Wow... check out this guy's rig..
 

quite the setup....

http://www.overclockers.com/tips915/

this part really got my attention. so higher velocity at block=good, lower velocity at radiator=good
BUT
this way all the heat from the pump goes straight into the block too! isn't it better that the coolest water, albeit, not as fast-flowing, goes straight from the radiator to the block?

WOW. can somebody confirm this? anybody had any real-world experience with the order? (and reverse order?)

Go back and read the article... he said that when he reversed the order, the temps went up 3C ...
The water isn't in the pump long enough to absorb much heat. Rads work best w/ high deltas (I think that means high temperature differences). If ambient is 23C, it would take longer for a rad to cool 24C water back to ambient than it would take it to cool 40C water to say, 35C. (those numbers are made up) The amount of heat added to the water in a given period of time would be minimal, so I'm willing to bet that water coming from the pump would be about as cool as water coming from the rads. The main difference then is flow rate and pressure.
Many tests have shown that WB prefer high flow rates, while rads prefer slower rates.

but aren't flow rates constant in an entire rig. the only thing changing is velocity and hence pressure?

I'm still working on learning the correct vocab...

But, Here's an analogy. Imagine that you've got a 100' length of garden hose connected to your outdoor spigot (sp?). The valve is wide open. The water coming out of that hose isn't going to have a whole lot of pressure, due to the resistance of the hose. It follows that the velocity of the water coming out of the hose isn't going to be very great, either. If the hose were removed from the spigot, or a shorter hose was used, there would be a much higher pressure & velocity, therefore a higher flow rate.
Now I know that there cannot be more water going into that hose than what is coming out of it. (at least after the water's been running for a bit, and the pressure differences w/in the hose have been established) Speaking of which...
Back to our 100' hose. If it sprung a leak close to the spigot, the leak would spray out w/ great force. If the leak happened at the end of the hose, you'd be lucky to get a trickle. Hence, a pressure difference.
You know... I really don't know how to quantify what I am thinking, but it makes sense to me.
Here's a thought. The volume of water going into a circuit/pump has to be the same as the volume coming out. Water moves slower as it passes through large spaces and faster when it travels through tight spaces. Perhaps this exlains why some pumps' intakes are bigger than their exhausts. What if you had a system the had a smallish hose from the pump to the WB for max velocity, and medium size hose from the WB to the rads, and a large hose from the rads back to the pump. The water would continuously move from an area of higher resistance/velicity to an area of lower resistance/velocity (ideal for WC setups). I suspect that this setup would yield higher flow rates than a system that was setup in reverse (slow, low pressure water moving into areas of high resistance). The second system would be less efficient than the first.
Does this help? I'd like the pro's input too.

that is pretty sexy, dual huge rads!!

--Matt

chalk up another good review for the Gemini block.. way to go peter.

okok..i THINK I get what you're saying. thanks :dome:

ok, is this the new consensus then? that best order is pump->block->radiator->pump is best?

No :)

I think before you get too caught up in that write up you need to find out the answer to a simple question :)

"How was the flow reversal done?"

I suspect he simply switched connections at the pump based on the difference he saw....

What are you suggesting? and are my conclusions incorrect?


"Here's a thought. The volume of water going into a circuit/pump has to be the same as the volume coming out. Water moves slower as it passes through large spaces and faster when it travels through tight spaces. Perhaps this exlains why some pumps' intakes are bigger than their exhausts. What if you had a system the had a smallish hose from the pump to the WB for max velocity, and medium size hose from the WB to the rads, and a large hose from the rads back to the pump. The water would continuously move from an area of higher resistance/velicity to an area of lower resistance/velocity (ideal for WC setups). I suspect that this setup would yield higher flow rates than a system that was setup in reverse (slow, low pressure water moving into areas of high resistance). The second system would be less efficient than the first.
Does this help? I'd like the pro's input too. "

here's a thought, when he conducted the second test the ambient had dropped by 3C

...and? he has a temp delta listed there as well, so the decrease in ambient temp shouldnt really matter.

and just to clarify, what second test are you talking about? im assuming you mean his watercooled temps (compared to his previous air-cooled temps), but i just want to be sure...

I mean his ambient dropped 3C between pump-rad-cpu and rad-pump-cpu

I asked this question on a thread and Mr Evid said that he had measured this and came out with no real difference. With an inline pump the water is not that near the heat bits and does not linger too much. It should only make a real difference with a submurged pump. This does need to be tested by someone with the nohow.

please quote me where he said that; i didnt see it.
actually, he doesnt say that. all he said (regarding the flow pattern of his system), was that his temps (im assuming CPU temps @ full load) increased 3C when he "reversed the configuration" (which im assuming means he changed it to pump -> rad -> WB).

now that i look at it, he was somewhat vague since he doesnt directly state "my CPU temps rose 3C" or anything like that. but he also doesnt give any ambient temps while he tested the different patterns; he only gave ambient temps of his old (HSF cooled CPU) and current watercooled CPU.

ECU - I'm suggesting that he might have inadvertently altered more than just the order of his hookup if he really had that kind of delta. From looking at his hookup pictures, I see ways he could get that kind of delta if for example he simply swapped the connections at the pump. Bottom line - there is waaaaay to little information supplied about his setups and methods to be jumping to any conclusions :)


As far as your conclusions, let me just give you some basic info -


The total mean mass flow through any component in a serialized, closed, single phase fluid system is the same at any point in the system.


The total mass flow in a closed system is determined by the system's overall flow restriction and the pump. It doesn't matter what order the components are in, it is the sum of the individual component's losses added together without regard to their order. You then take that value and the pump's curve to see what the mass flow will be. (it actually requires a numerical solution involving iteration for turbulent flow regimes, but for discussion purposes we can ignore how the precise value is arrived at.)


The mean velocity at any point in a closed system is determined by the hydraulic diameter at that point and the mass flow rate.


End result - you want the lowest possible flow restriction in your tubing at ALL locations to get the maximum mass flow rate in the system - which will give you the highest velocity in your block - which will give you the highest heat coefficient - which will give you the coolest CPU. ( In other words - don't use progressively sized tubing in your system ;) )

Oh, and the only way you are going to affect the mass flow rate in the radiator section (given you are not modifying the radiator or lowering the flow restriction in some other area of the system) is by using multiple radiators in parallel.

I was being semi sarcstic, semi smart. just joking really. I still can't see how the temps can change that much, unless he had a couple of sharp turns or a kink, or something like that without realising it

In all of this, bruces, keep in mind that this is a user submitted article. Don't put the stamp of approval on it so easily like we do to Joe and the rest of the staff's work over there. Take it all w/ a grain of salt. He is mostly likely as accurate as an in socket thermal probe...
-Kev

:p maybe he has some air trapped somewhere in the system which got bleed out after he changed it? :shrug: