Dual loops. Why?

[JonJ678]

Hey. I see a lot of references to dual or triple loops reducing temperatures, I'd like to know why this is so. To clarify, when is it better to run reservoir => pump => blocks => rad twice over rather than to run res => pump => all the blocks => all the radiators?

I can see two possible benefits.
Higher flow rate as there are two pumps rather than one. Arguments against are the large pressure heads available on DDC's, and the DDC tops which run two alongside each other if you need more pressure. Negative is slightly more tubing, needing more reservoirs etc.

Second, you can run one loop at a higher delta t than the other. So run the gfx loop at 50 degrees, the cpu at 40. This doesn't seem ideal, as tubing and pumps have temperature tolerances as well. I can see it being useful, but not enough to justify two loops.

What have I missed?

As an alternative to two loops, I'm considering the following. One reservoir, with four holes in it. Two pumps drawing water from this. One pumps through water blocks, perhaps one radiator, then empties back into the reservoir. The other pumps through a couple of radiators then empties directly into the reservoir.
Relies on turbulent mixing of hot and cool water in the reservoir, which should be fine. Problems with this idea as well of course, hence why I'm asking why people normally do it

Cheers

 

[Jokester]

As you say in the ideal world you want to at least common up your loops so that you make maximum use of your cooling performance from your rads. Though I would still go with the norm of each loop having a pump, blocks and radiator back to a common manifold of some sort. In the event say you're GPU loop isn't having to deal with a large heat load it's able to help out with the CPU loop if it's got a high heat load.

 

[Firestar_3x]

Or just spec each loop with the correct size rad in the first place, i've done it because i wanted two liquid colours running in my case

And i was fedup of building single loop systems so wanted a change.

 

[Bubo]

As an alternative to two loops, I'm considering the following. One reservoir, with four holes in it. Two pumps drawing water from this. One pumps through water blocks, perhaps one radiator, then empties back into the reservoir. The other pumps through a couple of radiators then empties directly into the reservoir.
Relies on turbulent mixing of hot and cool water in the reservoir, which should be fine. Problems with this idea as well of course, hence why I'm asking why people normally do it

This is a quite interesting idea. It would allow you to have two pumps in your system which would not need to have identical characteristics, which you would need if you put them in series in a single loop system, but also has the benefit of potentially get the best from all the rads, if one of the rads in an equivalent dual loop system was overspeced and the other under specced. Sure it it stabilise at some midpoint temperature compared to a true dual loop but it sounds like an idea worthy of experimentation, to see if you can get the same cooling performance overall with perhaps fewer/smaller rads compared to a traditional dual loop system.

 

[jolie77]

cheaper

 

[w3bbo]

As an alternative to two loops, I'm considering the following. One reservoir, with four holes in it. Two pumps drawing water from this. One pumps through water blocks, perhaps one radiator, then empties back into the reservoir. The other pumps through a couple of radiators then empties directly into the reservoir.
Relies on turbulent mixing of hot and cool water in the reservoir, which should be fine. Problems with this idea as well of course, hence why I'm asking why people normally do it

Nice idea in principle but I doubt it would make any difference. Perhaps the only gain would be by having less restriction on the 'working loop' due to it not having a rad. Hmmmmm....

 

[Bubo]

Nice idea in principle but I doubt it would make any difference. Perhaps the only gain would be by having less restriction on the 'working loop' due to it not having a rad. Hmmmmm....

Hey, you got a "Hmmmm" out of Webbo, that is pretty good going.

 

[Audigex]

Make sure that the working loop draws from the bottom of the rad, and the rad loop from higher up, so you get the coolest and warmest possible water.

Personally I think you'd just end up drawing warm water round the main loop, and the cooling loop just keeping it at a middling temperature.

 

[w3bbo]

Hey, you got a "Hmmmm" out of Webbo, that is pretty good going.

Hehe. Well he may be on to something there, especially with blocks such as the supreme and cuplex Di creating so much restriction and benefiting from high flow. It's certainly worthy of experiment if you have a restrictive block. The temps of both loops would equalise over time but the higher flow in the working loop might actually have a slight improvement in temps.

The simplest ideas are usually the best.

 

[Bubo]

I'm asking why people normally do it

The main reason why I maintain a dual loop system, with the cpu separate from the gpu, is primarily for the different equilbrium temperatures. If I decide to upgrade my graphics for example, then the potential higher heat output in the graphics loop will not interfere with the cpu loop. If it was all in one loop then the higher heat load would potentially lead to a higher equilibrium temperature, which could potentially force me to lower my stable cpu clock. Same may happen to the gpu clock if I upgrade my cpu. Basically it is easier to predict and control what happens. Your idea will almost certainly endup with an intermediate temperature anyway, its only debatable whether it would be the cpu or gpu would see a higher or lower equilibrium temp compared to having an entirely separate dual loop system, and whether this actually affects your max overclocks in practice anyway. Like I say, I do it as a traditional dual loop for the certainty of isolation, and if I do have any spare capacity in my separate rads I know it will cope with a fair number of upgrades over the years.

 

[ecat]

Several years ago I tried this out. The res was made from a waterproof electrical junction box, so perhaps lacked volume. The pumps probably were not matched, I think I only had two at the time. From memory I set up...

[FONT="Fixedsys"]
            -----------
Loop1 In ---|          |--- Loop2 Out -> pump2 -> rad -> Loop2 In
            |          |
Loop2 In ---|          |--- Loop1 Out -> pump1 -> cpu -> Loop1 In
            -----------
             
[/FONT]

As the in and out were on opposite sides of the res I deliberately crossed the flow within the res. to try and force the paths.

iirc, with the lid off the junction box things looked promising. Once the lid was on and the system was sealed the cpu temp started to rise and rise.

Why ?

My theory is pump2 was starving pump1, or maybe I just messed up. I never did repeat the experiment to find out. So if you try it out I'd be interested in seeing the results.

 

[MacGruber]

well here is my 2cents worth, i did my first watercool pc on monday and it all went very good i'm glad to say (^_^)

like i said i don't know much about watercooling loops and such got me thinking of the loop pictured.



what do you guys think? would this loop be any good? i know people say that having the pump from the res to the blocks wouldn't be good as it might warm the water up but not sure where else to effectively put it to have a good flow rate.

anyways ridecule away please ^^

 

[Bubo]

anyways ridecule away please ^^

What you've got there is a single loop with two pumps. Firstly, there is no need to have two reservoirs in any single loop water cooling system so one of the two you have is redundant. Secondly the with the two pumps in the loop they have to be identical and usually one straight after the other. The driving force behind people putting two pumps in a single loop is to overcome highly restrictive blocks and thus maintain a decent flowrate. Unless you have such blocks the two pump idea is largely overkill. Usually it is better to simply get a more powerfull single pump anyway.

 

[JTrickle]

If a pump breaks, can it impede water flow? Depending on how its failed, I suppose.

 

[ngfireball]

well here is my 2cents worth, i did my first watercool pc on monday and it all went very good i'm glad to say (^_^)

like i said i don't know much about watercooling loops and such got me thinking of the loop pictured.



what do you guys think? would this loop be any good? i know people say that having the pump from the res to the blocks wouldn't be good as it might warm the water up but not sure where else to effectively put it to have a good flow rate.

anyways ridecule away please ^^

think you messed up your drawing, if i aint mistaken it went:

Reservoir: Out, Pump, Radiator, Motherboard, Cpu, Reservoir: In

you dont have your GPU block yet.

the motherboard has a cooler for the Northbridge or something

 

[JonJ678]

Good responses, cheers. ecat's set up is pretty much exactly what I had in mind, need to have a think about why it didn't work.
The idea is that the mass flow into the pump in one loop is equal to the mass flow out into the reservoir, so once the system is bled each loop should have a constant volume of water inside. I fear the problem may have been that the two flows of water didn't mix well enough, and so the hot water stream and cold water stream stayed largely independent. Imaginative design of reservoir could solve that, I'll keep thinking though.

I think MacGruber's set up is flawed. You don't only need matched pumps, you also need matched resistance to get the same flow rate. As that isn't practical, one loop will flow faster than the other, one reservoir will attempt to empty itself and the other overfill itself. Even with the system full, you'll have significant back pressure from the overfilled reservoir, so one pump is starved of water and the other overworked.

Separate loops for thermal isolation makes sense. I suppose one of the benefits to what is suggested here, were it to function properly, is that if components get hotter you just add another radiator to the external box.

I'm a little concerned about flow rates. At the moment I have an 18W laing ddc w/ reservoir top driving two 120mm radiators, an ek supreme and an ek 8800gt full cover block. Temperatures are pretty good but nothing amazing, waiting for the cpu block to empty of air before I decide if anything has to change. It's an absolute mess of tubes already, taking all the radiators external would definitely make the case neater inside.
Starting to have quite a lot planned for this summer

 

[Bubo]

It's an absolute mess of tubes already, taking all the radiators external would definitely make the case neater inside.
Starting to have quite a lot planned for this summer

+1 for external rads. I have an old style Cheiftec Dragon full tower and I still mounted my rads externally. I am extremely fussy about having as minimal clutter inside the case a possible. So what if it looks like skyscraper with a helipad, it works. Its still beige for gods sake, shows you how much I care about asthetics

 

[Jokester]

If a pump breaks, can it impede water flow? Depending on how its failed, I suppose.

Most pumps will allow flow through them when they're not operating. Running 2 pumps in series is largely pointless, the second pump makes such little difference you're better off using it in a separate loop.