Water Flow Question

[Neptune]

I've got a question about the optimal flow pattern of a Water System. I have seen people run there water systems in two different patterns:

1. Start at Pump/Resevoir --> Water-block --> Radiator --> Back to Pump/Resevoir
2. Start at Pump/Resevoir --> Radiator --> Water-block --> Back to Pump/Resevoir

Which is better and why?

On my main PC i am using option 2, on the PC i'm typing this on now, i just switched from 2 to 1 hoping to see an improvement or decrease in performance.

-Neptune

 

[RJ]

Neptune (June 19, 2001 09:05 p.m.):
I've got a question about the optimal flow pattern of a Water System. I have seen people run there water systems in two different patterns:

1. Start at Pump/Reservoir --> Water-block --> Radiator --> Back to Pump/Reservoir
2. Start at Pump/Reservoir --> Radiator --> Water-block --> Back to Pump/Reservoir

Which is better and why?

On my main PC I am using option 2, on the PC I'm typing this on now, I just switched from 2 to 1 hoping to see an improvement or decrease in performance.

-Neptune

Im going res/pump>radiator>cpu>mbo chip>res/pump. I had it the other way didnt notice any difference I figure you should cool the water the pump heats before the cpu

 

[TT120]

I have mine going throught tha radiator before going to my CPU. I figure I want the water as cool as I can get it before it hits the chip.

 

[FizzledFiend]

Has anyone played with water flow rates or a thermostat like a car? Do cars get to hot from the water flowing thru the radiator to fast and not cooling off enough to disipate the heat in the Block. Hence the thermmostate does its job by limiting water flow thru the radiator to alow more time to cool from the air passing over the coils. Hmmm I do believe that older cars it works in reverse close to heat open to cool in fact my junker works this way! On this note if the water was coold say by ICe or by a fridge wouldn't you want the water flow somewhat slow in order for the water to cool the CPU and then spend time in the cooler getting cold again? Is this how tose systems work? I am asking cause I don't know.

 

[Spewn]

FizzledFiend (Jun 20, 2001 04:55 a.m.):
Has anyone played with water flow rates or a thermostat like a car? Do cars get to hot from the water flowing thru the radiator to fast and not cooling off enough to disipate the heat in the Block. Hence the thermmostate does its job by limiting water flow thru the radiator to alow more time to cool from the air passing over the coils. Hmmm I do believe that older cars it works in reverse close to heat open to cool in fact my junker works this way! On this note if the water was coold say by ICe or by a fridge wouldn't you want the water flow somewhat slow in order for the water to cool the CPU and then spend time in the cooler getting cold again? Is this how tose systems work? I am asking cause I don't know.

I just wanna state that I don't mean to sound rude or anything, I'm just gonna state some facts...First, you cannot achieve a lower flow rate in the radiator than in the cpu block, which means that the water must flow at a constant rate(but variable pressure) throughout the system, just like electricity. Second, if the water is flowing slower thru the cpu block, it will "have time" to pick up more heat, BUT, will lose efficiency as it gets warmer. See in a watercooling setup like this, since the water isn't directly contacting the core or anything, we're really cooling the copper waterblock and not the cpu. The reason you want a higher flow rate is this: there will always be a constant volume of water in the waterblock, and as this water flows through it will heat up. If it flows through faster, it will heat up less, and maintain a higher thermal efficiency between itself and the block. The cooler you can keep the volume of water at any given instant in time, the better your setup will work. The only way to cool the water down better after it goes through the block is to have a larger radiator, having bigger pipes won't change your flow rate(unless your radiator is the restricting element, in which case it'll cause flow-rate to increase), it will only change the pressure.

As far as the original post goes, I think it'd be best to go resevoir/pump ->radiator, then wherever you want after that. Even though most/many pumps claim to be "low heat" etc, they give off a small amount, and it's best to cool that out of the water if you can before it gets to the cpu, right?

Matt

 

[AMDGuy]

Mine goes from Pump/Res -->Radiator-->Block-->Pump/Res. As others said, I want the coolest water going to the block.

 

[cjtune]

I think this argument is really moot. No matter how, your radiator water cooling system MUST pass on as many watts of waste heat from your CPU to the ambient (unless you bought/built a lousy WC system).

Assuming that your reservoir is well-insulated (no heat transfer to the ambient- if it is not, all the better!), then your system really has only two component participating in the heat transfer process (out of three- the waterblock, the radiator, and the reservoir/pump). Then it really doesn't matter what is upstream of what, right? Since RJ mentioned that he did not notice any difference, well, there you have it.

 

[Neptune]

*bump*

 

[cjtune]

CpuWaterCool (Jun 21, 2001 09:39 a.m.):
As a general rule, you want your pump to be 'pushing' cooler water through the waterblock, rather than 'pulling'.

True, pumps are VOLUMETRIC displacement devices -which means they will displace only a certain gallon of fluid per unit power supplied. Once water or any fluid has expanded due to heating, the pump will move less of a MASS of a fluid for a same power input, therefore your flow rate will be less. Your pump will also cool off better with the cooler water, and perhaps it'll have a longer service life.

However, given water heated from 25'C to 35'C, the volumetric expansion of water is only about 0.3% (Expansion coeff. @ 305K=276E-6 K^(-1)).

 

[Beggar]

So you were wondering why the placing makes no difference.

this is probably due to the type of cooling passive(no fan)/active(fan)


Start at Pump/Resevoir --> Radiator --> Water-block --> Back to Pump/Resevoir
With this the water will be hottest at the reservior, and in the reservior, it will be loosing thermal energy and will therefore will be cooler when in the radiator than if using the
Start at Pump/Resevoir --> Water-block --> Radiator --> Back to Pump/Resevoir
as with this, water is hottest just after it has come from the heat block and entered the rad.

With the first of these if the cooling of the water was the same then the water would be cooler at the chip, however if cooling is passive, or possibly what i like to cool (couldnt resist the pun) 'loose' active cooling i.e. partial fan usage (what most people seems to use) then the hotter the the water in the radiator the faster it will be cooled, concentrate here comes the science bit:

Rate of thermal loss Is proportional to (Temperature difference)^5/4

(Note with true active cooling: The rate of thermal loss is pretty much constant)

This effectively means if the water looses thermal energy in the reservior, the radiator will not be as effective as if it were first in line (i.e. less thermal energy is dissipated in the same time), and as the reservior is no doubt not as good radiator as erm the radiator then the order effects of thermal loss cannot be rectified.

So there u have it

---With the pump before the rad the thermal energy from the pump is partially dissipated before the block.

---With the pump after the rad more energy is dissipated in the rad so the water before the pump is cooler and the addition of pumps warmness makes no difference.

This is at least what i think is the reason, if ne one else has a better idea explain plz.

Also chose your flow rate carefully, faster the flow rate, higher the temp difference at all times therefore better cooling/heating, but more hotness from pump. Lower the flow rate, less hotness from the pump, but less temp difference so less cooling/heating. Also never constrict the flow from the pump, as gives more warmess at a less flow rate, do it electronically instead.

Hope this cleared some mysteries thou probably created some.

 

[cjtune]

Ok, I'll backup Beggar here: (1) cooler water before a radiator simply has less thermal energy to loose, and (2) warmer water before a radiator will definitely loose more thermal energy because of the temperature difference. The exit water of case (2) will still be warmer than the exit water of case (1) from the radiator but then will cool down anyway in the reservoir/pump section. Either way, the water will be of the same temperature before entering the waterblock -assuming the pump is a submersible pump (and is in the reservoir) and not an inline pump, which will dispose of its waste heat anywhere it is placed in the line.

Passive cooling? Who ever uses natural convection to cool down his WC system either has a damned big radiator or has never crossed the 1 gig limit for his CPU. Active cooling (by using fans) can be hundreds of times more effective (ie. less radiator fin area needed) than natural convection/passive cooling. Beggar, I assume you discounted the heat transfer from radiation when you mentioned passive cooling. Anyhow, the type of cooling does not matter as long as any method used removes the required amount of heat from the WC system at operating conditions.

I assume, in describing passive cooling, you used a horizontal, isothermal pipe surface temperature gradient correlation (with 10⁴<Ra<10⁷) to show that heat loss from a radiator pipe segment is proportional to deltaT^5/4?

BTW, heat exchanger equations usually employ log-mean temperature difference (involving three temperatures and natural logarithm) rather than plain old deltaT because the either length of piping that makes up a radiator is not at constant temperature throughout.

 

[Beggar]

yeah so i cant really be bothered to put this point across any better than i have already + im sure theres loads of stuff that i missed out when writing it, so its pretty cannok, rem its only an idea, i moght think it through better later now im not half drunk, but oh well... actually now im thinking on the opposite lines but who gives a frell?

 

[cjtune]

Well, you did bring up a good point when considering passive cooling...

One question: what's a cannok?

 

[Beggar]

Right it didnt take as much thinking as I thought it might, (if that made sense) im still gonna go on bout the active and passive cooling stuff but hear me out this time it makes (more) sense. Bear in mind I've totally changed my mind on the whole subject.

Right plain and simple active cooling occours at the pretty much same rate regardless of temp diff (dT)
This happens in peoples radiators right?

Passive cooling occours faster when dT is bigger (So the newtonian eqn above can apply here as tis a body not a pipe)
This happens in peoples reserviors right?

So based on this for an optimum system water must be at its hottest in reservoir (except just out of block), and what ever in the radiator as long as cooler than reservoir.

So optimum system with pump in reservior is.
Pump/Reservoir>>>>>Rad>>>>>>>Waterblock>>>>>>>>Pump/Reservoir

Optimum with pump inline (better i reckon as u can air cool it a bit)
Pump>>>>>Reservoir>>>>>Rad>>>>>>>Waterblock>>>>>Pump

Im not actully gonna run the bottom system my self even tho my pump is inline cause means pressure sealing the reservoir, and thats too big a mod for my system, also have fun trying to add extra water!

Hope u agree with this one cjtune! (cannok = bad)

 

[cjtune]

Okay, Beggar, I can detect another mechanical engineer in my midst... Heh heh, I think the rest of the other fellas have long since abandoned this topic to us.

But I still think ANY kind of process of heat transfer can be increased with higher deltaT, passive or active (isn't dQ=h*A*dT?, where h=f(Ra) for passive cooling). But then again who wants a higher deltaT? This can only mean the source temp is much higher than the ambient which is the bane of all overclockers...