What would [H] do?

[undertheradar]

Okay, WWHD? I am contemplating the flow in my water cooling system. These are the blocks I have...

CPU - Watercool Heatkiller 3.0

GPU - Watercool heatkiller GPU-X2 gtx285

NB - EK-NB ASUS HP

SB - Koolance CHC-36-D06

I have to run the SB in parallel with something since its only 1/4", so I figure it should be in parallel with the NB. These two I have to run in parallel, but I wonder if I should run the others in parallel or in series. If I put them in parallel, should I put some valves on the split lines so I can adjust the flow between the two main branches? I suppose a third option would be to run the CPU and GPU in parallel, but then put them back together before the NB/SB split.

Anyways, would it be better to keep the flow up rather than split it considering the 'high flow' nature of these blocks (and get the most nozzle pressure)? Or, would it be better to run them in series to make for the most flow through everything...???

What would you do?

 

[lilfiend]

i like the second option, you get max flow to your cpu and cpu then it splits where it has to but only to be ran through the radiator and res before it is pumped back up to full speed

 

[dealmaster]

Air cool the SB, there's no reason to watercool it. Water will go mostly in the path of least resistance so the SB may not even get much water flow.

 

[undertheradar]

I have been wondering that myself dealmaster. As of right now, if I watercool the CPU and NB, the SB is the hottest part on the board by ALOT. (Rampage 2 Gene). Its a small SB heatsink... but it should be better than the stock solution. That stock sink is very small. You are right about 'taking the path of least resistance' though... I might have to add some sort of additional restriction to the NB... I wont know until I set it up and try it though...

After looking around at more systems like mine (high flow blocks like the GTZ and Heatkiller with 1/2" piping) it seems like running them all in series (except where I cant) is going to be the way to go.

 

[Tahoe916]

By default, running in a series is almost always the best way to go. A lot of people that water cool don't really understand thermodynamics. There are no spots in a loop with an appreciable temperature difference. The water coming out of the rad is not any appreciable difference in the temperature than the water coming out of the CPU block. Same with flow, for our purposes.

Just set it up in a series in a way that uses the smallest amount of tubing and is easy for you to setup and maintain. Good luck! and most importantly, have fun

 

[undertheradar]

By default, running in a series is almost always the best way to go. A lot of people that water cool don't really understand thermodynamics. There are no spots in a loop with an appreciable temperature difference. The water coming out of the rad is not any appreciable difference in the temperature than the water coming out of the CPU block. Same with flow, for our purposes.

Well, I have a formal education in thermodynamics, and that is why I asked. Unlike most systems, mine actually does show an appreciable difference in temperature throughout the system... http://www.hardforum.com/showthread.php?t=1399592 The water spends over 1/2 a minute in the radiator alone. I also know enough about fluid dynamics to know that putting all those blocks in series MIGHT reduce the flow of the pump enough that I would be better off with running the blocks in parallel.

This is my 'hypothetical' comparison...

Running the blocks in series, could result in a lowering of my flow due to back pressure... restricting the pump from its usual 315gph down to say... 150lph. We will say 150gph is the 'break even' scenario.

Likewise, the 'break even' scenario could also be if I run them all in parallel, and due to the lower back pressure of such a setup, maybe I will end up with 150gph going through both branches of the loop.

What it really comes down to is if say, running the blocks in parallel allows me to keep more flow through the GPU and CPU blocks since even though I am splitting the flow, and I would end up with something more than 150gph going through both blocks. OR, lets say I run them in series and I get more flow through each block... thats the basis of my question.

 

[ekuest]

oh youre the massive passive guy. thought the name was familiar. havent seen any updates to your log in a while, figured you were finished. i would say that yes, if you have a block or blocks that arent meant for very high flow, then you could be slowing down the entire loop with one little chipset block or whatever. id say just spend a day messing with it. try this, try that, measuring cpu and gpu temps each time, and see what works the best. having the most low wont necessarily be the best solution, because maybe the slower flow through the rad is what is making the loop so cool in the first place. just trial and error it out. keep a spreadsheet open on your laptop, and go with whatever gave you the best results. might take some time, but at least you will get real answers instead of just guessing at what may or may not happen.

 

[undertheradar]

Thats what I figured... just have to tinker with both. I dont think the chipset blocks will hinder much compared to usual since I will be running them in parallel no matter what (since that SB one is only 1/4").

Yep, Im the 'massive passive' guy. Its not finished, and it works very well so far... I just have to finish putting the rest of the blocks on. My i7 920 is sitting at a cool 31C right now (according to the Gene's LCD). Room temps seem to have more impact on the temps than the actual CPU activity or overclocking.