperature, current being the water flow.

Consider a radiator is a battery. It has a certain cooling potential (voltage) and a certain flow allow (current).

Consider two 3 volt and 1 amp batteries in parallel. Combined, you still have 3 volts, but now you have 2 amps worth of power.

The same two batteries in series (like a flashlight) have a voltage of 6v, but the same number of amps (1a).

If radiators act like batteries, having two radiators in parallel will give you double the flow rate, but the same cooling potential. Two radiators in series give you double the cooling potential, but the same flow rate.

Radiators aren't batteries. There are flaws with the above comparison.

1.) Radiators restrict waterflow (ie, have resistance).
2.) Radiators are more effecient at higher water to ambient differentials.

Lets ignore the first flaw for now, the restrictive waterflow. We can assume we have a pump powerful enough to provide adequate waterflow, so this doesnt become an issue.

Flaw #2, the bane of this entire arguement. Efficiency loss. The second radiator in series will not do as much work as the first radiator. That is a given. What people don't take in to consideration is that the efficiency loss is usually NOT enough to make the series radiators to function worse than parallel.

Lets proove this mathematically. I encourage anyone out there to counter this proof. Its a friendly debate...can someone prove me mathematically incorrect?

Lets take a fake radiator. Our sample radiator has the ability to cool water at a rate of 5 degrees at some standard flow rate (which is not important). Ambient temperature is 20c. Water temperature just before it hits the radiators is 40c.

Okay, we take a single sample radiator and hook it to the system. Water leaving the radiator will be 20% cooler, so the water temperature will be 32c after the radiator. (20% of 40c is 8c).

Take two sample radiators in parallel. Each radiator is equal in height, tube lenght, and water at the same flow rate is split with a Y.

1/2 of the 40c water enters each radiator and looses 20%. The water exiting each radiator is 32c degrees. Recombined with the Y, the water is still 32c degrees. This is the same temperature as the single radiator provided, however the radiators in parallel can handle 2x the water flow (ie, cooling more water, but at the same potential).

Lets go series. As we know water leaving the first radiator will be 32c degrees. 20% of 32c is 6.4 degrees. So the radiator looses an additional 6 degrees and ends up being 26c. Advantage: serial.

The funny thing about this is...even if the second radiator is only 1% efficient because of entropy, the water coming out of two radiators in series will still be 1% cooler than the radiators in parallel. In fact, unless the water coming out of the second radiator is less than or equal to ambient temperature, series will always provide colder temperatures.

In practise, testing my various radiators (Black Ice, AquaCoil, DangerDen Cooling Cube, 12x5 Aluminum, Hayden, etc), series has always worked better. I have been using dual radiator configurations for over two years now. Beleive me, theres alot of experience here.

Yes, Ive been ignoring flow rate, by assuming you have a powerful pump so that you loose very little by going in series. If you have a week pump, then yes, parallel radiators will work better, because you simply will loose too much flow rate in series. However, if your serious about water cooling, you have a serious pump, and will beable to take advantage of the series configuration.

This also doesn't mean parallel doesn't have its place. CPU water cooling, eh...its not so important. But in larger scale operations, where you need to worry about quantity of water cooled, rather than amount of water cooling (like a nuclear cooling tower) then you need to use a parallel configuration. It just cools 2x as much water. Again, it doesnt cool it as MUCH as a series configuration, it just cools 2x as much water at the same speed.

I welcome all challenges to my theory. Theres been so much debate on this subject...Id really like to see someone prove my math wrong for once :)

[AFroDYme]

seems that i could calculate what is best in series or in parallel

but i just had my exam "thermodynamics" and i'm full if for now all that babbeling about cold sides , closed loop systems , radiation , turbine system, carnot processes :-)


maybe later :-)

hmmz i will be needing my fluido dynamics book too :-(

yuk !!!!!!!

[kms]

I wouldn't be so quick to say that I have less experience than Bill Adams, or anyone else water cooling right now for that matter. But Im not going to use credentials, Im going to use some cold hard numbers. I have done many tests with different radiatior configurations to find out which one is best. Series has always worked best for me in practice.

thats w/c P0, every setup is very different even if the specs seem the same, whats good for one isnt always good for all, hence i dont doubt for a second that your best results come from series :)

But lets look at the mathematical theory. Thermodynamics is fun because it sort of emulates electronics, voltage and current. Voltage being the temperature, current being the water flow.

agree

Consider a radiator is a battery. It has a certain cooling potential (voltage) and a certain flow allow (current).

sorry but i'd say a radiator in electrical anology would be a resistor, it limits flow (current), creates a pressure drop (voltage) and its conversions is heat (same). The mech analogy of a battery is a pump

Consider two 3 volt and 1 amp batteries in parallel. Combined, you still have 3 volts, but now you have 2 amps worth of power.

agree

The same two batteries in series (like a flashlight) have a voltage of 6v, but the same number of amps (1a).

agree

If radiators act like batteries, having two radiators in parallel will give you double the flow rate, but the same cooling potential. Two radiators in series give you double the cooling potential, but the same flow rate.

sorry pump cant compare, having 2 equal resistors in parallel would 1/2 the current (flow) down each branch, the volt drop (pressure drop) across each would also be identical

Radiators aren't batteries. There are flaws with the above comparison.

agree

1.) Radiators restrict waterflow (ie, have resistance).
2.) Radiators are more effecient at higher water to ambient differentials.

1.agree 2.definately agree as is the case for any known rate of energy transfer

Lets ignore the first flaw for now, the restrictive waterflow. We can assume we have a pump powerful enough to provide adequate waterflow, so this doesnt become an issue.

contrary to popular belief high water flow is not the be all & end all though high turbulent flow within the waterblock does aid see here (http://www.pclincs.co.uk/forums/showthread.php?threadid=166) for more info especially the top link :)

Flaw #2, the bane of this entire arguement. Efficiency loss. The second radiator in series will not do as much work as the first radiator. That is a given. What people don't take in to consideration is that the efficiency loss is usually NOT enough to make the series radiators to function worse than parallel.

sorry to disagree again bud, though high turbulent flow is desired in the water block so more molecules come into contact with the surface of the block and in turn carry more heat away, low flow is desired in the rads to give time for heat to be dissipated ..... so splitting the rads into a parallel config, will by the resistor anaolgy reduce the flow through each leg by 1/2 giving the desired low flow required. It aslo ensures max dT on each radiator rather than on just the 1st

Lets proove this mathematically. I encourage anyone out there to counter this proof. Its a friendly debate...can someone prove me mathematically incorrect?

always freindly m8, lifes to short :)

Lets take a fake radiator. Our sample radiator has the ability to cool water at a rate of 5 degrees at some standard flow rate (which is not important). Ambient temperature is 20c. Water temperature just before it hits the radiators is 40c.

cooling is not linear & "standard flow rate" can not be used to compare series & parallel configs as in series it is twice that of the desired low flow in a parallel config

Okay, we take a single sample radiator and hook it to the system. Water leaving the radiator will be 20% cooler, so the water temperature will be 32c after the radiator. (20% of 40c is 8c).

Take two sample radiators in parallel. Each radiator is equal in height, tube lenght, and water at the same flow rate is split with a Y.

1/2 of the 40c water enters each radiator and looses 20%. The water exiting each radiator is 32c degrees. Recombined with the Y, the water is still 32c degrees. This is the same temperature as the single radiator provided, however the radiators in parallel can handle 2x the water flow (ie, cooling more water, but at the same potential).

as above, all i'd add is that the w/c pumps used are direct drive, no fluid coupling, no speed control, so they give there all, all the time hence flow is dependant upon circuit resistance. Take the series circuit and give each rad a resistance unit of 10 :- 10+10 units = 20 units of resistance, the parallel circuit has (1/(1/10 + 1/10)) = 5 units, so though the parallel circuit splits flow between the rads, the resistance is reduced so more flow through your block too

Lets go series. As we know water leaving the first radiator will be 32c degrees. 20% of 32c is 6.4 degrees. So the radiator looses an additional 6 degrees and ends up being 26c. Advantage: serial.

up above

The funny thing about this is...even if the second radiator is only 1% efficient because of entropy, the water coming out of two radiators in series will still be 1% cooler than the radiators in parallel. In fact, unless the water coming out of the second radiator is less than or equal to ambient temperature, series will always provide colder temperatures.

no, your maths cant be applied :)

In practise, testing my various radiators (Black Ice, AquaCoil, DangerDen Cooling Cube, 12x5 Aluminum, Hayden, etc), series has always worked better. I have been using dual radiator configurations for over two years now. Beleive me, theres alot of experience here.

i'm fully aware of, & respect your experience m8, your posts are often my 1st source of info as you always seem to get the new stuff long before we can in the UK

Yes, Ive been ignoring flow rate, by assuming you have a powerful pump so that you loose very little by going in series. If you have a week pump, then yes, parallel radiators will work better, because you simply will loose too much flow rate in series. However, if your serious about water cooling, you have a serious pump, and will beable to take advantage of the series configuration.

big / more isnt always best

This also doesn't mean parallel doesn't have its place. CPU water cooling, eh...its not so important. But in larger scale operations, where you need to worry about quantity of water cooled, rather than amount of water cooling (like a nuclear cooling tower) then you need to use a parallel configuration. It just cools 2x as much water. Again, it doesnt cool it as MUCH as a series configuration, it just cools 2x as much water at the same speed.

I welcome all challenges to my theory. Theres been so much debate on this subject...Id really like to see someone prove my math wrong for once

challenged :)

[Player0]

thats w/c P0, every setup is very different even if the specs seem the same, whats good for one isnt always good for all, hence i dont doubt for a second that your best results come from series :)

This is certainly true. Even though the general rule might be that one set up is the best way, there always seems to be exceptions to those rules. Its still my opinion that most setups would benefit from serial configuration, but there may be exceptions to that, although I have not seen hard evidence of one.

sorry but i'd say a radiator in electrical anology would be a resistor, it limits flow (current), creates a pressure drop (voltage) and its conversions is heat (same). The mech analogy of a battery is a pump

You are absoultely correct, a radiator is a resistor. I used the battery because I felt that more people would understand the terms serial and parallel in terms of battery. As I mentioned down further in my post, I gave reasons why the battery anology was wrong. But you are correct, a radiator acts like a resistor.

Consider two 3 volt and 1 amp batteries in parallel. Combined, you still have 3 volts, but now you have 2 amps worth of power.


"If radiators act like batteries, having two radiators in parallel will give you double the flow rate, but the same cooling potential."

sorry pump cant compare, having 2 equal resistors in parallel would 1/2 the current (flow) down each branch, the volt drop (pressure drop) across each would also be identical

Chalk this up to bad wording. I should have said 'double the potential flow rate'. Two radiators in parallel can handle double the water flow, which is what I meant. Not that two radiators would give you double the flow, only that they can HANDLE double the flow. Again, miswording on my part. We agree here I think.

Lets ignore the first flaw for now, the restrictive waterflow. We can assume we have a pump powerful enough to provide adequate waterflow, so this doesnt become an issue.

contrary to popular belief high water flow is not the be all & end all though high turbulent flow within the waterblock does aid see here (http://www.pclincs.co.uk/forums/showthread.php?threadid=166) for more info especially the top link :)

No, I agree, waterflow seems to have very little impact on temperatures. Turbulance is more important, I agree. Since waterflow has little impact, this is why I decided to ignore it.

sorry to disagree again bud, ...low flow is desired in the rads to give time for heat to be dissipated ..... so splitting the rads into a parallel config, will by the resistor anaolgy reduce the flow through each leg by 1/2 giving the desired low flow required.

Please do not confuse volume of water with speed of the water. Each radiator gets 1/2 the VOLUME of water, but the speed is NOT cut in half. In fact, the speed of the water can stay the same. There is NO cooling benefit from sending 1/2 the waterflow through a radiator. Slowing the water down in the radiator (via turbulance) CAN improve performance, but again, you mustnt confuse 1/2 the volume with 1/2 the speed. Thus, there is no added benefit here.

It aslo ensures max dT on each radiator rather than on just the 1st

This I agree with. But I must conclude that the efficiency of the second radiator isn't lowered enough to be unuseful. The dT of the second radiator would have to be reduced to 0 for the second series radiator to have no benefit at all.

always freindly m8, lifes to short :)

Agreed, hot debate yes, but friendly :) Not trying to argue in a negative way :)

cooling is not linear

Cooling is a curve because of the radiator efficiency loss because of entropy, but it is a fairly steady curve. Didnt say differently. My math takes this in to account, the percentage formula I used creates a curve.

"standard flow rate" can not be used to compare series & parallel configs as in series it is twice that of the desired low flow in a parallel config

I honestly do not understand this. I think you are saying that flow rate is doubled in parallel configurations. This is not true. The rads in parallel have twice the POTENTIAL flow rate, but realistic flow rate is not doubled. Radiators in series to not flow 1/2 as much water as parallel. There is a reduction, but the flow rate is not cut in half. As we both agreed on before, flow rate is not that critical anyway.

as above, all i'd add is that the w/c pumps used are direct drive, no fluid coupling, no speed control, so they give there all, all the time hence flow is dependant upon circuit resistance. Take the series circuit and give each rad a resistance unit of 10 :- 10+10 units = 20 units of resistance, the parallel circuit has (1/(1/10 + 1/10)) = 5 units, so though the parallel circuit splits flow between the rads, the resistance is reduced so more flow through your block too

Correct, parallel has 1/4th the water restrictions of series. But it doesnt meant the series flows only 1/4th the water. Its not that cut and dry. In series, you may loose...say 25% of your flow rate. However if you have 400GPH, and reduce that to 300GPH, you still have more than enough horsepower to create the much needed turbulance in your waterblocks. Waterflow theory is hard as hell to prove without physically testing. I have just added a third BIX radiator in series in my cube, and my hot water temperatures were reduced a further 6c degrees. Waterflow was reduced, but it didnt have a big enough effect to counter the productiveness of a third radiator.

up above
no, your maths cant be applied :)

Well, lets decide that after this post :)

i'm fully aware of, & respect your experience m8, your posts are often my 1st source of info as you always seem to get the new stuff long before we can in the UK

Thanks, I respect your ideas as well, and am simply making a freindly argument which i hope will beneift others :)

big / more isnt always best

In terms of pump? I dont have an opinion on this yet. I have yet to see any one not benefit from a larger pump. However, Im not convinced that bigger pumps will work better. Honestly, I think you would be hard pressed to notice any difference between a 200-300-400 or 500 GPH pump. However, turbullance in the waterblocks and the radiators is key. Too much pressure may eliminate this turbulance effect. Too little, and you create no turbulance. I use the Eheim 1250 pumps religiously, and they seem to be right on. I recommend 300-500GPH for best results in a 1/2" ID system.

Look forward to your responce now that Ive countered with the speed vs flowrate :)

[kms]

Sorry KMS...I accidently over wrote this post with my responce, hitting the edit button instead of the quote button. Erf.

Well, see below.

Player0

[MrP]

handbags at dawn????

toodles :xsofa:

[Player0]

I am sincerely sorry KMS. I went to click the Quote button, and I ended up hitting EDIT instead :( Didnt realize till i saw that I had edited over your post.

Really sorry...I tried to restore your older post, but I cant :(

Anyway, I hope theres enough left of your post in the one I replied back.

Guess it is the second time that thread was lost heh.

[Player0]

Understand about the loosing post thing. This is becoming a large topic. I think Ive seen articles from this guy before. I will read the link you sent me, although I dont see anything pertaining to dual radiators yet. This thread getting hard to follow, will cut it down some.

Its still my opinion that most setups would benefit from serial configuration, but there may be exceptions to that, although I have not seen hard evidence of one.
>>i'll try and find BillA's quote or do you mean 1st hand ? <<

See if you cant find his results for Dual Rads, I cant seem to find any. Id like to see his results.

>> {waterflow} does influence {temperatures} a fair bit since the heat dissipated by a radiator is inversely related to the flow rate, selecting a higher volumetric capacity unit (other things being equal) will result in increased cooling as the slower flowing coolant will have more contact time with the wetted tube surface. (increase using 1/2" rather than 3/8" tube for e.g), <<

Im not sure I agree with this. I beleive radiators are designed around an optimal flow rate. If the waterflow is too low, there is less turbulance created, and thus less internal resistance in the radiator. If the waterflow is too fast, there is too much backpressure, and again, turbulance is reduced. There are more variables involved than water spending more or less time in a radiator. Water turbulance/friction is also a big factor. With that said, more isnt always better for a radiator, less isnt always better for a radiator.

There was an article on overclockers.com about this as well, but I cant find it. They compared different flow rates in the system, and found that the middle of the road flow rate was best.

ANYwho. This has little to do with the serial versus parallel arguement, since waterflow is REDUCED when used in series. Since the water speed is slower in series, by your arguement, the advantge here actually goes to Serial. Are you helping my point? :)


>> even with 0 dT the additional surface area would assist but thats side tracking :D the 2nd rad still performs but to a much lesser degree than 2 with max dT like in parallel <<

Even if its a much lesser degree, it still helps a little. Even if it helps a little, its producing cooler final temperatures than the two radiators in parallel.

>> exponential e^x ? <<

I beleive its:

OutTemp = InTemp - Efficiency * (InTemp - Ambient)

Where In water temperatures are 50c, efficiency is 20% and ambient is 20c:

Out = 50 - 0.2 * (50 - 20)

Out = 44c

Hmm...thats not a curve, but a line. I need to double check this. It seems to be right.

>> :), how did you calculate loss of efficiency for each rad btw<<

Its a made up number, which should be close to the efficiency Ive seen for cube style radiators.

Quote from BillA Two small radiators in series will double the hydraulic flow resistance and have a marked effect on (reduction of) pump output. If two must be used, run them in parallel.

I think weve covered this. Yes, radiators in series will double the flow resistance (and 4x over the radiators in parallel). However I see no corrolation yet between reduced flow rate and poorer radiator performance. In fact the opposite might be true.

If radiators work better with more turbulance (friction) and less flow, and since series radiators create less flow, than it stands to reason that series would benefit more from this than parallel??

Yes, series radiators reduces flow rate. You seem to be arguing on my side on this...its confusing me ;)

[kms]

aaghhhh lol, you've no idea how long that took to type twice :D

will look for those articles and get back :)

>> {waterflow} does influence {temperatures} a fair bit since the heat dissipated by a radiator is inversely related to the flow rate, selecting a higher volumetric capacity unit (other things being equal) will result in increased cooling as the slower flowing coolant will have more contact time with the wetted tube surface. (increase using 1/2" rather than 3/8" tube for e.g), <<

Im not sure I agree with this. I beleive radiators are designed around an optimal flow rate. If the waterflow is too low, there is less turbulance created, and thus less internal resistance in the radiator. If the waterflow is too fast, there is too much backpressure, and again, turbulance is reduced. There are more variables involved than water spending more or less time in a radiator. Water turbulance/friction is also a big factor. With that said, more isnt always better for a radiator, less isnt always better for a radiator.


### hmmm your kinda disgreeing with the laws of physics then m8ee :) I merely quoted them :D The thing with radiator design is some quote oil or transmission cooler specs as if they were applicable to watercooling applications, ignoring the differences in fluid properties, flow rates, temperature differentials, etc. Be skeptical of all data generated for non-CPU watercooling applications - OCing conditions are rather different than the industry norm, specs arent what they appear (if they exist at all!). I agree that all systems are unique however even when specs appear the same, i.e. length of tube, angle of bend, ambients, case deisgn & air flow, heat source etc so an optimal balance has to be found for each setup. To quote BillA again CPU watercooling radiator "ratings" by the vendors (where such exist) are about worthless, and often quite misleading. This is due to a total absence of standardized testing conditions (unlike the pump and fan industries). It is up to the consumers to insist on product descriptions based on real data from actual testing. ###


ANYwho. This has little to do with the serial versus parallel arguement, since waterflow is REDUCED when used in series. Since the water speed is slower in series, by your arguement, the advantge here actually goes to Serial. Are you helping my point?

### no as above a unique balance "sweet spot" for flow is required for each individual system, a blance between desired high turbulent flow in the block and low flow to give time for heat to transfer in the rad is required. Your reduced flow in your series setup may provide the balance your system requies??. A reduced total flow rate in your system will benefit the rads but reduce effect in the block.

Even if its a much lesser degree, it still helps a little. Even if it helps a little, its producing cooler final temperatures than the two radiators in parallel.

### sure it will help a little but not a lot, sorry to go into quote mode again but ... Larry Browning, overclockers.com, Brute Force WaterCooling (http://overclockers.com/tips915/), The primary problem, as I see it, is that water-cooling is attended by more flash and folklore than science. This is easily observed in the 'Waterblock of the Week' hysteria featured in the forums, as this month's bull becomes next-month's hamburger. One week, some unobtainable metal is the absolute pinnacle of scientific achievement, and the next week the blocks are being used as ashtrays by the support-room staff. The web is full of convincing anecdotal tests and comparisons, but the lack of standards for test methods and criteria limits our ability to extrapolate results from one test to the next. The net sum of all the water-cooling experience in the world is a hefty chunk of knowledge, but until we all get together and compare notes, we toil separately in darkness. The brightest ray of science-light so far has been the breakthrough article by Bill Adams, "Radiator Heat Dissipation Testing" (http://overclockers.com/articles481/). Using standardized conditions and criteria, Bill analyzed the most fundamental element of water-cooling, the heat exchange process itself. Bill is a controversial guy, but if you don't do anything else this year (the taxes can wait), please read this article.

beleive its:

OutTemp = InTemp - Efficiency * (InTemp - Ambient)

Where In water temperatures are 50c, efficiency is 20% and ambient is 20c:

Out = 50 - 0.2 * (50 - 20)

Out = 44c

Hmm...thats not a curve, but a line. I need to double check this. It seems to be right.


###not a line m8, just a single point, the "work" that the radiator is performing is calculated:

Q = Ww Cp (T - Ti)

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
A more convenient restatement is:

Heat Extracted = Btu/hr = 499 x flow rate in gpm (T - Ti)
Note that the above is NOT concerned with how close to ambient the "cooled" coolant temperature is. A radiator's "rating" simply describes its heat dissipation capability (in Btus or Watts) at specific air and coolant flow rates. ("Btus" as in Btus/hr and can be converted to Watt*hours by multiplying by 0.2931). The goal of CPU watercooling is to cool the baseplate of the waterblock as much as possible. To do this, a coolant temperature of only slightly above ambient is sought, the thought being that the lowest possible coolant temperature will maximize the temperature difference - and therefore the heat transfer potential. As the efficiency of the radiator is greatest with the largest possible temperature difference[, this would suggest lower coolant flow rates to maximize the heat rejection by maximizing the contact time of the coolant and the radiator tube walls which admitedly is acheived in series config but check the bit in bold above & also remember the requirement for high velocity turbulent flow required for optimal block performace, reducing flow rate increases likelihood of more laminar flow in the block (not desired) ###

I think weve covered this. Yes, radiators in series will double the flow resistance (and 4x over the radiators in parallel). However I see no corrolation yet between reduced flow rate and poorer radiator performance. In fact the opposite might be true.


###yes the opposite is true, but outweighed by the combination of reduced block flow, reduced block turbulence, increased pump back pressure, and most of all by reduced potential (temperature differential) - you know electronics check the decay of charge from a capacitor thats exponential, see how the reduced potential affects rate of decay, its the same for cooling####

If radiators work better with more turbulance (friction) and less flow, and since series radiators create less flow, than it stands to reason that series would benefit more from this than parallel??


###does friction not increase the likelihood of laminar flow ?? slow moving non mixing radial flow and faster moving central flow not in contact with a heat exchanging surface ? need to check ###

Yes, series radiators reduces flow rate. You seem to be arguing on my side on this...its confusing me

### thats true but your forgetting the all important dT & you know how important flow is for the block, your maths is flawed i'm afraid m8, lets check the all important heat extracted from the system :

Heat Extracted = Btu/hr = 499 x flow rate in gpm (T - Ti)

lets stick with easy numbers, say initial temp Ti 50c, we'll use your 20% efficiency figure so 40C exit temp T for a single rad, flow rate 100:-

2 rads in series

= 499x100x(40-50) = -499000 rad1

= 499x100x(32-40) = -399200 rad2

= -898200 total

2 rads in parallel

= 499x100x(40-50) = -499000 rad1

= 499x100x(40-50) = -499000 rad2

= -998000 total

so for 2 rads parallel offers 10% better perfromance than series, yes reduced flow rate in series benefits rad performance but also decreases block performance so lets take these as negated.

lets check your 3 rad config :

3 rads in series

= 499x100x(40-50) = -499000 rad1

= 499x100x(32-40) = -399200 rad2

= 499x100x(25.6-32) = -319360 rad3

= -1217560 total

3 rads in parallel

= 499x100x(40-50) = -499000 rad1

= 499x100x(40-50) = -499000 rad2

= 499x100x(40-50) = -499000 rad3

= 1497000 total

so a whopping 19% increase with a parallel config over series using 3 rads ??

:) ###

[mdzcpa]

***crunch..crunch...crunch***

Wow...this is a great debate....

...popcorn needs more butter though...

crunch...crunch...crunch...

[Player0]

Ack! Getting...long. Im just going to cut to the meat here:

### no as above a unique balance "sweet spot" for flow is required for each individual system, a blance between desired high turbulent flow in the block and low flow to give time for heat to transfer in the rad is required. Your reduced flow in your series setup may provide the balance your system requies??. A reduced total flow rate in your system will benefit the rads but reduce effect in the block.

We both agree with this. There is a point of balance where the water is slow enough for the rads and fast enough for the waterblock. On a side note, my cube gets around this problem because it is dual loop, the CPU doesn't directly interact with the radiators, so one loops slow, one loops faster.

###not a line m8, just a single point, the "work" that the radiator is performing is calculated:

KMS, our equations are doing much different things. My equation is a simple efficiency equation, which takes an In temperature, and provides an out result based on radiator efficiency and entropy. Your equation already assumes you have these numbers. In effect, your equation requires one similar to mine to function, because you need to determine T, and thats the hard part.

Heat Extracted = Btu/hr = 499 x flow rate in gpm (T - Ti)

That seems about right. Ill have to assume thats the proper conversion for GPM. To convert for GPH, I beleive its:

BTU/Hr = 8.32 x GPH (T - Ti)

### thats true but your forgetting the all important dT & you know how important flow is for the block, your maths is flawed i'm afraid m8, lets check the all important heat extracted from the system:

In your equations, you provide a flow rate of 100GPM! Thats a bit much for my liking. Lets use a more realistic 400GPH, ill plug them in to your formulas:

Heat Extracted = Btu/hr = 499 x flow rate in gpm (T - Ti)

Ti =50c
T = 40c (20%)
GPH = 400

2 Rads in Series

= 8.32 x 400(40-50) = -33280 BTU/Hr
+
= 8.32 x 400(32-40) = -26624 BTU/Hr

= 59904 BTU/Hr

Now, for parallel. Here comes the flaw in your math. Only HALF of the flow is going to each radiator! This is important, because when doing these equations, GPH = 200, not 400! This is the key which will win me this arguement, as you will see below:

2 Rads in Parallel

= 8.32 x 200(40-50) = -16640 BTU/Hr
+
= 8.32 x 200(40-50) = -16640 BTU/Hr

= 33280 BTU/Hr

As you can see, because you cut flow rate in half which is what happens when you connect the radiators in parallel, Series wins. But Im glad you provided your own equation, I hope now this proves my side. I will skip the 3 radiators in series for now.

so for 2 rads parallel offers 10% better perfromance than series, yes reduced flow rate in series benefits rad performance but also decreases block performance so lets take these as negated.

I beleive you will find that series does outperform. You have some cool watercooling parts there, maybe a real life test is in order? :)

Looking forward to your response :) This has been a very well constructed arguement, for sure :) Some minor math flaws here and there, but I beleive this one is solid :)

[Player0]

Addendum:

What your equation does NOT take in to account is the flow resistance your radiators create.

Let us say, that each radiator reduces flow rate by 20%. Is that a good number? I will plug this back in to my math.


2 Rads in Series

400 GPH * (20% * 2 rads) = 240GPH Each Rad

= 8.32 x 240(40-50) = -19968 BTU/Hr
+
= 8.32 x 240(32-40) = -15974 BTU/Hr

= 35942 BTU/Hr Adjusted

400GPH * (20% / 2 Rads) / 2 = 180GPH Each Rad

= 8.32 x 180(40-50) = 14976 BTU/HR
+
= 8.32 x 180(40-50) = 14976 BTU/HR

= 29952 BTU/Hr Adjusted

Advantage: Series is 17% better than Parallel.

As you can see, calculating for flow rate is definately, because it in fact, has a HUGE inpact on radiators. Now, I dont know how much a radiator actually DOES reduce the flow rate of a system. 20% seemed like a good number.

The numbers are close, for sure. There isnt a huge difference between series and parallel, either way. This is why there is so much controversy...it doesnt really seem to make that much of a difference. I still think series is a little better. But if someone has inadequate flow to begin with, it would be very easy for them to test this and find parallel works better. But not for reasons they would be aware of. As you said, it would be their waterblock that would perform worse in a series scenario.

So, like I have said from day one, Series works better IF you have a pump which can handle it!

[kms]

Hi P0 it is getting long, we'll get there soon though :D

We both agree with this. There is a point of balance where the water is slow enough for the rads and fast enough for the waterblock. On a side note, my cube gets around this problem because it is dual loop, the CPU doesn't directly interact with the radiators, so one loops slow, one loops faster

dual loops definately the answer & wtg but not what we're discussing here :)

My equation is a simple efficiency equation

too simple as it leaves out many important factors, such as resistance, back pressure & dT and you assume it to be linear which it is not. Remember e^x, remember the time constant involved with heating & cooling, remember that the specific heat capacity of water is excellent! Realise that joules per second transfered are much less at lower dT. Q = h * DeltaT (heat flux equals heat transfer coefficient times the temperature delta)

which takes an In temperature, and provides an out result based on radiator efficiency and entropy.

i cant see how / where your caclulation includes entropy ?

Your equation already assumes you have these numbers. In effect, your equation requires one similar to mine to function, because you need to determine T, and thats the hard part

I used 50c 20% efficiency = 40c as they were the figures you had used, T can be measured. Again the equation i used has many other contributary variables missing but much more applicable than yours :)

As you can see, because you cut flow rate in half which is what happens when you connect the radiators in parallel, Series wins. But Im glad you provided your own equation, I hope now this proves my side. I will skip the 3 radiators in series for now.


sorry m8 i knew you'd instantly tally with that one was hoping you'd miss it, the reason being it opens another section of calcs that require doing, yes you split the flow in half when going parallel, but you also cut the flow resistance by almost half as well, meaning your pump's flow rate will go up with them in parallel. What the net result is regarding flow depends upon how restrictive the blocks are and where you are on the PV curve of the pump but as a rule unless your components are hideously mis matched it will be in the realms of 40-50% so for all intense and purposes i ignored it as it very nearlly negates anyway.

I beleive you will find that series does outperform. You have some cool watercooling parts there, maybe a real life test is in order?


i know that you believe m8ee :) otherwise this discussion would have ended ages ago :)lol As ive stated a couple of times i dont disbelieve in yours and many setups that series is the defining answer but for rads the majority of users report better results from parallel :) I'll do the tests, i dont think it will be soon but i will do them and you can count on me to be 100% honest, i'll provide full pics and data along with result validation and means of calibration. Just hope mines one of those that prefers parallel :D

Addendum:

What your equation does NOT take in to account is the flow resistance your radiators create.

Let us say, that each radiator reduces flow rate by 20%. Is that a good number? I will plug this back in to my math.


as above Tom, i purposely left it out as it is just under a 1/2 and cancels out, sorry 20% is not a good figure. There are numerous less contributary factors i could add also left out, i think they are negligable to an extent but dont have the understanding to add them in, my BSc study was in electronics & control not mech unfortunately, think we could do with some mecho input though :)

The numbers are close, for sure. There isnt a huge difference between series and parallel, either way. This is why there is so much controversy...it doesnt really seem to make that much of a difference. .


agreed :)

[Player0]

KMS,

It seems we are at an impass.

sorry m8 i knew you'd instantly tally with that one was hoping you'd miss it, the reason being it opens another section of calcs that require doing, yes you split the flow in half when going parallel, but you also cut the flow resistance by almost half as well, meaning your pump's flow rate will go up with them in parallel. What the net result is regarding flow depends upon how restrictive the blocks are and where you are on the PV curve of the pump but as a rule unless your components are hideously mis matched it will be in the realms of 40-50% so for all intense and purposes i ignored it as it very nearlly negates anyway.

Theres no way for either of us to prove our side without knowing what the heck happens to flow rate unfortunately.

Yes, if the flow rate is cut in half when the radiators are in series, they won't perform as well as parallel.

Thing is...just because there is 2x the flow resistance, doesnt mean there is 1/2 the flow.

You say the flow is reduced 40-50%, but I think its more like 20-30% max. Who's right depends on the real answer here.

So we are stuck until we can prove how much flow rate is really reduced in series and how much in parallel. Flow rate IS reduced in both accounts.

Series Resistance = Rad Resistance * 2
Parallel Resistance = Rad Resistance / 2

Looking for info...no luck yet.

[DexterHolland91]

Remember guys reduction in flow rate depends on which radiator you are using. :) I remember some seeing some data on the reduction in flow rates, I'll see if I can find it. This is quite the debate!