pump location in system flow

[shiltz]

I was wondering how important pump location is in the system, originaly I was gona run res > pump > rad > CPU1 > CPU 2 > video > res, thinking that with the pump before the rad any heat the pump may put into the water (gona be using a hydor 30) will get taken out in the rad before it goes to the CPU, however i've been playing around with placement of everything in the case and found a much better placement for both the amount of space the whole water cooling system will take up and also reduce the amount of bends in the tube making it have a slightly better flow however it would result in the flow going res > rad > pump > CPU1 > CPU 2 > video > res, would the added heat from the pump make a different or would the slightly higher flow from less bends in the tubes more or less make up for that?

[airspirit]

Try this:

res->pump->split to two CPUS blocks and rejoin->rad

Keep it simple. Try to keep your entire system other than the CPU blocks in 5/8" and try to run a higher powered pump (no cheapo Via Aqua or Rio).

[shiltz]

well the stuff I have and the stuff i'm ordering is all 1/2" so I can't use 5/8" and split down to 1/2" for the CPU's, could split down to 3/8" for the CPU's but i'd rather not go that small, one of the CPU's run's significaly cooler (5-6C less) than the other too for some reason so I can run thru the hotter one first then the cooler one, and eventualy i'm going back to a single CPU when the AMD 64 comes out so staying inline with all 1/2" I can just take one water block out and be go to go for it.

mainly i'm trying to decide between those to for placement wise, trying to minimize the runs having to go all over the place with lots of bends, here's what I was considering at first so you can get a visual picture of what it looks like, this is the res > pump > rad > CPU1 > CPU 2 > video > res setup, biggest problem with this is I will totaly lose 3 5 1/4" bays and will have to replace the current casters with taller ones to put an exaust duct under the rad to keep the hot exaust to come right back out the front and get sucked back thru the rad again.

[shiltz]

here's the second setup, this one is the res > rad > pump > CPU1 > CPU 2 > video > res, this one is a lot more space friendly, I only partialy lose 3 5 1/4" bays, but they are perfect still for my fan controller and temp guage and this one will suck cool air in from under the case and exaust out the rear, only thing I may have to do it attach a panel on the back of the system to make it less likely to suck th warm air from back there in again, not as much of an issue though since the exaust is above the intake and warm air rises.

[airspirit]

I would highly recommend going from your pump directly to your water blocks and having the rad after them but before the res. It seems to be the best configuration to maximise your flow, and since your blocks is what is going to cause your spaghetti mess, it allows you to mount your rad wherever is convenient.

[shiltz]

thanks, redid it some, now it's res > pump > CPU1 > CPU 2 > video > rad > res, was able to keep the locations of everything without really adding a bunch of bends and extra runs to the tubing as well.

[redleader]

You have ~ 10w going into the coolant from the pump. Furthermore a lot of that heat isn't generated at the pump. Thats a tiny fraction of a degree C added to your coolant.

IOW it doesn't make any difference.

[KlevrPhelo]

My current setup is:

Pump -> Rad -> CPU1 -> CPU2 -> Res -> Pump

I figured that's the best for cooling, since the rad is supposed to cool down the water, and you want the water going to your CPU's to be as cold as possible. Seeing as the pump adds some (but not much) heat to the cooling solution you'd want it "cooled down" before it touches your CPU's.

You might want to consider splitting the flow from the Rad to each CPU, then joining the flow again before it hits your res. I myself haven't tried it, but I've seen it done and it's even been suggested by some people too. I think it might cause some problems with flow.. but if done correctly it might not. Then again, I could be wrong.

Just a thought...

[8-Ball]

One thing I always think about here, is whether blocks perform differently depending on their position in the loop. That is C/W, regardless of temerature.

Consider this, on the pump outlet, the pressure is positive and the maximum in the system, whereas, at the outlet, it is negative, ie it is being "sucked" into the pump rather than "pushed" from behind. The degree to which the water is being sucked/pushed will vary throughout the loop, though I guess you have to visualise the loop as a series of restrictions rather than the actual real world "length".

Now consider a White water. Would the flow in the block be different if the water were sucked from the outlet rather than pushed into the inlet?

In other words, would water being pushed or pulled through the block affect its performance?

8-ball

[mad mikee]

I've done just enough tinkering to be dangerous now, and have gotten 2 things from experience on earlier systems.

1. Splitting up flow into the various parts so that they all get coolest water did make things run just a little better. (2 y -connectors from wherever to the CPUs wouldn't appear to clutter your setup from what I see here).

2. You want the max pressure into your CPUs (and if you are splitting this of course is even more true) and while the flow may even out around the loop, there should be a point of max pressure / flow right out of the pump head. Also you take a flow/pressure hit in every block (think of smashing into brick wall and then reversing direction). 2 of those in a row will slow down the flow more than 2 in parallel?:shrug:

Lack of sharp turns is very good.

3. Check out the Article on Maximizing flow rates as it would also apply here.

[Alchemy]

Originally posted by mad mikee
2. You want the max pressure into your CPUs (and if you are splitting this of course is even more true)

Why would it matter? Do you think you'd get better performance if you pressurized the cooling loop to a few psi before operating the pump? Do you think your waterblock knows or cares what its inlet pressure is compared to the atmosphere?

and while the flow may even out around the loop, there should be a point of max pressure / flow right out of the pump head.

This is true, but I still don't see why you think higher static pressure will give you better heat transfer. It's higher flow rate that gives you better heat transfer, and this is determined by pressure drops across various components. Absolute pressure in any single point in the cooling system is irrelevant.

Also you take a flow/pressure hit in every block (think of smashing into brick wall and then reversing direction). 2 of those in a row will slow down the flow more than 2 in parallel?:shrug:

It is the nature of flow resistances that two flow resistances in parallel will cause an overall pressure drop *less* than two flow resistances in series. This means that putting them in parallel will make the flow rate through each block more than half the flowrate that would go through the blocks if they were in series.

Lack of sharp turns is very good.

Absolutely.

Alchemy

[bigben2k]

I think Mikee meant that out of the total pressure drop, most of it should be at the block, and I agree.

[8-Ball]

I kind of answered this one for myself while chatting to a friend of mine who has just be studyin fluid dynamics for a module he is taking.

Water will flow from high pressure to low pressure. The total pressure drop between the outle and the inlet of the pump should not vary with arrangement of the components, provided the same components and lengths of tubing are used.

The pressure difference between the inlet and the outlet of the waterblock should be constant regardless of it's position in the loop. The flow through the block is determined by the pressure difference between the in and out, regardless of the actual values of the pressure. So while the pressure may be highest if the inlet of the water block were right next to the outlet of the pump, the pressure difference would be the same as if it were at the other end of the loop. The water flows down the pressure gradient, regardless of the actual values of the pressure, this pressure gradient will not vary regardless of the pumps location.

8-ball