My parallel set up - how many pumps and where to place 'em?

Help!

I'm at a turning point in my proposed parallel set up and have been given so many pointers as to how many pumps I'll need and where they should be in the circuit. I dunno ! So I'm putting it to the vote!

How many pumps and where to place them?

I've drawn up 5 options and illustrated them below:











Thank you for voting!

tbh... i know you have been through a lot of thought about this on the forums and sharing your work... but to me, it still seems flawed, unless the reason you are doing this is 'because you can'! and in that case, i'll say 'good on ya!'

You got it, 'because I can'! Where's the harm in a little experimenting Thanks for voting!

Just a note...your configuration 3 has the pumps in series not parallel...they are just separated by the rad and res.

Well spotted Erasmus, I knew I could trust you to see something I didn't!

Great ideas but what case will this be installed in?

I like configuration 3 for two pumps. Configuration 1 if you decide to just use one.

If I have to chose from the options your giving I would agree with Mark's choices - 3 or 1.

IMO, in a 2 pump system, it doesn't make sense to have 2 pumps placed one after another in the loop like in config 4 & 5. My instinct would be to split up the pumps like in config 3. My reasoning is that there is a pressure drop after coolant goes through a component in the loop, so, you want to use the second pump to boost the pressure back up before you send the coolant to the next component. I have no idea if that is really how it works, but, I'm interested to know your reasoning behind that setup.

I've read through all the past threads on your project and remember that one person suggested you get rid of the hot manifold and just put a reservior there. That made a lot of sense to me because it would simplify the loop, and I think it would remove the need for all those ball valves and make the system less restrictive.

Apologies if I'm talking out of my rear. My comments are all intuitive and not based on any knowledge.

Anyway, I think your project is a really cool experiment. Good luck with it.

Have you considered inserting pump 1 between the inlet manifold and the cpu block inlet. I wonder if this would act to compensate flow to the cpu block?

R1ckCa1n - it's a custom build, see the picture in my other thread about half way down

Mark305TBI - looks like you're not alone, it's my favourite of the 2 pump configurations.

CoolNQuiet - I agree with the two pumps next to each other being a bit odd. But it was a valid suggestion from someone. It looks even more strange in the diagram but this would provide a high amount of flow, arguably.
From what I have learned so far on this forum and other sites, there are 3 things that are important in a cooling loop:
- provide a good amount of flow at the input of the blocks
- 'push' as much flow as possible through the radiator (goes hand in hand with the above point)
- minimise the amount of backpressure (resistance) at or after the blocks
- 'pull' as much flow as possible from the reservoir (if fitted) (again, goes hand in hand with the point before)
- the ongoing argument about low vs high flow

Config 3 looks to satisfy all the above (low vs high unknown at this point). Both pumps may not operate at exaclty the same flow rate (manufacturing errors etc) so may vary. My intuition says that when pump 1 has a lower flow than pump 2, pump 2 will take up the slack by reducing the backpressure, effectively reducing the resistance pump 1 is combating. Pump 2 will be fighting the backpressure of the rad and then the water goes into the res where the chain starts again. However, if pump 2 is running lower than 1, then this may actually increase the backpressure at the front of pump 1, but hey, it's running faster anyway so it may negate this effect. In saying all that, the difference between each pumps flow should be negligible at the same setting (D5 pumps). Am I talking sense?

calvin_q - imo placing the pump there will have a detrimental effect on other blocks and the cpu block. The pumps is sucking on the inlet side right? So it could suck water from inlets of the other blocks rather than the res and create a little circuit all by it's own that never gets to the radiator.

i'd say you're well off with the "parallel" pumps, which as pointed out aren't really in parallel. seems that feeding the output of one directly to the input of another wouldn't help it as much as getting a boost halfway through would. then again, i'm not a fluids engineering expert

None of those pump layouts are parallel, merely relocating a pump in series farther down the line.

In the case of your parallel manifold cooling circuit, restriction most likely drops to a fraction of a normal loop, meaning adding a second pump in true parallel may help alot. Though to prevent bottlenecks, you may want to have a separate rad for each pump and multiple inlets/outlets on the manifolds for each pump to directly input and draw from the back end. A single reservoir could suffice if it can handle that many inlets and outlets.

Double Jesus said:

i'd say you're well off with the "parallel" pumps, which as pointed out aren't really in parallel. seems that feeding the output of one directly to the input of another wouldn't help it as much as getting a boost halfway through would. then again, i'm not a fluids engineering expert

Click to expand...

Basically you can think of a closed loop watercooling system as being very similar to an electrical circuit. Wires are tubing, blocks are resistors, and pumps are batteries.

If you connect two pumps in series you increase the maximum head pressure, but not the maximum flow. Similarly if you connect two batteries in series you increase the voltage, but not the mAh (how long it can run). If you connect the pumps in parallel you increase the maximum flow, but the head pressure is the same as one pump. If you connect two batteries in parallel the voltage remains the same but the mAh increases.

Now the question is...do you want higher free flow or higher head pressure? That I cannot give you a good answer to.

I say screw the manifold and engineer a pump to act like a manifold. That way you have equal pressure going to all your blocks. Otherwise you could possibly have flow issues to one block over another because of restrictions.

So set a multi outlet pump supplying an equal pressured water flow to all of your water blocks with a receiving manifold tapering everything back to one line for the radiator.

BioVader said:

I say screw the manifold and engineer a pump to act like a manifold. That way you have equal pressure going to all your blocks. Otherwise you could possibly have flow issues to one block over another because of restrictions.

So set a multi outlet pump supplying an equal pressured water flow to all of your water blocks with a receiving manifold tapering everything back to one line for the radiator.

Click to expand...

Mysterae has already thought about unequal flow...which is the reason for the ball valves. The ball valves are a good idea because it allows him to precisely control the flow to each of the parts. He can monitor the temperatures and adjust/tweak the flow to each of the parts until he gets the balance he desires.

oh....

Originally posted by Erasmus354
If you connect two pumps in series you increase the maximum head pressure, but not the maximum flow.

Click to expand...

Theoretically, does that imply if the 2 pumps are in series it shouldn't matter where you place them in the loop?

CoolNQuiet said:

Theoretically, does that imply if the 2 pumps are in series it shouldn't matter where you place them in the loop?

Click to expand...

yes, however certain configurations may make it easier to fill/bleed the loop and/or route the tubing.

I like the 3rd option best, but I believe there's another possibility you should consider. You got basicly 1 circuit that splits 6 ways and then comes together again. So you got the same fluid running. (pretty hot)
Why not use 2 independent circuits each one with 1 pump and 1 reservoir?
Fluid comes out of the rad into both the 2 res. Place each pump after each res and one pump feeds GPU1+GPU2+NB and the 2nd pump feeds CPU+HD1+HD2 (or other combo) and then each curcuit conects just before the rad and so forth.

Instead of using 1 or 2 large pumps you can use 2 weaker ones (so less noise) and the fluid will run cooler. I hoppe you choose a pretty big rad for this setup. How are going to cool the HD's?

Some good points and ideas guys, keep 'em coming.

Having 2 parallel loops is a good idea, but imo I'd need another rad to keep the heat from both circiuts separate to be of use - the bottleneck of the two circuits meeting at one rad would be just the same as the config 3. I can fit another pump no problem, but another res and/or rad isn't a possibility.

The idea of combining the hot manifold into a res is still in the back of my mind. So I'm designing the res to have 6 ports (1/4") that will be blocked off initialy. Then I can change the circuit real easily to test the benefits of the both 2 manifolds vs 1 manifold and combined reservoir (manivoir? )

Shame I can't edit the poll to add another option!

Forgot to mention the cooling hardware you asked:

pump(s) - D5 or MCP655 (same thing)
rad - ThermoChill PA120.2
hd coolers - Cooler Master AquaTurles

If you decide to turn the hot manifold into a resevoir I think you should do something like the new DD Fillport Resevoir. Basically you have a cylindrical (or hex to fit the manifold) resevoir that would be attached to the 1/2" hole in the middle. That way you wouldn't have to scrap one of your beautiful manifolds.

I can't visualise what you mean after looking at the DD fillport res..

If I end up having a spare manifold I'll punt it on ebay as a candle holder

well actually after thinking about the design of your manifolds again, screw the fill port res...you could just replace the top of one of your manifolds with something much deeper, and perhaps an additional outlet to act as a fillport. This would effectively turn your manifold into a resevoir.

Although I do not run a parallel setup, I would have the pump(s) before the radiator. My reasoning is that you want the maximum heat difference between the radiator and the air moving through it to maximize heat dissipation. Every pump is going to inject heat into the water it pumps, even if it is rather marginal, every bit helps.

Basheron said:

Although I do not run a parallel setup, I would have the pump(s) before the radiator. My reasoning is that you want the maximum heat difference between the radiator and the air moving through it to maximize heat dissipation. Every pump is going to inject heat into the water it pumps, even if it is rather marginal, every bit helps.

Click to expand...

But electric pumps are better at pushing fluids than pulling fluids. That's why you want the reservior to be plumbed straight to the pump's intake.

Basheron said:

Although I do not run a parallel setup, I would have the pump(s) before the radiator. My reasoning is that you want the maximum heat difference between the radiator and the air moving through it to maximize heat dissipation. Every pump is going to inject heat into the water it pumps, even if it is rather marginal, every bit helps.

Click to expand...

The difference the pumps would make would not be noticeable at all. Two D5's would raise the coolant temperature by *maybe* 0.1C which is hardly enough to make a difference in heat dissipation.