Water cooling loop temperatures -

[Waco]

Well I got my second temperature sensor and display in yesterday and installed them today (without draining the loop ).

 

I didn't take any pictures (go me!) because it's Sunday and I'm lazy, but the delta between the hottest point in my loop (after the CPU and all 4 GPUs) and the coolest point in my loop (after both radiators, measured in my reservoir) when running Kombustor is ~.6 C. It fluctuates a bit but after letting it get as hot as it would go (which takes about 30 minutes) one sensor was at ~8 C over room temperature and the other was at ~7.4 C.

 

So yeah...loop order matters less than a degree now matter how you want to arrange your loop. I do have fairly high flow rates even with all of my blocks due to my D5 being on maximum and all of my blocks being pretty non-restrictive so you could see a higher delta if you have less flow through your loop (or if it's exceeding long). But in general, I think we can say that loop order essentially doesn't matter.

 

Also, since that experiment was done, I packed up the second sensor and set up an air temperature sensor instead in the front air intake so I can see my ambient to water temperature delta. At idle I sit at ~3 C over room temperature. I think I have too much cooling...if that's even possible.

 

 

EDIT: It's closer to 2.5 C over room temp. Apparently it's still cooling down.

Edited November 6, 2011 by Waco

[El_Capitan]

That's pretty awesome info. Thanks Waco!

[Waco]

It's been bugging me a long time...so the peace of mind is worth the $10 I spent on the temp probe.

[Boinker]

Mmmmmm Data. NOM NOM NOM.

 

 

Thanks a bunch Waco.

 

So this proves that the flow rate that you have versus the heat the components (which to most water loops you have a lot) the amount of heat being produced is nil enough to drastically change the water temperature. This also proves that your radiators are working well. That's a cool deal Waco. Thank you very much for Verifying that theory. When i get my loop up and running ill try using a single radiator first and see if it can be managed... but after this data i'm pretty sure it can just fine.

[Waco]

I'm pretty sure it's dependent nearly completely on flow rate versus heat dump. If I wasn't lazy I'd try turning my pump down to the minimum but even on that setting it moves a LOT of water.

[Boinker]

It would be nice if we had a load maker. Then we could really measure the limit of water. Lol

[Waco]

I'm going to rename my computer "load maker".

 

I could rig up a heating element but I'd rather not melt my loop.

[Speedway]

Waco, don't you think that just adding another temp sensor into the existing loop only gives you info on the current delta not being much different, no matter where the temp sensors are? What I'm saying is of course in your current loop, that already has had the delta established, the water temps, no matter where the sensors are placed aren't going to be very different! Once the delta T in a stabilized loop has been established (usually about 30min after PC startup) then of course the water temps everywhere in the loop are going to be relatively the same, with maybe a very minimal difference. And yes, this proves that there isn't "hot" water coming out of the cpu/gpu section of the loop. But, it doesn't mean that a different loop order altogether couldn't effectively lower the delta overall. I think in order to really prove your theory that loop order doesn't make a difference in the delta T in general, it would be more of a PITA than just adding a 2nd temp sensor. You would probably have to actually change the loop order, and then measure the temps again to see if they were in fact any different, and then it would only be accurate to a point based on the ambient temp being as close to identical as possible.

 

One of the better acticles I have found about Delta T and watercooling.

[ARandomOWL]

I'm pretty sure it's dependent nearly completely on flow rate versus heat dump. If I wasn't lazy I'd try turning my pump down to the minimum but even on that setting it moves a LOT of water.

You would be right waco.

 

 

 

Q is energy in Joules, m is mass in kg, c is specific heat capacity which is a constant (4200 for water), delta theta is change in temperature.

 

Energy is power multiplied by time. So we can re-write the equation as:

 

 

 

and divide by t:

 

 

 

Now we have m / t which is the mass flow rate of the water. If we re-arrange for delta theta:

 

 

 

where delta theta is temperature change in Kelvin (or degrees C), W is power absorbed by the substance in Watts, m dot (previously m/t) is the mass flow rate in kg/s, and c is the specific heat capacity (4200 for water) in J/kg/K.

 

So in conclusion, the change in water temp as it passes over a heat-producing object is proportional to the power you put into it and inversely proportional to the flow rate. This is in an open loop situation. In your closed loops, things will be slightly different.

 

Oops, looks like the transparency on the images went a bit wonky, sorry about that. Opening them in a new tab should make them readable.

Edited November 7, 2011 by ARandomOWl

[Waco]

Waco, don't you think that just adding another temp sensor into the existing loop only gives you info on the current delta not being much different, no matter where the temp sensors are? What I'm saying is of course in your current loop, that already has had the delta established, the water temps, no matter where the sensors are placed aren't going to be very different! Once the delta T in a stabilized loop has been established (usually about 30min after PC startup) then of course the water temps everywhere in the loop are going to be relatively the same, with maybe a very minimal difference. And yes, this proves that there isn't "hot" water coming out of the cpu/gpu section of the loop. But, it doesn't mean that a different loop order altogether couldn't effectively lower the delta overall. I think in order to really prove your theory that loop order doesn't make a difference in the delta T in general, it would be more of a PITA than just adding a 2nd temp sensor. You would probably have to actually change the loop order, and then measure the temps again to see if they were in fact any different, and then it would only be accurate to a point based on the ambient temp being as close to identical as possible.

 

One of the better acticles I have found about Delta T and watercooling.

Wait what?

 

I'm talking about inter-loop temperatures, not about the delta from ambient. Measuring at the hottest and coldest points of the loop gets me the delta between them (and since I have all of my heat-producing components in serial and all of my heat-removing components in serial it's pretty easy to prove that there's no "worse" configuration for having an internal delta).

 

Changing the loop order wouldn't do anything in terms of the overall delta from ambient. It's a closed loop and the internal delta is less than a single degree even with my extremely hot-running components. The overall delta from ambient is based purely on the ability to remove heat (number of radiators, fan setup, etc) and the amount of heat dump (pumps, GPUs, CPU, etc). The loop order doesn't change any of this since it's a closed system.

 

You would be right waco.

 

 

 

Q is energy in Joules, m is mass in kg, c is specific heat capacity which is a constant (4200 for water), delta theta is change in temperature.

 

Energy is power multiplied by time. So we can re-write the equation as:

 

 

 

and divide by t:

 

 

 

Now we have m / t which is the mass flow rate of the water. If we re-arrange for delta theta:

 

 

 

where delta theta is temperature change in Kelvin (or degrees C), W is power absorbed by the substance in Watts, m dot (previously m/t) is the mass flow rate in kg/s, and c is the specific heat capacity (4200 for water) in J/kg/K.

 

So in conclusion, the change in water temp as it passes over a heat-producing object is proportional to the power you put into it and inversely proportional to the flow rate. This is in an open loop situation. In your closed loops, things will be slightly different.

 

Oops, looks like the transparency on the images went a bit wonky, sorry about that. Opening them in a new tab should make them readable.

There's the kind of math I like! Thanks for going through that - I had a hunch if I cranked through the equations that I'd find out the answer but I'm not that motivated. Remember, I'm lazy on Sundays.

[Speedway]

Wait what?

 

I'm talking about inter-loop temperatures, not about the delta from ambient.

 

My point here is this: You didn't need a 2nd temp sensor to find out that your water temps are almost identical regardless of what point in your loop the temp sensors are placed! Once the PC has been running and the heat output and cooling are constant, then the water temps will be basically the same throughout the loop, and a 2nd temp sensor wasn't needed to prove this.

 

Measuring at the hottest and coldest points of the loop gets me the delta between them (and since I have all of my heat-producing components in serial and all of my heat-removing components in serial it's pretty easy to prove that there's no "worse" configuration for having an internal delta).

 

Changing the loop order wouldn't do anything in terms of the overall delta from ambient. It's a closed loop and the internal delta is less than a single degree even with my extremely hot-running components. The overall delta from ambient is based purely on the ability to remove heat (number of radiators, fan setup, etc) and the amount of heat dump (pumps, GPUs, CPU, etc). The loop order doesn't change any of this since it's a closed system.

 

My point here is this: You never changed the order of the loop and IMHO the only way to prove that loop order doesn't make a difference is to actually change the loop order and see if in fact the Delta T is the same. For example you have your loop order currently with 2 temp sensors in 2 different locations in your loop, you have calculated the difference (Delta T) between the ambient (room) temp and the water temp, but all that has been proven is that the water temp, regardless of location is relatively the same. You have not proven that by changing your loop order that the Delta T would remain the same.

 

Now, if you changed your loop order from your current (and this may not be accurate, didn't see your actual loop order stated) cpu->gpu->gpu->rad->rad to rad->cpu->rad->gpu->gpu and your Delta T remained 2.5-3 C over room temp, then you would have proven that loop order doesn't matter in regards to Delta T.

 

But, maybe I am just misunderstanding your point of proving that inter-loop water temps are almost identical regardless of where in the loop you are getting a water temp reading, not that loop order has no bearing on Delta T.

[Waco]

My point here is this: You didn't need a 2nd temp sensor to find out that your water temps are almost identical regardless of what point in your loop the temp sensors are placed! Once the PC has been running and the heat output and cooling are constant, then the water temps will be basically the same throughout the loop, and a 2nd temp sensor wasn't needed to prove this.

I wasn't trying to prove that the delta between the water and the ambient was unchanged, I was proving that the difference within the loop was minimal. That's the whole point of this thread.

 

 

My point here is this: You never changed the order of the loop and IMHO the only way to prove that loop order doesn't make a difference is to actually change the loop order and see if in fact the Delta T is the same. For example you have your loop order currently with 2 temp sensors in 2 different locations in your loop, you have calculated the difference (Delta T) between the ambient (room) temp and the water temp, but all that has been proven is that the water temp, regardless of location is relatively the same. You have not proven that by changing your loop order that the Delta T would remain the same.

 

Now, if you changed your loop order from your current (and this may not be accurate, didn't see your actual loop order stated) cpu->gpu->gpu->rad->rad to rad->cpu->rad->gpu->gpu and your Delta T remained 2.5-3 C over room temp, then you would have proven that loop order doesn't matter in regards to Delta T.

 

But, maybe I am just misunderstanding your point of proving that inter-loop water temps are almost identical regardless of where in the loop you are getting a water temp reading, not that loop order has no bearing on Delta T.

You have X amount of heat dump and Y amount of dissipation capacity. Changing the order of the loop won't matter at all especially since we've just proven that the water temperatures are essentially the same anywhere in the loop.

 

I'd actually wager good money that if there is a difference at all any other loop orientation would result in worse loop temperatures. Because there is a very slight difference from the coolest to hottest point in the loop you'll want the very hottest water going through your radiators (to maximize the delta between the incoming cooling air and the radiator). Any other order would make the water temperature very slightly closer to ambient and make your radiators very slightly less effective.

 

My order is this, btw: pump -> CPU -> GPU1 -> GPU2 -> GPU3 -> GPU4 -> radiator 1 -> radiator 2 -> reservoir

Edited November 7, 2011 by Waco