Dual Pump Setup

[billbartuska]

With this setup pumps could be sized (flow rate and pressure) for a rad AND for blocks. Perhaps high flow/low pressure for a 1/2" rad and low flow/high pressure for restrictive blocks. Kinda like having a source of the coolest water possible and being able to ram it through the blocks. Or you could look at it as being able to select (tune?) pumps to specific requirements. And, why stop at two pumps?

Or, just buy an Iwaki20RZ(T)?

[csimon]

mine is setup that way ever since this weekend as an experiment. It worked just as well with the single pump though. I think I need more radiator or something.

[bobkoure]

You think a radiator has enough resistance to merit its own pump?

[billbartuska]

Kinda. I think that all radiators require a specific flowrate to function at peak eficiency, just as there is a specific fan cfm that is most efficient.

[Long Haired Git]

Check out billA's testing, and you'll see flow rate has a small impact on radiator efficiency once the rate gets "good enough". Given the heat added of the extra pump, and pumping losses through the complexity, I doubt it would make much/any beneficial difference.

[billbartuska]

Yes, I remember those curves and took another look at them. You're right, water flow makes little differance, air flow alotta differance. And as to more heat from two pumps, I agree too. But, flow resistance would be less for each pump than for one single pump. So, I guess that this setup would only be worth investigating if you're already going to be using two pumps anyway for whatever reason.

[Dave]

Why two pumps?

Size and noise, two small MAGS will give you all the flow you need, and 1/2 to 1/4 the noise of a big pump.

I use a combo unit on CPU, and a second pump / rad for GPU block, using 80MM rads on both loops.

I find my 6800 loves overclocking

[bobkoure]

Originally Posted by Dave

Why two pumps?

Size and noise

So... you run a separate pump to move coolant through your radiator? I would have expected you to have simply set them up in series to maximize your waterblock's heat transfer capability...

[Dave]

^ I find this is best set up, I should also state both my flows "mix" in a comman reserve as per suggested dual par-setup on Web pages.

[JamesAvery22]

Soo by that diagram pump2 and all the blocks/system are in parrallel with the line that goes from the rad to pump1's input???

Why would you want to do that?

[Dave]

Will provide max flow, as you have the least restriction.

Trying to force 200 gph through the whole system in series, is less effective then splitting the system and pushing 130 gph through each loop independant. This allows you to use small pumps.

The reason for the "mix", is to provide max rad eff.

[billbartuska]

The thought is for pump 1 to be an Eheim 1048 and pump 2 to be an Iwaki MD20RZ(T). Hopefuly both delivering something slightly less than 2GPM.

My question: Why not run just the Iwaki? Does the reduced restriction( and higher flow) on the Iwaki (no rad) outweigh the heat added by the Eheim? I think not based on BillA's rad flow rate testing.

Comments on this "fan": http://www.ebmpapst.us/allpdfs/D2E133.PDF
Of course at 55dBA variable speed would be required.

[csimon]

Originally Posted by bobkoure

You think a radiator has enough resistance to merit its own pump?

my radiator doesn't benefit from 2 pumps in series or in parallel ...I say dave has the right idea.

[JamesAvery22]

Originally Posted by Dave

Will provide max flow, as you have the least restriction.

Trying to force 200 gph through the whole system in series, is less effective then splitting the system and pushing 130 gph through each loop independant. This allows you to use small pumps.

The reason for the "mix", is to provide max rad eff.

I guess I just can't picture how the flow, namely in proportions of each section, would look like

[Dave]

http://www.c-systems.ca/MultiCSP.pdf

See the multi path, multi rad set up flow diagram

[JamesAvery22]

Originally Posted by Dave

http://www.c-systems.ca/MultiCSP.pdf

See the multi path, multi rad set up flow diagram

the multi block/multi path/ multi rad setup makes perfect sense to me. Basically two loops just sharing the same resevoir.

But the OP's path is like a single block multi path single rad.

How is this:



Better than this:


???

With that little section that isnt removed in the first pic it looks like the rad is going to get water that it just passed. Will the rad just cool that water down by another 0.01c or something?

[LPorc]

Originally Posted by JamesAvery22

With that little section that isnt removed in the first pic it looks like the rad is going to get water that it just passed. Will the rad just cool that water down by another 0.01c or something?

I looks like his idea is that half the water coming out of the rad goes back into the rad, a variation on having a separate loop with higher flow for the rad. The thinking seems to be that twice as much flow through the rad as the loop increases the effectiveness of the rad.

Due to delta T it hits diminishing returns quickly. The upper limit of improvement as if the ratio of flows (say 1.5 to 1 through the rad) were sent through that many identical radiators in series. Figure in the added pump heat and complicating the loop, and it starts looking like a very questionable improvement. That being said, the only real test is measurement. I'd guess that the only real way it would show improvement is if the rad was insufficient to begin with.

[billbartuska]

No, I don't think there would be multi pass water through the rad. Well, some of the water moecules may multi pass, but the heat wouldn't. The thought was to use a low pressure pump on a low restriction rad, and a high pressure pump on high restriction (high efficiency) blocks. As for your second picture, I don't like running two pumps of diferent flows and pressures in series. Wouldn't they cancel each other out, lower pressure for the high pressure pump and lower flow for the low pressure pump?

[Dave]

Hmm, you got me on this one, going to think about it awhile, maybe test.

[Joe Camel]

well, for what its worth, HERE is (was) my multi pump / loop setup.

although i never *really* tested it, i can say that the temps i got were @ lest as good as anyone else's "normal" loop set-ups and i had 3 pumps (2 were over volted)

GOOD LUCK!!

[bobo5195]

your first idea wont work (wont work as well). May even decrease perfromance as the pressure drop available accross the system is less so flow rate will decrease. Also water might just end up being circulated through your rad instead of being pumped around the system.

While water flow rate will not effect rad performance alot it will effect waterblock performance. Running two pumps in seris will double the maximum head but maximum flow rate will remain the same (as an example this is rougly equivlant to the step from a hydor l20 to a mcp600 here http://www.procooling.com/articles/h...haestus__1.php) . Two pumps in parallel will double the flow rate but max head will remain the same. So if your running a restrictive system seris is better (i think this would cover most water cooling loops) and via versa for a high flow system.

It doesnt matter where you put the pumps in the system the same amount of water will flow through the system as water is incompressible and mass flow rate is conserved. The one caveat to this is that pumps add heat to the water so it maybe better to add them to "Hot" parts of the loop ie not before the cpu block.

edit: above stuff is an ideal appoxiamation there are loads of little problems which means that some of it might not be true for a real pump loop but the results will be very close.

An easy way (depending on how you think) is that hydraulic systems are like electrical ones (if you squint and ignore some problems). Voltage is pressure, flow rate is amps. Resistance is a restriction causing a pressure drop (water block = lightbulb?) and pumps can be thought of like batteries.

I think overvolting a pump is probably a bad idea.

[billbartuska]

OK. One Rad, One Pump, one Fan. Getting the parts together.
Now only thing left is to put barbs on a res I'm making..