Pumps and heat
 

I've read a lot of comments lately about pumps adding heat to the water, and that rads should be after the pump and before the waterblock in order to get the coldest water possible through the block. I was curious if anyone has done any tests concerning how much heat the pump actually does dump into the cooling loop (on an inline setup of course). My first instinct is that even though various pumps are rated at 10W, 28W or whatever, very little of that heat makes it to an inline setup. Remember, most of our pumps are made from polypropylene, epoxy and ceramic, two materials which are not the best conductors. Also, our pumps are typically mag-drive, which means that the motor is isolated from the impeller and housing.

If it is the case that the pump does not add very much heat to the water, I think it makes since to have the rad follow the block as closely as possible, to maximize the thermal differential, and have the rad empty into a resevoir. The resevior can only help cool things down, since unless you're using a chiller or a phase-change setup we're not cooling below ambient anyways.

Comments? I am interested in hearing what you all have to say.

I placed a stainless steel reservoir (about 1/2 litre) and a BIX with two 120mm fans outside of my case (on its top) for the horisontal airflow. So I am going to use the setup "Eheim 1040 or 1250 (outlet) - WB - BIX - Reservoir - Eheim (inlet)". Thus the reservoir will cool the water after the pump (if it warms the water). If it will be necessary, I can place a 120mm fan near the reservoir.

Sorry, I meant setup "" Pump(outlet)- reservoir-WB-BIX-pump (inlet).

I tried to address this issue in another thread i started some time ago. http://forums.procooling.com/vbb/showthread.php?s=&threadid=3323&highlight=ambient or here's the gist of it.

Well for a little test last night, i hooked up my system with just the rad, res, and pump (I don't have a block yet.) And for those of you who thought more flow bigger pump is better, hmm.... it adds A LOT of heat to the water. Here is what I found

All temps were taken with dd5

setup is danner mag 5 rated 500gph @ 0 foot head.
4x4x6 underground electrical box for res, w/ probe mounted in it.
Black Ice Extreme w/ sunon 86cfm 120mm shrouded @ 12v.
1/2" vinyl
bout 3 quarts of tap filled system with about 1 cup anti freeze.


ambient 24c - 23c
pump left on over night without rad fan on water temp = 38c this morning.
with rad fan on 25c

Ok... sounds like an interesting idea... though your res sounds more like a really large section of pipe than a res. Do you think that the res will have a pocket of air at the top, or are you going to keep it full? Also, your res is not going to be open is it?

yeah it is more like a really big pipe. There isn't any air in the top of the res, and it is sealed.

Any decent pump is going to add heat to the system. Ambient's test clearly shows that as long as you have a decent radiator, this is not an issue, even with a Danner 500 gph pump which runs at 45 watts. His water temps. increased by 1-2C with the radiator. Most heater cores can tolerate a heckuva lot more heat than that. I have 4 pelts., two pumps and two waterblocks in my system and the heat from the pumps is easily handled by the radiator.

What are you basing this on? Have you tested it, or are you simply assuming that your pump drops that entire 45W into the coolant loop? If that were the case, then it would be cool to the touch...

The pump isn't dumping ALL of the energy into the water as heat. Some is used as mech energy, some as heat. Some of the heat gets transferred to the water through the casing of the pump, and about as much gets transferred to the air/water around it. IF you are using a submerged unit, ALL heat generated by the pump is being transmitted into your water (you were warned ;) ). By nature, centrifugal pumps are VERY inefficient, and quite a bit of the energy is converted to heat (which is why many require submersion, or else the engine would BURN UP).

Pumps add heat. It's a fact. If you are adding more than 3-4C to water temp, there is something drastically wrong. I'm using an extremely overrated pump, and it does raise the water temp by a couple degs C when in use with the rad fans on but the CPU off. Most rads can handle the excess heat easily, especially since they are more efficient with a greater temperature differential between the fluids (water/air).

just to note, there is an 80mm vantec stelth fan @21cfm 3 inches behind the pump blowing directly on the pump. If i had the means i would run the test with the pump submerged that would prove/disprove weather the entire 45w was dumped into the water. Anyone else have data on this?

Airspirit... how is this heat being put into the water though? The impeller and casing are made from polypropylene, and the impeller shaft is ceramic and is "attached" to the motor with a magnetic field... the motor, which is the thing making all of that damn heat is never in contact with the water. Like I said, I want some numbers. :)

That I concede... it's just a question of how much......

You could test, but I'll demonstrate the two scenarios:

Situ1 (inline): The heat bleeds from the motor to the housing, most of which is surrounded in open air, and part of which is filled with water. The water filled portion wicks heat FASTER than the air (not an assumption unless you have VERY large surface area on your pump) giving half, or maybe more of the heat energy to your water. A fan may help lower that transmission by putting cooler air over the pump allowing it to wick more heat to the air.

Situ2 (subm): All heat generated by the pump will go into the water. There is nowhere else for it to go. It doesn't just disappear.

The only caveat is that the inline pump will create more heat per W rating due to motor friction (the subm normally are water lubed as well while inline are NOT).

I don't know if that babbling helps, but it seems to apply ... heh.

Edit: Skulemate, damn near everything can transmit heat. Though the motor compartment is sealed, it does generate heat, and it must go somewhere. Heat does not just diappear. It ends up bleeding through the casing, part of which is in contact with the water. The rest is detailed above.

I know... don't get me wrong. While I am not an expert with heatflow by any means, I do remember some of what I was taught. And I don't think I disagree with a single point that you made in your post. Like I said, it is only a question of how much heat is transferred......

P.S. I never meant to include submerged setups in this thread, because... well... where else would the heat go? :D

The funny thing is the folks that say "the water is cooling the pump, but the pump is heating the water!?!?! WHY IS THAT!?!?!?"

People like that give me a kick, though I doubt you'd find any here. You have to be at least somewhat savvy to even ATTEMPT this kind of monstrosity in your machine.

Anyway, since water wicks heat so much better than air, I would guess, with neutral airflow, on most pumps I've seen you'd get about 40-60% transmission to the water (remember, the air in your case isn't usually the coolest in the world). With a fan blowing across the pump, you can probably shave off about another 10% if the air you're blowing is colder than the water passing through it.

That is all estimates, and could bear some testing, but I'm pretty sure I'm hitting it with about a 10% margin of error, and if anything, that 10% would be going to the water. Remember also, the motor is directly behind the impeller housing and is usually thinner than the housing itself.

does anyone know if the input impeller housing is insulated between the motor encasement? I was thinking about cutting the back end of the pump off to aid in cooling it. Any thoughts:cool:

I was thinking of running a test myself which involved using a sealed cooler as a resevior. I'd set up the pump to circulate the water, with the resevior approximately 4' above the pump centre line. I'd measure temperature using my Radio Shack indoor/outdoor thermometer. Any comments on this? I am especially interested in any obvious sources of error that I have missed.

About the insulation... mine is not. As far as I can tell, there is only the polyproplyene housing between the magnet and the fluid.

Are you going to run the test 2 ways, submerged and inline? As long as you let the water cool back to ambient in between testing the results shlould be valid.

No, I can't since my pump is inline only.

I have run a short test, and am in the process of interpreting the data. I will post both the data and my interpretation when I am done so you can make your own conclusions.

Friction is the other main way that pump adds heat to the water, in an in-line system 100% of the kenetic energy given to the water by the pump is converted into heat.

higher flow=more friction=more heat

NO!

All energy is coverted to heat. Always. If your pump uses 30w under normal conditions, its producing 30w of heat.

The casing is also made of insulating plastic, and clearly the heat is leaving the pump somehow, so how is relativly unimportant.

My guess is that since both the coolant and the air are about equally insulated from the motor, the coolant takes more of the heat then the air simply by virtue of its much higher flowrate and conductivity.

Another thing to consider is that not 100% of the pumps heat is dissipated at the pump. After all the coolant is moving, and thus carring away a good deal of kinetic energy that only becomes heat later.

Considering this I'd guess that about 2/3 to 3/4 of the the heat goes into the coolant typically, but this is just a shot in the dark. I'm probably off.

No, some of the energy is transformed in mechanical energy. Airspirit is correct.

The 45 Watts (for example) of energy that is supplied to the pump, in the form of electricity, is used for the most part to turn the impeller. Whatever is NOT used to turn the impeller, appears in other forms and include:
-noise
-heat
-fighting friction (thx Volenti)
-magnetic field (aka EMI, Electro-Magnetic Interference)

Very few, if any, of the things we use are 100% efficient. Of all the power provided to a device, there is always some kind of loss. The best example of this is the lightbulb: it also emits a lot of heat.

Overall, our physics laws apply: nothing is lost, nothing is gained. All of the energy supplied is transformed into something.

As for pumps, to me it's very simple: a pump is a motor. A motor is composed of coils of wire. These coils generate a magnetic field that drives the impeller. If the wire in the coils is a low grade, then it will heat up very easily, but will transmit a magnetic field more efficiently (because it's wound up closer to the impeller). If the grade of this wire is too high, then the magnetic field is not as efficiently transmitted to the impeller, but there's not as much heat.

I think you'll all find that every model of pump is different. This heat, from the motor coils, is transmitted to the water, through the impeller housing. This housing is usually made of some kind of plastic, which is a good insulator against heat, but never a perfect one (depends on the grade of plastic). Plastic is used here because it is cheap, and because it can be used in a (relatively) thin layer, between the electrical coils, and the impeller. This layer needs to be as thin as possible, otherwise there's a loss of efficiency in the transmission of the magnetic field that drives the impeller. Plastic also has this advantage where it does not interfere with magnetic fields.

In terms of efficiency, magdrive pumps are typically around 70% to 80% efficient (Someone correct me if I'm wrong here). So, about 80% is used to turn this impeller, and the rest is EMI, heat, noise and fighting friction. In which proportion, depends on the pump design.

I hope I've helped clarify this issue.

I'm looking forward to Skulemate's figures!

"ALL energy is converted to heat", wrong. a pump moves water by converting somewhere around 20% (varies greatly) of the electrical energy into fluid momentum, the rest creates heat. [edit: bigben beat me to it, my dial-up is buggy so it took a few tries to upload]

some of you might recall my fan duct for my ehiem 1250. i did a test. I fixed all my other fans typical operating RPM, turned off the pump's 80mm Antec fan, ran SiSoft CPU burn-in for 3hrs, & took temps. then turned on the pump fan to 30% duty cycle (~30% max RPM) & did the same:

w/o pump fan:
h2o=33.4C
ambient=24.6C
diff=+8.8C

w/ ducted 80mm pump fan @ 30% speed:
h2o=31.0C
ambient=25.2C
diff=5.8C

that's 3C cooler just by moving a small amount of air over the pump. so i may have proved that inlines need airflow, because without air-flow the heat builds up quickly on & around the pump & water temps generally will rise as a result. for even the most powerful inline pumps, a small amount of air flow is all that is required to sustain low temps, so the benefits of a high CFM pump can be realized (more CFM=more water velocity=more turbulence=better cooling). further, about 10CFM directed on the ehiem1250 is enough to peel the heat off & dilute with cool-air, your exhaust fan(s) will do the rest.

Good work Nemaste! That's a mod article all on it's own! Maybe mounting a 30 or 40mm fan right on the Eheim 1250, where the motor housing is cut open, would work well.

I completed one short test, lasting 30 minutes. My setup included my insulated resevior positioned approximately 3' above the centreline of the pump, with 5' of 5/8" vacuum PVC tube as an intake and 5' of 1/2"x5/8" PVC as the discharge. I had my temp probe sumberged in the resevior, and when running there was considerable mixing of the fluid in the resevior (at this head my pump moves approximately 580 GPH). I used 5L +/- 5% of tap water in the resevior, and siphoned from the resevior to prime the pump and tubes. After setup I waited five minutes to check that all was well, and to allow the pump to reach its operating temperature, before taking the first measurements. Here are the results from my short initial test. I do plan on repeating the test for a longer period of time to check whether or not these results are repeatable. Please bear with the ugliness of the table:

Time.........H2O °C...Ambient °C
21:18........23.2........22.3
21:22........23.8........22.1
21:27........24.3........22.0
21:35........25.1........21.8
21:42........25.8........21.7
21:58........27.2........22.2

I then plotted the results in excel, and added a trendline, which had R=0.9987 (omitting the first point since it was not exactly linear... guess I should have waited another minute or so before beginning the test). As you can see, there was a temperature increase of 4.0°C in 30 minutes. This temperature rise corresponds with a 83700J increase in energy, which is approximately 46.5W (at least, that's how I remember it being done... please correct me if I am off base). Since my pump is rated at 190W, that's 25% of the pumps rating going to the water.

Comments or questions?