Anyone run (or tried to run) blocks in parallel?

[JonJ678]

Note, not the same idea as parallel loops

It's not something I see done very often. It makes the tubing more complicated, and care must me taken or you end up starving parts of the loop of water since the current is stronger through the path of least resistance. However it decreases load on the pump, and not everything needs the full flow rate over it. So it is possible to gain higher flow rate through radiators and cpu block at the expense of flow rate elsewhere.

Even if perfectly balanced, you are splitting the flow of coolant which (probably) leads to less pressure in each arm and worse cooling. Flowrate decreases as well, which probably means putting radiators in parallel is a bad idea.

Sli graphics card blocks are almost designed for this. With inlet top left and outlet bottom right, connecting each card to the next with two tubes instead of one gets you a series of parallel sections of exactly equal resistance. I'm sure this will work well, and have seen a couple of photos of people doing just this.

Less simple would be splitting the loop so one arm goes through northbridge and the other through mosfets, which I think can be achieved but may need either care or testing.

So, introduction over, does anyone do this? I've asked thermochill to machine some Y pieces, but it's possible the lab at uni can do this as well.

 

[shadowscotland]

I've seen ram coolers run in parallel (due to it's small ID) but no 'before/after' test results
And in theory a typhoon 3 runs parallel loops of a kind (shairing pump and res)

In general I don't see any advantage over running a loop(s) in series (as the temp is constant throughout) apart for the two examples above.

 

[Audigex]

I don't see the advantage unless the blocks are more efficient at a lower rate. The pump needs to work as hard and there are risks of starving components.

You'll need just as much push, but you're leaving more to chance.

If you're going to do it, I'd suggest dropping the tubing size down at the same time, so you're keeping the pressure up on each side (hopefully) meaning that one side doesn't cause back-pressure in the other.

 

[JonJ678]

The theory is similar to that. It isn't that some blocks arwe more efficient at lower rate, just that some don't care so much. Cpu blocks tend to perform significantly better at higher flow rates, whereas mosfet blocks just wont care. So you can decrease the flow through the mosfets in order to increase the flow through the cpu block.

The pump needs to push less hard, parallel loops offer lower resistance. If it didn't decrease load on the pump I'd agree there's no point.

Now, if the split loops are at smaller tubing diameter, you lose the benefit of decreasing load on the pump. I think your suggestion would yeild the same results as a simple loop but with more effort.

I'm fairly confident in this being a sound idea. Certainly for graphics cards. The biggest problem with it presently is finding the Y pieces and deciding how to split up the loops. I know this is considered an odd thing to do, and before/after test results are probably required if I do so. For the moment I will hold out hope that someone on here has tried this, and if not shall search elsewhere.

Cheers