Pump Heat Dump

[coolmiester]

I stumbled across this ages ago probably like many others maybe when leak testing but just lately i have been doing a lot of reservoir building for other projects and on one occasion i left a res hooked up with a DDC 3.2 for a good few hours and noticed the temperature of the liquid was really warm which got me thinking about heat dump from different pumps.

I know Martin touched on this with his Danger Den CPX-Pro Pump Review, noticing at the 2 GPM mark, the DDC is consuming nearly 21 watts, where then DD CPX-Pro is only consuming about 15 watts, so the DDC at that point is consuming about 40% more power which in the end becomes added heat to the loop so i just thought i would do a simple comparison (as only i can given limited equipment) between pumps without any form of cooling, basically to see what sort of heat dump is generated from the pump alone.

Anyway, both tests were started with normal tap water which gave a reading of just slightly over 20 degrees on start up with the temp probe attached to the top of the rad then left the loop running for 3 hours.

Results after 3 hours:
DDC 18watt 3.2 = 37.9
CPX-Pro = 29.4


The results are quite amazing really from pump to pump and it would appear running a DDC 3.2 gives quite a clear disadvantage right from the get-go though admittedly i was only using an XSPC Temp Censor to monitor temps but it does give a good like for like comparison.

Now i'm not trying to cause pump wars in any way but i do however think that this needs more in depth research with better equipment as imo, 8 or 9degrees is pretty significant in the scheme of thing.

.........anyone want to take this further??

[RickCain]

I guess this shows you need more radiator as the heat dump increases to off-set the added flow?

[jas420221]

Interesting...

But is this really a "real life" situation? I mean, wheres the radiator?

EDIT: In a 300W loop (say o/c quad and vid card) is 18W really a make or break?

EDIT: Disregard... I can see the rad if I raise my anti glare shield.......

[jas420221]

If 6% is make or break, isnt it time to get bigger rad/higher CFM fans?

[coolmiester]

I found a good while back when making the Yamaha Stacker (2x TDX - 1x GTX8800 - 1x Aquadrive - 1x GTX240 Rad) that its cooling ability was 4 or 5 degrees better when running a 10watt DDC compared to the 18watt DDC.

My point is that if as Martin found with the CPX-Pro it give almost exactly the same flow as the DDC but with less heat dump to worry about - it comes with G1/4 thread so no aftermarket top needed - its cheaper so all in all i'm struggling to find a reason to carry on using a DDC

[coolmiester]

Oh yeah, forgot that one....... its quieter than a DDC

[Sparky]

I think it might have to do with how well the pump itself is cooled too. If the pump is in decent airflow, less of the heat is likely to get into the loop than if the pump has poor air circulation around it.

We are looking at a chipset's worth of heat approximately (depending on the chipset of course), so it isn't a huge deal. We all did know that pumps put some heat into the loop, that is unavoidable.

[IanY]

How realistic is it to have absolutely no fans on the radiator? The Germans who use passive cooling have humongous radiators, sometimes the size of car radiators, or even larger.

Naturally, with mediocre thermal dissapation, the less heat dump a pump has, the lower the temperature. Using this logic, then any Iwaki is definitely a horrendous pump, and we all should be choosing the most minimalist pump with the minimum viable flow level. You say 1 gpm? Then we should never exceed 1 gpm because the extra flow must equal extra heat.

That's tantamount to saying that engine heat is high in a powerful car, so less powerful cars are more desirable.

[Martinm210]

Great test!
I did something similar with my CPX-1 pump using a thermos and a known volume of water you can calculate the amount of heat.
The calculations go something like this:

q=(specific heat of H20)X(grams of H20) x dT
specific heat of water =(4.18 J/g-K)

1ml H20= 1 grams H20
dT is the change in temperature over an hour

There is 3,600 Joules per watt-hour, so
Watts = Joules/3600

This will tell you how many watts was entering the water. I did this for the Alphacool AP1510 and came up with 16watts, I did it for the CPX-1 and came up with 6 watts, and I even tried it on my Q6600 with Prime95 and came up with around 130watts. It's an interesting experiement.

I really like the new CPX-Pro pump, I'm thinking it very well might be optimal for your typical single loop double or triple rad type system with medium restriction type blocks.

From what I can tell, our blocks of today are about 4 times less impacted by lower flow rates than the more open channel types several years ago. That just means pump heat dump is also about 4 times more important. I don't think most people need to worry about it though, pick up a good pump like a D5 or DDC or CPX-pro and run it. Chances are it's more than enough pumping power.

But hey, it's water cooling and XS. We like our extreme setups and toys, even if the benefit isn't there, can't argue with having fun..

[coolmiester]

Cheers Martin, couldn't agree more, the day watercooling becomes a chore is the day to move on but its all fun these days

[Waterlogged]

How realistic is it to have absolutely no fans on the radiator? The Germans who use passive cooling have humongous radiators, sometimes the size of car radiators, or even larger.

Naturally, with mediocre thermal dissapation, the less heat dump a pump has, the lower the temperature. Using this logic, then any Iwaki is definitely a horrendous pump, and we all should be choosing the most minimalist pump with the minimum viable flow level. You say 1 gpm? Then we should never exceed 1 gpm because the extra flow must equal extra heat.

That's tantamount to saying that engine heat is high in a powerful car, so less powerful cars are more desirable.

Also, I'm very certain that if this exact test is done using a BIP I the temps will be a bit higher and if it's done with a GTX480, temps will be a bit lower...it's all relative to the size of the "exchanger" and as such...fodder.

[cobra_kai]

It seems to me that in a typical 2x120 or 3x120 loop that has several hundred watts of heat entering it through components such as a couple gfx cards or a quad cpu that a difference in 5 or 10 watts from the pump should be negligible... its only on the order of ~5% increase in energy to the system...

[Martinm210]

FYI,
You can use this estimator if you want to see how heat load affects the water temperature:
http://www.martinsliquidlab.com/img/...stimator06.zip

Of coarse it all depends on the radiator itself and fan combination.

An MCR220 with 1350RPM fans for example will raise in water temperature by .65C with a heat load as small as 18watts.

Or folks that have gone from a CPU to a CPU plus NB loop might have seen something like a 30Watt increase due to the NB chip which is enough to bump the water temperature 1.1C.

That scales down as the radiator size increases though. An MCR320 which is about on par with a PA120.3 is only going to see a .75C gain in water temperature from a 30watt chipset.

Now if you look at really strong pumps..
An RD-30 at 24V is really dumping some heat, getting up there in the 60 watt range, that's good for about a 1.5C increase in water temperature just from the pump.


If pump heat was completely insignificant, everyone would be running RD-30's at 24V, yet that has never been the case even with the older blocks.

Like anything in life, there's an optimal for everything...

[coolmiester]

Fantastic, cheers Martin, though i must admit that did get a bit heavy for what few grey cells i got left.

Anyway, my good pal Rig did suggest - how about adding the D5 into the mix.

.................obviously i told him due to size, noise, expense and versatility due to the G1/4 thread it was straight out the reckoning

[BreeSpree]

Hmmm, interesting, I never realized that the DDC had so much heat dump. Currently I only use D5's so what would I know. Now I know not to buy a 18 watt DDC.

[dionysio]

if you lower d5's speed to say 2 heat dump almost dissapears... imho...

[bigslappy]

seems to me a test should be done of just the heat dump of all the pumps

IE; closed loop = pump > Xlarge resevior > temp probe > back to pump

start the loop at the same water temp each time for each pump let run a set amount of time then measure the heat gain .. you could even do 2 tests of each pump = one with a large fan blowing on it & then a 2nd test with no fan ...

I'd do it but all I got are Iwaki RD-30's

but I do also have one DDC/petra top that's on the test bed .....

as i have done a test with the Iwaki RD-30 in closed loop...... pump>GTX480 rad (no fans) > EK400 RES > pump > ....
and after 4 hours of run time at 18volts the loop tubing/rad top was warm to the touch ..I'm guessing here on temp !

about as warm as a toddlers forehead with the Flu

water temp at the start was 65F air temp was about the same the water in the loop was sitting in my garage & the setup was runnig also in the same gargage(NO WIND factor) as it was a cool NorthBay day
didn't do this test for temps just was checking for leaks , flow , vortexing & bleedout times of all the air bubbles of the setup

if I had all the pumps I'd do the testing .. SOO send pumps !
but then agin Been eyein' another 4870X2

if this has been done already then my life is Really tough & I'm stupid as hell !

[IanY]

bigslappy,

Wow.. you have really horrible pumps! The worst kind imaginable.

[Rapt0r]

Hmm I would still get the DDC3.2 if its more powerful with a modded top...so which is more powerful the DDC3.2 with mod top or DD CPX-Pro?

[Martinm210]

A DDC3.2 with top is quite a bit stronger than a CPX-pro.

In the end of all this, it probably depends on the waterblock/rad setup. Naeku said he saw good gains with the RD-30 at around 21V.

A DDC3.2 with XSPC res top is pretty close in performance to an RD-30 at 15V, where dual DDC 3.2s with XSPC tops is closer to an RD-30 at 21V.

From the little bit of discussion I have heard on the RD-30, 24V is almost always worse, and back in the day most people were running them around 18V. Somewhere in the 25-21V range is probably optimal.

Think of it like an air/fuel mixture ratio on your motorcyle, too much air and you can lean out your motor and fry a piston, too much and you'll foul plugs....neither of which produces the best horsepower

That's similar here, there is not any bad pump or perfect pump, it's all about tuning....the only way to know whats perfect is to test and tune your own specific setup....there may not be any sort of rule of thumb to follow..

All we know at this point is:

-There are thermal gains to flow rate, more is always better but that gain is quickly flattening beyond the 1-1.5 GPM range
-Pump heat dump plays a significant role and will quickly cross the lines of the thermal gain at some point
-Open channel blocks from several years ago had nearly 4X the thermal gains from flow rate in the 1-2GPM range than some of todays blocks
-Pumps of today such as the D5, DDC, CPX-Pro, etc...have significantly higher pressure capabilities that pond style pumps from several years ago.

The best pump and setting for your system will be one that gives you the most delta between flow rate thermal gain and heat pump heat dump loss. It's likely not going to be any set number and will heavily depend on the radiator fan combination, the waterblock used, and the amount of heat generated by the pump.

I never said an RD-30 was a bad pump, all I'm saying is heat is important and you need to spend the time to balance flow rate gains from pump heat dump losses if you truely want the "perfect" setup. We can't generalize like that, we need the patience and will to test our own setups if you really want the best possible performance possible. An RD-30 at 24V is likely well over optimal for any system, but it is an awesome pump and can run clear down to 12V if needed, so it's tunable and an excellent pump. If we don't want to bother with the work, we need to accept it's probably not perfect. For most people that don't care about a degree, which may mean the larger audience, I wouldn't worry about it...pick something and go with it.

[Kibbler]

thank you martin for your excellent info and advice!

[SNiiPE_DoGG]

Just a quick thought based on the first post, and I may be way off base here...

but if the DDC is an 18w pump and its consuming 21 watts while running, doesnt that mean that ~18 watts of those 21 watts are going into spinning the impeller and ~3 watts of it is going into sound and heat??... just thinking about it from a conservation of energy perspective

[hotdun]

A DDC3.2 with top is quite a bit stronger than a CPX-pro.

Different pumps for different.....umm...strokes, I guess. If you're in the market for a new pump, gotta use Martin's calculator for best results.

Just a quick thought based on the first post, and I may be way off base here... just thinking about it from a conservation of energy perspective

Good question...

[MomijiTMO]

Oh yeah, forgot that one....... its quieter than a DDC

This is good to know. My pump is the loudest component in my pc when my pc is at idle.

[Martinm210]

Just a quick thought based on the first post, and I may be way off base here...

but if the DDC is an 18w pump and its consuming 21 watts while running, doesnt that mean that ~18 watts of those 21 watts are going into spinning the impeller and ~3 watts of it is going into sound and heat??... just thinking about it from a conservation of energy perspective

The energy of the pump consumption leaves in three forms.

1. The biggest is motor heat energy lost into the water, which is generally somewhere between 80-90%, small pumping motors are just very inefficient.
2. The second is in the form of water horspower, you can covert the energy of water flow rate and pressure into wattage. This energy leaves the pump and is then lost in the restriction of the system. Most people don't think about this, but you do actually gain heat by the restrictions. Pressure drop in a waterblock for example is energy lost. That lost energy is transmitted into heat energy, most of which is absorbed by the water in the loop.
3. The third is motor heat lost in the air from the pump motor casing. If you've felt the bottom of a DDC pump, you know what that is. Some pumps loose more heat in the air than others and it depends on the pump motor design and how the motor does or doesn't transfer heat into the fluids. It likely also depends on how much air movement occurs around the pump.

But yes, all that energy goes somewhere, unfortunately most of it ends up getting watercooled...