Tube Routing Question

[MiRai]

I have this idea that has been kicking around in my head for awhile, and I'd like to hear from those of you with some expertise in water cooling, to see what you all think.

The reason behind this layout is purely aesthetic, but do you see any potential issues with it? Is the, assumed, additional restriction from the GPU section going to negatively affect anything else?

Additional Information:

1) I may not be able to fit the top-most pump/reservoir since I'm unsure of clearance at this time. If I cannot, then I may need to split directly from the multi-port radiator at the end. Would this cause any issues?

2) The pumps/reservoirs after the split will have water fed into their reservoirs in a sort of "waterfall" way. I do not expect to keep the reservoirs filled to the very top so that I can visibly see that there is proper flow happening, but I also do not expect to have the water line too far down so that air bubbles are not introduced into the loop.

3) All radiators are multi-port.


Please let me know if I'm crazy, or if this will work, thanks.

 

[ReignQuake]

I'd like to know how many radiators you're planning on and what sizes they might be. Then I'd like to know what pumps and reservoirs you're thinking of and why you'd like three of them? I assume your plans are in the earlier stages and open to change? I don't consider myself to be an expert but I've watercooled for long enough in enough different computer cases to be interested in your project.

 

[MiRai]

I'd like to know how many radiators you're planning on and what sizes they might be.

There will be, at least, 1,800mm of radiator (480x3 + 360 (or another 480 if it fits--I don't know if it fits at this time)).

Then I'd like to know what pumps and reservoirs you're thinking of and why you'd like three of them?

I'll be using Alphacool D5 pumps, and my reasoning behind the third, top-most pump is, in my fairly uneducated water-cooling opinion, so that I can provide redundancy to each of the parallel routes in case one of those pumps fails. Without that third pump up top, then I assume that if one of the others fails, then that particular route/"branch" will have a reduced flow, and I am unsure of how that will, if at all, affect the water loop, as a whole.

However, the top-most pump/res combo will also allow for me to more easily fill the loop and bleed the system since the other reservoirs are more-or-less in the middle of the loop itself (the diagram doesn't do the best job of showing where everything should be).

I'm also under the impression that running pumps in parallel can help increase flow, and while I do have some fairly un-restrictive radiators (or so I'm told--Alphacool UT60), there is still a good bit of them to push water through.

I assume your plans are in the earlier stages and open to change? I don't consider myself to be an expert but I've watercooled for long enough in enough different computer cases to be interested in your project.

I expect that this thing will be built within the next 3-4 weeks, but my original thoughts were that I would just run two pumps in series and route it like any other water loop (pump/res > pump/res > CPU > GPU, radiators). Then I got this idea to make things more complicated for the sake of aesthetics and color-scheme matching, which is not out of the ordinary for me and what I like to do with my hobbies.

 

[aigomorla]

yes...

water travels where there is the least amount of resistance.

gong by your diagram that would be bypassing the radiator completely...

edit:

oh wait i see your paralleling at the radiator... using a pump to drive each path on the parallel... ummm... i still wouldnt do it.
parallel's are very trickey to setup, and sometimes will not work out as intended.

 

[MiRai]

yes...

water travels where there is the least amount of resistance.

gong by your diagram that would be bypassing the radiator completely...

edit:

oh wait i see your paralleling at the radiator... using a pump to drive each path on the parallel... ummm... i still wouldnt do it.
parallel's are very trickey to setup, and sometimes will not work out as intended.

I've updated the image to show the expected flow, but why would they be any more tricky than a standard setup? Is there some sort of special "balance" that needs to happen between each route which is more difficult to attain?

 

[MongGrel]

Redacted, I do not have enough personal experience with water involving PC's to offer advice.

Even though I've been reading builds for decades.

 

[MiRai]

I could go another route with it, but this drawing may be difficult to read because I'm trying to depict things that are going to be in 3D space, in 2D space.

Again, while it looks more complicated, I think it would be easier for me to pull off routing-wise due to me having to use less tubing.

If I had the case here in front of me, I could just do a mock-up with pipe cleaners or something so that it would be easier to understand, but unfortunately, I do not.

EDIT:

Here is maybe a better visualization of both choices:

 

[aigomorla]

ok because of how again water works follows the least obstructive path.

going by that notion, we can all agree depending on the cpu block one has, it is going to be a lot less obstrusive then 3 gpu blocks, so naturally water flow will be increased in the cpu loop, which i feel is kinda moot because you would want the greater flow in the gpu loop as yo have more heat which needs to be moved.

this wont matter on the fact you have a pump each parallel, if anything it could lead to cavitation problems at the first pump in the t section because your still under the constraints of a G1/4 outlet.

In honesty you would be better doing it in a 3 parallel loop style.
Have a pump res cycle itself from the radiators in a very short loop so the turn over rate is very high. (call this the radiator loop)

Then have the cpu loop tap from the res directly, and not from the pump, to feed the cpu back to the res loop on its own pump. (cpu loop)

Then another loop tapping from the res in the rad loop feeding the gpu's and then back to the res on its own pump again. (gpu loop)

This may give marginal performance gains because the turnover rate in the radiator will be much greater then if you had everything in a series.

 

[MiRai]

In honesty you would be better doing it in a 3 parallel loop style.
Have a pump res cycle itself from the radiators in a very short loop so the turn over rate is very high. (call this the radiator loop)

Then have the cpu loop tap from the res directly, and not from the pump, to feed the cpu back to the res loop on its own pump. (cpu loop)

Then another loop tapping from the res in the rad loop feeding the gpu's and then back to the res on its own pump again. (gpu loop)

This may give marginal performance gains because the turnover rate in the radiator will be much greater then if you had everything in a series.

I'm going to have to request a drawing via MSPaint like I've done above in order to understand this three loop parallel setup.

 

[aigomorla]

bleh...

loop1:
Rad -> Rad -> Rad -> Res(A) - > Pump(1) -> back to rads

Loop2: -----------------Res(A) -> Pump(2) - CPU -> Res(A)

Loop3: -----------------Res(A) -> Pump(3) - Gpu -> Gpu -> Gpu -> Res (A)

RES(A) is shared by all 3 loops.

You would need 6 ports on Res(A) a inlet and outlet for each loop.
It would also be in your favor to use a very BIG res, a DYO from koolance using the 80mm tubes. However a BIG res also does look very eye candy.

Since the cooling loop with the rads is the shortest loop, ideally the coolant level inside your res will not be that great because the turn over will be a lot faster since its a shorter loop from warm water to radiators. However you will need volume which wont be a problem for you seeing as you have 4 radiators.

This is just a theory i had but never implemented so your millage may vary.