Measured pump heat transfer temps?

rocketmanx · 02-02-2004, 12:32 AM

I've seen al the posts about pump heat and understand why it's a concern but haven't seen any measured temps yet.
Has anyone measured these yet? Depending on desing it would seem the difference between wattage used and actual temps to the water may differ greatly.
I did an external measurement, with my digital pyrometer, of the casing on my MCP600 which was 87F (varied depending on how close to the impeller of course) with a room temp of 74F. Lines measured 76F in and 78F out. This is of course useless as far as water temp is concerned but it can be guessed that the internal temp is higher than the case, especially since the case has a fan blowing across it.

wicka · 02-02-2004, 08:00 PM

I dunno...but it seems to me that it'd be kinda hard to measure the temp of the water inside the pump...

rocketmanx · 02-03-2004, 07:36 AM

Yeah, it appears that jaydee's having a beech of a time measure cpu wattage.
I'm asking about acroos the pump at the in and out. An inline sensor wouldn't be too bad but I'm just askin' as I was curious to the true amountof actual heat tranferred to the water. A "baseline" of sorts could be done though by using comparison relaitve to a certain pump. But, still, just wonderin' as I'm not about to do it. Just wnodering if the test existed.

wicka · 02-03-2004, 04:28 PM

Ah...well it wouldn't seem too hard to do it that way, hell if you wanted to you could just slice a little hole in the tubing and drop a probe in there, then seal it up real nice. That'd be uber ghetto, hehe.

fhorst · 02-03-2004, 05:26 PM

as far as I'm aware, there is not such test availible.

All submerged pumps will transfer their full wattage into the water.
All inline pumps just a small portion (3 to 5 % ?) as only the axis and the impeller + impeller housing are in contact with the water.

I did a test once, with an inline pump, 35 watts, made a closed loop with only a small res.
After 2 days of running, the water was only slightly above ambient (total of water 300cc)

I also did a test with submerging the pump. After 1 hour, the water was about 20 degrees above ambient. (this time it was little over one litre)

After this home test I decided that for me I will use only small low wattage pumps submerged. (like the small ones from Koolance 3 watt's each)
Bigger pumps only go inline!

wicka · 02-03-2004, 08:33 PM

Quote:

Originally Posted by fhorst

as far as I'm aware, there is not such test availible.

All submerged pumps will transfer their full wattage into the water.
All inline pumps just a small portion (3 to 5 % ?) as only the axis and the impeller + impeller housing are in contact with the water.

I did a test once, with an inline pump, 35 watts, made a closed loop with only a small res.
After 2 days of running, the water was only slightly above ambient (total of water 300cc)

I also did a test with submerging the pump. After 1 hour, the water was about 20 degrees above ambient. (this time it was little over one litre)

After this home test I decided that for me I will use only small low wattage pumps submerged. (like the small ones from Koolance 3 watt's each)
Bigger pumps only go inline!

Ah, well that was helpful. Thx dude!

freeloadingbum · 02-03-2004, 08:35 PM

If I had an inline pump that drew a true 35 watts, then I'd generally expect about 28 watts to go into the water and 7 watts (20%) of heat to be dissipated off the motor

If my system water temp was 5C above ambient under load and the total load was a 100 watts (cpu and pump combined), then I'd expect submerging the pump to increase the water temp around .35C (everything else being the same).

AngryAlpaca · 02-03-2004, 09:37 PM

Let's do some math! MATH IS FUN!!! The procore has a thermal resistance of approximately 0.02 C/W, and let's assume that ALL 35 watts from the pump went into the water. 35 X 0.02 = 0.7 degree increase in total water temperature. Not really all that harmful, hmm? Submerging a pump will allow a slightly higher flow rate, and that may offset the heat a little bit, but not enough. However, if what freeloading says is true about the inline one putting 28 watts into the water, then the difference is 7 watts, or 0.14 C cooler water. Cathar, I believe, estimates it at around 20 watts, or 0.30 degrees cooler than a submerged pump, that is impossibly transferring 35 watts to the water. By the way, fhorst, it sounds like the entire setup is different, and that makes ANY results you got irrelevant. Also, saying the water was "slightly" above ambient in the first case, and 20 degrees above it in the second case also reduces the relevancy, as "slightly" could, in fact, mean 73972 degrees Celcius.

Sorry, rocketmanx, I know that wasn't the answer you were looking for, but I was trying to bring a little bit of fact into the discussion.

fhorst · 02-04-2004, 03:17 AM

LOL. By slightly i mean 25 degrees at an ambient of 20 degrees.

To make it more sipmle, if you feel the temp of your inline pump after you wrap it with a towel e.g. tou are happy that you don't bring that temp into your water.

The funny thing I see all over the place, is that you all expect the radiator to get rid of all the heat. A radiator is far from 100% effective. If that was the case, a BIX micro would be enough to keep your CPU loop at ambient. Well, I can tell you, it's not.

Cathar · 02-04-2004, 05:42 AM

I've measured the amount of heat that various pumps dump into the loop. I typically achieve this in a bit of a kludgey way, but seems to be effective.

I use a foam esky, exactly 6 litres of water, the pump and some tubing. I estimate the thermal capacity of the system, and pour in water that's slightly colder than ambient. I then measure the time it takes for the pump to heat the water from 1C below ambient to 1C above ambient.

Using a 12V in-line pump I have here, and running it through the multi-meter to establish its exact power draw, I was able to match up what I measured the heat input into the water to be, to the pump's power input, and the two matched within 3% (which was incidentally 14.6W). The pump in question was a mag drive, but instead of the impeller sitting inside the electromagnet motor body, the magnetic impeller mechanism is shaped like a hole-saw bit and surrounds the electro-magnet mechanism, resulting in a pump that transfers very close to 100% of its power draw into the water as heat.

Different pumps transfer different amounts of heat, dependent upon their design.

ALL pumps (in-line or nor) will transfer at least as much heat into the water as the shaft power of the motor, which is typically around 50-60% of the actual motor power draw. Of that shaft power for centrifugal pumps, anything from 10-50% gets used as motive force for pushing the water, the rest is simply wasted as frictional energy (heat) due to the fairly large inefficiencies of the centrifugal pumping action. Closed impeller designs will get to the 50% value. Open impeller designs with sloppy clearances will be down around the 10% mark (or even lower).

On top of that, most pumps that give a single power rating will typically draw around 50-70% of their rated power with 100% of their rated power only being drawn at startup, unless an explicit power draw curve is given with the PQ curve, in which case that's the correct power consumption to trust.

It is highly unlikely that you will ever see less than about 30% of a in-line centrifugal pump's peak rated power being added as heat to the water.

A random test of how warm the water gets is useless. A test of how quickly the water temperature changes over a fairly small deviation from ambient in an insulated environment is a somewhat accurate method.

The Iwaki pumps do dump a lot of heat into the water, despite the motor being mostly separate/insulated from the pump body. The reason being that for my Iwaki MD30-RZ, the shaft power is about 45W, and immediately ALL of that is going to end up as heat in the water through one way or another. Then there's the actual pump motor heat, for which some of it will make it into the water. Even if it's just 10% of the pump motor's heat, that takes it up to the ~50W of heat output that I measured that the Iwaki MD30-RZ does dump into the system - and it's an in-line only pump.

AngryAlpaca · 02-04-2004, 08:46 AM

"The funny thing I see all over the place, is that you all expect the radiator to get rid of all the heat. A radiator is far from 100% effective." I used a thermal resistance of 0.02... Far from 100%, speaking for a radiator. So that was inline, and you measured the same amount of energy being transferred to the water as drawn? What exactly was this pump?

wicka · 02-04-2004, 02:33 PM

Personally, I think it'd be quite useful for some people to test the heat that their pumps added to the water, and then make a little graph or something...

But that's just me.

freeloadingbum · 02-04-2004, 04:34 PM

Quote:

Cathar: I use a foam esky, exactly 6 litres of water, the pump and some tubing. I estimate the thermal capacity of the system, and pour in water that's slightly colder than ambient. I then measure the time it takes for the pump to heat the water from 1C below ambient to 1C above ambient.

Did you ever try to use that method to calculate cpu wattage?

Les · 02-04-2004, 07:05 PM

Quote:

Originally Posted by freeloadingbum

Did you ever try to use that method to calculate cpu wattage?

This ,to my knowledge, is Cathar's last posted work in this field.
Although nearly 2 years old, the technique not as refined, and our understanding of heat transfer was less advanced it is still instructive.

Cathar · 02-04-2004, 07:48 PM

Quote:

Originally Posted by Les

This ,to my knowledge, is Cathar's last posted work in this field.
Although nearly 2 years old, the technique not as refined, and our understanding of heat transfer was less advanced it is still instructive.

Thanks for digging up the old link Les.

That experiment I did back then was horribly inaccurate due to the radiators still being attached (even though they were insulated).

I will attempt an estimation of my MD30-RZ wattage, and then another with the CPU running at full load.

At best I figure this method is good for +/- 10% accuracy, and most depends on how accurately one can estimate the thermal capacity of the system. The more water the better though as the thermal capacity of the water will greatly "overshadow" any estimation of thermal capacity of the rest of the loop.

freeloadingbum · 02-04-2004, 08:37 PM

Very interesting link Les. Alot of good info. I'm sorry for distracting from the original thread topic, but I was wondering if BillA ever did the TIM test with the copper spacer in the middle?

Back on topic: Cathar, how did you arrive at the 50 to 60% figure for pump motors? I had read in another thread where myv65 said motor efficiency was around 75 to 85%. Is there something different with regards to pump motors that decreases it to your value, or was myv65 just wrong?

Cathar · 02-04-2004, 08:48 PM

Quote:

Originally Posted by freeloadingbum

Back on topic: Cathar, how did you arrive at the 50 to 60% figure for pump motors? I had read in another thread where myv65 said motor efficiency was around 75 to 85%. Is there something different with regards to pump motors that decreases it to your value, or was myv65 just wrong?

Hmmm, when it comes to pumps I would always defer to Dave (myv65) over myself.

I got those ratios from browsing through the Iwaki catalog and looking at their mag drive pump motor output vs motor input ratios. Mind you, I'm talking mag-drive motors, and myv65 is probably talking direct-drive motors. It may be possible to get a 75-85% efficient mag-drive, but I doubt it, or at least Iwaki can't achieve it and mag-drive pumps don't come much better than the Iwaki's (plenty of makers around about the same level of efficiency though).

Edit: Some of the larger Iwaki mag-drive do appear to get to 70% efficient, but we're talking 100W+ motors here, not 10-30W mag-drive motors. All Iwaki motors below 100W or so see a fairly large drop off in efficiency, with the smallest models being just 15% efficient.

My guess is that it becomes extremely hard to make efficient mag-drive motors once you get down to the levels we want (<50W).

Les · 02-05-2004, 01:35 AM

[quote=freeloadingbum].......was wondering if BillA ever did the TIM test with the copper spacer in the middle?.....
QUOTE]

http://forums.overclockers.com.au/sh...threadid=67401

freeloadingbum · 02-05-2004, 03:50 AM

Thanks Les. That link answered alot of questions I had bouncing around in my head for the last year or so.

xgman · 02-05-2004, 01:14 PM

All I know is that if my stove ever breaks, I can always fry an egg on top of my inline pump.

02-02-2004, 12:32 AM	#1
rocketmanx Cooling Savant Join Date: Nov 2003 Location: Indiana Posts: 127	Measured pump heat transfer temps? I've seen al the posts about pump heat and understand why it's a concern but haven't seen any measured temps yet. Has anyone measured these yet? Depending on desing it would seem the difference between wattage used and actual temps to the water may differ greatly. I did an external measurement, with my digital pyrometer, of the casing on my MCP600 which was 87F (varied depending on how close to the impeller of course) with a room temp of 74F. Lines measured 76F in and 78F out. This is of course useless as far as water temp is concerned but it can be guessed that the internal temp is higher than the case, especially since the case has a fan blowing across it. __________________ "The more I learn the less I know, but I never knew as much as when I was sixteen"

02-02-2004, 08:00 PM	#2
wicka Cooling Neophyte Join Date: Feb 2004 Location: Bloomington, IN Posts: 23	I dunno...but it seems to me that it'd be kinda hard to measure the temp of the water inside the pump... __________________ Project Ares 1700+ \| NF7-S 1.2 \| 256MB Buffalo PC3700 \| Chaintech Ti4200 Dtek WhiteWater \| HWLabs Black Ice Xtreme \| Via Aqua 1300

02-03-2004, 07:36 AM	#3
rocketmanx Cooling Savant Join Date: Nov 2003 Location: Indiana Posts: 127	Yeah, it appears that jaydee's having a beech of a time measure cpu wattage. I'm asking about acroos the pump at the in and out. An inline sensor wouldn't be too bad but I'm just askin' as I was curious to the true amountof actual heat tranferred to the water. A "baseline" of sorts could be done though by using comparison relaitve to a certain pump. But, still, just wonderin' as I'm not about to do it. Just wnodering if the test existed. __________________ "The more I learn the less I know, but I never knew as much as when I was sixteen"

02-03-2004, 04:28 PM	#4
wicka Cooling Neophyte Join Date: Feb 2004 Location: Bloomington, IN Posts: 23	Ah...well it wouldn't seem too hard to do it that way, hell if you wanted to you could just slice a little hole in the tubing and drop a probe in there, then seal it up real nice. That'd be uber ghetto, hehe. __________________ Project Ares 1700+ \| NF7-S 1.2 \| 256MB Buffalo PC3700 \| Chaintech Ti4200 Dtek WhiteWater \| HWLabs Black Ice Xtreme \| Via Aqua 1300

02-03-2004, 05:26 PM	#5
fhorst Cooling Savant Join Date: Oct 2003 Location: Almere, The Netherlands (Europe) Posts: 262	as far as I'm aware, there is not such test availible. All submerged pumps will transfer their full wattage into the water. All inline pumps just a small portion (3 to 5 % ?) as only the axis and the impeller + impeller housing are in contact with the water. I did a test once, with an inline pump, 35 watts, made a closed loop with only a small res. After 2 days of running, the water was only slightly above ambient (total of water 300cc) I also did a test with submerging the pump. After 1 hour, the water was about 20 degrees above ambient. (this time it was little over one litre) After this home test I decided that for me I will use only small low wattage pumps submerged. (like the small ones from Koolance 3 watt's each) Bigger pumps only go inline! __________________ If it get's hot, it needs to be watercooled!

02-03-2004, 08:35 PM	#7
freeloadingbum Cooling Savant Join Date: Nov 2002 Location: winnipeg Posts: 129	If I had an inline pump that drew a true 35 watts, then I'd generally expect about 28 watts to go into the water and 7 watts (20%) of heat to be dissipated off the motor If my system water temp was 5C above ambient under load and the total load was a 100 watts (cpu and pump combined), then I'd expect submerging the pump to increase the water temp around .35C (everything else being the same).

02-03-2004, 09:37 PM	#8
AngryAlpaca Cooling Savant Join Date: Jan 2004 Location: Alberta Posts: 631	Let's do some math! MATH IS FUN!!! The procore has a thermal resistance of approximately 0.02 C/W, and let's assume that ALL 35 watts from the pump went into the water. 35 X 0.02 = 0.7 degree increase in total water temperature. Not really all that harmful, hmm? Submerging a pump will allow a slightly higher flow rate, and that may offset the heat a little bit, but not enough. However, if what freeloading says is true about the inline one putting 28 watts into the water, then the difference is 7 watts, or 0.14 C cooler water. Cathar, I believe, estimates it at around 20 watts, or 0.30 degrees cooler than a submerged pump, that is impossibly transferring 35 watts to the water. By the way, fhorst, it sounds like the entire setup is different, and that makes ANY results you got irrelevant. Also, saying the water was "slightly" above ambient in the first case, and 20 degrees above it in the second case also reduces the relevancy, as "slightly" could, in fact, mean 73972 degrees Celcius. Sorry, rocketmanx, I know that wasn't the answer you were looking for, but I was trying to bring a little bit of fact into the discussion. Last edited by AngryAlpaca; 02-03-2004 at 09:47 PM.

02-04-2004, 03:17 AM	#9
fhorst Cooling Savant Join Date: Oct 2003 Location: Almere, The Netherlands (Europe) Posts: 262	LOL. By slightly i mean 25 degrees at an ambient of 20 degrees. To make it more sipmle, if you feel the temp of your inline pump after you wrap it with a towel e.g. tou are happy that you don't bring that temp into your water. The funny thing I see all over the place, is that you all expect the radiator to get rid of all the heat. A radiator is far from 100% effective. If that was the case, a BIX micro would be enough to keep your CPU loop at ambient. Well, I can tell you, it's not. __________________ If it get's hot, it needs to be watercooled!

02-04-2004, 05:42 AM	#10
Cathar Thermophile Join Date: Sep 2002 Location: Melbourne, Australia Posts: 2,538	I've measured the amount of heat that various pumps dump into the loop. I typically achieve this in a bit of a kludgey way, but seems to be effective. I use a foam esky, exactly 6 litres of water, the pump and some tubing. I estimate the thermal capacity of the system, and pour in water that's slightly colder than ambient. I then measure the time it takes for the pump to heat the water from 1C below ambient to 1C above ambient. Using a 12V in-line pump I have here, and running it through the multi-meter to establish its exact power draw, I was able to match up what I measured the heat input into the water to be, to the pump's power input, and the two matched within 3% (which was incidentally 14.6W). The pump in question was a mag drive, but instead of the impeller sitting inside the electromagnet motor body, the magnetic impeller mechanism is shaped like a hole-saw bit and surrounds the electro-magnet mechanism, resulting in a pump that transfers very close to 100% of its power draw into the water as heat. Different pumps transfer different amounts of heat, dependent upon their design. ALL pumps (in-line or nor) will transfer at least as much heat into the water as the shaft power of the motor, which is typically around 50-60% of the actual motor power draw. Of that shaft power for centrifugal pumps, anything from 10-50% gets used as motive force for pushing the water, the rest is simply wasted as frictional energy (heat) due to the fairly large inefficiencies of the centrifugal pumping action. Closed impeller designs will get to the 50% value. Open impeller designs with sloppy clearances will be down around the 10% mark (or even lower). On top of that, most pumps that give a single power rating will typically draw around 50-70% of their rated power with 100% of their rated power only being drawn at startup, unless an explicit power draw curve is given with the PQ curve, in which case that's the correct power consumption to trust. It is highly unlikely that you will ever see less than about 30% of a in-line centrifugal pump's peak rated power being added as heat to the water. A random test of how warm the water gets is useless. A test of how quickly the water temperature changes over a fairly small deviation from ambient in an insulated environment is a somewhat accurate method. The Iwaki pumps do dump a lot of heat into the water, despite the motor being mostly separate/insulated from the pump body. The reason being that for my Iwaki MD30-RZ, the shaft power is about 45W, and immediately ALL of that is going to end up as heat in the water through one way or another. Then there's the actual pump motor heat, for which some of it will make it into the water. Even if it's just 10% of the pump motor's heat, that takes it up to the ~50W of heat output that I measured that the Iwaki MD30-RZ does dump into the system - and it's an in-line only pump.

02-04-2004, 08:46 AM	#11
AngryAlpaca Cooling Savant Join Date: Jan 2004 Location: Alberta Posts: 631	"The funny thing I see all over the place, is that you all expect the radiator to get rid of all the heat. A radiator is far from 100% effective." I used a thermal resistance of 0.02... Far from 100%, speaking for a radiator. So that was inline, and you measured the same amount of energy being transferred to the water as drawn? What exactly was this pump?

02-04-2004, 02:33 PM	#12
wicka Cooling Neophyte Join Date: Feb 2004 Location: Bloomington, IN Posts: 23	Personally, I think it'd be quite useful for some people to test the heat that their pumps added to the water, and then make a little graph or something... But that's just me. __________________ Project Ares 1700+ \| NF7-S 1.2 \| 256MB Buffalo PC3700 \| Chaintech Ti4200 Dtek WhiteWater \| HWLabs Black Ice Xtreme \| Via Aqua 1300

02-04-2004, 08:37 PM	#16
freeloadingbum Cooling Savant Join Date: Nov 2002 Location: winnipeg Posts: 129	Very interesting link Les. Alot of good info. I'm sorry for distracting from the original thread topic, but I was wondering if BillA ever did the TIM test with the copper spacer in the middle? Back on topic: Cathar, how did you arrive at the 50 to 60% figure for pump motors? I had read in another thread where myv65 said motor efficiency was around 75 to 85%. Is there something different with regards to pump motors that decreases it to your value, or was myv65 just wrong?

02-05-2004, 01:35 AM	#18
Les Cooling Savant Join Date: Oct 2001 Location: Wigan UK Posts: 929	[quote=freeloadingbum].......was wondering if BillA ever did the TIM test with the copper spacer in the middle?..... QUOTE] http://forums.overclockers.com.au/sh...threadid=67401

02-05-2004, 03:50 AM	#19
freeloadingbum Cooling Savant Join Date: Nov 2002 Location: winnipeg Posts: 129	Thanks Les. That link answered alot of questions I had bouncing around in my head for the last year or so.

02-05-2004, 01:14 PM	#20
xgman Cooling Neophyte Join Date: Nov 2003 Location: Maryland Posts: 27	All I know is that if my stove ever breaks, I can always fry an egg on top of my inline pump.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)