Pump heat?

[Blackeagle]

I've seen many times posts/threads regarding the heat added by pumps to the systems water.

Iwaki pumps are often refered to as adding less heat (percentage of pump use) that other pumps on the market.

How large is the differance between free flow and under the resistence of a system with 2 blocks, a example being a Cascade CPU block and a Fusion GPU block, with 1/2" lines.

If a particuler pump needs to be spec'd, then let's use the MD30rlzt, as it's about as big as anyone would use. Or for a smaller pump the MD15, still strong, but more reasonable.

The MD15 has a 28 watt draw & the MD30 has a 90 watt draw.

Perhaps a short version being: What would be the maximum amount of added heat be from these pumps under a heavy load/resistence?

[Cathar]

I use the 50Hz version of the MD-30RZ. It's rated as a 70W power draw.

The amount of heat that it adds to the water is not insignificant at all. As a rough guesstimate, around 40W of heat is added to the water, which percentage wise, is about the same as a submerged Eheim pump adds to the water.

By itself it pumps pretty much bang on 15LPM in free-flow mode. Pumping through two heater-cores in series sees over 14LPM still flowing.

With a Cascade (SS) in the loop as well, we're somewhat below the 11LPM mark, and with a moderately restrictive micro-channel GPU block in the loop, down to around the 9.5LPM mark.

For the 60Hz version of the MD-30RZT, I'd estimate >50W of heat being added to the water. I'd also being expecting around 10% higher flow rates than the 50Hz model on average.

[nikhsub1]

I wish there were a way to actually quantify this. Personally, I don't think as much heat makes it to the water as Cathar does, but that is not saying much. If I am not mistaken, pumps expel the most heat at zero restriction, as flow resistance builds up, they expel less heat. Please correct me if I am mistaken.

[pHaestus]

You can quantify it but you need calibrated temp probes on the inlet and outlet of pump and a flowmeter. You can then calculate W from the temperature rise of water across the pump. You'd probably want to characterise this as a function of flow rate with a ball valve though as pump power will vary with restriction.

[freeloadingbum]

By pHaestus: You can quantify it but you need calibrated temp probes on the inlet and outlet of pump and a flowmeter. You can then calculate W from the temperature rise of water across the pump. You'd probably want to characterise this as a function of flow rate with a ball valve though as pump power will vary with restriction.

I don't think that would be enough because you would only be measuring a portion of the pump heat. There would also be heat converted throughout the whole system. The lower the resistance, the higher the portion would be throughout the system and the less you would measure across the pump

As I understand it, only about 20% of the consumed wattage is lost at the motor. The rest goes into the water as a combination of both heat and flow but it all becomes heat in the end.

[bigben2k]

Do I need to remind everyone about the different sources of heat?

There's heat from the motor coils, as the current passes through them, which is passed, to some extent, to the water.

But more importantly, there's the heat from the actual pumping action, which will appear at each of the restrictions throughout a loop, as well as inside the pump housing.


Oh never mind... I'll just run the test myself (when I have the time and equipment) and report results.

[pHaestus]

Sorry Ben. Didn't mean to offend the experts.

Wouldn't any rise in temperature from restriction and friction and pumping action be independent of pump? For these types of temp rises then pushing x gpm through a loop with any pump should yield the same temp rise. As a practical matter, can you actually measure these rises? What magnitude? Quite small I'd guess across any given component. What resolution is needed? 0.001C would be my guess for MINIMUM resolution to detect such.

If they are just a function of flow rate though then pumps can still be compared to one another by the delta T across the pump itself (the only variable in the system would be how much of a pump's heat is dumped into water).

I seem to recall someone measuring the volume of water in loop, taking a stab at thermal conductivity of all the parts in loop, and then making an educated guess by looking at dT/dt. Sorta fuzzy on details though; was long ago...

[freeloadingbum]

As long as you know the flow and pressure drop you should be able to calculate the system's portion of pump heat. Add that to the wattage calculated from the temp rise across the pump and you should have the total amount of pump heat. I'll see if I can find the formula for the system heat

[bigben2k]

I don't believe that it would be independant, no: remember that our poor mag drives are poorly efficient and that some water actually recirculates within the impeller housing.

If measuring, I'd do it across a restriction, not the pump.

Instead of measuring deltaT, wouldn't it be easier to measure the water temp increase over a longer period of time?!?:shrug:

[pHaestus]

Why would the flow resistance part of the pump's heat change from pump to pump Ben? I don't follow. Wouldn't that only be a function of flow rate and the resistances in the loop? Now there is a portion of the heat dumped into water directly by the pump. THAT would vary from pump to pump. I don't see how a pump's efficiency could affect an equation for secondary heat production due to friction though. I also don't think you can measure the deltaT across a restriction with any kind of accuracy at all: across a waterblock or a radiator you can do ok (and probably across a pump) but across something that isn't actively dumping or removing heat then I don't think you'll see a difference with any equipment we can afford. You'd need better than 0.01C res at any rate to pick up any change.

[pHaestus]

You can measure dT/dt, but then you have to account for all of the heat losses from the system due to heat transfer through hose, fittings, wb tops, etc etc. That's why steady state conditions are desireable in the first place. Initial rates would be best bet, but such modeling is notoriously unreliable because the number of points you choose as being "far from equilibrium and suitable for initial rate calcs" will affect the observed rate.

[bigben2k]

Yes, it would be a function of the flow rate. I'm assuming various pumps with the same loop, resulting in different flow rates.

No, measuring a temp increase across a restriction isn't anywhere near practical.

There's a formulae in this article that could come in handy here.


What I'm trying to get to (and for some reason, keep circling around) is that there's heat generated by the pump within the housing, and it's in two forms:
1-the recirculating water within the housing
2-the motor coil heat

It's late... I'll pick this up tmo

[Cathar]

I'm basing my values on a (possibly flawed) ratio of the water rise above ambient with the pump turned off (0.0C funnily enough eh?), the pump turned on but the computer off, and finally with the computer on an running at what I believe to be about a real 100W CPU heat load (Barton AthlonXP @ 2.6GHz/1.95v running BurnK7).

Not saying that's the most accurate thing. Like I said, just a guesstimate. Mind you, the Iwaki specs say that the motor is actually rated at 45W of power at the 70W power draw, and as myv65 will tell you, pretty much all of a motor's energy will make it into the water in one way or another, whether that be motor heat or friction.

[myv65]

Basic thermodynamics would tell you to draw a "box" around the system and do an energy balance. We're not nuclear here, so there can be no creation nor destruction of energy. You put juice in the "box" (power to the pump with an electrical cord)? Then every bit of that juice must somehow "leave the box". Some exits off the motor housing (which may or may not be in the water). Some initially shows itself in the form of pressure rise * flow, but it must still leave the "box".

If it isn't electricity, there are only so many other ways that energy may enter and leave the box. The only other one of consequence is heat transfer.

Anyway, energy in must equal energy out at steady-state.

The two items related to this that seem to cause the most confusion are the actual power draw of a pump and how this relates to flow. Pumps draw more power at higher flow, but the power vs flow decreases with increasing flow. This means even though you're putting more energy into the water at higher flow, the temperature rise over the pump will decrease with increasing flow.

Lord have mercy but this topic seems to have been talked to death.

[jaydee]

Here is one for you. A little experiment I did this weekend because I couldn't stand the noise of my HydroThruster 500 wide open any longer.

I got the adjustable switch on mine which lets you turn the motor up and down similar to a fan controller. I have been running the pump full speed for a week now with my new pin block I made and this weekend I took some rought temps. I was getting 41C for the CPU pretty steady for the week. Being I did some gaming this weekend and my gaming rig is in the same room I couldn't take the noise any longer and turned the pump all the way down.

I was expecting the computer to lock up after a few hours and thought I would have to un-overclock the system. Well to my suprise the temps DROPPED 5C. :shrug: Don't know what to make of it. All the other variables stayed the same....

[pHaestus]

By any chance is the fan that cools the Thruster pointed at the motherboard? I know the air coming off that thing gets pretty warm sometimes...

[jaydee]

Originally posted by pHaestus
By any chance is the fan that cools the Thruster pointed at the motherboard? I know the air coming off that thing gets pretty warm sometimes...

No, the mobo is sealed off from air currents, but the radiator isn't. I got the system in a box to keep a more stable ambient temp. The extra heat from the pump was going through the rad... I don't have a working water temp probe so I didn't catch it. I bet that is what it is.

But even still the flow rate is WAY slower than before. You can see the rotation of the impellor and almost count the revs. I am just suprized the system is staying so cool at such low flow rates and such a high overclock. XP1700+(A core)@1800mhz 1.85V. I never seen such low temps on that system. Maybe that block is better than I thought.

Man these little discoveries sure throw a wrench in the works... I am going to have to re-evaluate the whole thing. I got a VIA 2600 the other day. I am going to swap it in as it doesn't seem to have a fan on it like the HydroThruster.