Watercooled System Theory

[Emo_Boy]

Okay, we all know that the slower water is moved through the radiator the more heat the radiator can remove. Also, the faster the water is moving through the cpu the more heat will be removed from the processor. Now using this, can it be safe to assume that the following setup would be very effective (but not cost efficient)?

reservoir >> radiator >> pump >> 2nd smaller radiator >> waterblock >> reservoir

maybe put a valve (to act as a flow limitor) between the reservoir and first radiator. This would allow the water to be moving very slowly through the first radiator removing all the heat from the water that was gathered from the CPU. The pump (which does create some heat) would then pump water really fast through the second radiator (much smaller non-flow-restrictive radiator) and onto the CPU for supremely effective CPU cooling. You can even put another valve between the second radiator and the CPU to have a fully adjustable flow rate to maximum performance. The water is then pumped to the very high part of my full-tower case and into the reservoir where the cycle starts over again.

Now if this makes absolutely no sense at all I apologize and invite you to make jokes about my mother (joking).
Thank you very much and good night

[Joe]

I would actually put the smaller one before the rese, and the larger radiator right before the CPU.

Heat in your rese, and in your pump isn't an issue, both can take much more heat that you will have in the system. And also letting the coolant sit in the rese will only help it to disspate more extra heat.

in all honesty you dont need 2 radiators for any one CPU loop if you use radiators of good enough quality.

[Kevin]

Well you can't have the water flowing really slow through the radiator and fast throgh the rest of the system. Flow needs to be constant.
-kev

[Emo_Boy]

Actually I wasnt too worried about the effectivness of the radiator or the cooling of the water (I plan on getting the Black Ice II). My main concern is flow rates. Trying to get a very low flow rate through the first radiator (the longer the water spends in there the closer to ambient it will become) and optimal fast rates through the CPU block (too a certian degree, the faster the water moves though the block, the more heat will be able to transfer to the water). I could slow the water going into the main (and larger) radiator but then it would be traveling even slower throught the CPU (which would be in-line right after the radiator). The way I figured I could beat this would be to put the pump right after the radiator so the water going through the CPU would be very fast. I was then concerned that the water exiting the pump would be slightly heated (because of the pump) so i though of putting a small single-pass radiator and small fan to dissipate that small ammount of heat. I only wonder if the heat generated by the pump will adversely affect the waters cooling ability through the waterblock.

Thanks for the feedback btw. Even though I've been case modding for a few years (back when rounding your cables was the coolest thing to do ) I've only been on the watercooling scene for a couple months. I just like to design everthing and plan it out too much before I actually do it. I dont actually have a complete watercooled system yet or I'd test the theory out myself.
Thanks again everyone.

[deeznuts]

like kevin said, flow rate is constant. no such thing as fast in one area, and slow in another, unless you have multiple loops are something.

deez

[Emo_Boy]

Okay, I'm seeing my misunderstanding of the subject. For some reason I look at the coolant as being more like a strait line where a limitation of the flow at one point will only affect the flow rate of all points after that point. I'm not realizing that when flow is limited that creates a bottleneck in the whole system. Thank you all for your posts! It has been very helpful even though the idea was unsucessful . Oh well, back to the design board. <<opens up MS Paint >>

[Butcher]

It can be fast in one area and slow in another. The gph is the same throughtout but the actual speed of the water can be different, eg. thin pipe, fast moving liquid flows into wide pipe - slows down, flow rate is unaffected but speed isn't.

[Tweekster]

Butcher is right...that is one of the reasons a heater core works better than the multiple coil/loop type radiators....as the coolant enters the core it splits up to several passages, increasing the overall area, decreasing the coolant velocity and allowing for better heat transfer.....you can have too much coolant velocity....in car racing (NASCAR for example), they actually gear down the water pump to slow the flow for better cooling.....there is an optimum flow range

[Kevin]

You could do this... have two heatercores. Water goes, hits a "Y" or "T" and splits into two heatercores. The flow is cut in half cuz the volume of water traveling through the hose is cut in half. After it goes through the heatercore, the lines meet up again and flow speeds up.
-Kevin

[LiquidCool]

You really want to know the simplest way to do this? The REALLY simple way?
.
.
.
.
.
One high flow pump and a low flow pump, pumping from the same reservoir. High flow pump goes to the block and back to the res. Low flow pump goes to the radiator and back to the res. Problem solved 8P

[futRtrubL]

That wouldn't help at all. That system is a an unseparated hart system which has the disadvantage of mixing the two inputs in the res. You have cool water coming from the rad loop mixing with the hot water from the block loop. This mixture then goes into both loops so you still have warm water going to the block.
The best looking system is the Y to 2 rads, if you can get flows equal in each rad.

Edward

[ 11-18-2001: Message edited by: fut ]

[Emo_Boy]

How would someone go about getting the flows equal? Would using 2 of the same radiator do the trick?

What are some other options for an efficiant watercooled system?

And one other thing. How effective are those peltier water chillers (Like the one sold on overclock watercool)? What is there effectivness on a non-peltier cooled CPU versus a peltier cooled one?
Any help would be much appreciated here. Thanks everyone!

[futRtrubL]

I haven't used a Y system so this is just theory but, if you want equal flows then both rads have to be the same, beyond that I don't know if length of tubing and bends will cause enough a difference in resistance to cause most of the warer through one rad. If you had a flow meter on both lines you could check and use a hose clamp to alter the resistances to get it equal but that is way too expensive and time consuming for me.
The most popular system seems to be 1 big rad, 1 big pump and one big block. My system adds another rad in series, I feel this allows the water to get closer to ambient, but if you get a big enough rad this isn't nescesary. I do 2 rads as I wanted mine to be low profile.
Pelting the CPU can get you great gains if your WC system can handle the added heat. I have conflicting reports about water chilling but personally I believe it gets you little to no gains for a lot of pain( 2 water loops instead of one, if it chills well then you need to worry about condensation not just on the CPU but also on the cooling line.)
I can't comment on the pelt/non-pelt effectiveness as I haven't got a pelt system and because it depends on too many factors. However a well done pelt system beats any non-pelt system due to below ambient temps. and overall better Watts heat/Kelvin even if the system isn't designed to go below ambient(if you don't want to deal with condensation.

Edward