1 Big Loop VS 2 Loops?

[shunail]

So, here is what I have. I've Laing DDC DDCT-01s pump, EK-RES250 reservoir and Thermochill Pa 120.2 Dual 120MM and Thermochill Pa 120.3 Dual 120MM connecting:

Res => Pump =>CPU => 8800GTX => 8800GTX => Rad 120.3 => Rad 120.2 =>Res.

Now that I'm upgrading my Board and CPU, I thought EK-Bay Spin Reservoir would look cool in front bay. I've 2 of these Laing DDC DDCT pumps (petra tech's). What should I do to utilize this EK bay res and pump (not necessarily both, but one if is enough like it is now). Since EK-Bay Spin has three outlet, I was thinking atleast using 2 outlets straight like

Reservoir Out1 => CPU => Rad 120.2 =>Res

Reservoir Out2 => GPU => GPU = Rad 120.3 =>Res

or should keep it same as before and keep 2 other outlets blocked?

 

[shunail]

May be I wasn't clear... no replies???

well, what I was thinking is

Res => PUMP => CPU => Small Rad and back to Res.

Res => PUMP => GPU1 => GPU2 = Big Rad and back to Res

where back to Res with Y-Plug since Res has only 1 inlet. So both Radiators outlets to 1 inlet of Reservoir.

Since CPU to GPU 1, water may become hot and from GPU1 to GPU2, it become even more hot so the GPU2 (poor thing) is taking the most heat

What do you think?

Single (big) loop is simple and easy and I think 1 pump should be sufficient also... right?

 

[NickJames]

No replies?

Your first post Today, 12:42 PM
Your second post Today, 12:43 PM

 

[shunail]

Yeah, I don't know why. My post was last night at 8PM and the reply post was an hour ago. You're right!! That's weird but I seriously posted at the difference of more than 12 hours

 

[sok0]

If you had single cpu and graphics id do 1 loop with the triple. But since you have double video cards, id do the double rad on the cpu (i have same radiator and it used to cool cpu/gpu just fine) , and triple on the video cards/motherboard blocks.

 

[Tsumi]

I would go with what sok0 said. Daisy-chaining more than three heat sources on a single loop is just asking for higher temps. If you want to, I would consider water cooling your board along with your CPU, and run the GPU's on a separate loop.

 

[shunail]

Here's my current setup. Pardon my dust

I think if I run tubing CPU - 2 GPU and Board, it would be too congested...

 

[DraganUS]

I would do this.

Res => Pump =>Rad 120.2 => CPU => Rad 120.3 =>8800GTX => 8800GTX => Res.

 

[Rossi~]

I would do this.

Res => Pump =>Rad 120.2 => CPU => Rad 120.3 =>8800GTX => 8800GTX => Res.

I concur!

 

[shunail]

Whoa, Res => Pump => Rad 120.2 => CPU
so Res is in the Bay. Pump will be next to it pulling water up where Rad 120.2 is. Than go to CPU and back UP to Rad 120.3. Coming down to GPUs.

I'm using 1/2" ID tubes with 3/4"OD. It'll be 1 tube going UP + 1 down + 1UP = 3 Tubes hanging!!! Wow!! Is that practically feasible? Can 1 pump handle all that? Someone told me I should consider water blocks on motherboard as well!! I'm pulling my hairs as I'm writing

 

[Hades16x]

In my case, I have the same kinda thing
Res>Pump>PA120.3>CPU>Swiftech 220>gtx480>res

 

[shunail]

So here's what I've decided

XSPC Acrylic Dual 5.25in. Reservoir for Two Laing DDC with 2 Pumps conencted =RES+PUMPs
RES+pump 1 => CPU => MofSet => NB/SB => Rad 120.3 => Res
RES+Pump2 => GPU1 => GPU2 => Rad 120.2 => Res

2 seperate loops

 

[DraganUS]

Not sure about two pumps in same loop. If u are about to use 2 pumps I guess u can invest few more $$ and buy another res and go with dual loop set up.

 

[shunail]

That's exactly what I'm doing - 2dual loops. But someone said:

You cant create or destroy energy. Making a second loop does not make your heat magically go away.

If you have a CPU that dissipates 200Watts of heat and three GPUs that dissipate 100 watts each, regardless of how many loops you have, your total heat input is still 500Watts. And if you assume the same amount of rads, the amount of heat dissipated is also the same, no matter how you cut it. And dont forget, radiators are more efficient the hotter the water is.

By making a second loop, you are adding to the over all resistance by increasing the overall tube length. This is overcome by the fact that you now have two pumps, but it will be more beneficial to just add a second pump to your single loop, keeping you tube length (and therefore resistance) better optimized.

When setting up a loop, first install the pump and reservoir followed by the rads and blocks as best as you can. Then plan out the most direct path for your tubes, avoiding un-necessary bends, loops, etc... If you are worried about heated water from your CPU affecting your GPUs, its pretty easy to put a red between then (most cases/moboards can internally fit a 120mm rad on the 120mm exhaust). If you are able to stagger the rads just right, and have a radiator that dissipates 200+Watts between your CPU and GPU's (and a 300+watt rad after your GPUs), then you just isolated your GPUs and CPUs thermally (just like you had two loops; but better).

If you dont have an efficient set of radiators (ie, the amount of heat input is equal to or greater than the ability of the radiator to dissipate efficiently) making two loops you can optimize the performance of one of your components at the price of degrading the cooling on the other. However this is stupid as for the same price of a new pump, res, and tubing, you can probably just get a new radiator.

Dont fall for the 2 loop nonsense.

Add a second pump to the single loop and compare. Otherwise you are comparing apples to oranges.

Realize, depending on how you split up the loop, one component may be cooled better, but at the expense of the others. Its physically impossible for both components to be cooled better if you keep the other variables exactly the same. Its the laws of thermodynamics

 

[dizolit]

That's exactly what I'm doing - 2dual loops. But someone said:

If it's worth anything, this guy knows what he is talking about. Though, dual loops would not perform exactly the same as a single loop. Why? Analyze it doing a closed volume around each heat exchanger (cpu, gpu) and they will use different delta Ts due to different initial temperatures of the fluid. The poster also address this with the idea of a 1x120mm radiator between the CPU and GPU. The temperature profile on a graph for these sorts of heat exchangers are inversely exponential (initial steep drop into a smaller slope value). How low it goes depend son the initial Tc (temperature of cold fluid). The difference is negligible, relative to the real limiting factor: the cooling capacity (radiator).

As the poster you quoted suggested, the only way to make a noticeable difference in temperatures is to add more radiators, or introducing another pump. 1st law of thermodynamics states Q = m(h1-h2) [assume delta KE, PE, Work = 0] for an open system (when looking at a single heat exchanger) so either increase the mass flow rate (pump) or the cooling capacity (radiator).

disclaimer: I didn't exactly get an A in heat transfer, may not be completely correct.

tl;dr No big difference between single and dual loop. If you want lower temperatures, introduce more radiators or pumps.

 

[0xygenthief]

There are many advantages to having a single loop over dual loops. With the same equipment (i.e. dual pumps) you can have a level of redundancy and piece of mind should you be worried about a pump dying. Plus you may find yourself wasting some of the performance of a radiator in a dual loop setup.

If you have two separate loops both cooling the same type of heatsource then this isn't an issue. But most of the time CPUs get put into one loop and GPUs in another. So you have to weigh how much cooling performance to place in each loop. Though you could provide the "same" cooling performance in a single loop by placing the radiators in the right locations you would also be benefiting from the total/combined cooling power of your rads.

If your CPU loop needs 200w of heat dissipation capability and your rad can strip away 300w then that 100w is wasted when it could be put towards the other heat sources.

If you are limited by noise, money, or space then special consideration should be made to maximizing your rad use. Dual loops are for those who can safely say they don't need to worry about these things. That or they are particularly focused on performance in only one of the loops (CPU loop most likely).

 

[Olle P]

A single loop will always maximise the cooling efficiency, simply because all radiator area will be used to cool off all heat, no matter if the heat is currently produced in the CPU or GPUs. (The CPU and GPUs don't work at full load at the same time, other than at very rare occasions.)

My suggested loop would be:
Reservoir => Pump1 => CPU => (NB?) => GPU1 => GPU2 => Pump2 => Rad1 => Rad2 => Reservoir

- It's a pretty long loop creating a big pressure drop. Therefore it's nice to add the second pump roughly in the middle, pressure wise, to even out the difference and add more flow.
- Since the GPUs can handle temperatures much higher than the CPU, and GPU1 is normally working with less load than GPU1, the blocks should be connected in this order.
- It doesn't matter in which order the rads are mounted. (An alternative solution, that might be better depending on where the rads are positioned in the case, is to have the smaller radiator placed in the loop between pump1 and the CPU block.)

 

[BillParrish]

No big difference between single and dual loop.

There it is. (other than cost and complexity).

I too recommend a single loop and arrange the components to make the tubing runs as simple as possible and about the only rules that are critical are res just before pump and if possible mount the pump below the res. Not for cooling but just to help ensure the pump can never run dry. Unless you are using a jet impingement block with significant flow restriction, I do not even see the need for a second pump, keep it for a spare.

But hey, its your stuff and experimenting is more than 1/2 the fun.